Report No. 56747-LV Latvia From Exuberance to Prudence A Public Expenditure Review of Government Administration and the Social Sectors (In Two Volumes) Volume II: Analytical Report September 27, 2010 Human Development Sector Unit Europe and Central Asia Region Document of the World Bank Latvia Currency Equivalents (Exchange Rate Effective August 31, 2010) EUR 1.00: US$ 1.267: LVL 0.709 ABBREVIATIONS, ACRONYMS & TERMS BoL Bank of Latvia HBS Household Budget Survey CEE Central and Eastern Europe IMF International Monetary Fund (CDS Credit Default Swap LFS Labor Force Survey CoM Cabinet of Ministers LVL Latvian Lats CIT Corporate Income Tax MoES Ministry of Education and Science CSB Central Bureau of Statistics MoF Ministry of Finance DC Defined Contribution pension plan MoH Ministry of Health EC European Commission MoW Ministry of Welfare EU SILC Survey of Incomes and Living MoRDLG Ministry of Regional Development and Conditions Local Government EBRD European Bank for Reconstruction and NDC Notional Defined Contribution, pay-as- Development you-go pension plan ECB European Central Bank OECD Organization for Economic Cooperation and Development ESA European System of Accounts PER Public Expenditure Review ESF European Social Fund PFM Public Finance Management ESSNS Emergency Social Safety Net Strategy PHC Primary Health Care EU European Union PISA Program for International Student Assessment FDI Foreign Direct Investments PIT Personal Income Tax GDP Gross Domestic Product PPP Public Private Partnerships GMI Guaranteed Minimum Income means SEA State Employment Agency tested social assistance benefit GP/PHC General Practitioner/Primary Health SOE State Owned Enterprises Care ECA Vice President: Philippe H. Le Houerou ECCU5 Country Director: Peter Harrold ECSHD Sector Director: Mamta Murthi (Acting) Country Manager: Thomas Blatt Laursen Sector Manager: Jesko Hentschel Task Team Leader: Truman G. Packard ii Acknowledgements The World Bank team would like to acknowledge the extensive and professional collaboration of the institutions of the Government of the Republic of Latvia. In particular, the team would like to thank the Ministry of Finance, the State Chancellery in the Office of the Prime Minister, the Ministry of Regional Development and Local Government, the Ministry of Health, the Ministry of Education and Science, the Ministry of Welfare, the agencies affiliated with the ministries, and the Association for Local Governments and Municipalities for their input and discussion during the preparation of this review. Additional thanks are extended to the State Audit Office of the Republic of Latvia for providing critical input from its own assessments of the areas covered by this review. The Public Expenditure Review (PER) team comprised Charles Griffin (Sr. Advisor and Health Economist), M. Ihsan Ajwad (Sr. Economist, Human Development Economics), Asta Zviniene (Sr. Pension Economist), Michael Mertaugh (Education Economist, Consultant), Clelia Rontoyanni (Public Sector Specialist), Stanislav Polak (Country Economist), Emily Sinnott (Sr. Economist, Human Development Economics) and Johannes Koettl (Economist, Human Development Economics). The PER benefited enormously from the tireless efforts of Lars Sondergaard (Sr. Economist, Human Development), Igor Kheyfets (Economist, Human Development), Malinee Yasmin Rachel Am Valenzuela and Matthew Louis Gyory (Government Expenditure Consultants) who prepared and applied the BOOST government expenditure tool for the team. The team was advised by Mihails Hazans (Professor, Faculty of Economics, University of Latvia), and Inga Vilka (Faculty of Economics, University of Latvia) who made extensive analytical contributions to the review. Regina Nesiama (Program Assistant, Human Development) provided inputs and support to the team. The PER team was lead by Truman G Packard (Lead Economist, Human Development Economics). The PER team worked under the direction of Tamar Manuelyan Atinc and Mamta Murthi (Prior and Current Directors, Human Development, Europe and Central Asia); Peter Harrold (Country Director, South Central Europe and the Baltic States); Jesko Hentschel (Sector Manager for Human Development Economics, Europe and Central Asia); Thomas Blatt Laursen (Country Manager, Poland and the Baltic States), and Alberto Rodriguez (Country Sector Coordinator for the South Central Europe and the Baltic States, Human Development). The Peer Reviewers of the report were Karlis Smits (Country Economist), Emiliana Vegas (Sr. Education Economist), and Alejandra Gonzales (Health Economist, WHO Europe). Tamas Evetovits (Sr. Health Finance Specialist, WHO Europe) also contributed with comments on earlier drafts of this report. The World Bank team is grateful for the comments and suggestions made by the reviewers. The data/information used for this review is collected during the first 2 quarters of 2010. This document represents the opinion of the World Bank. (This document has not been approved by the Government. The Cabinet of Ministers only took note of the informative note prepared on the basis of this PER.) iii Table of Contents ACKNOWLEDGEMENTS ................................................................................................................................................... III CHAPTER 1: INTRODUCTION AND MOTIVATIONS FOR FISCAL CONSOLIDATION IN LATVIA ................................... 1 I. Objectives Scope and Summary of the 2010 Public Expenditure Review ..........................................................1 II. Latvia’s Economic Boom and the Fiscal Consequences of High Growth .........................................................11 III. Emergency Measures and the Need for Permanent Fiscal Adjustment ........................................................15 CHAPTER 2: PUBLIC ADMINISTRATION REFORM, SUBSIDIES, AND STATE OWNED ENTERPRISES ........................ 28 I. Introduction and Main Messages ....................................................................................................................28 II. Latvia’s Wage Bill and Subsidies to Enterprise Prior to the Crisis ...................................................................29 III. Changes in Administration and Enterprise Support Since the Onset of the Crisis .........................................34 IV. Options for Achieving Budget Savings in the Immediate Future ...................................................................41 V. Options for Reforming Public Sector Remuneration Policy ............................................................................52 VI. Other Medium-term Options to Increase Efficiency in the Public Sector ......................................................55 CHAPTER 3: MUNICIPAL SPENDING AND INTERGOVERNMENTAL FINANCE......................................................... 65 I. Introduction - Latvia: Intergovernmental System............................................................................................65 II. Design of the Intergovernmental System .......................................................................................................66 III. Local Government Expenditure Structure......................................................................................................72 IV. Local Government Revenue Structure ...........................................................................................................75 V. Local Government Revenue Equalization .......................................................................................................78 VI. Macroeconomic Control and Expenditure Effectiveness ...............................................................................80 VII. Summary of Recommendations ...................................................................................................................85 CHAPTER 4: SOCIAL WELFARE IN LATVIA ............................................................................................................. 87 I. Introduction and Main Messages ....................................................................................................................87 II. The Structure of and Spending on Social Welfare in Latvia ............................................................................89 III. Performance of Social Welfare ......................................................................................................................96 IV. Critical Issues in Social Welfare ...................................................................................................................106 CHAPTER 5: EDUCATION ................................................................................................................................... 151 iv I. Introduction and Overview ............................................................................................................................151 II. Public Expenditure on Education ..................................................................................................................153 III. Improvements in the Education Sector’s Efficiency .....................................................................................171 IV. Teacher Salaries and Teacher Earnings.......................................................................................................173 IV. Interpreting the Expenditure Evidence & Implications for Education Policy ...............................................179 V. Education Sector Policy Options for Government Consideration .................................................................184 CHAPTER 6: HEALTH .......................................................................................................................................... 198 I. Health Conditions ..........................................................................................................................................198 II. Cross Country Comparisons of Health Resources, Utilization, and Expenditure ..........................................210 III. Recent Trends in Latvia’s Health Policy and Expenditures ..........................................................................218 IV. The Private Health Insurance Market in Latvia ...........................................................................................240 V. Conclusion ....................................................................................................................................................248 VI. Recommendations ......................................................................................................................................253 CHAPTER 7: LONG-TERM CARE .......................................................................................................................... 255 I. Introduction ...................................................................................................................................................255 II. Aging and Its Long-term Impact on the Dependent Population ...................................................................255 III. Current Public Expenditures on LTC .............................................................................................................258 IV. Future Policy Directions...............................................................................................................................262 v Chapter 1: Introduction and Motivations for Fiscal Consolidation in Latvia I. Objectives Scope and Summary of the 2010 Public Expenditure Review A Public Expenditure Review for Latvia’s Post-Crisis Objectives 1. This Public Expenditure Review (PER) was conducted at the request of the Ministry of Finance (MoF) on behalf of the Government of the Republic of Latvia. The objective of the PER is to identify potential areas of further budget savings in public administration and the social sectors that could help restore fiscal balance, speed Latvia’s recovery from the current crisis, and help it to meet the Maastricht Criteria by 2012 so it can adopt the Euro in 2014. In this way, the PER responds directly to the Government’s request to international agencies, including the World Bank, to help it meet its goals for fiscal consolidation. Perhaps more importantly than opportunities for further fiscal consolidation in 2011 and 2012, the measures recommended in this review could lead to improved performance and greater economic and social returns on Latvia’s investment of public resources. One important aspect in future sectors are their equity impact -- prior to the crisis, the impact of public spending on income equality in Latvia was among the lowest in the European Union. 2. Pushed by the devastating consequences of the global financial crisis of 2008 and Great Recession of 2009, Latvia has achieved years worth of difficult structural reforms in the short space of just a few months, and for its efforts in the face of the most adverse circumstances, deserves praise. However, more can be done not only for the country to achieve the goal it has set for itself of entry to the Euro, but also to improve the public services that Latvians depend on. This PER offers a menu of policy options from which the Government -in close consultation with stake-holders- can consider and choose the measures it thinks are not only appropriate but also timely. It is not a prescription. 3. Implementation of any of the options identified in this review will require broad consensus around difficult political choices. Most of the suggestions can help the Government achieve further reductions in the state budget in the immediate term. But for the most part, the options identified in this review are desirable in their own right in order to provide a more efficient and effective public administration, and better incentives for sustainable and responsive social-sector programs. Here, as with many of the structural reforms already implemented in 2009 and 2010, the current economic crisis and the need for fiscal consolidation that it brings, offers an opportunity to adjust public spending with a clear rationale that will increase the likelihood of improved performance in the future. Main Messages for Policy Makers 4. There is little, if any, room for further reductions in the budgets for health and education. The PER concludes that the health sector and the education sector have borne the brunt of fiscal adjustment in the 2009 and 2010 budgets. With very few exceptions, further cuts in the budget allocations to these sectors could put in jeopardy the progress achieved so far. Indeed, the speed of reductions in the in- patient care budget have surpassed the ability of the authorities to close hospital beds and 1 decommission hospital buildings, and supplemental financing in 2010 to sustain an orderly, and de- politicized process of hospital closures phased over the next three years will be required. 5. Structural reforms of health and education are on track and are showing measurable improvements, but they need to be continued with care and commensurate financing. On the whole, the structural reforms in the health and education sectors initiated in 2009, although rapidly put in place, took the form of fundamental changes in incentives that need time to “take rootâ€? before they can fully deliver expected gains. Yet already there is clear evidence of a shift in health spending which favors more intensive use of preventive and day-care procedures, and which is likely to better address the health problems Latvians suffer. The heavy burden on the budget of over-capacity of schools and teachers has been lightened significantly and financing for primary and general secondary education has been fundamentally altered to better adapt to demographic trends and increase efficiency. However, there is still much to be done in the health and education sectors to make them perform to the high- standards of an EU member state, and the sections of this review on both sectors are rich with suggestions for how to accelerate convergence. But while these suggestions can indeed speed improvements in performance, they are not likely to deliver the savings the Government seeks in 2011 and 2012. 6. The opportunities for the savings in 2011 and 2012 that Latvia seeks to meet its Maastricht objectives, lie in reducing spending on social insurance pensions and efficiency gains from further public administration reform. In contrast to the review of the health and education sectors, the PER finds that bolder action could be taken by the Government to capture substantial budget savings from further structural reform of public administration, reductions in subsidies, and reform of the social welfare sector. The most immediate fiscal savings could be had from well targeted measures to recover at least part of the windfall gains made by the recipients of social insurance pensions, while protecting the lowest pensions. The actions proposed in this review would be starkly different in substance and in form from the Government’s flat, across the board pension cut in 2009, which regressively penalized all pensioners regardless of their income or their age, and inadvertently punished pensioners who preferred to keep working. Despite some consolidation in 2009, the public administration could be further streamlined at the national and municipal levels of government, although to ensure it retains skilled and motivated civil servants, the PER warns against further reductions in public sector salaries at the national government level. As was the case with the structural reforms to education and health in 2009 and 2010, the PER’s recommendations in these areas are made to guide the process of fiscal adjustment with a clear rationale that will increase the changes of improved performance. 7. The three main messages of this review, stated above, are the most important conclusions policy makers should take away from this review. These messages summarize the findings of an intensive period of work by the World Bank team with Government counterparts. The three main messages are complemented with important points summarized in the sections that follow. 8. However, readers should note that the PER is only a partial analysis of government expenditure. It is important to consider the options identified in this review in the broader fiscal context. This review focuses only on the three social sectors and the Government’s efforts to improve public administration. There may be opportunities for immediate expenditure savings in other sectors that impact less directly 2 on household welfare. Furthermore, the PER does not cover issues of public financial management and the budget process, which are critical to prudent and sustainable public finances, and are being examined in depth through a program of technical assistance provided by the IMF. A broad view should be taken when looking for savings particularly in a period of economic contraction and high unemployment. Changes in how the Government uses revenue instruments -being discussed between the Government and other international partner agencies- will be critically important to achieving the goals of fiscal consolidation, and are only summarized briefly in this report. 9. As the conclusions of the social welfare sections of this review make clear, Latvia does not have a particularly generous social welfare system when compared with its partners in the EU. For this reason, and especially until there are signs of recovery in the labor market, this review suggests ways in which the modest amount that Latvia spends on state social benefits and municipal social assistance could be redistributed to much better effect. By taking these suggestions, the Government can confront the high cost of social insurance pensions with greater confidence that it does so in a way that protects vulnerable households. A similar emerging policy approach can be seen in the health sector: more parsimonious and targeted use of co-payment exemptions, and strategic deployment of public health measures, allows the authorities to confront the difficult process of rationalizing the hospital network with greater confidence. The Government has set a strong precedent in this regard, with the rapid design and deployment of the cross-sector Emergency Social Safety Net Strategy, which it continues to monitor closely, and plans to sustain throughout 2011. There are lessons from the design and deployment of the safety net that could inform more permanent reforms to welfare, health and education policy. Noteworthy Items from the PER’s Menu of Policy Options 10. In this section, the three main messages of the PER are further substantiated with more specific policy suggestions selected by the PER team for particular attention and consideration by the Government. Two criteria have been used by the team to make this selection. These are the measures that are (i) most likely to deliver the savings the Latvia seeks in 2011 and 2012, and/or (ii) are critical to improving equity as well as efficiency. Public Administration and Subsidies 11. Remuneration structures in the public sector could be further improved and more effective establishment control could be put in place, but further reductions in salaries in 2010 and 2011 would pose a risk to the quality of the civil service. The PER suggests further amendments to the new remuneration grid that was introduced in 2009, that would eliminate salary bands and reduce the number of public service job families. The remuneration grid remains overly complex and contains many inconsistencies that reduce predictability in managing the wage bill, and undermine the principle of ‘equal pay for work of equal value’. Short-term savings can be achieved through a reduction of the maximum salary levels to bring them closer to the current average salaries paid by ministries and agencies. This would force agencies that pay salaries on the higher end of the bands (e.g. Ministry of Health, Ministry of Regional Development, Ministry of Welfare) to reduce their wage bill. The benefits from a more simple grid would be greatly enhanced with the establishment of a centralized personnel 3 and payroll system and introduction of central controls for the establishment of new posts. Stronger establishment control would ensure consistency in human resource management and remuneration policies and result in further savings in administrative costs through reduction of HRM staff. In the period 2010-2011 further reductions in salary levels does not seem feasible or advisable, as they could lower morale and would carry a very high risk of qualified staff leaving the civil service. 12. Further staff reductions to pre-2007 levels, particularly in agencies subordinated to ministries and local government bodies, are possible and, if implemented carefully, would not risk the quality of public services. Overall, public sector employment in Latvia remains higher than in some comparator countries, and staffing levels remain higher than 4-5 years ago. The wage bill as a share of GDP is also on the high side of a range of comparator countries, Thus there may be some room for adjustment to pre- 2007 staffing levels without a negative impact on policy functions and service delivery. It is important to note that the central government has borne the brunt of staff reductions thus far. For the reason, the most substantial budget savings from this option in 2011 and 2012 would come from reducing staff at the level of local government. 13. Significant savings can be found from phasing out direct subsidies to loss-making enterprises at both the central and municipal levels. Currently, subsidies are not targeted to supporting access to services for low-income citizens, but come in the form of ‘compensation for losses’ to providers, without adequate cost-control mechanisms. As a result, subsidized enterprises have an incentive to provide unnecessary services, to understate revenues, and to inflate operational costs. Reduction of central budget subsidies –e.g. for passenger transport – could be achieved by targeting the subsidy to low- income persons or for rural routes only. The risks of this option could be substantially lowered by accompanying the reduction in subsidies with a compensation strategy, along the lines of current ESSNS support for disabled passengers. User fees (e.g. passenger transport prices) could be allowed to increase. This would allow subsidies to be targeted to protect low-income beneficiaries, instead of protecting firms from losses. Services considered socially essential, but which would not be commercially viable even if providers were able to charge market prices, could be specifically identified and prioritized based on cost-benefit analysis. As it reviews the merit and justification for subsidies, the Government could also scrutinize subsidies for SOEs in the areas of culture, broadcasting, and sport. Many advanced countries subsidize these activities. The policy question for consideration is whether set against the objectives of further fiscal consolidation and other priorities for support (such as social services for vulnerable groups) these subsidies are affordable. This question can only be answered by the Government in consultation with stakeholders. Social Welfare 14. The Government could lower the amount of pension income exempt from taxation. The unprecedented economic growth until 2007 translated into a rapid growth in wages and consequently of the wage bill covered by the social insurance system. This boom was reflected in the notional interest rate applied to pension benefits from the public pension pillar. Between 2005 and 2009 the average pension for new retirees grew 69 percent. The cohorts that retired prior to 2010 have benefited from a windfall from the pre-2007 bubble that the Government could recover, applying the principle of burden sharing. A broader “taxation basedâ€? measure is probably justified (over, for example, a measure 4 targeting only recent retirees) as the increase in revenue to the state social insurance special budget motivated other costly pension policies during the high-growth, such as the indexation of pensions to wage growth. It is important to note that this suggestion is a far more equitable and incentive compatible policy option and different in both form and substance to the flat cut in pensions in mid- 2009. The 2009 pension cut was both regressive and penalized working pensioners, and was overturned by a Constitutional Court decision of December 21, 2009. However, the recommended measure is far more progressive, affecting lower-income pensioners relatively less, and it does not distort work incentives. The Constitutional Court’s decision indicates how the measure might be more appropriately implemented by the Government: with (i) careful presentation of alternative options, and why they are less effective; and (ii) sufficient prior notice between passage of the measure and when it takes effect to allow households dependent on pension income to prepare for the change. 15. The pre-1996 service pension supplement could be used more effectively to support the poorest pensioners. This option is the measure that holds the greatest potential for fiscal savings among the all the options presented in the PER. The pension supplement pays LVL 0.7 (70 centimes) for each year of service an affiliated worker had accrued before 1996, and on average amounts to 14 percent of the overall pension spending. The supplement was originally introduced as a measure directed at pensioners with the lowest benefits. However, in the general exuberance that gripped fiscal policy prior to the crisis was extended to all pensioners in 2008. Extending the supplement to all pensioners was an unsustainable policy. The Government could save a substantial amount of resources from repealing the extension of supplements to all pensioners, and still continue to supplement the pensions of the 10 percent of retirees with lowest incomes. 16. The age of retirement could be increased starting in January 2015, but will only bring savings in the medium term. An increase in the statutory and effective retirement ages is essential to the long- run financial sustainability of the social insurance special budget, and to help raise replacement rates. Among the cohorts turning 62 in 2012 and 2013, around 70 percent will already be receiving old age or disability benefits. This defeats the purpose of trying to increase retirement ages before 2014. Furthermore, the cohort turning 62 in 2014 will already experience a significant increase in the effective retirement age, so it is not equitable to increase their burden further. Thus, the most logical time to start increasing retirement age is 2015. An increase in the retirement age earlier than January 2015 will probably precipitate a faster “rush to retireâ€? that could destabilize the state social insurance budget. It is frequently argued that raising the retirement age, particularly in a recessionary period (or when the labor market is very slack) exposes older workers to greater risk of unemployment. However, this concern seems misplaced. Workers who would be affected by an increase in the retirement age, are also more likely to have coverage of unemployment insurance, as well as relatively greater job security (than younger cohorts). Indeed, in the current contraction, younger workers have suffered unemployment at relatively higher rates than older workers. 17. Also critical to the sustainability of retirement income security in the medium and long term, are effective and viable means of diversifying risks, such as a strong funded pension pillar. Chapter 4 in Volume 2 explains why Latvians will have to rely on the funded pillar of their multi-pillar pension system for a greater share of their income security in old age. To ensure it could keep paying pensions 5 during the unprecedented economic contraction – which play an essential safety net function for most Latvian families - in 2009 the Government diverted a portion of contributions to individual accounts in the funded pillar of Latvia’s pension system to shore up the finances of the pay-as-you-go, NDC pillar. While regrettable, this decision is understandable in the circumstances. The measure is also temporary, and passed with a clear “exit strategyâ€? that envisions a phased restoration of contributions to the funded pillar. The funded pillar was "small" relative to the NDC pillar even when the contribution rate was 8 percent of workers' salaries. So it was struggling to function as a viable risk-diversification instrument even before the diversion in 2009. Now at 2 percent of workers' salaries it is even smaller. It is difficult for fund managers to keep administering such a small trickle of contribution capital without a firm expectation of when that flow will be restored to the volume they expected to be managing. A negative outcome could be -without the certainty of a clear time-table for restoration of contributions- a withdrawal of market participant that would increase the concentration of the industry and limit the benefits to affiliates of competition between fund managers. A more immediate fiscal danger could arise from how any departure from (or delay in) the planned restoration of contributions to the funded pillar would be interpreted by markets. A loss of market confidence could put in jeopardy the hard-won gains that the Government has made in navigating the financial crisis. 18. To improve the equity impact of the welfare system, the Government could replace the earnings-related Parental Benefit with an augmented flat Childcare Benefit of LVL 100 paid from State basic budget. The relatively recently created Parental Benefit was shifted into the social insurance special budget in 2008. The Parental Benefit is earnings-related, and paid only to those with a history of contributions to the social insurance. However, when it was introduced, no increase in contribution revenue to the state social insurance budget was made. The benefit is, for all intents and purposes, a non-contributory benefit. For this reason, the recent shift of the Parental Benefit spending to the social insurance budget is inappropriate. Furthermore, as with all other “non contributoryâ€? benefits, linking the benefit amount with the recipients’ earnings, introduces regressive redistribution to people with higher incomes. A flat Childcare Benefit is already granted from the State basic budget for the same social welfare objective. The coexistence of the earnings-related Parental Benefit alongside the flat Childcare Benefit augments inequality, and imposes a high fiscal cost. The progressive flat benefit could be raised to its 2005 level, LVL 100. 19. Family State Benefits (FSB) could be targeted to poorer households with children, to improve their impact and reduce waste. Family State Benefits constitute about 37 percent of state social benefits. FSBs are monthly payments of LVL 8 to all families with a child between the ages of 1 up to 18 years of age. The Government could achieve a higher social return from its social welfare spending, by reducing coverage of FSB for non-poor households, but maintain assistance for poor households. Means-testing all state social benefits (i.e., those financed from the national basic budget for the Welfare sector) would be advisable – with the exception of benefits related to disability. FSB could be the best benefit to start with, given its currently wide coverage; large volume (and thus potential savings); and that the public is “sensitizedâ€? to reform of this benefit (earlier public discussion of possible reform. FSBs have a small impact on non-poor households and are more symbolic than welfare enhancing: households in quintiles 4 and 5 spent about 2,300 and 4,200 LVL per person per year in 2008 and hence the added 72 LVL per child per year has little impact on household welfare. Given how little 6 Latvia spends on social assistance, savings from targeting FSB could be re-distributed to strengthen targeted social assistance programs. 20. To strengthen the safety net for the poorest households, the Government could increase co- financing from the State budget for GMI and Housing benefit. GMI and Housing benefits are increasing in importance as safety-net instruments, and appear to be reasonably well targeted to the lowest income groups. Both benefits are mandatory for local governments to pay to eligible households, but are financed from municipal budgets. The mismatch between a national mandate and local financing causes perverse incentives at the local level, and disparities in provision of these mandatory benefits across wealthy and poor municipalities. For this reason few EU and OECD countries retain decentralized financing of mandatory targeted benefits. Raising co-financing and eventually recentralizing financing for mandatory targeted benefits from the State budget, will ensure a robust, uniform safety net for the lowest-income, needy households wherever they happen to live. The co-financing introduced in the ESSNS (50 percent of spending on GMI and 20 percent of spending on housing benefit) helps mitigate the problem of perverse incentives, particularly in the face of higher demand for benefits from the crisis. Savings from targeting FSBs could be used to increase the share of co-financing of mandatory benefits from the State budget, and still leave room for overall fiscal savings. Education 21. In order to retain and attract good teachers, the Government could maintain the level of salaries of general primary and secondary teachers and the parity of teacher remuneration with workers with similar qualifications in the public sector. The formula financing reform in primary and general secondary education is on-track, and already delivering efficiency gains. A critical part of ensuring the reform continues to deliver benefits is adequate and competitive teacher remuneration. Teachers in Latvia were substantially underpaid even before the reductions in the education budget in 2009. During the first year of implementation of the education financing reform, steps have been taken to bring greater equivalence between teachers’ pay and the pay of similar positions elsewhere in the public sector. A salary increase in January, 2010 brought average teacher earnings (including overtime) to parity with employees of state budget institutions. Maintaining teacher salaries at a comparable level with employees of state budget institutions will help attract and retain competent teachers. The Government could monitor this parity closely, as a reversion to former low salaries for teachers would endanger education quality by dissuading the most competent teachers. Competitive pay levels would be easier to monitor and maintain if the system of teacher remuneration were shifted towards full-time concept of teachers’ profession, based on learning outcomes rather than time inputs. This shift would require strengthening of assessment instruments. 22. Finance vocational education places more adequately, but ensure a “level playing fieldâ€? between vocational and general secondary education, so households make better informed choices about their investment in education. There is a legitimate policy question about the appropriate scale of vocational education relative to general secondary education. Streaming students into vocational education at early ages could risk locking them into lower-lifelong earnings, and close off other opportunities. This risk can be mitigated by ensuring a “level playing fieldâ€? between vocational and general secondary, and keeping opportunities open for movement from one stream to the other. The 7 level of per-student expenditures by government on vocational secondary education in Latvia is below those of international comparators with high-quality vocational education programs. To the extent that vocational education is justified by skill needs in the labor market, it should be funded adequately to provide high-quality and relevant skill training. A key step towards ensuring a “level playing fieldâ€? would be to ensure that students in general secondary and vocational secondary are equally eligible for the same welfare benefits, so that decisions about what sort of secondary education households chose are not influenced by considerations other than the quality of instruction and likely future earnings. With major EU support, all secondary vocational schools are preparing and implementing restructuring plans. All of the restructured vocational schools are scheduled to be transferred to local governments by 2015. Local management and financing of vocational schools would help ensure that vocational programs are cost-effective (in comparison to general secondary education), and provide good value in terms of improved job prospects for graduates. 23. The Government could reduce budget-financed student places in higher education, financing these at a more adequate level, and prioritizing them for the poorest applicants (including those interested in part-time courses that allow them to work during their study), Government expenditure on higher education is very low in Latvia by comparison to most countries. Budget-financed places are rationed, solely on the basis of student performance and not on need. Half of full-time students pay full cost-recovery fees, which vary by specialization. The other half pay no fees; their places are financed by the budget. The number of budget-financed places in higher education increased over the past three years, despite large overall budget reductions in higher education – implying a deterioration of teaching and learning conditions. As currently used, budget financing of higher education is inequitable because it disproportionately benefits students from higher-income families. Greater reliance on fee financing and targeting budget assistance to the neediest students of otherwise equal academic merit, would improve equity and fiscal sustainability of higher education, and promote quality by encouraging competition based upon relevance to changing skill needs in the labor market as reflected by changing patterns of earnings. Current university students have acquired rights to continued financing under the arrangements that were in place when they entered higher education. Thus, this shift in policy would need to be introduced gradually, applying to newly registered students. Health & Long Term Care 24. The most important opportunities for structural reforms created by the economic crisis have already been seized by the Ministry of Health: prioritizing the right services and creating a means tested safety net that did not exist for Latvian households previously. However, the path ahead demands a much stronger effort to rationalize the hospital sector, to improve the prevention of the health conditions that contribute the most to mortality in Latvia, to reform the reimbursement system to improve incentives for secondary outpatient services and hospitals, and to make the health sector a more dynamic contributor to Latvia’s economy. 25. The Government will need to prepare and implement a technical plan for closure of hospital beds and to decommission hospital buildings. As a result of budget cuts and priorities set in 2009, Latvia now prioritizes adequate funding of general practitioners, emergency care, health services for mothers and children, dental services for children, covered prescriptions, referred secondary outpatient 8 services, a safety net for needy families, and unplanned hospitalizations. Hospitals have become a residual in the budget, which provides funding for continued catastrophic protection at that level. However, the hospital infrastructure has – not surprisingly – not adjusted in one year to substantially lower funding. Therefore funds for hospitals in 2010 would need to be restored, if the funding is available, but only with the proviso of an agreed technical plan for closure of hospital beds and buildings – a plan that would ideally be implemented from 2010 through 2013. The Government could consider a 5000 bed reduction that takes place at a measured pace from 31 October 2010 to 31 December 2013. By 2013 the goal would be to achieve an overall ratio of 350 beds per 100,000 population. Although there would be some additional spending required over this period, by 2013 the goal would be to realize spending for inpatient hospital services at 35 percent or less of total medical care expenditure. Hospitals need to be closed or they will come back on the budget in the future, and as long as buildings are kept open much of the expected savings will not materialize. 26. To support rationalization and improve health outcomes, the Government could further reduce average length of stay in hospitals. There has long been a structural bias in Latvia’s health sector toward provision of relatively costly in-patient care. The Government has reduced this bias. Challenges remain. Latvia’s average hospital length of stay has been stuck at a high level for many years; cutting it is essential if fewer hospital beds are to be adequate. In 2009 the average length of stay dropped from 9.5 to 8.7 days. The Government could set itself the target of reducing average length of stay by 1.2 days in 2010 (to 7.5), 0.5 days in 2011 (to 7.0), and 0.5 days in 2012 (to 6.5). The high length of stay may reflect reimbursement policies and could be tackled with conversion of secondary outpatient services and inpatient care reimbursed based on diagnostically related groups. But it will take time to change reimbursement policies and for providers to adjust; in the meantime, direct action will to be taken to reduce lengths of stay. It is unclear that the global budget system – a first attempt at prospective payments- put in place in 2010 will be sufficient to change current incentives for hospitals to keep patients in beds. In addition, to reduce incentives for costly behavior by providers, prospective payment methods should replace fee-for-service payments to the degree possible for hospital and secondary outpatient services, as soon as is feasible. 27. The Government could further improve financial protection from the cost of health care, with more equitable and targeted use of subsidies and copayment exemptions. Copayments are a variable that the health sector authorities have control over. Copayment policies can be adjusted further so that those who can pay, pay more, while the poor are better protected. Close attention needs to be paid to annual data on the percent of income devoted to health care by quintile in the household consumption survey data. In particular, pharmaceutical reimbursement policies should be adjusted as needed to reduce the burden of these expenses on the poor and chronically ill. A reasonable goal is to maintain in the household consumption of health care as a flat or increasing function of expenditure per capita, so the poorer quintiles are devoting a smaller proportion of their expenditures to health care than are the wealthy. This outcome is more likely if the Government were to shift how it targets exemptions from social groups to completely means testing. 28. To increase the efficiency and equity of health finance, the Government could clearly codify the State-financed benefit package of health coverage and the “rules of the gameâ€? for non- 9 government providers for coverage of additional health risks. There is no compelling argument for structural reform of health finance in Latvia at this juncture, such as a shift to a social insurance model, mandatory private insurance, or a mix of public and private insurance beyond what exists today. Latvia has a “single-payerâ€? health finance system with very low administrative costs.. However, to encourage public debate about the future of health finance in Latvia, the PER provides analysis of one of many possible options for development of the insurance market. The PER argues for the development of competitive contracting, which would allow the State, as it gains experience, to extend this to the competitive contracting1 of insurers if so desired. However, in making these arguments, the PER is not advocating this policy. A clearer definition of what is the core level of services the State will finance (and limits on providers’ ability to game multiple payers) would encourage the development of a more robust private supplementary insurance system that would open possibilities for the future. 29. As Latvia’s elderly population continues grows, the Government could contain the cost of long term care by improving discharge management in hospitals and eliminating biases in current financing model that encourage medical in-patient treatment for the elderly over long term care. Due to the overcapacity in the hospital sector it is conceivable that a large part of long term care (LTC) is currently provided in hospitals, but in the guise of in-patient services. This is both expensive and to the determent of the wellbeing older people. A more cost-efficient solution would be to discharge the patient when medical treatment is no longer necessary, and to provide the social support through community-based LTC services. Many countries in Europe have converted redundant hospitals into residential LTC institutions. However, this introduces a costly bias towards expensive institutional care. The final chapter of the PER proposes a new “Community Centersâ€? LTC model which could be at the center of care coordination for patients. Community centers could house daycare facilities for elderly and disabled people (or even childcare), but also outpatient services like physical therapy. They could also be the hosting facilities for home-based services like care assistants or community nurses who support dependent people in their homes. The authorities could convert redundant municipal hospitals into LTC Community Centers—privately or publicly owned—that could provide a whole range of LTC and rehabilitative services, and thus avoid the costly model of institutional care. 30. The matrices included in this volume present the World Bank team’s best estimate of what the savings from the particular measures suggested are likely to be. However, these estimates are in some cases based on secondary sources, such as household survey data, or vulnerable to even small changes in assumptions. This is particularly the case in trying to calculate the budget savings from changes in support to state-owned enterprises (SOEs) and subsidies to transportation. For these reasons, the World Bank team’s estimates are undoubtedly second best. Ideally, the estimates of likely immediate and longer term savings will be verified and further refined with better informed estimation by counterparts in each line ministry and in the Ministry of Finance, who have superior knowledge of Latvia’s programs and a far better command of administrative data sets. 1 The health sections of the PER suggest that the Ministry of Health pilot competitive contracting, which could help to develop several instruments that are currently missing in the system: public competitive procurement of medical services based on tenders transparently defining the services to be provided; tenders encouraging providers to cooperate to integrate care and compete for patients on that basis; contracts exceeding a year; much clearer rules of the game for providers and other insurers on what the government will cover, on service levels, quality of care, and so on. 10 II. Latvia’s Economic Boom and the Fiscal Consequences of High Growth 31. In the years leading up to the global financial crisis, Latvia’s economic growth significantly outpaced potential growth rates. Following Latvia’s accession into the European Union in 2004, the annual average rate of growth – which was already high at 7 percent from 2000 to 2003 - accelerated to more than 10 percent through 2007. In contrast to actual GDP growth, potential output (which grew on average by around 6 percent at the beginning of the decade) began to decline after 2004. In 2009, potential GDP growth was estimated to be negative 1.4 percent due to low gross fixed capital formation and the decline of employment as a result of the economic contraction. Real GDP growth reached its peak of 12.2 percent in 2006 (Figure 1.1) while the output gap culminated in 2007 (at around 18 percent). Figure 1.1 GDP and potential output growth Figure 1.2 Composition of FDI 15.0 6000 Real estate, renting and business activities Financial intermediation 10.0 Transport, storage and communication 5000 Retail trade 5.0 Construction 4000 Electricity, gas and water supply 0.0 Manufacturing GDP growth rates in constant 3000 -5.0 prices, % Agriculture, hunting and forestry Potential GDP growth rates, % -10.0 2000 -15.0 1000 -20.0 2000 2002 2004 2006 2008 2010 2012 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Source: Ministry of Finance Source: Bank of Latvia 32. Double digit growth after Latvia’s EU accession stemmed from a vigorous financial integration with the EU15 countries which lowered the price of borrowing and encouraged an unprecedented increase in private debt. The expansion of credit was largely financed by short-term external borrowing from foreign private banks, resulting in the highest loan to deposit ratio of any banking sector in the region (248 percent in 2007).2 Fast credit growth led to a consumption and real estate boom, and eventually to a bubble. The largest increase in foreign direct investment during 2004-2008 was observed in the financial intermediation, real estate, and retail trade sectors (Figure 1.2). 33. Accelerated economic growth, driven almost entirely by excessive domestic demand, translated into large macroeconomic imbalances. Consumer inflation rose from 6.2 percent in 2004 to over 10 percent in 2007. Rapid wage increases outstripped productivity growth undermining Latvia’s external competitiveness. The current account deficit widened from 13 percent of GDP in 2004 to almost 23 percent of GDP in 2007. By 2008, the need for a macroeconomic correction was increasingly evident. 2 Latvia experienced rapid credit growth, due to broadly held expectations of fast convergence with EU living standards, and a strict peg to the Euro which eliminated exchange rate risk. 11 Against this backdrop, in September 2008, the re-pricing of risk in emerging markets with the onset of the global financial crisis put Latvia’s currency peg under pressure.3 34. To correct this rapidly deteriorating financial situation, the Latvian authorities agreed to a stabilization program in December 2008. The program package of €7.5 billion - more than 35 percent of GDP - was supported by the EC, the IMF, EBRD, the World Bank, and bilateral support from Nordic and Central European countries, and is conditioned on strong fiscal adjustment to achieve internal devaluation and restore the country’s competitiveness under the fixed exchange rate arrangement. Given the extent of overheating of Latvia’s economy, the macroeconomic adjustment strategy was implemented at the cost of a sharp contraction of output.4 35. Latvia’s post accession boom fueled excessive public expenditures (Figure 1.3). Growth above potential during the period 2004-2007 generated buoyant tax revenues. A pro-cyclical fiscal stance turned these windfalls into rapidly increasing government spending which further fuelled the domestic boom. The fiscal deficits were small (not exceeding 2.5 percent of GDP) from the beginning of the current decade and reached a small surplus (0.6 percent of GDP) in 2007, which is partly responsible for disguising the problem. The extent of the fiscal expansion became apparent with the economic downturn when the financing of public expenditure became unsustainable in the face of much lower revenues. Figure 1.3 General Government Total Revenue and Expenditure (LVL millions) 3 By November 2008, the Bank of Latvia’s foreign exchange reserves had fallen by 20 percent to euro 3.4 billion. Latvia was soon after downgraded by rating agencies and its Eurobond spread increased to around 600 basis points, while the 5-year Credit Default Swap (CDS) spread jumped to around 1000 basis points in November 2008. 4 The sharp output contraction is expected to lead to a large negative output gap of almost 10 percent in 2010. 12 Figure 1.4 General Government Expenditure by function (COFOG, LVL millions) Source: Ministry of Finance 36. All public sectors of the general government (using the international COFOG functional classification) benefited from the pro-cyclical stance of fiscal policy (Figure 1.4). Expenditures in all governmental functions at least doubled between 2004 and their peaks in 2008. Despite this accelerated increase in public spending, prior to the crisis in 2007, Latvia had one of the lowest reductions in inequality due to government taxes and spending when compared to high-income OECD countries. The Gini coefficient for household income was 0.44 before and 0.35 after government taxes and transfers. Moreover, expenditures for social protection and general public services increased further in 2009 when nominal GDP fell by 20 percent. 13 Figure 1.5 Impact of Taxation and Expenditure on Income Equality in European and OECD Countries Source: Staff estimates using data from EUROSTAT and OECD 37. Fiscal consolidation in 2009 and the planned adjustment in 2010 are most apparent in health, education and defense, bringing nominal expenditures to or below the 2007 levels. Comparing Latvia’s COFOG expenditures with EU10 and EU27 over 2004 and 2008 suggests that Latvia’s expenditures as percent of GDP were below the averages of these groups in all sectors, except for education and economic affairs. In this context, Latvia’s expansion was largely in line with its European peers. However, given its growth rate being one of the highest in the region and driven entirely by domestic demand, it should have been targeted and partly channeled towards savings. 38. Such budget expansion led to large structural fiscal imbalances (Figure 1.6). During the period 2000-2008, general government expenditure grew on average, in real terms by 6.5 percent annually (deflated by GDP deflator). The structural (cyclically adjusted) fiscal deficit reached negative 7.4 percent of GDP in 2008 and minus 8 percent of GDP in 2009. The recent sharp recession represents a permanent shock to revenues. As such, permanent expenditure cuts and/or sustainable increase of revenues are required to correct the structural fiscal imbalances. 39. The Government’s main objective is to eliminate structural deficits and to return the economy to a sustainable fiscal path. The Government’s Stabilization and Convergence programs aim at designing both revenue and expenditure policies to reduce the cyclically-adjusted structural general government 14 budget deficit to below 1 percent of GDP after 2012. This would enable automatic stabilizers to smooth cyclical swings caused by factors outside the control of the government. 5 Figure 1.6 Output Gap, Nominal and Cyclically Adjusted Budget Deficits Source: Ministry of Finance 40. In the years ahead, the negative output gap is expected to close only gradually and both actual as well as potential growths are expected reach 4 percent in 2012. This expectation rests on the current macroeconomic outlook and the successful implementation of structural reforms agreed to in the stabilization program. The current program, aimed at the adjustment of domestic costs through fiscal consolidation and structural reforms, should stimulate new investment in export-oriented industries. However, this adjustment is likely to come about only gradually. III. Emergency Measures and the Need for Permanent Fiscal Adjustment 41. Fiscal adjustment has been a key macroeconomic tool for stabilizing the economy and putting it on sound footing. With a fixed exchange rate, there are few options for macroeconomic management other than through fiscal consolidation. The authorities chose to maintain the exchange rate peg after taking into account the balance sheet effect of a potential devaluation, severe corporate, banking and social implications, as well as fear of contagion beyond Latvia’s borders. Most importantly, maintaining the exchange rate peg has been considered a natural exit strategy of the stabilization program and the best way to prepare the country for the adoption of the euro. 42. The 2009-2010 fiscal correction came about primarily from “emergencyâ€? measures on both the expenditure and revenue sides. Tax revenues started declining significantly in the autumn of 2008. The fiscal consolidation launched in 2009 and continued in 2010 aimed at correcting the fiscal position mainly with immediate emergency measures. A comparison of baseline and projected fiscal balances 5 A Fiscal Responsibility Law that would allow counter-cyclical fiscal policy is currently being discussed by the authorities. One of its four rules permits a deficit of 0.4 times deviation of growth from its potential and has been supported by the IMF. This rule is a pragmatic approach to a cyclically adjusted budget balance aiming at balanced public finances during “normalâ€? growth of real GDP. 15 (Table 1.1) shows that fiscal measures amounting to 13.3 percent of GDP have been implemented, of which 2.2 percent result from tax policy changes (some of them simply reversing previous tax cuts, i.e., the VAT reduction in 2009).6 Expenditure cuts over a wide range of institutions and functional areas made up the remaining 10.4 percent of GDP suggesting that the bulk of adjustments have already taken place on the expenditure side. Table 1.1 General Government Operations 2008-10 (Percent of GDP) 2008 2010 Baseline Projection Measures Total revenue and grants 35.2 36.8 39.6 2.8 Tax revenue 29.1 25.1 27.3 2.2 Direct taxes 18.5 16.0 16.7 0.7 Indirect taxes 10.6 9.1 10.6 1.5 Non-tax 3.4 4.2 4.9 0.7 Total expenditure 38.5 58.2 47.8 10.4 Current expenditure 33.5 49.6 42.2 7.4 Remuneration 10.3 13.7 9.7 4.1 Goods and Services 5.7 7.6 5.6 2.0 Subsidies and transfers 16.5 27.0 24.8 2.2 Capital expenditure 4.6 6.9 3.9 3.0 Fiscal balance -3.3 -21.4 -8.1 13.3 Source: Latvian authorities and the IMF Staff estimates 43. After the supplemental budget was passed in June 2009, which introduced measures totaling LVL 500 million, the authorities managed to get the fiscal accounts under control. The initial spending cuts, which were unevenly implemented for most of the first half of 2009, proved insufficient in curbing past spending trends in light of more severe economic conditions. The supplementary budget introduced revenue measures, which amounted to 0.6 percent of GDP (higher excise tax rates, a lower threshold on income exempt from taxes, and increases in state revenue from dividends) and expenditure cuts, which reduced spending by 5.5 percent of GDP. These expenditure cuts came mainly from a lower wage bill (reflecting a nominal wage reduction and a fall in public sector employment) and lower social transfers (including a 10 percent nominal pension cut and a 70 percent reduction of pensions for working pension recipients, which have since been reversed7), but less through structural adjustments. 44. On December 1, 2009, Latvia’s Parliament (Saeima) approved the 2010 budget with a programmed deficit of 8.5 percent of GDP. The budget is designed to lower Latvia’s 2010 general government deficit by LVL 500 million (4.2 percent of GDP) as agreed in the revisions to the stabilization program in June 2009. The 2010 budget includes measures to broaden the revenue base (2.3 percent of GDP) and further cuts in expenditure (1.9 percent of GDP). Following on measures introduced in 2009 to protect the poor, the 2010 deficit target allows for an additional maximum ½ percent of GDP of social 6 Baseline scenario provides estimates of the budget without incorporating fiscal measures needed to achieve programmed fiscal targets. 7 On December 21, 2009, the Constitutional Court ruled that the government's decision to reduce pensions was unconstitutional. The ruling requires the government to (i) refund the withheld pensions (more than LVL 74 million) by 2015, and (ii) start to pay pensions in full from March 1 (LVL 111 million). 16 safety net spending if the authorities deem it appropriate to ease social hardship if the contraction is more severe than expected. In addition, the 2010 deficit accommodates the Constitutional Court ruling of December 21, 2009 that the pension cuts of June 2009 were unconstitutional.8 45. Implementation of the 2010 budget indicates that the approved adjustment of 8.5 percent of GDP is largely on track. Weak personal income taxes have been offset by strong indirect, corporate and nontax revenues. Since the government has been in minority since March 2010, several less desirable decisions have been also taken, including lowering hotel VAT from 21 to 10 percent, extending low VAT rates to books and the print media, exempting gifts for medicine and education from personal income tax and exempting unused infrastructure from real estate taxation. These measures will increase the deficit by 0.3 percent of GDP in 2010. However, the government intends to compensate for these slippages by cutting central government net lending. 46. Despite the substantial fiscal consolidation achieved in 2009-2010, Latvia’s fiscal policy continues to face unprecedented challenges due to still high fiscal imbalances compared to the authorities’ fiscal target of meeting the Maastricht fiscal criterion in 2012. Under the current macroeconomic framework, the required fiscal adjustment is estimated at LVL 800 million, or 6 percent of GDP. Of this amount, about LVL 420 million or 3.3 percent of GDP needs to be achieved in 2011. This is a formidable adjustment given the programmed deficit for 2010 is 8½ percent of GDP. In addition, revenues to the budget will be reduced by the reversal of pension contributions from the social insurance special budget back to the funded pension pillar (1.2 percent of GDP), declining non-tax revenues (0.6 percent of GDP), and increases in interest payments (1 percent of GDP).9 47. The new fiscal measures have to be prepared with more time giving consideration to their quality and sustainability. The agreement of the authorities to work with the World Bank on the new public expenditure review of social sectors is aimed at identifying areas for expenditure savings that will be sustainable over the medium term. Given that 2010 is an election year, the objective is to design a menu of available options for the next government and possibly, to determine more savings than are actually needed so that the new government can select the options that are the most politically palatable. 48. The authorities extended the Stabilization Program with international lenders until the end of 2011. This extension, made possible by better than expected financing conditions in 2009,10 is recognition that sustainable fiscal adjustment will take longer than initially programmed. The extension will cover the period of 2012 budget preparation which is key to bringing the deficit within 3 percent of GDP, as required by the Maastricht criterion. Continued involvement of international lenders beyond the elections in October 2010 should also provide assurances to financial markets that the authorities will continue with consistent and sustainable policies over the medium term. Recently, the authorities 8 The Government resumed payment of pensions in full in February 2010 (at a cost of around LVL110 million, or 1 percent of GDP) and refunded pensioners for the reduction with a lump sum in April 2010 (around ½ percent of GDP). 9 These estimates are subject to change depending on uncertainties related to economic growth, restructuring costs, domestic short term interest rates, etc. 10 The NIR target for 2009 will likely be exceeded by a very substantial margin and further accumulation of reserves is anticipated in 2010. Therefore, there is room for shifting some portion of the financing access into 2011. 17 expressed their intention to treat future disbursements as credit lines, thus lowering interest costs and public debt. With the exception of about euro 300 million from the EC and IMF and euro 100 million from the World Bank, other disbursements will be decided on a review-on-review basis. 49. While the consensus view is that most of the fiscal adjustment should come from the expenditure side, revenue measures will also still be required given the nature of fiscal imbalances and the significant size of the adjustment needed to fulfill fiscal deficit criterion for euro adoption in 2014. The Ministry of Finance and the IMF are working to identify scope for further revenue measures, which may include: (i) an increase in residential real estate tax combined with mechanisms to protect the poor; (ii) an alignment of reduced VAT rates with standard VAT rate on heating, natural gas, medicines, electricity and press; (iii) a review of the generous taxation of pension and insurance funds; and (iv) higher taxation on capital income and cars. These measures are intended to improve the quality and composition of the overall tax system as well as to increase incentives for households and firms to pay taxes. They will complement fiscal savings that will need to come from further reductions in spending to eliminate structural deficits and to improve efficiency of the public sector. 50. This section ends by summarizing the current policy discussion on the revenue side of Latvia’s fiscal consolidation, and places this discussion in a regional context. Revenue policy is being covered by technical assistance from other international agencies. However, some mention of the scope and structure of tax amendments is important, given the discussion of structural expenditure cuts to help fulfill the Maastricht fiscal criterion by 2012. 51. The tax system in Latvia is characterized by a relatively low tax burden and a reasonably wide tax base. In 2007, the tax burden as a percentage of GDP placed Latvia in the fourth lowest position in the EU. In 2009, the tax base was expanded with changes to the real estate tax rate. Despite the tax increases in 2009-10, the tax burden has still decreased due to structural changes such as a sharp fall in consumption, anemic exports, an increase of undeclared “shadow economyâ€?, as well as the decline of wages and corporate profits. In the next three years, when tax revenue to GDP is likely to rise slightly, the share of the tax burden will probably still remain below 29 percent. 52. The economic downturn led to changes in both tax structure and revenue, prompting the Government to consider future efforts in improving the composition of taxes. Until 2008, the share of direct taxes (the largest in total tax revenues) continued increasing. In 2009, due to personal income tax (PIT) rate reductions, an increase of non-taxable minimums, and higher VAT and excise rates, the overall tax burden shifted towards indirect taxes. The bulk of total tax revenues in Latvia is comprised of labor taxes, which rose steadily until 2008 (from 48.5 percent in 2005 to 49.8 percent in 2008). In tandem, the share of consumption taxes in total tax revenues has decreased (from 42 percent in 2005 to 36.4 percent in 2008), suggesting this as an area of revenue for future improvement. Especially under conditions of diminishing employment and increasing public debt, increasing and shifting the tax burden towards property and consumption taxation would be appropriate given that these taxes pose less of a harmful distortion to economic activity and less vulnerable to evasion. 53. The undeclared shadow economy poses a formidable challenge to any effort by the Government to increase tax revenue. Measuring the shadow economy is difficult, by the very nature of the 18 phenomenon. The estimates of shadow economy from the Central Statistical Bureau and State Revenue Service range from 12-16 percent, while other estimates are as high as 40 percent of GDP. As proxies for undeclared economic activity, according to labor force and other expert surveys, 11 percent of the employed population between 15-74 years old work without an employment contract, while another 12.7 percent receive additional “wages in envelopesâ€?. Table 1.2: Tax revenue in general budget (% of total tax revenue) 2004 2005 2006 2007 2008 2009* Direct taxes 62.66 59.49 59.20 61.21 65.37 63.34 Personal income tax 21.28 19.77 19.67 19.84 20.57 19.82 Social insurance contributions 32.37 30.31 29.94 30.78 33.33 36.16 Corporate income tax 6.25 7.02 7.60 8.93 10.06 5.36 Real estate tax 2.77 2.39 1.99 1.66 1.41 1.98 Indirect taxes 36.43 39.60 39.85 37.90 33.88 35.9 Value added tax 23.78 26.29 27.85 26.88 22.33 21.7 Excise duty 11.46 12.19 10.96 10.01 10.81 13.7 Tax on cars and motorcycles 0.36 0.37 0.41 0.37 0.19 0.07 Electricity tax - - - 0.01 0.02 0.02 Customs duty 0.84 0.75 0.62 0.62 0.53 0.41 Other taxes 0.90 0.91 0.96 0.90 0.75 0.76 Tax on natural resources 0.40 0.40 0.33 0.23 0.18 0.2 Lottery and gambling tax 0.50 0.51 0.63 0.66 0.57 0.56 Source: Ministry of Finance 54. A comparison of the implicit tax rates by consumption, labor and capital with Latvia’s neighbors and other EU countries shows both potential for and constraints to further tax reform. Official data as of 2007 show that the implicit tax rates in Latvia were about the EU average on consumption and labor, however significantly below the average on capital and business corporate income. Increases of VAT, excise and PIT rates in 2009 have contributed to Latvia’s changed relative position in the consumption and labor category. In addition, the broadening of the personal income tax base to include capital income, which came into effect on January 1, 2010, will increase the implicit tax rate on capital. This will be the only category of income where there is still some room for further improvement. However, the Government is deliberating any change with close attention to tax policy in neighboring countries since huge swings can contribute to capital flight and other unintended consequences. With the changes introduced in 2009-2010, room for extensive tax reform is limited. 19 Figure 1.7: Comparison of implicit tax rate Figure 1.8: Tax revenue by economic function Source: Eurostat Source: Convergence Program 2009-2012 55. A more detailed comparison of Latvia’s selected taxes with the EU, Lithuania and Estonia (Table 1.3) shows that considerable changes have been already made in the various tax categories but further amendments could be warranted. It also reveals that there are few taxes that have room for additional adjustment in the short-term. Among the few options that can result in short term savings are gradual increase of personal taxable minimums, elimination of the CIT tax reliefs (since they do not represent a significant share of business expenditure), elimination or increase of VAT reduced rates, or increase of the tax rate for expensive cars. In addition to these measures, longer-term impact on revenue could be achieved by expanding the RET rate on residential property -without due care not to overburden low- income citizens- and further increases in excise taxes. Table 1.3: Comparison of Selected Taxes Latvia Lithuania Estonia Comparison to EU and Other Comments Direct taxes Personal Min/Max Max/Min 21 Max/Min 21 EU maximum tax rate is higher than in Latvia income tax 2009- percent; percent; while the minimum tax rate is lower. Latvia has (PIT) 23percent; introduced a number of reliefs, the most 2010-26 significant being non-taxable minimum and percent allowance for dependents. The changes made in 2009/2010 have increased this tax burden and made the tax regressive. Currently, the non- taxable minimum is three times lower than in Estonia/Lithuania. Although the tax burden is not so great compared to EU, given the low purchasing power of Latvian citizens, wage decline, increasing unemployment and shadow economy, it would not be useful to further increase the tax burden from PIT. However, a gradual increase of the taxable minimum may be considered. 20 Latvia Lithuania Estonia Comparison to EU and Other Comments Direct taxes Corporate Reduced 15 percent 21 percent EU CIT rate is higher than in Latvia. In addition to income tax from 25 a very low CIT rate, Latvia has many tax relief (CIT) percent in incentives. This combination makes Latvia one of 1995 to 15 the most favorable countries within the EU. percent Usefulness of CIT tax relief is being considered since it does not represent a significant decrease in business expenditure. Real estate As of Jan. RET imposed RET is Despite the 2010 significant RET rate increase and tax (RET) 2010, RET on on land (at applied only base expansion, the RET burden is not excessive, land and about 1.5 on land and especially for residential areas. Further medium- buildings for percent) and the term solutions are expected to be developed economic buildings for municipality while not burdening low-income citizens. activity economic may fix a increased activity rate from 1 to 1.5 (municipality between percent; tax may set rate 1.1-2.5 base was between 0.3 percent widened by to 1 percent) including engineering construction (1.5 percent) and residential buildings (0.1-0.3 percent). Indirect taxes Value Standard Standard rate Standard The weighted average VAT rate in Latvia has been added tax rate 21 21 percent; rate 20 reduced from 17.28 percent in 2004 to 15.51 (VAT) percent; reduced rate percent ; percent in 2008, suggesting that the tax burden is reduced rate 5/9 percent reduced rate substantially lower than the standard rate due to 10 percent 5 percent exceptions and reduced rates. While the standard rate is among the highest in the EU (increased by 3 percent in 2009), it was unable to fully offset the tax revenue loss associated with decrease in demand. Since VAT affects consumer prices/ consumption, it would not be advisable to increase the VAT rate during economic downturn. A better solution would be eliminating or increasing the reduced rate or exceptions. 21 Latvia Lithuania Estonia Comparison to EU and Other Comments Direct taxes Excise duty Fuel (on Fuel – on Fuel – on Main focus of excise tax policy in Latvia has been – as of unlead petrol unlead petrol unlead harmonization with EU requirements on energy, 1.07.2009 379 Euro per 434 Euro per petrol 398 alcohol, and tobacco (Latvia only reached 100 Litre, on 100 Litre, on Euro per 100 minimum levels in 2009). Oil and tobacco rates diesel 330 diesel 330 Litre, on increased substantially contributing to high excise Euro), Euro diesel 369.9 revenue growth. However, increase on all excises alcohol Euro in 2009 was affected by sharp drop in trade (ethyl Tobacco 49.6 volumes. Further increases would contribute to alchohol Euro per 1000 Tobacco higher share of illegality in the market. Such 1255 Euro cigarettes. 63.8 Euro increases could be considered at a later time in a per 100 Litre, per 1000 wider EU context and when economic conditions intermediate cigarettes. improve. products Excise on ethyl 98.7 Euro, alcohol 1278 Excise on wine 56.4 Euro per 100 ethyl alcohol Euro), Litre, on 1291 Euro tobacco intermediate per 100Litre, (67.7 Euro products on per 1000 88Euro, on intermediate cigarettes) wine 57.3 Euro products are 142 Euro, on harmonized wine 66.5 among the Euro EU. In 2009, Latvia reached EU minimum for unlead petrol and diesel. In addition, Latvia charges excise tax for coffee and non-alcoholic beverages. Tax on cars As of January There is no There is no Some other EU states do not have such tax, and 2010, tax on such duty such duty although Latvia's tax burden in this category is motorcycles unregistered relatively low. Given the drop in the car market, cars or unless the tax burden is increased on registered exclusive/expensive cars, it would not be good to abroad was raise this tax for other car categories. introduced based on CO2 emissions ranging from 0.3 to 5 lats per each 22 Latvia Lithuania Estonia Comparison to EU and Other Comments Direct taxes gram per kilometer. Source: Ministry of Finance IV. Municipal Spending and Intergovernmental Fiscal Structures 56. Latvia is a unitary state with a single-level local government system. Municipal governments are key providers of social services in Latvia, being responsible for delivering about 70 percent of the overall education budget and for substantial supplementary social assistance programs and housing benefits, particularly for the poor. While health does not feature as a large expenditure under local government budgets as health services are largely funded through regional health funds, local governments have an important role to play in health care management as employers and operators of medical facilities. Thus, this review includes a chapter that focuses specifically on municipal spending and intergovernmental fiscal structures, the role of local authorities is touched on in almost all chapters of the report. 57. Prior to June 2009, sub-national governments in Latvia were organized along two levels. The first level (closest to citizens) consisted of republican cities, towns, and municipalities. The second level (county level) was formed by 26 districts and republican cities. Seven republican cities were included under both local government levels as they provided both first-level local government and district government functions. First level local governments were not subordinated to district governments. Following the implementation of local government reforms in 2009, there is a single level sub-national government system in Latvia consisting of republican cities and municipalities. The district level of government has been abolished. 58. Before the reform, the total number of local governments was 548, including 26 districts (rajons), 7 republican cities (republikas pilsÄ“ta), 50 towns (pilsÄ“ta), 424 rural municipalities (pagasts) and 41 reformed municipalities (novads). The average population at the first-level of local government before the reform was 4,260 thousand, but the number of residents in the smallest municipality (Kalncempju pagasts) was 251. The population stood at fewer than one thousand in more than one third of local administrative units. After the reform, the number of local government units declined to 118 comprising 9 republican cities and 109 municipalities (novads). 59. Despite the reform, there are significant disparities in population size between local governments, ranging from the capital city Riga with almost one third of the country’s population (709,145) to the municipality of Baltinavas with 1,365 residents. Thirty-seven municipalities have a population of less than five thousand residents. There is some evidence that further municipal mergers may be warranted in order to create municipalities of a sustainable size. Further mergers may lead to budget savings with regard to administrative functions. Preliminary evidence points to possible 23 inefficiencies due to the small scale of many municipalities. There is a tendency for per capita administrative costs to decline the higher the population size serviced by local government in Latvia. Out of the 13 municipalities with an administrative budget share of total spending of over 20 percent (compared to an average of 12 percent), 10 have a population of less than 10,000.11 It should be noted that international evidence is not conclusive as to the optimal size of local governments to achieve economies of scale. Some studies have suggested that there is a U-shaped relationship between population size and administrative costs and others have found that amalgamation do not lead to a reduction in administrative costs. 60. Beyond the weight of administrative expenses, there may be strong arguments for amalgamation related to increasing the quality of municipal service provision. Administrative capacity is likely to be constrained for local governments that cover very low populations. This was the main motivation for the recent reform of municipalities in Denmark. Denmark restructured its municipalities in 2005 by merging 270 municipalities into 98 units (enforced in 2007). Now there are few municipalities with less than 20,000 in population. The objective of this reform was chiefly to ensure municipalities have adequate administrative capacity for service delivery. Using the Danish population threshold, even following the reform Latvia still has 94 municipalities out of the total 109 with under 20,000 in population. 61. Even if the current number of municipalities (novads) is to be maintained, semi-amalgamation models could be considered. Novads may be too small for some of their functions, i.e., organizing public transport services, managing roads funding, operating secondary schools with wide curricula, operating orphanages and homes for the disabled or retirees. A semi-amalgamation model whereby small municipalities are combined into large units for service delivery would be an option for taking on these functions (perhaps using the five planning regions). 62. Local governments differ not only by size, but also by the demographic and socio-economic situation of the population. At the beginning of 2010, among municipalities the unemployment level varied from 7.8 percent (Ä€dažu novads, Garklanes novads) to 28.3 percent (ViļÄ?nu novads). Average unemployment in the republic cities at 10.7 percent was lower than that of municipalities, where the mean was 13.4 percent in 2010. There is also a large difference in the increase in unemployment that occurred in local government territories from January 1, 2009 to January 1, 2010. The rise in unemployment ranged from 4.3 to 13.5 percentage points. Among those municipalities that had to cope with the greatest increases in unemployment are those with small and largely rural populations. Dependency ratios vary across local government areas in Latvia, with differences in the share of under- and over-working age of the population. Such differences highlight the varied needs for service delivery of local government territories in Latvia. 63. The economic recession and rising unemployment led to an increase in the demand for social benefits for the poor. This showed the procyclical social protection spending pressures that arise at a local level in bad times. In first half of 2009 the expenditures for social benefits was lower than in the same period of previous year, but since August 2009 those expenditures outgrew those of the same 11 The data on administrative expenditures is based on budgeted amounts for 2010. 24 period in the in previous year. The impact of the crisis also is reflected in the increase in spending on the GMI. In 2008, the GMI equaled 3.4 percent of all social benefits; in 2009 it had increased to 8.9 percent. In the first three months of 2010, GMI benefits had risen to comprise a 16 percent share of overall social benefits given by local governments. 64. Individual local governments face very different conditions with regard to social protection spending. For the group of republican cities social benefits per capita varied from LVL 16 (JÅ«rmala) to LVL 38 (LiepÄ?ja) in 2009. For municipalities, it differed from LVL 7 (Rojas novads) to LVL 42 (Aizkraukles novads). This data and the different underlying social economic conditions (unemployment level, personal income tax revenue) show that local governments have very different needs for social assistance and also often different local social assistance policies. A significant proportion of local governments have problems in maintaining adequate social services. This situation was exacerbated in the aftermath of the recent crisis. As most household revenues have decreased, so have local government revenues, but the necessity for social support has been increasing. 65. Up until 2008, local budget revenues rose each year as did the share of local government revenues in overall general consolidated budget revenues. Local government revenues were 2.5 times higher in nominal terms in 2008 than in 2004 and their share of overall revenues had increased to 29.3 percent. The fall in government revenues in 2009 led to a nominal contraction of 20.7 percent in local government revenues. The share of local government consolidated budget in general budget revenues also decreased somewhat to 28.2 percent in 2009. 66. Local governments do not have the legal authority to impose taxes in Latvia. Only the national- level government has the power to set taxes. The local government shares revenues from state taxes. Tax revenues and transfers made up almost equal shares of local government revenues in 2009, combining to account for over 90 percent of total local government revenues. The majority of tax revenues came from personal income taxes (PIT) (40 percent of total local government revenues). Real estate taxes contributed under 5 percent to total local government revenues. Of state budget transfers, earmarked grants are the most significant, contributing 18 percent to total revenues. Inter-local government transfers account for 11 percent of overall basic budget revenues. Transfers here included the substantial transfer for teachers salaries recorded as a local government transfer as it went to municipalities through district government (but it went directly to republican cities). The Local Government Finance Equalization Fund (LGFEF) makes up a 5 percent share of total revenues. The main contributors to the LGFEF are local governments themselves. Own revenues remain relatively low at a 7 percent share of overall revenues. 67. The Government sets the PIT base and rate. Although Latvia allocates a large share of PIT to local governments, this is because there is a relatively low yield from the PIT, which has a low, flat rate. Administration, except in the cases of the republican cities of Riga and Ventspils, is carried out by the State Revenue Service. The local government share of PIT constantly increased from 2005 until 2009, going from 71.6 percent before 2005 to 73 percent in 2005, 75 percent in 2006, 79 percent in 2007, and 80 percent in 2008 to 83 percent in 2009, but in 2010 it decreased to 80 percent. As between local governments there is a big difference in the growth of PIT revenues in periods when revenues are expanding, with the increase in the PIT local government tax share during good times, the difference 25 between local governments’ financial capacity increased. Rich municipalities benefitted from the increase in the PIT tax share, whereas for poor municipalities there was almost no impact. 68. Chapter 3 of this review suggests that giving the ability to local governments to set and raise their own tax revenues could be considered. Tax pressure in general is relatively low in Latvia; and Latvia together with Malta is one of only two countries in the EU-27 without local government taxes.12 A challenge is that such a tax would need to be a stable revenue component if local governments were to rely on it. The tax would need to provide an adequate revenue base, and not be overly procyclical. More cyclical taxes, such as corporate taxes, would possibly promote procyclical revenues. A local sales tax is not allowed under EU regulations. A local income tax or real estate tax is a possibility. Taxes on immovable property are most appropriate for the financing local government. In the medium-term such a development is likely not to occur, as the national government may need to rely on any increase in real estate taxes rather than devolving them to lower levels government in the current tightly constrained fiscal environment. An additional possibility is to consider giving local governments more rights to charge local service fees. 69. As pointed out in Chapter 3, the equalization system is procyclical in that it is fully dependent on contemporaneous government revenues. The amount of central government funding that goes to the Local Government Finance Equalization Fund has remained unchanged at LVL 7.2 million since 2001. Therefore, the system relies mostly on the revenues received from personal income tax (with a smaller share coming from the real estate tax). Given that the benefits of the equalization structure are likely to be of greatest value to households during economic downturns, the building up of buffer resources to be used to fund transfers to needy local governments in bad times (to finance social spending requirements) could be a valuable countercyclical tool. Currently, when local governments’ tax share amounts decrease, so does the amount available for redistribution. 70. Small municipalities rely considerably on the equalization fund for revenues as their revenue share from PIT is low. A new law to alter the equalization system is due to be submitted to parliament next year (January 1, 2011). The ministry responsible has appointed a committee to examine the issue. One idea being discussed is that the mechanism may be deepened through additional resources from the State. 71. As the discussion of how to improve equalization takes place, it would be useful to clarify the objective and definition of terms used in the equalization framework, and the role of the equalization grant, given that other budget mechanisms (such as several of the earmarked grants) also contribute to vertical equalization. A clarification of the methodology and the order in which calculations are to be made is also merited. The current system is complicated and not understood by many local governments. To sustain the effectiveness of the equalization mechanism, the indicators used for the equalization formula could be regularly updated, at least every 4 to 5 years. 12 Source: Sub-national governments in the European Union: Organization, responsibilities and finance. Dexia, July 2008, page 97. 26 72. The mechanism for redistributing revenues across local governments would also benefit from a system of monitoring and evaluation. For example, there is no regular analysis of how the mechanism has performed. The criteria used to determine each unit of local government’s spending needs are connected with providing basic local services to the population (education, social services, etc.). It would be useful to examine whether the redistributed funds result in an equal provision of such services across local governments. In addition, the performance of the revenue forecasts used to calculate the equalization needs are not compared with the actual situation that evolves. 73. There may be some room for redesigning the equalization mechanism to increase the incentives for local governments to raise revenues. Under the current system, a government that benefits from the equalization grant has little motivation to increase tax and other revenues if increased revenue effort would result in them losing the equalization grant. One method that could be used to increase the returns to collecting revenues is to reward local governments for increased revenue effort by setting aside a portion of revenues raised and not subjecting it to the equalization calculation. 74. Chapter 3 makes several suggestions for improving the efficiency, equity and predictability of the municipal revenue and spending. The discussion and suggestions are general in character, and require refinement to take account of the political complexities of Latvia’s administrative structure. However, the discussion in Chapter 3 is of core importance to any attempts to sustainably improve the performance social policy in Latvia, as discussed in the other chapters of this review. Local government expenditure has a substantial social sector element, and the activities financed from social sector budgets are in most cases, those that benefit households most directly. 27 Chapter 2: Public Administration Reform, Subsidies, and State Owned Enterprises I. Introduction and Main Messages 75. This chapter of the PER reviews expenditure trends in the public administration with a focus on the remuneration costs of public employees (the public sector wage bill), and subsidies to publicly and privately owned enterprises.13 Latvia’s spending on these categories is compared with other countries in Central and Eastern Europe (CEE) and elsewhere in the European Union (EU) based on available data. The chapter examines the Government’s response to the crisis and economic contraction, by reviewing the distribution of the spending cuts in the 2009 and 2010 budgets. The chapter proposes options that the Government may wish to consider as a means of achieving further budget savings and increasing revenues to meet its fiscal targets for the years 2011 and 2012. Potential savings are calculated based on 2009 expenditure levels. Finally, the chapter outlines some short- and medium-term reforms that could help the Government improve efficiency and predictability in public expenditure. 76. The savings that may be available in the next two years involve implementing further consolidation and/or elimination of ministries, agencies, and functions; greater simplification and consistency in the application of uniform salary and human resource rules; the reduction and even elimination of subsidies to loss making enterprises; and measures to strengthen public financial management. 77. There is considerable scope for further consolidation of administrative bodies, which could reduce duplication of functions. At the national government level further staff reductions are feasible among categories of public employees that appear to be generously staffed, compared with other countries in Europe. However, further across-the-board reductions in public sector salary levels are not advisable at this time, as they may erode staff motivation and risk a loss of the best people once the labor market begins to recover. This said, the Government would be better served by a coherent and fiscally sustainable public sector salary policy, characterized by firm establishment and fiscal controls to avoid another uncontrolled expansion of public employment and the wage bill. Further simplification on the heels of recent reforms could be pursued, and a central unit with a robust mandate to monitor application of uniform human resource policies, including pay policy, across government bodies would help. 78. Furthermore, the Government could find substantial savings from a review of spending on subsidies paid to enterprises, with the aim to substantially reduce or phase these subsidies out. Where there may be a social welfare justification for subsidies, these would more effectively be delivered as targeted transfers as part of the welfare system, which augment the demand of low-income households for important services. Subsidizing demand is a far more nimble and equitable approach than the current subsidies to the supply of services that encourage inefficiency. 13 The chapter does not cover subsidies in the health or education sectors, which are examined in separate chapters, and those related to EU programs, which are protected expenditures. 28 II. Latvia’s Wage Bill and Subsidies to Enterprise Prior to the Crisis 79. Latvia’s wage bill14 as a share of GDP appears to have been somewhat higher than in comparator countries prior to the crisis. According to public expenditure data from the IMF GFS Yearbook of 2008, Latvia’s wage bill represented just below 10 percent of GDP in 2007 (Figure 2.1), the last year before the onset of the crisis. This was above the average for Central and Eastern Europe (8.5 percent), but below the average for Eurozone countries (11.5 percent). In 2008, Latvia’s wage bill actually increased to 10.3 percent of GDP. According to the EUROSTAT GFS report published in May 2010, Latvia’s wage bill before the crisis appears even higher: 10 percent of GDP in 2006, 10.6 percent in 2007, and rose to 12 percent in 2008.15 Figure 2.1: Wage Bill as a Percentage of GDP in 2007 12 10 8 6 4 local central 2 0 Note: Data for Romania refer to 2006 Source: IMF, GFS Yearbook, 2008 80. As a share of public expenditure, Latvia’s wage bill seemed disproportionately high prior to the crisis, but its share of public expenditure has gradually declined. Benchmarking the wage bill as a share of public expenditure gives a better measure of its affordability, as this takes into account the large differences in the share of public expenditure across countries. Comparative GFS data suggest that Latvia was an outlier in spending 31 percent of primary expenditure on its wage bill (Figure 2.2).16 This 14 The wage bill comprises all elements of remuneration, including employer’s social security contributions. It therefore corresponds to Latvia’s Treasury economic code 1000 (remuneration) and the IMF GFS category ‘compensation of employees’ (table 2). 15 Eurostat reports higher wage bill figures than the IMF GFS and Latvia’s Treasury. For example, for 2007, Eurostat reports Latvia’s wage bill as LVL 1,574 million, compared to LVL 1,393 million reported by the IMF GFS and LVL 1,391 million reported by Latvia’s Treasury. 16 Primary expenditure corresponds to the category ‘expense’ (table 2) of the GFS, which excludes external debt servicing costs. 29 was considerably above the averages for both Eurozone members (25.3 percent) and CEE countries (24.5 percent).17 Figure 2.2: Wage Bill as % of Primary Expenditure in 2007 35 30 25 20 15 local 10 central 5 0 Note: Data for Romania refer to 2006 Source: IMF, GFS Yearbook, 2008 81. Latvia’s wage bill grew steadily in the pre-crisis period and continued to expand as late as 2008, despite cuts introduced in the second half of 2008. Some of this increase was due to growth in staffing (see below), but growth in salary levels was the main driver. As shown in figure 2.4 until 2008, the wage bill continued to rise above inflation. Figure 2.3: Trend in Wage Bill Expenditure (LVL Figure 2.4: Trend in Wage Bill Expenditure (LVL million), Nominal million), Nominal vs. in Real Terms (Deflated by CPI) 1,800 1,800 1,600 1,600 1,400 1,400 780.7 1,200 1,200 616 1,000 1,000 656.5 800 800 local 449.5 490.9 600 600 365.4 central 400 400 895 775 580 200 200 413 521 469 0 0 2006 2007 2008 2009 2010 (plan) 2005 2006 2007 2008 2009 2010 (plan) nominal real terms For 2005: GFS data; for 2006-2009: Treasury data 17 Eurostat data show Latvia’s wage bill as 30.6 percent of primary expenditure in 2007, compared to 24.8 percent for the EU average and 23.9 percent for the Eurozone average. 30 82. Public employment in Latvia expanded during the boom years that preceded the crisis to levels exceeded only by neighboring Estonia and Lithuania. Though data on the size of public employment in Latvia for the years before 2008 is patchy and unreliable, estimates by the State Chancellery suggest that general government employment grew from 7.5 percent of the population in 2005 to 8 percent in 2008, declining to 7.4 percent in 2009 after staff cuts that were implemented in response to the crisis. 18 According to staff estimates for this PER based on data from the State Chancellery and the Ministry of Finance, in the last quarter of 2009 government employees at the central and the local levels represented 7.4 of the population or 14.3 percent of the labor force. This level of public employment is higher than in most Central and Eastern European countries, but lower than in neighboring Estonia and Lithuania, even allowing for the fact that in Latvia healthcare personnel are not considered public employees (Figure 2.6). It is worth noting, however, that public employment levels in many Central and Eastern European countries are above the OECD and EU averages. Moreover, Latvia has a considerable segment of employment in state-owned enterprises (SOEs). If staff in SOEs are included, public employment in Latvia reaches 10 percent of the population or 19.5 percent of the total labor force. According to the EBRD, in 2008 the public sector accounted for 32.1 percent of total employment in Latvia, which is the highest share in CEE (figure 2.7).19 Figure 2.5: Government Employees as Percentage of Figure 2.6: Government Employees (Excluding Health the Population Sector Workers) as Percentage of the Population and Labor Force 9 20 8 18 7 16 3.5 3.6 14 6 3.3 3.4 3.3 12 5 10 4 8 3 6 2 4.2 4.3 4.4 4.4 4.1 4 1 2 0 0 2005 2006 2007 2008 2009 Lithuania Estonia LATVIA Hungary Poland Slovenia Croatia local govt central govt % of population % of labor force Source: State Chancellery Source: National Statistical Offices; Staff calculations (data for Latvia refer to 2009 and 2008 for other countries) 18 However, Chancellery data appear to understate total government employment because the data exclude a number of central institutions that have the status of ‘public agencies’ (e.g. some higher education institutions) that are neither ‘state institutions’ nor are they subordinated to a ministry. 19 EBRD, Transition Report 2009: Transition in Crisis?, p. 186. In the EBRD definition, public sector employment includes SOEs. 31 Figure 2.7: Public Sector Employment as Percentage of Total Employment 35 30 25 20 15 10 5 0 Source: EBRD Transition Report, 2009 Figure 2.8: Public Sector Employment as Percentage of the Labor Force in OECD Countries Latvia: 19.55% Source: OECD, Employment in Government, 2008 (data for most countries refer to 2005) 83. Like in other countries in the region, even before the crisis the salaries of government employees were somewhat higher than the average salary in the Latvian economy. According to data from Latvia’s Central Statistics Bureau, in 2008 the average salary in the government was 14 percent higher than the average salary in the economy. However, as in other countries in the region, salary levels in the private sector may be under-estimated in official data due to the allegedly widespread practice by firms of under-reporting employee salaries to reduce payroll tax liabilities. 32 Figure 2.9: Ratio of Average Government Salary to Average Salary in the Economy (2008) 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Lithuania Estonia Slovenia Hungary Poland LATVIA Croatia Source: National Statistical Offices 84. Latvia’s expenditure on subsidies to enterprises was also comparatively high. Latvia’s system of health financing, whereby most healthcare providers have the status of commercial entities, which are financed primarily through government subsidies, partly accounts for the high share of subsidies in public expenditure. According to the GFS classification, in 2007 Latvia’s spending on subsidies represented 7.23 percent of GDP. This was the second-highest in the EU behind the Czech Republic and far above the average in the Eurozone (1.39 percent). EUROSTAT data on “state aidâ€? to enterprises, which is a narrower definition than subsidies as an economic category,20 show that state aid in Latvia (at 1 percent of GDP in 2007) was also higher than the EU average (0.5 percent). In Latvia state aid peaked in 2006 at 1.3 percent of GDP, but declined afterwards until the crisis necessitated the rescue of two major banks by the Government. Figure 2.10: Subsidies as a Share of GDP Figure 2.11: State Aid as a Share of GPD 8 3.00 7 2.50 6 2.00 5 1.50 4 1.00 3 0.50 2 0.00 1 0 2005 2007 Source: IMF, GFS Yearbook 2008 Source: Eurostat GFS, 2010 20 The Eurostat definition of state aid includes subsidies to firms (both state- and privately owned) in specific sectors (agriculture, fisheries, manufacturing, coal, transport except railways and other services) as well as those given on ad-hoc basis to individual firms (e.g., for rescue and restructuring) and for horizontal objectives such as research and development, safeguarding the environment, energy saving, support to small and medium-sized enterprises, employment creation, the promotion of training and aid for regional development. Therefore it does not include subsidies for public services such as utilities, healthcare, or public transport. 33 III. Changes in Administration and Enterprise Support Since the Onset of the Crisis 85. Starting in the second half of 2008, the Government took decisive actions to cut the wage bill. This was done by reducing remuneration levels and by cutting staff numbers by 14.8 percent between the first quarter of 2008 and the fourth quarter of 2009.21 Base salaries of central government employees were cut by 30 percent on average. These salary cuts were differentiated so that salaries above LVL 300 per month were reduced by 32 percent, while those below LVL 300 were reduced by 28 percent. 86. The wage bill was also reduced by freezing the payment of performance-related bonuses (as of November 2008) and most other allowances such as holiday pay through 2011. This measure has meant that many civil servants, especially those in senior positions in central government bodies for whom performance-related bonuses accounted for a significant share of remuneration, have suffered a steep decline in their earnings. At the same time, the freeze on bonuses and allowances has produced substantial savings, as bonuses and allowances declined from 11 percent of the total wage bill in 2007 to 3.7 percent in 2009.22 As a result, in 2009 the wage bill declined by 26.2 percent compared to 2008 and is projected to decline by a further 23.4 percent in 2010. As a result of these cuts, the ratio between the average salary in the economy and the average government salary declined from 1:1.14 in 2008 to 1: 1.02 in 2009.23 As a share of GPD, the wage bill declined from 10.3 percent in 2008 to 9.3 percent in 2009 or according to EUROSTAT data from 12 percent in 2008 to 11.9 percent in 2009.24 Figure 2.12: Wage bill as % of GDP Figure 2.13: Wage bill as % of primary expenditure 14 14 12 12 10 10 8 8 6 6 4 4 2 2 0 0 2006 2007 2008 2009 2006 2007 2008 2009 Source: Eurostat, GFS 2010 edition 21 This estimate is based on Chancellery data, with 16 percent reduction at the central level and 13.7 percent reduction at the local level. The Prime Minister’s report to the Saeima of March 2010 notes that by December 2009 staffing in central government bodies had been cut by 19 percent compare to average staffing levels for 2008. 22 Bonuses were much more widespread in the central government, where they made up 14.7 percent of the wage bill, as opposed to only 7 percent of the local government wage bill in 2007. 23 This ratio is based on data by the Central Bureau of Statistics. Data by the State Revenue Service show higher average earnings for public administration employees. 24 The difference is due to higher wage bill figures reported by Eurostat: LVL 1,947 million in 2008 (compared to LVL 1,676 million by the MoF of Latvia) and LVL 1,578 million in 2009 (LVL 1,237 million reported by the MoF). 34 87. The crisis has spurred an effort to expand the coverage of the unified remuneration principles, which were first introduced in the central government in 2007 to reduce pay differentials between staff in different government bodies and improve wage bill management. The Law ‘On Compensation of Officials and Employees of State and Local Government Institutions in 2007’, which came into force in January 2009, extended the grading system to all public administration institutions (including most independent agencies). The grading system comprises 16 vertical grades (monthly salary groups), which are based on 43 families, each divided into levels of responsibility. Each grade (monthly salary group) is divided into 6 horizontal steps (qualifications grade). The vertical grade determines 67 percent of salary, while the horizontal salary step (qualifications grade), which accounts for 33 percent of salary, is defined based on length of service (10 percent) and the employee’s performance appraisal rating (23 percent). 88. The application of these principles helped to limit but not eliminate pay differentials among public administration bodies, as the remuneration system established overlapping salary bands that allow for significant pay disparities.25 The law that came into force in January 2009 also abolished performance contracts for civil service managers, which had previously accounted for high shares of total remuneration. The Law ‘On Compensation of Officials and Employees of State and Local Government Institutions in 2010’ and the Government regulation no. 1651, which came into force in January 2010, further reduced base salary levels for civil servants and other employees of government bodies by 5 percent and made some further progress in streamlining the salary system. However, different remuneration systems continue to apply to ministries and subordinated agencies, while municipalities remain largely beyond the scope of this remuneration policy. 89. Local governments’ wage bill has declined far less sharply than that of the national government. Reflecting this, there was an increase in local governments’ share of the total wage bill. Treasury data show that local government spending on employees’ remuneration declined by 16 percent in 2008- 2009, compared to 35.2 percent decline in remuneration for employees of the national government. As a result, local governments’ share of the total wage bill exceeded the national government wage bill in 2009 for the first time, growing to 53 percent from 46.6 percent in 2008. 25 As of January 2010, the ranges are narrow at the lower levels of the administration, but present wide gaps at some of the higher levels, with grade 13A step 6 being the most extreme case (minimum of LVL 605; maximum of LVL 1,347). 35 Figure 2.14: Trends in Wage Bill Expenditure (Central Figure 2.15: Local Government as Percentage of Total vs. Local Government), LVL million Wage Bill 1,000 54 900 52 800 50 700 600 48 500 46 400 300 44 2005 2006 2007 2008 2009 42 central local 2005 2006 2007 2008 2009 Sources: GFS (2005); Treasury reports (2006-2009) 90. The national government has shouldered a disproportionate share of the salary cuts in the public sector, as well as sharper staff cuts compared to local government. According to data from the State Chancellery, national government bodies (ministries, agencies subordinated ministries, and other bodies subordinated to the Council of Ministers) reduced staffing by 16 percent between the start of 2008 and the end of 2009, while local governments did so by 13.7 percent. MOF data, which do not include local government, but do include independent public institutions and commercial entities owned by national government, show a reduction of 11.4 percent of employees between the first and fourth quarters of 2009 – or 11.9 reduction of full-time staff positions. Figure 2.16: Government staffing by quarter Figure 2.17: Public employment cuts in 2009 120,000 120,000 100,000 100,000 80,000 80,000 60,000 60,000 40,000 40,000 20,000 20,000 0 0 Q1 2008 Q2 2008 Q3 2008 Q4 2008 Q1 2009 Q2 2009 Q3 2009 Q4 2009 central govt local govt central SOEs local SOEs local central Q1 2009 Q4 2009 Source: State Chancellery Source: Staff estimates based on MoF and State Chancellery data 91. Staff cuts have so far been achieved largely by merging administrative bodies and consolidating support functions, and do not appear to have had a tangible impact on core policy and service delivery functions. The Government took steps to reduce the number of administrative bodies, primarily by merging bodies with related functions. As a result of this consolidation, by January 2010 the number of 36 central government bodies and subordinated agencies was reduced to 13626, from 229 a year earlier. The Government also consolidated some support functions such as human resource management, accounting, and audit among ministries’ departments and subordinated institutions. This reflects an effort to minimize disruption to core government functions and the delivery of public services. However, in the absence of a central personnel database, the Government lacks detailed information on staff composition in different administrative bodies, which makes it difficult to identify areas where further staff reductions could be undertaken without a significant impact on core functions or public services. To address this information gap, the Government has conducted targeted functional reviews focused on areas of apparent duplication of functions (e.g. support functions, vehicle registration, forensic services) to assess the scope for further efficiency gains and identify options for reform. Such options for achieving permanent savings include combinations of delegating, outsourcing, or eliminating certain non-core functions – to be decided on a case-by-case basis. 92. The scale of staff cuts has varied among budget organizations. In the course of 2009, the Ministry of Environment let go 23 percent of its staff, the Ministry of Foreign Affairs 22 percent, and the Ministry of Economy 20 percent. The Ministry of Welfare lost 9 percent of its staff, the Ministries of Justice and Finance 11 percent and the Ministry of Culture 13.5 percent. The Ministry of Health lost 26 percent of its headquarters staff, but staff levels at the subordinated agencies actually increased by 11.4 percent, mostly due to the absorption of functions previously exercised by sub-national authorities. In contrast, staffing at the Ministry of Regional Development and Local Government and the subordinated Agency for Regional Development actually increased by 18 percent in 2009. This might be related to the local government reform (e.g. absorption of regional coordination functions previously exercised by the recently abolished rajons), but it nevertheless a marked exception to the overall trend. While it is not necessary to reduce staffing in all ministries equally, the Government might consider whether it is possible to identify further room for staff reductions in ministries and agencies with large staff numbers, especially those that have seen relatively limited cuts so far. 93. Likewise, the introduction of the new salary system may have narrowed previous divergences in salary levels among ministries and agencies, but significant disparities remain. The average salary in the headquarters of the highest-paying ministry (the Ministry of Welfare) is 25 percent higher than that in the lowest-paying ministries (Ministry of Education and Science and Ministry of Culture). Payroll data suggest that indeed different ministries continue to pay different salaries for employees of the same grade and salary step (within the permissible bands for each step) and that they use the grading system, particularly the horizontal qualifications grades, inconsistently.27 26 This includes 39 central government bodies and 97 subordinated institutions. 27 For example, the average salary for employees in grade (monthly salary group) 11, step (qualifications grade) 1 in the Ministry of Justice was 25 percent higher than in the Ministry of Finance. 37 Figure 2.18: Average Monthly Salary in Ministries’ HQ (in LVL) as of January 2010 1000 900 800 700 600 500 400 300 200 100 0 Source: State Chancellery 94. SOEs and commercial entities funded primarily from the state budget have not avoided staff cuts, but appear to have largely escaped salary reductions. According to MoF data, in 2009 staffing in SOEs owned by the central government declined by 8 percent and average salaries by 7 percent. Labor survey data from the Central Statistics Bureau suggest that average public sector salaries, which include SOE employees, were 3.8 percent higher than the average government salaries in 2008, with the gap widening to 7.2 percent in 2009. As SOEs accounted for 27 percent of public employees at the end of 2009, the average salary in SOEs was 65 percent higher than the average government salary and 79 percent higher than the average salary in the private sector. Salaries in SOEs are not currently subjected to legal restrictions. While – from a fiscal point of view - this may not be a concern in the case of those SOEs that are profit-making and do not represent a burden on the budget, in the case of SOEs that receive direct financing from the budget or enjoy a monopoly position, press reports on their managers’ and employees’ salary levels can undermine the credibility of the Government’s program of fiscal austerity. Figure 2.19: Average monthly salary by sector of employment (LVL) 600 500 400 300 2008 200 2009 100 0 Public sector General govt Private sector Average salary sector in the economy Source: Central Statistics Bureau 38 95. Subsidies to enterprises have been reduced and are now primarily targeted at healthcare expenditure and the implementation of EU policies and programs. The high level of subsidies to health care providers reflects the structure of healthcare financing in Latvia, which has a “purchaser-provider splitâ€?. Because providers of healthcare are commercial entities that provide services to households paid for by the Ministry of Health, the bulk of Latvia’s healthcare spending is reflected in the budget as subsidies. Healthcare as a sector therefore absorbs the largest share of subsidies -- 38.6 percent of all subsidies paid out of the central government budget in 2009 (down from 52.6 percent in 2006). Between 2008 and 2009, however, subsidies for health declined by 16.2 percent (as is explained in Chapter 5 of the PER). 96. Latvia’s subsidies to firms related to the implementation of EU policies and programs have continued to rise since the onset of the crisis. The main beneficiary sectors were agriculture, environment, and economic affairs. This trend is in line with the Government’s policy of maximizing the absorption of EU funds, which requires co-financing from Latvia’s budget. At the same time, Latvia’s spending on the EU definition of state aid increased from 1 percent of GDP in 2007 to 5.1 percent in 2008 reflecting funds spent to recapitalize two distressed financial institutions. Figure 2.20: Trend in subsidies Figure 2.21: EU-related subsidies as % of total 1400 55 1200 50 1000 45 800 600 40 400 35 200 30 0 2007 2008 2009 2010 budget 25 2006 2007 2008 2009 central local Source: Treasury 39 Figure 2.22: Functional composition of central Figure 2.23: Trends in central government subsidies government subsidies (excluding health and agriculture) 100% 60 90% 80% 50 70% 40 60% 50% 30 40% 30% 20 20% 10 10% 0% 0 2006 2007 2008 2009 Health Agriculture Economic Environment Transport Culture & affairs broadcasting Health Agriculture Economic affairs Environment Culture & broadcasting Transport 2006 2007 2008 2009 28 Source: Treasury 97. After transfers for health services and EU programs, transport and culture are the sectors that receive the largest volume of subsidies. Subsidies for culture and broadcasting declined by 30 percent between 2008 and 2009. However, at LVL 13 million in 2009, subsidies for radio and television continue to be substantial. By contrast, subsidies to the transport sector increased between 2008 and 2009, though most of this increase was related to co-financing of capital projects supported by EU structural funds.29 Subsidies for passenger transport, which are meant to cover the operational losses of carriers, continue to account for the bulk of transport subsidies and declined by only 4 percent, from LVL 74 million in 2008 to LVL 71 million in 2009. This includes both subsidies paid by the national government to the railway company and to inter-city road transport providers (LVL 43 million in 2009), and subsidies paid by municipalities to bus companies out of an earmarked grant from the national budget (LVL 28.5 million in 2009). Passenger transport subsidies are set to decrease by a further 21 percent in 2010 (see figure 2.25) with savings being achieved through cuts in services rather than fare increases. 28 Data for transport include the subsidy for passenger transport (Treasury code 3300), which is not included in the general subsidies (code 3200) and also the targeted grant to local government for passenger transport (which is used to pay subsidies to firms) but appears as a ‘transfer’ in the central budget classification (under code 7300). 29 According to Treasury data, total subsidies in the transport sector increased from LVL 86 million in 2008 to LVL 98 million in 2009, of which LVL 13 million was related to EU programs. 40 Figure 2.24: Trend in Transport Subsidies Figure 2.25: Trend in Passenger Transport Subsidies 100 80 27 70 80 28 60 31 29 60 50 11 27 9 27 40 6 40 16 14 12 30 20 40 47 43 20 17 13 22 18 0 10 13 2006 2007 2008 2009 0 MoT subsidies for passenger transport 2008 2009 2010 other MoT subsidies Urban bus routes Regional (rural) bus routes Inter-city bus routes Source: Treasury Source: Ministry of Transport and Communications 98. The Government has also increased revenues in the form of dividends from SOEs. Thanks to a legal amendment that raised the minimum dividend paid by profit-making SOEs to the budget from 15 percent to 80 percent in 2009 the revenue to the state budget from dividends reached LVL 138.3 million compared to 89 million in 2008 and 59 million in 2008. In the short term this has proven a useful mechanism for increasing revenues, but the provision may need to be revisited in the future to ensure that it does not damage the enterprises’ investment strategies. IV. Options for Achieving Budget Savings in the Immediate Future Remuneration policy in the public administration 99. Further across-the-board reductions in salary levels are not recommended, as they may erode staff motivation and risk the loss of highly qualified staff as the labor market starts to recover. Public employees began to experience pay cuts in the second half of 2008, when the payment of bonuses and allowances was frozen, and much sharper cuts in 2009, when base salaries were cut by an average of 30 percent. High-level officials in the national government, for whom performance-related bonuses represented a higher share of their total pay, experienced particularly sharp losses, which in many cases exceeded 40 percent. Due to the continuing economic decline and the depressed labor market, so far these sharp pay cuts do not seem to have led to a significant out-flux of staff from the public sector. However, they may have negatively affected staff motivation and productivity – an impact that the Government may wish to investigate through a survey and/or focus groups of public employees. As the Latvian economy and those of other EU countries gradually recover, the competitiveness of remuneration in the public administration may lag behind and staff may leave the public administration for other opportunities. The Government may therefore wish to closely monitor salary trends in the private sector and, once the fiscal space is available, review salary levels to ensure the public administration’s capacity to attract and retain qualified staff. 100. However, pay levels in ‘outlier’ organizations that pay disproportionally high salaries can be further adjusted by reducing the ceilings of the salary bands, thereby narrowing pay differentials. 41 Adjusting pay levels in some of the high-paying ministries (e.g. Ministry of Agriculture, Health, Welfare, Regional Development) and some agencies (e.g. the Social Integration Foundation, State Education Development Agency, National Film Center), where average pay levels outstrip those of other ministries, could yield minor but significant savings (approximately LVL 1.5 million). At the same time, this step could pave the way for further reform of remuneration aimed at equalizing salary levels across the public administration (for example, by eliminating the salary bands). Scope for further staff reductions, consolidation of agencies, and other efficiency gains 101. At the national government level, further staff reductions appear feasible among certain categories of public employees that seem comparatively generously staffed. For example, police staffing has been cut considerably in recent years, with the number of police officers per 1,000 residents declining from 6 in 2006 to 3.5 in 2009. However, a further reduction of some 5 percent could be considered to bring staffing levels closer to those of leaner forces in countries with comparable crime rates (e.g. Lithuania and Estonia). The prison administration also remains overstaffed with some 2,800 staff for some 6,500 prisoners.30 Likewise, Latvia has a comparatively generously staffed court system – with 22 judges and 63 support staff per 100,000 residents.31 Figure 2.26: Police Officers per 1,000 Residents (Data for Latvia refer to 2009; 2006 for other countries) 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 32 Source: UNODC, Tenth CTS (2008) 102. The tax administration (State Revenue Service) appears overstaffed but underpaid, with 4,500 staff and an average salary of LVL 329. The most recent Business Environment and Enterprise Performance Survey (BEEPS), which was conducted in 2008, shows that tax inspections are more frequent than in other CEE countries. The Government could reduce the number of SRS offices (from 38 at present to 5-6) and reduce the frequency of tax inspections, increasing the use of risk-based audits 30 Data on prison population refer to 2006 (as reported in UNODC CTS 2008). 31 Data refer to 2006 based. For full comparative data see Council of Europe Commission for the Efficiency of Justice (CEPEJ), Evaluation of European Judicial Systems, Third Report, 2008 32 UN Office of Drugs and Crime, Tenth UN Survey of Crime Trends and Operations of Criminal Justice Systems, 2008 42 and third-party information. In 2008, 61 percent of firms surveyed in Latvia reported having been visited by tax officials (compared to 48 percent in EU10 countries) 2.9 times on average (compare to an average 2.5 times in EU10 countries). Therefore, further streamlining of the revenue administration could improve the business environment and result in efficiency gains. However, short-term savings from staff reductions may need to be used to increase salary levels, which currently seem low (lower than salaries in the police). Cutting staff numbers by two thirds and doubling salary levels would still yield substantial savings. The ability of the State Revenue Service to attract and retain qualified staff should be a key consideration in setting salary levels to ensure the integrity and effectiveness of revenue administration. A reduction in the number of SRS offices would also produce substantial savings, but a reduced physical presence across the country would likely require some business process re-engineering and investments in upgrading the IT infrastructure to increase efficiency.33 103. Considerable scope remains for further consolidation of administrative bodies, which could help reduce duplication of functions, though the savings may be relatively modest after the consolidation achieved in 2009. Some further efficiency gains could be achieved by merging institutions, including ministries. Compared with other small countries in Europe, Latvia does not have a particularly large number of ministries (see figure 2.27), though it does have a high number of subordinated agencies, which offer wider scope for consolidation. Figure 2.27: Number of Ministries in Small Countries (Population Under 6 Million) 20 LU DK 18 HR 16 SK LV LT SI IE 14 EE 12 FI CY IS 10 MT 8 6 104. However, it would seem possible to further reduce the number of ministries, for example by considering options such as the following: ï‚· The Ministry of Culture could merge with the Ministry of Education and Science; ï‚· the Ministry of Justice with the Ministry of Interior; 33 Cutting staff by 2/3, while doubling salaries (to LVL 660 per month) would lead to a saving of LVL 3 million in the first year, even taking into account severance pay of LVL 2.9 million. In the second year, the savings would reach LVL 5 million. 43 ï‚· the Ministry of Finance with the Ministry of Economy; and ï‚· the Ministry of Environment could be merged with the Ministry of Regional Development. 105. There is greater scope to consolidate subordinated agencies that perform related or duplicative functions, but large-scale elimination of non-core functions may not be feasible in the short term, as it requires a clear strategic vision and substantial preparatory work. Agencies that perform related (but not overlapping) functions could be combined into larger agencies to maximize synergies and economize on resources, though their functions and expertise would be preserved. For example, agencies that manage EU funds (e.g. Regional Development Agency; Rural Development Agency) could be combined on this basis. The consolidation of various inspectorates could not only produce savings, but also result in improved coordination of inspections on businesses. At the same time, this process requires a thorough examination of the functions on a case-by-case basis to ensure that essential functions are not impaired. In the case of duplicative functions, the merger of the agencies concerned should result in the elimination of unnecessary duplication (e.g. various forensic services; vehicle registration), many of which have been identified by functional reviews conducted by the State Chancellery. Examples of subordinated agencies that could be consolidated include the following: ï‚· The Center for Curriculum Development and the Agency for Education Development under the Ministry of Education and Science; ï‚· The State Forest Service and the State Plant Protection Service under the Ministry of Agriculture; ï‚· The Road Traffic Safety Directorate and the Transport Accident Investigation Bureau under the Ministry of Transport; ï‚· The Public Health Agency and the Health Inspectorate under the Ministry of Health; ï‚· The State Forensic Science Bureau (under the Ministry of Interior) with the State Center for Forensic Medical Examination (under the Ministry of Health); and ï‚· The State Language Agency (under the Ministry of Culture) and the National Language Center (under the Ministry of Justice). 106. Other subordinated agencies or SOEs performing government functions could be absorbed by the supervising ministries, devolved to the municipal level, or - in case where their functions are deemed unnecessary - eliminated. For example, the Government might consider devolving most institutions currently subordinated to the Ministry of Culture (e.g. music schools, libraries, museums) to municipal authorities, which could then decide which to maintain and which to close; other agencies (e.g. the Cultural Heritage Agency, the National Film Center, State Language Agency) could also be absorbed by the Ministry. Likewise, the Ministry of Agriculture could streamline the number of the territorial divisions of the food and health service management and regional agricultural administrations. 107. In the short term, the cost savings from such mergers is likely to be modest, especially if many support functions have already been consolidated. Further savings would come primarily from the reduction in administrative departments and managerial posts as well as reduction in office space. So far 44 cost savings from non-salary administrative costs (e.g. rental of premises) have been modest following the mergers of administrative bodies in 2009, which indicates that there may be scope for additional savings. 108. In the medium term, potentially substantial permanent savings could be achieved with the elimination of unnecessary functions. Such reforms would require substantial preparatory work, which should ideally be guided by a clear concept of the appropriate role of the State in Latvia’s economy and society. This could provide a basis for determining which functions lie at the core of the Government’s mission and which ones can be considered of secondary importance or unnecessary. In the absence of adequate preparatory work, the abolition of functions (e.g. through deregulation, delegation, or outsourcing) such as inspections could lead to unintended consequences. Such concerns have been raised with regard to the recent delegation of construction inspections to the sub-national level, which may have resulted in diluted accountability and increased the risks to the public that state regulation is intended to mitigate. 109. There seems to be considerable scope for the Government to increase the efficiency of public expenditure on ICT without much additional investment. As a functional review by the State Chancellery indicates, expenditure on the establishment and maintenance of ICT hardware, databases, and proprietary IT systems (software) managed by different government bodies could be significantly reduced through a combination of centralization and outsourcing. In particular, the Chancellery’s report suggests that the efficiency of new ICT investments could be improved through multi-annual investment planning for the sector, supported by cost-benefit analysis and monitoring of performance indicators. Likewise, it finds that efficiency gains could be achieved by centralizing the development and maintenance of ICT hardware, databases, and software in a specialized government agency. However, in the short term, only simple systems such as e-mail could be easily centralized; establishing a new agency with adequate capacity and developing a whole-of-government ICT policy and technology standards would necessarily be a medium-term endeavor. According to the report, outsourcing could potentially reduce the personnel costs of systems maintenance by half. This option is worth considering also at the municipal level, particularly for smaller municipalities that could not efficiently maintain their own units staffed with ICT specialists. 110. In addition to the recommendations of the Chancellery’s report, in the short term, efficiency gains could be achieved through the integration of the Government’s databases (of which there are some 175 at present) to ensure that they exchange data in real time so that data errors can be minimized. This would not only reduce database maintenance costs, but would also likely produce savings from other public expenditures. For example the integration of tax and social assistance databases could produce savings by reducing benefits fraud. Options for data exchange among separate databases may be the most affordable solution in the short term, while the consolidation of databases used by government agencies with inter-connected functions could also be considered. Consolidating various government bodies’ data backup facilities into a national datacenter with business continuity and disaster recovery functions could produce savings at least in the medium (e.g. three-year) term, while enhancing the security of government systems. 45 Promoting efficiency gains at the sub-national level 111. Given that most of the cost savings have so far disproportionately affected the national government, there may be further scope for efficiency savings in public administration at the municipal level. The constitutional division of responsibilities and functions between national and sub-national authorities, as well as Latvia’s carefully negotiated revenue sharing arrangements between the national government and municipalities will shape exactly how these efficiencies could be achieved. The authorities might consider creating incentives for these gains by reducing earmarked transfers from the central budget (with the exception of teachers’ salaries which are now governed by the per-student financing formula) or by reducing municipalities’ share of tax revenues. 112. As a result of the reform of territorial administration that was introduced based on the Law ‘On Administrative Territories and Settlements’ of December 2008, following the local elections of June 2009, the number of local government units was reduced from 548 to 118 municipalities and republican cities. The reform and the consolidation that ensued seem to have coincided with the acceleration of staff cuts at the local government level. Municipal government staffing declined by 8.5 in the second half of 2009, compared to 5.2 percent in the first half of 2009 (compared to a reduction of only 0.06 percent in 2008). However, the consolidation of local government administration has not yet reached its full potential in terms of achieving efficiency gains. ‘General government services’ continue to represent a relatively high 18.3 percent of expenditures (compared to 18.4 percent in 2009), suggesting room for further savings in this area. According to State Chancellery data, by the end of 2009 municipalities employed about 87,000 staff, of whom 37 percent worked in education. In the absence of a personnel database that can provide information on staffing composition at the municipal level, a functional review of selected municipalities could help identify areas in which staffing could be reduced without a significant impact on service delivery. 113. The Government might wish to review the expediency of maintaining the five planning regions as an administrative structure after having abolished regional administrations (rajons) in 2009. The five planning regions, whose members are appointed by municipalities, are financed by the Ministry of Regional Development and Local Government to provide coordinating functions, particularly related to accessing and spending EU funds. However, these functions would seem to overlap with those of the Ministry as well as those of specialized agencies (e.g. Agency for Regional Development, and the Agency for Rural Development). Moreover, currently their financing is not very transparent, as it does not appear as a distinct item under the spending of the Ministry of Regional Development and Local Government. The alternative argument of maintaining these planning regions for the purpose of managing the delivery of some public services that cannot be efficiently provided at the municipal level (e.g. vocational education; administration of subsidies for inter-city public transport) does not seem compelling, either, given that such functions could either be provided through inter-municipal cooperation or be kept under sectoral ministries at the national level. 46 114. The national Government could seek ways of reducing inefficiencies and addressing current spending disparities, which may also be reflected in disparities in municipalities’ provision of public services. According to MoF data, administrative expenditure34 on average accounted for 8.7 percent of total municipal expenditure budgeted for 2010, but accounted for 25 percent or more for the five highest-spending municipalities. Municipalities’ administrative expenditure varied enormously - from LVL 192 per capita to LVL 10 per capita. There are significant divergences in municipalities’ per capita administrative expenditure even among municipalities of comparable size (Figure 2.28). A possible explanation might be that municipalities with higher levels of administrative expenditure provide higher quality public services. However, the national government does not seem to have consolidated data on public service delivery at the municipal level, which would allow for a meaningful comparison of municipalities’ performance. 115. In future gains could be made by requiring municipalities to report publicly on their performance on the basis of a standardized set of indicators related to key services. Such a system, which could be developed by the national Government in cooperation with the Association of Local Governments, should focus on few key indicators (e.g. school enrolment and educational achievement; percentage of low income residents receiving social assistance from the municipality) to avoid imposing high administrative costs on the municipalities. Apart from enabling the national Government to gain a fuller understanding of municipalities’ efficiency in the use of public resources, such a system would also enable municipalities to benchmark themselves with their peers and seek to improve their efficiency and performance in delivering services. Public disclosure of this information would also allow residents to evaluate their municipality’s performance in providing services and assess its efficiency in spending public funds. Figure 2.28: Administrative Expenditure Per Capita - Municipalities with 25,000-35,000 Population (2010 budget) 120 100 80 60 % of total expenditure LVL per inhabitant 40 20 0 0 5 10 15 Source: Ministry of Finance 34 This refers to expenditure on the municipal administration (personnel and other operating costs) and excludes expenditure on direct service delivery (e.g. teachers’ salaries). 47 116. The national Government could also promote good practices such as inter-municipal cooperation in service delivery, which can help municipalities achieve efficiency gains. Groups of municipalities may achieve efficiency gains by concluding cooperation agreements (contracts) to pool their resources and delegate the provision of services to a neighboring municipality. Such practices have already started to emerge in Latvia in response to the local government reform and the crisis, e.g. in the provision of primary and secondary education. Back-office functions could also be pooled or outsourced. The Government and the Association of Local Governments could disseminate and encourage such practices with a view to promoting further efficiency gains. Reviewing the use of subsidies for savings 117. The Government could review spending on subsidies to enterprises with a view to phasing them out. Subsidies to enterprises tend to be an inefficient form of public expenditure, especially when they displace public procurement processes or are intended to cover operational losses. When used selectively, subsidies can be a useful policy instrument to address demonstrable market failures and to support the provision of public goods (e.g. introduction of new, environmentally friendly technologies; or the provision of postal services in rural areas), which the private sector could not otherwise provide profitably. In the interests of social equity, governments may also wish to subsidize the access of the poor to some goods (e.g. medicines) and/or services (e.g. public transport), which are considered essential to ensure an adequate quality of life or income generation opportunities. 118. However, Latvia currently provides subsidies to enterprises that do not seem to fully correspond to these principles. In particular, the Government continues to substantially subsidize the cost of public transport to keep fares low for all passengers, irrespective of their ability to pay higher fares that would ensure cost recovery for transport providers. Such subsidies are not targeted to benefit the poor nor do they seem necessary to maintain the provision of services (with the possible exception of rural routes). They are therefore inefficient. 119. The Government could review the composition of spending on subsidies with a view to substantially reduce or eliminate subsidies intended to cover enterprises’ operational losses and maintaining consumer prices below cost recovery level. Such subsidies create perverse incentives for enterprises to provide unnecessary services, and tend to keep their operational costs high. The Government could further reduce the amount of the passenger transport subsidies, not by cutting services, but by liberalizing passenger fares for most passengers, and continuing to subsidize of low- income persons (and possibly other beneficiaries such as school children or the disabled). For example, targeting the subsidy to the 40 percent of the lowest-income persons would result in annual savings of LVL 33 million below the planned subsidy for 2010. A targeted subsidy could be administered by issuing beneficiaries with cards proving their eligibility to discounted fares. The beneficiaries of the activities included in the Government’s recently implemented Emergency Social Safety Net package are issued with identification cards confirming their low income and needy status. Such a reform would need to be piloted and supported by a public information campaign to ensure adequate outreach to eligible beneficiaries. 48 120. Data from the most recent HBS show that rates of public transport usage do not differ significantly across income groups (see Figure 2.29), which suggests that most transport users have the capacity to pay higher fares. At the same time, households’ expenditure on public transport as a share of household consumption does vary from 3.6 percent for the lowest-income quintile (Q1) to 0.7 percent for the highest-income quintile (Q5), with rural households spending more than urban households (Figure 2.31). This suggests that there may be considerable efficiency gains from targeting the subsidies to low-income groups and rural routes. Moreover, the survey shows that the share of households that reports using long-distance railways and buses (0.1 and 0.3 percent respectively) are negligible, which calls into question the social rationale for subsidizing these services. 121. However, it may be necessary for the Government to continue subsidizing some rural routes, where passenger numbers would not be sufficient to ensure cost recovery for carriers. These routes need to be identified based on an analysis of passenger flows and surveys to assess passengers’ ability to pay higher fares. Passenger transport subsidies are currently paid on the basis of long-term contracts between government authorities and carriers. These contracts oblige the Government to cover the operating losses of the carriers who undertake to service the specified routes and charge passengers fares set by the Government. However, if the Government were to liberalize fares (while protecting vulnerable passengers from the increased costs), carriers’ operating losses should decrease, as would the amount of the required subsidies. This assumes, of course, that liberalizing fares for the non-poor would not reduce passenger volumes sufficiently to reduce carriers’ profitability. If the Government were to consider liberalizing fares, it would be advisable to do so on a pilot basis first. Figure 2.29: Public Transport Usage Rates (Percentage Figure 2.30: Rates of Public Transport Usage by Region of Households) by Income Quintile 47.5 47.3 49 48.4 47.0 48 47.4 46.5 46.5 47 46.1 46 45.5 46.0 45.7 45.1 45 45.5 44 45.0 44.8 44.7 43 42.4 44.5 42 44.0 41 43.5 40 43.0 39 Q1 Q2 Q3 Q4 Q5 Riga Pieriga Vidzeme Kurzeme Zemgale Latgale 49 Figure 2.31: Expenditure on Public Transport as Figure 2.32: Public Transport Usage Rates by Income Percentage of Household Consumption by Quintile Quintile and Means of Transport 4.5 4.1 30 4.0 3.5 25 3.5 3.0 20 3.0 2.4 2.5 2.2 15 1.9 2.0 1.4 1.3 10 1.5 1.0 0.8 0.7 5 0.5 0 0.0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Local passenger transport by railways Tram Rural Urban Local bus, coach, trolleybus Source: Household Budget Survey 2008 122. Enterprises’ use of public funds should be closely monitored and subsidies that remain unspent at the end of the fiscal year could be recovered in the following year. External audit reports suggest that subsidies to some commercial entities subordinated to ministries have been under-executed and/or misappropriated. The Government might consider legal amendments to strengthen the supervising ministries’ responsibility for ensuring sound use of public funds by subordinated agencies, including SOEs receiving budget financing. Unspent or misappropriated funds should be returned to the budget and/or deducted from an entity’s financing for the following year. Reviewing the status and management of SOEs and other state property 123. The Government’s recent initiative to review the status of SOEs and define key policy principles to inform the Government’s exercise of its ownership function could result in reforms that would not only have a positive fiscal impact in the short term, but also improve the competitiveness of the economy in the longer term. A key task of the working group led by the Ministry of Economy which is charged with this review has been to compile an inventory of SOEs. So far the Government has exercised its function as an owner of SOEs in a decentralized manner, whereby sectoral ministries have supervised SOEs subordinated to them without reference to a common policy framework or set of guidelines. At the same time, no institution at the center of government has had a comprehensive picture of the Government’s overall ownership of SOEs, including their functions, financial performance, and the Government’s return on its equity. The current degree of decentralization contradicts OECD guidelines for the governance of SOEs, which recommend the centralization or at least strong coordination of the Government’s ownership function in one agency, which should be directly accountable to Parliament and report annually on SOEs’ aggregate performance. Moreover, decentralized exercise of the 50 ownership function by sectoral ministries poses the risk of blurring the Government’s ownership and regulatory functions, which could distort competition in the market in favor of SOEs.35 124. The ongoing review should allow the Government to differentiate among SOEs with different functions and identify options for reform, including privatization, change of legal status, or closure. In doing so, the working group will need to define some parameters of an overall policy of SOE ownership, most notably the appropriate purposes of the Government’s ownership of SOEs. Typically, governments tend to maintain (majority or minority) stakes in SOEs for reasons of national security (e.g. in the defense industry), the provision of public goods in sectors of market failure (e.g. healthcare), or in sectors that represent ‘natural’ monopolies (e.g. water sector utilities, and railways). Yet many of the 60 SOEs owned by the Government of Latvia would not easily fit in these categories. Privatization could be an appropriate option for those SOEs that are not ‘natural monopolies’, but operate in an actually or potentially competitive market environment (e.g. banking; electricity distribution). The privatization of these SOEs, including through the possible reduction of the state’s ownership in them, could be a significant source of additional revenues, which could be estimated on the basis of each SOEs’ financial performance and assets, including stakes in subsidiary companies. 125. On the other hand, the Government might consider integrating those SOEs that exercise government functions into the public administration. Examples of functions currently performed by SOEs include market regulation (e.g. the Standardization, Accreditation, and Metrology Center), administration of public funds (e.g. the Road Transport Administration) or public procurement (e.g. Latvia State Roads). A possible option for transforming the status of these SOEs could be to close them as commercial entities and re-establish them as government bodies –either as agencies or departments within a ministry. The argument that these functions require highly specialized skills that the public administration cannot attract due to the salary ceilings is not convincing. These skills are also required for the effective exercise of the related policy functions and oversight of these entities by the Government. Moreover, Latvia’s flexible remuneration policy, which differentiates among different professional groups and tracks the pay levels each group commands in the national labor market, should allow for sufficient room to attract and retain professionals with the requisite skills.36 Likewise, the status of not-for-profit organizations may be more appropriate for some SOEs that provide public goods but do not operate in a competitive market (e.g. libraries, museums). 126. It would be expedient for the Government to also review which SOEs receive subsidies from the budget and which ones (if any) have received public procurement contracts. Reviewing subsidies to SOEs could also assist the Government in clarifying the basis on which such subsidies may be provided. According to the OECD guidelines, the obligations of SOEs to provide public services -beyond what would be in their corporate interests- should be clearly defined by law or regulation and related costs should be covered by the state budget in a transparent manner. Such compensation may take the form of direct subsidies or service contracts. The OECD guidelines recommend contractual arrangements, 35 OECD (2005), Guidelines on the Corporate Governance of State Owned Enterprises, pp. 26-27 36 Some exceptions that might require special salary scales based on international comparisons (e.g. for air traffic controllers) should be very few and based on prior analysis proving a demonstrable scarcity of skills. 51 whereby the costs are structured in a transparent manner, particularly for SOEs that operate in competitive sectors of the economy (e.g. passenger transport) to avoid distortion in competition.37 However, potential cases of SOEs acting as vendors of goods, services, or civil works procured by the supervising line ministry should be closely examined, as they may pose a conflict of interest between a ministry’s public procurement and ownership functions and may reduce efficiency of public procurement as well as distorting competition. 127. In this context, the Government might also consider capping salaries and/or recurrent expenditures in SOEs that are financed from the budget or have monopoly status. While such a measure does not in itself reduce public expenditure (though it may increase dividends in the case of profit-making monopolies), it would strengthen accountability and transparency in public sector management. In times of an acute and painful fiscal adjustment, the high salaries paid by many SOEs that either benefit from public funds or are sheltered from market competition threaten to erode public support for the fiscal adjustment. 128. In addition to the possible privatization of some SOEs, the Government might also compile and scrutinize a register of publicly owned assets and consider selling real property that is not used for a public function. Privatizing assets held by the ‘State Real Estate’ agency and various ministries could also reduce associated maintenance costs. Priority could be given to assets that are no longer used for a public functions. However, as with SOEs, the sale of such assets would need to proceed only after their valuation and exploration of potential interest from investors to ensure that the assets are not sold below their value on account of depressed demand due to the ongoing crisis. V. Options for Reforming Public Sector Remuneration Policy 129. The Government would greatly benefit from a more coherent and fiscally sustainable system of remunerations, which should ideally be anchored in a clear human resource management policy aimed at attracting and motivating staff with the appropriate skills for the needs of the public administration. Latvia has committed to establishing a central body with a mandate to formulate and monitor the consistent application of human resource management policy across the public administration by the end of 2010. This commitment need not imply a full reversal of the current decentralized model of human resource management in Latvia’s public administration (e.g. recruitment may remain decentralized). However, there could be benefits from coordinated staff planning and coherent policies on career development and staff incentives, including from a fiscal planning perspective. The body responsible for human resource management policy would also have a key role in designing and managing the pay system reform and would therefore need to be established before the Government embarks on further reform of the pay system. 130. A fiscally sustainable remuneration policy would need to be supported by firm establishment and fiscal controls to avoid a new uncontrolled expansion of public employment and the wage bill. This 37 OECD (2005), Guidelines on the Corporate Governance of SOEs, p. 20 52 may require the Government to develop longer-term fiscal targets for the size of the wage bill and principles for containing its growth, once the crisis has passed. The national Government would benefit from strengthened establishment controls to avoid an uncontrolled expansion of public employment in the future. At the same time, legislation on remuneration could be further strengthened to improve coherence and predictability (for example, to avoid a return to the proliferation of bonuses once the current restrictions expire). Linking salary levels to labor market trends may be an appropriately flexible way of managing pay policy, in conjunction with rules designed to achieve fiscal sustainability (for example, ensuring that the wage bill does not rise faster than the GDP and inflation). It is worth noting that a well-developed human resource management policy, which combines monetary with other incentives (e.g. training and career development opportunities) and other benefits (e.g. higher job security) could significantly contribute to the public administration’s ability to attract and retain qualified staff. In such a context, the public sector may not necessarily need to match salary levels paid by the private sector. An average remuneration of 80 percent of salaries paid by the private sector for equivalent positions (differentiated by professional group to keep this ratio constant) tends to be considered adequate to ensure the competitiveness of public sector remuneration in most countries. 131. The Government might further reform remuneration policy with the aim of simplification, equity, transparency, and fiscal predictability. As noted in previous sections, Latvia currently has several public sector salary systems operating in parallel, of which three are based on the Law on Remuneration of Officials and Employees of State and Local Government Authorities of December 2009. Salaries of civil servants and other employees of ministries and other national government bodies directly subordinated to the Council of Ministers are established based on a system that combines a salary grid with salary bands. A similar system applies to agencies subordinated to ministries. The salaries of local government staff are not covered by the ‘salary grid’, but they are capped by law at 3.64 times the average monthly salary in the economy – the level at which the Law on Remuneration caps the salary of presidents of municipal councils. Teachers, military officers, border guards, police officers and firefighters, diplomats, and medical personnel employed by government bodies have their own pay systems. Harmonizing all these distinct pay systems may not be feasible even in the medium term. However, the Government might consider abolishing the current bands within the salary grid applicable to the civil service and unifying the grids applicable to ministries and subordinated agencies. There may also be scope for simplification by reducing the number of job families (monthly salary groups) on which the job classification is based. Many job families that require similar skills could be merged (e.g. information management in libraries, archives, and museums), resulting in a simpler, more transparent job classification. 132. After the central government salary system has been streamlined into a single salary grid, municipal staff and various public sector agencies could also be integrated in this grid. Integrating municipal staff in the salary grid would further the principle of ‘equal pay for work of equal value’ and could also contribute to staff mobility among municipalities and across levels of government. The fiscal implications of including municipal staff in the grid should be thoroughly assessed. 133. The Government could also consider permanently abolishing all performance bonuses other than the performance-related element that is built into the salary grid, which already provides adequate 53 rewards for high performers. In the current salary system applicable to central government bodies (salary grid), 23 percent of pay is determined based on the employee’s annual performance appraisal rating.38 This share of the performance-related pay element is very high compared to OECD countries, where performance-related increments or bonuses typically represent less than 10 percent of base salary.39 There is no clear rationale for increasing this further, or for supplementing it with other forms of financial incentives such as performance contracts.40 The current system also has the advantage of allowing the performance-related element of an employee’s pay to fluctuate from year to year – as opposed to typical salary grid systems, which allow well-performing employees to advance to a higher salary step that becomes part of future year’s base pay. At the same time, the Government may consider reducing or eliminating the length of service increment and allowing the base salary (i.e., the element determined by the position’s classification) to increase as a share of the total pay. The Government may wish to limit the share of staff in each institution who may receive the top two performance ratings, not least to contain the fiscal impact of performance-related pay. Finally, there does not appear to be a need for salary decompression, as the current decompression ratio of 1:9.5 (based on current ceilings within the salary bands) is among the highest internationally, even considering that the Latvian salary system includes non-professional grade staff. 134. In Latvia the performance appraisal system appears to be implemented somewhat inconsistently across civil service organizations. Data on salary levels in different ministries suggest that most ministries do not use all the six horizontal salary steps (‘qualification grades’) in the salary grid. Instead, some ministries use only one salary step in some vertical grades (‘monthly salary groups’); others use only some of the available steps within each grade. Curiously, there are even a few cases where lower salary steps (i.e., those for employees with less experience and/or lower performance ratings) actually seem to correspond to higher pay levels than the higher qualifications grades within the same ministry.41 The monitoring of human resource policies by a central body should help reduce these type of inconsistencies in the application of the grading and performance appraisal systems. 135. Streamlining allowances would also help increase the transparency of the remuneration system and facilitate wage bill management. Various categories of public employees are legally entitled to a wide range of allowances. Most of these allowances (e.g. moving allowance, holiday allowance) can be eliminated or, where appropriate, integrated into base pay. The Government may also wish to limit 38 Base salary, which is determined based on the grading of the post in the job classification system (‘monthly salary group’ or vertical grade), accounts for 67 percent of total pay. The remaining 27 percent of pay is determined by an employee’s ‘qualifications grade’ (horizontal salary step), which is calculated based on the length of service (10 percent of total pay) and performance rating (23 percent). 39 OECD (2005), Performance-related Pay Policies for Government Employees. 40 At present the Law on Remuneration allows for annual bonuses of up to 120 percent of monthly salary. 41 For example, in grade (‘monthly salary group’) 13 of the Ministry of Education and Science salary step 4 has an average salary of LVL 1,328 compared to LVL 1,142 in step 5 in the same ministry. Likewise, in the Ministry of Finance grade (‘monthly salary group’) 12, the average salary for step 4 (LVL 907) is higher than that for salary step 5 (LVL 890). Such inconsistencies might result in cases where employees with lower performance ratings are concentrated in the upper salary steps by virtue of their length of service. For example, all staff with at least 11 years of service and a performance rating of B are placed in the salary step 6 (the highest). 54 overtime pay as a share of base salary and abolish eligibility for overtime pay for professional-grade staff, restricting compensation to additional leave. VI. Other Medium-term Options to Increase Efficiency in the Public Sector 136. The Government could proceed more quickly with its plans to eliminate unnecessary functions, especially with the aim of reducing the regulatory burden on businesses. In this regard, the Government may consider consolidating inspectorates into a single state inspection body with specialized divisions (e.g. a reform that has been implemented in Croatia and Slovenia) and/or delegating inspection functions to certified professionals in the private sector (third-party inspections), e.g. in the construction sector. These changes could reduce the regulatory burden on businesses as well as the cost of inspections to the state budget. Plans to streamline licensing requirements would have similar benefits. 137. Plans to streamline public service delivery should also result in efficiency gains in the medium term. Such plans include the introduction of one-stop shops to deliver services to citizens and businesses. These reforms would involve simplification of procedures, further automation of business processes, and the integration of databases and service delivery mechanisms among multiple agencies. As such, many of these measures could involve an initial investment, but one which would result in reduced administrative costs in the longer term. 138. The Government might consider developing a system of key indicators of organizational performance, supported by a monitoring mechanism at the center of government, with a view to informing future decisions on resource allocations. Ideally such a performance monitoring system should be based on a national policy framework (strategy) that would define the Government’s objectives and associated indicators in key policy areas. The national policy framework should also form the basis of expenditure prioritization. The recent reforms aimed at reducing personnel and other administrative costs have been disadvantaged by the absence of a national performance management system, which could inform expenditure prioritization and the selection among alternative cost-cutting measures. Such a system would have provided the Government with essential information about the services that each agency provides and enabled policy makers to estimate more accurately the impact of cuts in expenditure on service delivery. If extended to the municipal level, such a system could facilitate efficiency gains and performance improvements by making it easier to identify high performing municipalities and help others to emulate their practices. The possible introduction of a performance management system could be considered in the context of broader reform of the budgeting system, which may include other elements such as reviving medium term and program-based budgeting as well as moving toward performance-informed budgeting. Such reforms require a medium-term effort, as they need to be introduced gradually to allow the public administration to develop the necessary implementation capacity and also to allow for reform evaluation and adaptation so that a country can find the most appropriate arrangements for its own circumstances. A recent technical assistance note by the IMF provides more details on budget reform options. 55 139. Finally, the Government might consider developing a coherent policy that defines the objectives for the Government’s ownership of SOEs as well as specific guidelines for the exercise of this function to ensure high standards of corporate governance in SOEs. The centralization of the Government’s ownership function in a single agency with high standards of transparency and accountability could be an option for achieving coherence in the Government’s exercise of its ownership function and its separation from regulatory and public procurement functions. It will be particularly important to ensure that the Government regularly monitors the performance of SOEs with reference to appropriate indicators that should correspond to a SOEs’ key functions and related policy objectives. Likewise, good international practices identified by the OECD suggest that the Government should closely monitor the material risks (including investments in subsidiaries, public-private partnerships, etc) undertaken by SOEs, while refraining from interference in SOEs’ day-to-day management. The Government might also consider imposing legal limits on the scope of SOEs’ activities to prevent them from undue diversification into sectors unrelated to their core functions and reduce the risk of misuse of public funds or excessively ambitious expansion.42 140. In considering a policy on SOEs, the Government might wish to consider the longer-term benefits of limiting the scope of state ownership. In Latvia the private sector accounted for an estimated 70 percent of GDP in 2009, which is lower than most comparator countries. Privatizing or closing many SOEs that do not serve a clear public policy objective and breaking up current monopolies (e.g. in electricity distribution) would not only increase budget revenues in the short term, but – more importantly – would promote competition in Latvia’s market. In turn increased competition could result in higher enterprise productivity and lower prices for consumers, including for the Government as a purchaser of goods and services through public procurement. Figure 2.33: Private Sector as a Share of GDP 82 80 78 76 74 72 70 68 66 64 Source: EBRD Transition Report 2009 42 OECD (2005), Guidelines on the Corporate Governance of SOEs, p. 19 56 Table 2.1: Options for immediate term savings Action Estimated annual budget savings (from 2009 Longer-term implications expenditure level) Public administration – organizational structure and staffing Mergers of ministries and consolidation ï‚· LVL 3.5 – 5 million – from staff cuts depending Elimination/ reduction in scope of of subordinated agencies; elimination/ on the number of agencies to be consolidated certain functions could result in reduction of scope of some functions and the degree to which support services may additional savings that need to be 43 (e.g. inspections) have already been consolidated identified based on the expenditure of the agencies concerned ï‚· LVL 6 million in savings at the central government level from more efficient use of premises (including rental of state-owned Ensuring that mergers and vacant premises), e.g. consolidate police consolidation of regional branches of stations, tax offices, and other regional central institutions also result in more branches of central institutions efficient use of premises ï‚· LVL 1.5 million by abolishing the five planning Transfer most cultural facilities to regions municipalities Further staff reductions, targeting ï‚· LVL 2-4 million (approximately 2 million, if selected services police, prison administration and court personnel are cut by 5 percent) ï‚· A 5 percent reduction in local government staffing should save some LVL 30 million; however such savings may be more effectively achieved by reducing fiscal transfers to municipalities (see below) Phasing out salary bands and ensuring ï‚· LVL 1.5 million – through a reduction of salary The gradual harmonization of salaries consistent application of the salary grid levels in agencies where pay levels are above may be designed in a manner that is across public administration bodies. average (e.g. Ministry of Health, Ministry of fiscally neutral (without salary Regional Development, Ministry of Welfare) protection for those whose salaries and ensuring that average salary levels in will be above the level for their grade subordinated agencies do not exceed those of and step) or may have an additional ministries fiscal cost that will need to be carefully estimated (if the ï‚· The consistent application of the salary grid Government decides to combine the (use of salary steps or ‘qualifications grades’) abolition of salary bands with salary could potentially save up to LVL 0.5 million growth). ï‚· The launch of a HRMIS linked to payroll will The additional cost may be incur a start-up cost (up to LVL 0.5-1 million) compensated by the abolition of bonuses and most allowances. The measure will contribute to improve wage bill management, increased transparency, and a more equitable pay policy. Subsidies to firms (both private and state-owned) Reduction/ elimination of central LVL 28 million (local roads) Municipalities have a high share of earmarked grants to municipalities tax revenue that does not warrant 43 The Bank mission has not received expenditure data broken down by subordinated institutions to more accurately estimate the potential amount from the abolition or merger of such agencies; approximate estimates are based on salary level and employment data from the State Chancellery. In principle, mergers of institutions (e.g. consolidation of inspectorates in Croatia) may result in savings of up to 20 percent of operating costs. 57 (other than education grant) earmarked grants for basic local government functions without re- distributional impact (e.g. maintenance of local roads) 44 Reduction/ elimination of central up to LVL 33 million Subsidized travel should be limited to government subsidies for passenger low-income persons entitled to social transport; maintaining subsidy only for assistance and essential rural routes. selected rural routes and for the poorest 40 percent of the population Reduction in subsidies for SOEs in the up to 5-10 million areas of culture, broadcasting, and sport Privatization/ closure of SOEs that ï‚· Subsidies to these SOEs Productivity gains in the economy as distort competition and shift to ï‚· Savings from shift to competitive procurement well as cost savings for the budget competitive public procurement (e.g. (international competitive bidding for civil will result from increased for road maintenance) works) require an in-depth analysis of costs but competition in Latvia’s domestic are likely to be very substantial market. Temporarily extending requirement to Data on subsidiaries and their profits were not This measure should be reviewed pay 80 of profits as dividends to the available to the authors after the achievement of fiscal budget to SOEs’ subsidiaries adjustment targets to ensure adequate investment levels. Privatization of selected assets and Estimates of revenue potential require detailed Reducing the share of the public SOEs inventory and valuation of assets sector in the economy should increase competition and productivity Strengthen public financial up to LVL 10 million from closer oversight of Strengthen legal framework on management practices, including: subsidies and recovery of unspent/ misappropriated financial responsibility. ï‚· Closer oversight of subsidies; funds ï‚· Recovery of unspent or misappropriated funds (especially subsidies) ï‚· Measures to improve value-for- money in procurement (e.g. expand use of consolidated procurement) 44 This includes LVL 26.5 million earmarked passenger transport grant to municipalities. Data from the most recent household survey indicate that public transport users are fairly evenly spread among income groups, which a slightly higher rate of usage among rural residents. This estimate is based on the assumption that the current level of user subsidization (approximately 50% of the ticket price for bus services; 60% for railways) would be maintained only for the two poorest quintiles of the population (i.e., 40% of transport users), leading to a 60% reduction in subsidy. If the current level of subsidy for rural routes (LVL 13 million in 2010) were maintained, the estimated savings fall to LVL 25 million. 58 Matrix 1. Options for the Government to Consider in Public Administration Reform, Subsidies and SOEs Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Public Sector Remuneration 45 Further amendment of a new 1.5 1.5 - 2 As of January 2010, the The remuneration grid It may be difficult to only Increased consistency (e.g. remuneration system The consistent reduced remuneration remains relatively complex cut the salaries of those elimination of salary bands (elimination of salary bands & application of levels have been and contains many ministries/ agencies that and transition to a pure reduction in the number of public the salary grid reflected in a new inconsistencies that reduce pay higher than the salary grid system) and services job families) (use of salary remuneration grid, predictability in managing national government transparency (e.g. Short-term savings can be steps or which attempts to the wage bill, as well as average (currently LVL reduction in the number of achieved through a reduction ‘qualifications bring more coherence undermine the principle of 756 per month), and to allowances; increasing of the maximum salary levels grades’) could across civil service and ‘equal pay for work of resist pressures to base salary as a share of to bring them closer to the potentially save the public sector more equal value’ (especially increase the salaries in total remuneration from current average salaries paid up to LVL 0.5 broadly. salary bands and remaining those ministries and 67 percent to e.g. 80 by ministries and agencies. This million. differences in average pay agencies that pay less percent). would force agencies that pay levels in different than the average. In the salaries on the higher end of ministries and agencies). medium term, equalizing the bands (e.g. Ministry of salary levels across Health, Ministry of Regional ministries and agencies Development, Ministry of could lead to an increase Welfare) to reduce their wage in the overall wage bill. bill. Establishment of a centralized The launch of a MoU with the This will ensure The consolidation of the personnel and payroll system HRMIS linked to European Commission consistency in human HRM function would also (Human Resource Management payroll will incur requires the resource management and result in further savings in Information System, HRMIS) and a start-up cost Government to develop remuneration policies. The administrative costs introduction of central controls for(up to LVL 0.5-1 a strategy for unifying current situation lowers (through reduction of HRM the establishment of new posts million) the human resource the likelihood that staff). management (HRM) ministries, agencies, and function across public local governments administration accurately report payroll institutions. and personnel data (but continuing to hire staff not included in the payroll 45 This amount is calculated by assuming that average salary levels in ministries and agencies that currently pay above the average salary (LVL 756 per month) is reduced to this average. However, a more precise calculation could be made by estimating the number of staff whose pay levels are above the average within each monthly salary group and qualification grade. This information (number of staff within each monthly salary group and qualifications grad in each ministry/ agency) is not currently available to the World Bank team. 59 Matrix 1. Options for the Government to Consider in Public Administration Reform, Subsidies and SOEs Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit database). The introduction of a mandatory personnel ID number should increase the accuracy of payroll and personnel data.. 46 Further staff reductions, 4 5 In the period 2010-2011 Overall, public sector There is a risk that quality Optimization of the scope, particularly in agencies at the central at the central a further reduction in employment remains of public services may network and productivity subordinated to ministries and level level salary levels does not higher than some suffer if staffing levels are of public administration local government bodies (e.g. 30 through a 5 37.5 seem feasible or comparator countries and cut further. A review of institutions. police, revenue percent advisable, as it would staffing levels remain functions and appropriate administration) reduction in carry a very high risk of higher than 4-5 years ago. staffing levels within each municipal qualified staff leaving The wage bill as a share of agency would seem 47 the civil service and/or staffing GPD is also on the high side appropriate. Otherwise, losing motivation. of a range of comparator short-term savings may countries, especially if one have to be reversed later considers that healthcare to restore an acceptable staff are not included in quality of services. Latvia’s public sector wage The tax administration bill (which they are in most seems clearly overstaffed, countries). This suggests but salary levels also that there may be some appear too low. It is room for adjustment to therefore recommended pre-2007 staffing levels to consider cutting staff without a negative impact while increasing 48 on policy functions and salaries. A transition to service delivery. Specific a leaner tax 46 An estimated LVL 2 million would be saved in 2011, if staffing levels in the police, prison administration, and court personnel were cut by 5%. These savings are estimated by assuming that the redundant staff are currently earning the average salary paid by their service and will receive severance pay equal to 3 months’ salary. These savings would increase to LVL 2.5 million in 2012, as severance pay would not have to be paid. The higher figure can be reached by sharper staff cuts in these services (up to 10%) and/or by staff cuts in additional agencies. 47 Again, this amount is based on a 5% reduction of the municipal wage bill achieved through redundancies. The savings are expected to increase in 2012, as severance pay to employees laid off in 2011 would not have to be paid. 48 For illustration purposes, cutting staff by 2/3, while doubling salaries (to LVL 660 per month) would lead to a saving of LVL 3 million in the first year of the reform, even taking into account severance pay of LVL 2.9 million. In the second year of the reform, the savings would reach LVL 5 million. However, the reform of the revenue administration should be carefully designed to ensure that staff reductions are accompanied by efficiency gains and do not result in reduced compliance rates and revenue collection. 60 Matrix 1. Options for the Government to Consider in Public Administration Reform, Subsidies and SOEs Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit services (e.g. police, prison administration however administration) seem would be a medium-term overstaffed. reform. Consolidation of the Public Administration Further 3.5 – 5 Latvia continues to Optimizing the scope of -- Improved public elimination/consolidation of have what appears to service to the size of the perception of the public functions, through mergers of from staff cuts be too many agencies, economy and available administration and ministries and agencies. depending on which in many fiscal resources. increased quality of public the number of countries either do not The abolition of SOEs, goods and services. agencies to be exist or are which compete with consolidated consolidated. Some market operators may also and the degree entities subordinated be expected to increase to which to ministries could be competition in the support services amalgamated (e.g. domestic market, thus may have various inspectorates) improving the investment already been climate and resulting in 49 consolidated potential further savings for the budget (through increased competitiveness in public procurement). Reducing non-salary 6 Mergers of institutions/ -- Greater efficiency administrative costs (including at central reduction in staffing by raising revenue from the government should result in savings 50 rental of vacant state-owned level in rental of premises; premises), e.g. consolidate state-owned premises police stations, tax offices, and could be offered for other regional branches of lease central institutions Introduce salary caps for Savings would SOEs, including those in Control of administrative -- Greater efficiency monopolies and SOEs that be achieved receipt of subsidies, costs and performance 49 The Bank mission has not received expenditure data broken down by subordinated institutions to more accurately estimate the potential amount from the abolition or merger of such agencies; approximate estimates are based on salary level and employment data from the State Chancellery. In principle, mergers of institutions (e.g. consolidation of inspectorates in Croatia) may result in savings of up to 20 percent of operating costs, however lower savings are estimated here in view of efforts already undertaken to cut costs in 2009. 50 This is based on a 5% reduction in rent, lease, and non-capital maintenance costs (total cost in 2009: LVL 122 million for central government), which should be feasible through a more efficient use of premises. 61 Matrix 1. Options for the Government to Consider in Public Administration Reform, Subsidies and SOEs Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit receive subsidies from the through a have cut staffing and targets should be budget. reduction in salaries less sharply developed for all subsidies (to be than in the rest of the monopolies and firms estimated, and public sector. receiving subsidies. can be complemented with an increase in dividends) Support to Enterprises Phase out direct subsidies to Currently, subsidies are Contracts should be closely Impact on the poor of a Efficiency gains by allowing loss-making enterprises at not targeted to monitored to ensure cost rise in transportation prices to rise to cost both the central and supporting access to control. costs (this risk can be recovery levels, and municipal levels. services for low-income User fees (e.g. passenger minimized by developing targeting subsidies to low- citizens, but come in transport prices) could be a compensation strategy income users of the 51 Reduction of central budget 25- 33 the form of allowed to increase. This prior to cutting subsidies). services concerned (e.g. subsidies for passenger ‘compensation for would allow subsidies to be Allowing prices for railway and bus transport). transport by targeting the losses’ to providers, targeted to protect low- passenger transport to Increased budget revenue subsidy to low-income without adequate cost- income beneficiaries, increase may reduce and equal treatment by persons or to rural routes only control mechanisms. As instead of protecting firms passenger numbers, nature of the institution. (a compensation strategy can a result, subsidized from losses. which may reduce be developed before enterprises have an Services that are transport operators’ elimination of subsidies, along incentive to provide considered essential from a overall revenues and thus the lines of current ESSNS for unnecessary services, social perspective (e.g. reduce savings. disabled passengers). to understate transport routes to rural Transition to market revenues, and to inflate communities) but would (unsubsidized) ticket 52 Reduction in subsidies for up to10 operational costs. not be commercially viable, prices could be piloted in SOEs in the areas of culture, even if providers were able a few routes to measure 51 Data from the most recent household survey indicate that public transport users are fairly evenly spread among income groups, which a slightly higher rate of usage among rural residents. The higher figure is based on the assumption that the current level of user subsidization (approximately 50% of the ticket price for bus services; 60% for railways) would be maintained only for the two poorest quintiles of the population (i.e., 40% of transport users), leading to a 60% reduction in subsidy. The lower figure is based on maintaining the current level of subsidy for rural routes (LVL 13 million in 2010). 52 The estimate is based on a one third reduction from 2009 subsidy levels. There is no dispute that the State has a role to play in supporting culture, broadcasting and sport. The policy question that needs to be considered is whether the current level of subsidies to these activities is affordable, given the severity of fiscal constraints and what Latvia wants to achieve. This is only something the Government can decide. Based on Treasury data on expenditure by functional classification, it appears that subsidies to "recreation, culture, and religion" were LVL 9.8 million in 2008 and LVL 8.3 million in 2009. 62 Matrix 1. Options for the Government to Consider in Public Administration Reform, Subsidies and SOEs Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit broadcasting, and sport to charge market prices, the impact on passenger could be specifically flows and operators’ identified and prioritized revenues. based on cost-benefit Another risk concerns the analysis. identification of low- income transport users who could continue to benefit from the subsidy. Access to subsidized (discounted) transportation could be granted to persons who are eligible for other types of social assistance (e.g. free medical care; housing benefit; GMI) who are already issued with a document proving their status. The impact of removing subsidies for the non- poor in rural routes could be assessed by a survey of public transport users in a sample of these routes. Tightening public financial Savings depend Concerns about the Ensure recovery of -- Improved transparency management (PFM) controls on level of management of misappropriated or and efficiency of PFM, in SOEs and agencies under- subsidies to SOEs under unspent funds, including including improved budget executed/ Ministries of Education, subsidies (e.g. LOC). execution. misappropriated Culture, and Transport funds in 2010. revealed by the State Audit Office (SAO) in 2008. Revenue Raising from SOEs (and other state assets) Extend to SOE subsidiaries the -- -- A positive step was It would be expedient to -- Increase in revenue to the 63 Matrix 1. Options for the Government to Consider in Public Administration Reform, Subsidies and SOEs Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit requirement to increase the taken with legislation tighten the legislation to state budget. Consistency minimum dividend paid by adopted in 2009 that ensure that these SOEs’ in rules applying to SOEs profit-making SOEs to the increased the minimum subsidiaries are also and SOE subsidiaries. state budget to 80 percent. dividend paid by profit- included in the measure. making SOEs to the state budget to 80 percent. Extending this requirement to SOE subsidiaries would increase revenue to the state budget. Privatize SOEs that are not Savings depend -- The private sector Examine whether some Valuation of assets could Competition in Latvia’s providing essential public on valuation of accounts for 70 percent SOEs could generate more be affected by current internal market would be services. assets to be of GPD; this level could budget revenue by being contraction, continued strengthened. Savings to privatized be raised to 75 percent privatized. recession and weak the budget could also Prepare an inventory of real (e.g. as in Lithuania, recovery in the world accrue through increased property owned by the state Poland) or even 80 economy. competition in public and identify assets suitable for percent (e.g. as in procurement. privatization (those not used Estonia, Hungary). for public functions). A Government Working Group has been formed to review the expediency of state ownership of SOEs on a case-by-case basis. 64 Chapter 3: Municipal Spending and Intergovernmental Finance 53 I. Introduction - Latvia: Intergovernmental System 141. Latvia is a unitary state with a single-level local government system. Municipal governments are key providers of social services in Latvia, being responsible for delivering about 70 percent of the overall education budget and for substantial supplementary social assistance programs and housing benefits, particularly for the poor (see Figure 3.1). While health does not feature as a large expenditure under local government budgets as health services are largely funded through regional health funds, local governments have an important role to play in health care management as employers and operators of medical facilities. Figure 3.1: Latvia: Structure of government expenditures by function and government level, 2008 Social protection Education Recreation, culture, and religion Health Housing and community amenities Environmental protection Economic affairs Public order and safety Defense General public services 0% 20% 40% 60% 80% 100% Central Local Source: BOOST Treasury Database, World Bank. Note: The classification follows the United Nations Statistics Division’s Classification of the Functions of Government (COFOG) of government expenditure. 142. The analysis builds on the World Bank (2003) report Latvia: Beyond Territorial Reform, which examined the intergovernmental system and carried out a substantial assessment of local government sectoral spending. The 2003 report outlined the expected benefits that would ensue from the amalgamation of municipalities, anticipating the recent 2009 implementation of territorial administrative reform (which was somewhat delayed since its 1998 inception). This chapter updates this analysis by asking whether the recent reduction in local government units has been sufficient, and if so, whether there are other outstanding issues to be tackled in terms of local government organization. Many of the 2003 report’s recommendations remain relevant. Two of the report’s main suggestions are reiterated: 53 This section of the PER draws extensively on a background paper written in 2010 by the Latvia local government specialist, Inga Vilka. 65 ï‚· To provide sufficient financing for social assistance benefits, it may be necessary to provide additional financing in areas where poverty levels are unusually high. ï‚· After amalgamation, the newly consolidated units of local government—the novads—may be too small to take on certain functions, such as organizing public transport services, managing funding for roads, and operating retirement homes and orphanages. The government may consider assigning these functions to the recently-created five planning regions might be an appropriate basis for such a level of government. 143. The chapter looks at five aspects of the intergovernmental system in Latvia: the design of the intergovernmental system (section II); the structure of local government expenditures (section III) and revenues (section IV); the local government revenue equalization mechanism (section V); and macroeconomic control and expenditure effectiveness at the local government level (section VI). The recent implementation of the local government administrative reform is a critical departure point for the analysis. The reform is too recent for its impact to be fully discerned. The global financial crisis that hit Latvia in 2008, reversing a decade of impressive growth and exerting severe downward pressure on government budgets, additionally informs the direction of the analysis. The paper concludes with a number of recommendations for discussion in section VII. These are certainly not intended as prescriptive, but simply as open topics for debate regarding the future development of the intergovernmental system in Latvia. II. Design of the Intergovernmental System 144. The division of local government in Latvia was recently reformed. The local government reorganization originated in 1998 with the passing of the Administrative Territorial Reform Law. The Law set out a consolidation process for local governments and a change to the proposed territorial division. A transition period until 2005 during which the amalgamation of local government units would be implemented was initially envisaged in the Law. But implementation was delayed and the reform finally was put in place following local elections in June 2009. 145. Prior to June 2009, sub-national governments in Latvia were organized along two levels. The first level (closest to citizens) consisted of republican cities, towns, and municipalities. The second level (county level) was formed by 26 districts and republican cities. Seven republican cities were included under both local government levels as they provided both first-level local government and district government functions. First level local governments were not subordinated to district governments. Following the implementation of local government reforms in 2009, there is a single level sub-national government system in Latvia consisting of republican cities and municipalities. The district level of government has been abolished. 146. Before the reform, the total number of local governments was 548, including 26 districts (rajons), 7 republican cities (republikas pilsÄ“ta), 50 towns (pilsÄ“ta), 424 rural municipalities (pagasts) and 41 reformed municipalities (novads). The average population at the first-level of local government before the reform was 4,260 thousand, but the number of residents in the smallest municipality (Kalncempju pagasts) was 251. The population stood at fewer than one thousand in more than one third 66 of local administrative units. After the reform, the number of local government units declined to 118 comprising 9 republican cities and 109 municipalities (novads) (Figure 3.2).54 147. Local government councils are elected for four years by direct universal suffrage, under a proportional representation list system. The number of councilors in municipalities (novads) ranges from 13 to 19 depending on population size. Eight of the republican cities have 13 to 15 local government elected representatives, while Riga City has 60 councilors. Local government councils form at least two standing committees: a finance committee and a committee of education, social and culture affairs. The chair of the council is elected from among the council members. Formally, the council chair leads policy- making at the local government level. Figure 3.2: Administrative territorial division in Latvia after implementation of the Administrative Territorial Reform in June 2009 Source: Vilka 2010, and Ministry of Regional Development and Local Government. 148. The principles of local government in Latvia are based on the European Charter of Local Governments and the local government legal framework is set out in the Local Government Law passed in May 1994. Municipalities and republican cities have an extensive set of responsibilities under the 1994 local government law, which specifies that local governments are responsible for the social, economic, cultural and educational needs of the population. The law stipulates 22 autonomous functions for local governments. Added to this are further obligations and a range of temporary and 54 The new administrative territorial division (118 local governments) was approved by the “Administrative Territories and Settlements Lawâ€? (December 18, 2008). 67 voluntary tasks laid out in other laws and regulations. Thus, local governments are tasked with a wide range of functions from the provision of basic infrastructure, housing and transport services to ensuring social assistance is provided to its needy population to provision of preschool, primary and general secondary education. This array of functions for local government is broadly in line with that of other recent European Union member states (as noted by Dillinger, 2007). A list of the specific local government functions identified in the local government legislation is given in Annex 1. Municipal Size 149. Despite the reform, there are significant disparities in population size between local governments, ranging from the capital city Riga with almost one third of the country’s population (709,145) to the municipality of Baltinavas with 1,365 residents. Thirty-seven municipalities have a population of less than five thousand residents (Figure 3.3). Figure 3.3: Latvia: Population distribution of municipalities 45000 45 40 37 39 40000 35 35000 30 Number 25 30000 Population 20 25000 15 11 9 10 7 5 20000 4 3 2 5 1 15000 0 10000 5000 0 1 9 17 25 33 41 49 57 65 73 81 89 97 105 Population size Number of Municipalities Source: World Bank staff calculations based on data from the Latvian authorities. 150. There is some evidence that further municipal mergers may be warranted in order to create municipalities of a sustainable size. Further mergers may lead to budget savings with regard to administrative functions. Preliminary evidence points to possible inefficiencies due to the small scale of many municipalities. There is a tendency for per capita administrative costs to decline the higher the population size serviced by local government in Latvia (Figure 3.4). Out of the 13 municipalities with an administrative budget share of total spending of over 20 percent (compared to an average of 12 percent), 10 have a population of less than 10,000.55 It should be noted that international evidence is not conclusive as to the optimal size of local governments to achieve economies of scale (Brynes 2002). Some studies have suggested that there is a U-shaped relationship between population size and 55 The data on administrative expenditures is based on budgeted amounts for 2010. 68 administrative costs and others have found that amalgamation do not lead to a reduction in administrative costs. 151. Beyond the weight of administrative expenses, there may be strong arguments for amalgamation related to increasing the quality of municipal service provision. Administrative capacity is likely to be constrained for local governments that cover very low populations. This was the main motivation for the recent reform of municipalities in Denmark. Denmark restructured its municipalities in 2005 by merging 270 municipalities into 98 units (enforced in 2007). Now there are few municipalities with less than 20,000 in population. The objective of this reform was chiefly to ensure municipalities have adequate administrative capacity for service delivery. Using the Danish population threshold, even following the reform Latvia still has 94 municipalities out of the total 109 with under 20,000 in population. Figure 3.4: Latvia: Local government administrative expenditures decrease as municipal population size increases 120 Muncipal population ('000s) 100 80 60 40 y = -0.1389x + 21.896 20 0 0 50 100 150 200 Per capita administrative expenses (Ls.) Source: World Bank staff estimates based on data from the Latvian authorities. Note: For exposition purposes Riga is excluded. 152. Even if the current number of municipalities (novads) is to be maintained, semi-amalgamation models could be considered. Novads may be too small for some of their functions, i.e., organizing public transport services, managing roads funding, operating secondary schools with wide curricula, operating orphanages and homes for the disabled or retirees. A semi-amalgamation model whereby small municipalities are combined into large units for service delivery would be an option for taking on these functions (perhaps using the five planning regions). The verbandsgemeinde in some German Lander are an example of such a system. The planning regions in the Latvian intergovernmental system are one obvious avenue for facilitating municipal cooperation in service delivery. They currently have such a function in the realm of coordinating public transport delivery across municipalities.56 56 Prior to the recent amalgamation, local governments in Latvia were involved in inter-municipal agreements for providing services to increase economies of scale. 69 153. The large variance in municipal administrative costs warrants further study. For 109 municipalities in Latvia, the share of total spending going on administration ranges from 2 to 30 percent. While administrative costs do increase with the geographical area covered by the municipality, there is no clear relationship with the economic development or socioeconomic needs of the municipality. Two very similar municipalities in terms of socioeconomic characteristics—and even population and area size—can have widely varying administrative budget burdens. Additional analysis on the factors that impact on the costs of service provision at the municipal level would be useful. It may also be necessary to look at whether administrative costs are recorded consistently across municipalities. Planning Regions 154. Apart from local governments there are five planning regions in Latvia. The main function of the planning regions is to ensure regional development planning, coordination and cooperation of local governments and state institutions. Since 2003, government regulations define five planning regions as set out in the Regional Development Law. These are Kurzeme, Latgale, Riga, Vidzeme, and Zemgale (Figure 3.5). 155. Planning regions were developed in Latvia from 1996, but their mandate has been constrained by a lack of clarity on their legal basis. Until 2006, the legal status of planning regions was unclear. For example, some of them were classed as non-governmental organizations. The situation has been somewhat clarified by the amendments to the Regional Development Law passed in 2006, which attached to the planning region the legal status of a derived public person. According the Law the decision making institution of the planning region is the Planning Region Development Council. The members of the Planning Region Development Council are elected from among the councilors of the relevant local governments by the general meeting of the chairpersons of local governments (located in the planning region). The budgets of planning regions are included in the state budget as subprograms under the Ministry of Regional Development and Local Governments. The administrative staff of planning regions are state employees. 70 Figure 3.5: Latvia’s five planning regions 156. Yet, the responsibilities of planning regions are very narrow. Since 2008, planning regions have to prepare resolutions about the conformity of local government territorial development plans with legal acts and other binding territorial plans. This is a task that used to be under the remit of the Ministry of Regional Development and Local Governments. Recently in 2010, planning regions also became involved in the organization of public transportation service for municipalities (but not for republican cities). This followed from the elimination of the 26 districts (rajons), which had in the past taken care of this function. Planning regions also actively participate in projects financed by EU and other funds. In February 2010, the parliament (Saeima) passed the amendments in the Regional Development Law that introduces the right of state administration institutions to delegate tasks to planning regions. With the abolition of local authorities (districts), discussion has increased regarding the need for planning regions to evolve as strong arms of the state administration concentrated on service delivery at the territorial level. But the unclear status and future of the five regional planning regions does not promote their effective operation. Socioeconomic Differences 157. Local governments differ not only by size, but also by the demographic and socio-economic situation of the population. At the beginning of 2010, among municipalities the unemployment level varied from 7.8 percent (Ä€dažu novads, Garklanes novads) to 28.3 percent (ViļÄ?nu novads). Average unemployment in the republic cities at 10.7 percent was lower than that of municipalities, where the mean was 13.4 percent in 2010. There is also a large difference in the increase in unemployment that 71 occurred in local government territories from January 1, 2009 to January 1, 2010. The rise in unemployment ranged from 4.3 to 13.5 percentage points. Among those municipalities that had to cope with the greatest increases in unemployment are those with small and largely rural populations. Dependency ratios vary across local government areas in Latvia, with differences in the share of under- and over-working age of the population. Such differences highlight the varied needs for service delivery of local government territories in Latvia. III. Local Government Expenditure Structure Local Government Budget Structure 158. The local government budget is made up of two discrete parts: the basic budget and the special budget. The basic budget is the main part of the local government budget and is formed by general grants; earmarked grants; revenues from services and other own revenues; transfers; foreign financial aid; and the planned expenditures to be made with these revenues; as well as state budget loans and repayment of state budget loans. Basic budget revenues were a 95 percent share of local government revenues in 2009. The special budget is made up of earmarked sources of revenues and their distribution, with the most important component being the State Road Fund, which is used to fund public transportation. 159. The division of local budgets into basic and special budgets makes budget analysis complicated and outcomes less transparent. It also presents an obstacle for medium-term planning and to comparison between the local government budgets. Therefore, it may be appropriate to consider amalgamating the basic and special budgets to make the budget more straightforward and increase budget transparency. Expenditure Structure 160. Local governments account for about a quarter of overall consolidated government spending in Latvia (Figure 3.6). Many of the activities financed from this spending are described in detail in the chapters of this PER that follow, but a general picture is provided here first. By economic classification, staff costs account for the largest portion of local government spending, equivalent to 47 percent of gross expenditure in 2009. The capital share of local government expenditure fell during the crisis, with its share falling from 24 percent in 2008 to 19 percent in 2009. Education is the most important expenditure item in local government budgets. Breaking down government spending by functional classification, education accounted for just over 40 percent of local government expenditure in 2009 (Figure 3.7). The financing of education expenditures comes both from state budget earmarked grants as well as from local government general revenues. The transfer from the state budget for teachers’ salaries is included in this category. The education category is followed in importance by general government services and economic affairs. 72 Figure 3.6: Latvia: General government and local government consolidated expenditures (in million lats, unless otherwise stated) 10000 29% Local government 8000 25% expenditure share (in 28% percent) 6000 26% Local government 25% 4000 26% 2000 General government 0 2004 2005 2006 2007 2008 2009 Source: World Bank estimated based on Vilka (2010). 161. Social protection made up 7.4 percent of total local government spending in 2009. An important autonomous function of local governments is to ensure social assistance (social care) for inhabitants. This includes social assistance to low income families and socially vulnerable people, the provision of places in old people’s homes, the provision of places in educational establishments and homes for orphans and children left without parental care, and the provision of night shelter for homeless people. However, in 2008, only the provision of guaranteed minimum income (GMI) social benefits was compulsory for local governments, and since 2009 housing benefits have been added as compulsory. The expenditure share for social protection in local government budgets in 2008 was 7.4 percent, with 40.7 percent of this being accounted for by expenditures for social benefits (social allowances). In 2009, the share of social protection expenditures increased to 8 percent. The share of social expenditures in local governments in 2009 varied from 1 percent (IkÅ¡Ä·iles novads, Cranikavas novads – rich Riga dormitory towns) to 30 percent (TÄ“rvetes novads). 162. The economic recession and rising unemployment led to an increase in the demand for social benefits for the poor. This showed the procyclical social protection spending pressures that arise at a local level in bad times. In first half of 2009 the expenditures for social benefits was lower than in the same period of previous year, but since August 2009 those expenditures outgrew those of the same period in the in previous year. The impact of the crisis also is reflected in the increase in spending on the GMI. In 2008, the GMI equaled 3.4 percent of all social benefits; in 2009 it had increased to 8.9 percent. In the first three months of 2010, GMI benefits had risen to comprise a 16 percent share of overall social benefits given by local governments. 73 Figure 3.7: Latvia: Functional classification of local government expenditures Social protection Education Recreation, culture & religion Health care Community amenities & housing Environment protection 2009 Economic affairs 2008 Public order and safety Defense General gov services 0 10 20 30 40 50 Local government expenditure share (%) Source: World Bank staff calculations based on data from the Latvian authorities. 163. Individual local governments face very different conditions with regard to social protection spending. For the group of republican cities social benefits per capita varied from LVL 16 (JÅ«rmala) to LVL 38 (LiepÄ?ja) in 2009. For municipalities, it differed from LVL 7 (Rojas novads) to LVL 42 (Aizkraukles novads). This data and the different underlying social economic conditions (unemployment level, personal income tax revenue) show that local governments have very different needs for social assistance and also often different local social assistance policies. A significant proportion of local governments have problems in maintaining adequate social services. This situation was exacerbated in the aftermath of the recent crisis. As most household revenues have decreased, so have local government revenues, but the necessity for social support has been increasing. 164. With the implementation of the Emergency Social Safety Net Strategy (ESSNS) since October 2009, local governments receive from the state co-financing for GMI (50 percent), and housing benefits (20 percent). For unemployed people who meet eligibility criteria, stipends are paid on condition of full time work (100 LVL per person per month), as part of the Workplaces with Stipends public works program, administered by deconcentrated offices of the State Employment Agency. The ESSNS has been a satisfactory interim solution, but in the long term more radical reforms of municipal social assistance are appropriate to consider, in the context of broader welfare sector reform. 165. Health spending out of local government budgets is low at just 1.3 percent of overall spending in 2009. Besides this, local governments also operate local government health care entities (hospitals) that are in many cases legally organized as companies. This means that these (company) revenues and expenditures are not included in the local government budget reports. The function of local governments is to guarantee the accessibility of health care and to promote a healthy lifestyle and sport. This does not mean that local governments have to ensure the health care services. Health care provision is a state function and there is no clear formal division of the responsibilities it entails. For this reason, purely in formal terms, local government activities in health care could be considered voluntary 74 initiatives. However, local governments are often closely involved in health care management as employers and operators of hospitals and medical facilities. IV. Local Government Revenue Structure 166. Up until 2008, local budget revenues rose each year as did the share of local government revenues in overall general consolidated budget revenues. Local government revenues were 2.5 times higher in nominal terms in 2008 than in 2004 and their share of overall revenues had increased to 29.3 percent (Figure 3.8). The fall in government revenues in 2009 led to a nominal contraction of 20.7 percent in local government revenues. The share of local government consolidated budget in general budget revenues also decreased somewhat to 28.2 percent in 2009. Figure 3.8: General government and local government consolidated revenues (in million lats, unless otherwise stated) 8000 29% 27% Local government 7000 28% revenue share (in 6000 26% percent) 5000 25% Local government 4000 27% 3000 2000 General government 1000 0 2004 2005 2006 2007 2008 2009 Source: World Bank estimated based on Vilka (2010). 167. According the Law on Local Government Budgets, local budget revenues are: ï‚· Shares from state taxes and fees; ï‚· Local government fees; ï‚· State budget grants and earmarked grants; ï‚· Grants from the Local Government Finance Equalization Fund (LGFEF); ï‚· Transfers from local government budgets; ï‚· Payments for services; ï‚· Shares from companies profit; ï‚· Revenues from property (rent); ï‚· Revenues from property sale; and ï‚· Other revenues according laws. 168. Tax revenues and transfers made up almost equal shares of local government revenues in 2009, combining to account for over 90 percent of total local government revenues. The majority of tax revenues came from personal income taxes (PIT) (40 percent of total local government revenues). Real estate taxes contributed under 5 percent to total local government revenues. Of state budget transfers, 75 earmarked grants are the most significant, contributing 18 percent to total revenues. Inter-local government transfers account for 11 percent of overall basic budget revenues. Transfers here included the substantial transfer for teachers salaries recorded as a local government transfer as it went to municipalities through district government (but it went directly to republican cities). The Local Government Finance Equalization Fund (LGFEF) makes up a 5 percent share of total revenues. The main contributors to the LGFEF are local governments themselves. Own revenues remain relatively low at a 7 percent share of overall revenues (see Table 3.1). Table 3.1: Local government basic budget revenue structure, 2009 Type Gross revenue share (%) Tax revenues 45.5 Personal income tax 40.3 Real estate tax 4.9 Other tax revenues 0.3 Nontax revenues 1.8 Commercial activities & property 0.6 Fees 0.2 Penalties and sanctions 0.2 Other 0.6 Revenues from property sales 0.3 Revenues from service fees and other own revenues 6.7 Transfers 45.7 State budget transfers 34.5 Transfers for operational purposes 27.3 Grants 2.7 Earmarked grants 18.1 Grants from LGFEF 5 Transfers for capital expenditures 1 Revenues for EU SF projects 6.2 Local government transfers 11.2 Source: Vilka (2010). Note: LGFEF denotes Local Government Finance Equalization Fund Revenue-sharing 169. Local governments do not have the legal authority to impose taxes in Latvia. Only the national- level government has the power to set taxes. The local government shares revenues from four state taxes as follows: 76 ï‚· Personal income tax (PIT) (share going to local governments was 83 percent in 2009 and 80 percent in 2010); ï‚· Real estate tax (100 percent share); ï‚· Lottery and gambling tax (25 percent share of gambling taxes, and 100 percent share from local scale lottery); and ï‚· Natural resource tax (60 percent share in special budget for pollution, 30 percent share for radioactive waste tax, and 100 percent share for burning of dangerous waste and for mineral deposits). 170. The Government sets the PIT base and rate. As Dillinger (2007) points out, although Latvia allocates a large share of PIT to local governments, this is because there is a relatively low yield from the PIT, which has a low, flat rate. Administration, except in the cases of the republican cities of Riga and Ventspils, is carried out by the State Revenue Service. The local government share of PIT constantly increased from 2005 until 2009, going from 71.6 percent before 2005 to 73 percent in 2005, 75 percent in 2006, 79 percent in 2007, and 80 percent in 2008 to 83 percent in 2009, but in 2010 it decreased to 80 percent. As between local governments there is a big difference in the growth of PIT revenues in periods when revenues are expanding, with the increase in the PIT local government tax share during good times, the difference between local governments’ financial capacity increased. Rich municipalities benefitted from the increase in the PIT tax share, whereas for poor municipalities there was almost no impact. 171. The financial capacity of local governments varies substantially. Total tax revenues, tax revenues per capita, as well as the share of tax revenues in total revenues are positively correlated with welfare levels for local governments. The average tax revenue per capita in 2009 was LVL 300. The lowest tax revenue was LVL 100 per capita (Riebiņu novads), a third of the average, and the highest was LVL 509 per capita (Garkalnes novads), equivalent to 169.7 percent of the average. The portion of tax revenues in basic budgets varies from 15 percent (Baltinavas novads) to 79 percent (MÄ?rupes novads). The richest municipalities in Latvia (if tax revenues per capita are compared) are the dormitory towns around Riga city. Local Taxation 172. Giving the ability to local governments to set and raise their own tax revenues could be considered. Tax pressure in general is relatively low in Latvia; and Latvia together with Malta is one of only two countries in the EU-27 without local government taxes.57 A challenge is that such a tax would need to be a stable revenue component if local governments were to rely on it. The tax would need to provide an adequate revenue base, and not be overly procyclical. More cyclical taxes, such as corporate taxes, would possibly promote procyclical revenues. A local sales tax is not allowed under EU regulations. A local income tax or real estate tax is a possibility. Taxes on immovable property are most appropriate for the financing local government. In the medium-term such a development is likely not to occur, as the national government may need to rely on any increase in real estate taxes rather than 57 Source: Sub-national governments in the European Union: Organization, responsibilities and finance. Dexia, July 2008, page 97. 77 devolving them to lower levels government in the current tightly constrained fiscal environment. An additional possibility is to consider giving local governments more rights to charge local service fees. V. Local Government Revenue Equalization 173. Significant differences exist with respect to the degree of financial resources available to local governments in Latvia. Since 1995, a local government finance equalization system has been in place to mitigate these disparities. The existing law on equalization was passed in 1998. This law outlines the mechanism to be used to calculate the resources to be distributed to local governments for equalization purposes. 174. First, the aggregate spending needs for local governments is supposed to be calculated based on the previous year’s total municipal expenditures adjusted for inflation. This estimate of aggregate spending needs is then divided between republic cities and other local governments (novads) on a 47/53 percent basis. Next, the relative expenditure needs are calculated for each local government within these two groups using four criteria: population; the number of children aged 0 to 6; the number of children aged 7 to 18; and the number of the population above retirement age. Using projected revenues, excluding the equalization fund, the gap between projected revenues and estimate expenditure needs is then calculated. A local government’s revenues are capped if its projected revenues exceed its estimated expenditure needs by more than 10 percent. Where this occurs, 45 percent of the estimated budget surplus is taken away and allocated to the equalization fund. For those local governments where revenues are estimated to be less than 95 percent of estimated expenditures needs, the deficit is covered using proceeds from the equalization fund. The central government was, according to the legislation, to cover are shortfall between contributions into and payments out of the equalization fund. 175. In practice, the order of calculation that is employed differs from that stated in the Law (Vilka 2010). While the formula to be used is set out in the law, in reality a negotiation process takes place between local governments and the central government. There is considerable negotiation regarding revenue forecasts centered on the state grant amount, forecasted total and individual tax revenues. The central government’s nominal contribution to the fund has remained static for a number of years, rather than varying with the needs of local governments according to the process set out in the legislation. 176. The equalization system is procyclical in that it is fully dependent on contemporaneous government revenues. The amount of central government funding that goes to the Local Government Finance Equalization Fund has remained unchanged at LVL 7.2 million since 2001. Therefore, the system relies mostly on the revenues received from personal income tax (with a smaller share coming from the real estate tax). Given that the benefits of the equalization structure are likely to be of greatest value to households during economic downturns, the building up of buffer resources to be used to fund transfers to needy local governments in bad times (to finance social spending requirements) could be a valuable countercyclical tool. Currently, when local governments’ tax share amounts decrease, so does the amount available for redistribution. 78 177. As noted above, the proportion of resources received by local governments through the equalization mechanism is relatively low on average. However, smaller municipalities rely considerably on the equalization fund for revenues as their revenue share from PIT is low. A new law to alter the equalization system is due to be submitted to parliament next year (January 1, 2011). The ministry responsible has appointed a committee to examine the issue. One idea being discussed is that the mechanism may be deepened through additional resources from the State. 178. As the discussion of how to improve equalization takes place, it would be useful to clarify the objective and definition of terms used in the equalization framework, and the role of the equalization grant, given that other budget mechanisms (such as several of the earmarked grants) also contribute to vertical equalization. A clarification of the methodology and the order in which calculations are to be made is also merited. The current system is complicated and not understood by many local governments. To sustain the effectiveness of the equalization mechanism, the indicators used for the equalization formula could be regularly updated, at least every 4 to 5 years. 179. The mechanism for redistributing revenues across local governments would also benefit from a system of monitoring and evaluation. For example, there is no regular analysis of how the mechanism has performed. The criteria used to determine each unit of local government’s spending needs are connected with providing basic local services to the population (education, social services, etc.). It would be useful to examine whether the redistributed funds result in an equal provision of such services across local governments. In addition, the performance of the revenue forecasts used to calculate the equalization needs are not compared with the actual situation that evolves. 180. There may be some room for redesigning the equalization mechanism to increase the incentives for local governments to raise revenues. Under the current system, a government that benefits from the equalization grant has little motivation to increase tax and other revenues if increased revenue effort would result in them losing the equalization grant. One method that could be used to increase the returns to collecting revenues is to reward local governments for increased revenue effort by setting aside a portion of revenues raised and not subjecting it to the equalization calculation. 181. A more fundamental reworking of the framework could be done based on recommendations of the Council of Europe:58 a. Add three new criteria to the demographic measures currently used in the calculating spending needs: (i) Area of territory; (ii) Distance to Riga; and (iii) Indicators for service spillover centers, i.e., population centers that serve broader territory (Riga, other republican cities and novads with former district center towns); and 58 Recommendation Rec(2005) of the Committee of Ministers to member states on the financial resources of local and regional authorities. Council of Europe. (https://wcd.coe.int/ViewDoc.jsp?id=812131&BackColorInternet=B9BDEE&BackColorIntranet=FFCD4F&BackColorL ogged=FFC679) 79 b. Progressive rates for calculating the contribution of each local government to the equalization fund. VI. Macroeconomic Control and Expenditure Effectiveness Predictability 182. Another criterion by which to measure intergovernmental financial structures is whether they might allow local government fiscal behavior to threaten macroeconomic stability. In Latvia, the design of the intergovernmental system makes it is unlikely that such a situation would emerge. First, local governments are unlikely to encounter unforeseeable shortfalls due to the deviation between actual and estimated revenues. The revenues of the local government are guaranteed by the central government through compensation mechanisms. During the budget process, the central government gives implicit guarantees on the PIT tax income that is forecast (with the exception for the two republican cities that collect their own PIT). Second, restrictions are in place regarding local government borrowing. Each year a limit is set for total government borrowing, and the breakdown of local versus central. The country has not ratified the article of the European Charter that allows local borrowing directly. Local governments have to borrow from the Treasury or under better conditions than they would get from the Treasury. Third, there is a system whereby local governments with unsustainable fiscal frameworks are identified and an amelioration plan put in place. Local governments may be put under a process to stabilize their finances according to a law passed in 1998 if at least one of the following conditions: (i) The debt service of the local government exceeds 20 per cent of the total amount of total revenues; (ii) The local government is unable or, due to circumstances that can be proved, shall be unable to settle its debt obligations; or (iii) Local government debt exceeds the market values of its assets. A medium-term (3-year) budget framework would further increase budget predictability at the municipal. Implementing such a framework was interrupted by the crisis, but the plan is to put it in place at a later stage. Fiscal Responsibility Frameworks at the Sub-national Level 183. The Latvian authorities are considering putting in place a fiscal responsibility framework. The question is how this would be extended to local government levels. Local government expenditure made up 25 percent of the total general government spending in 2009. Therefore, any fiscal responsibility framework would need to include local governments in its design or risk being weakened by municipal fiscal slippage. The fiscal responsibility framework can take advantage of and build on the current policy of negotiation and consensus building between the central and local governments in making budget decisions. In examining whether fiscal rules are not sufficient for ensuring fiscal discipline at a sub-national level, Ter-Minassian (2005) points out that they often may not suffice as a fiscal rule can neither solve a badly designed intergovernmental system of government nor engender political commitment to fiscal responsibility at the sub-national level of government. Any numerical or procedural fiscal rules would have to involve an agreement between the government and association of municipalities on maximum spending increases. Setting expenditure limits for local levels of government may be one good way of exerting central control. Expenditure limits could be agreed upon and then the 80 lower level of government would be allowed to decide on how they would be implemented. There is adequate transparency regarding sub-national finances in Latvia to support any fiscal responsibility framework, with detailed data provided to the central government in a timely manner. 184. Under the current system, revenue transfers to local governments in Latvia are procyclical since they are mostly based on tax revenue sharing and transfers funded by current government revenues. In good times, the central government then shares its booming revenues with local governments, and does not set resources aside to be used to fund local government spending in bad times. In bad times, the result is a fall in revenues often partially compensated by reducing capital spending. Such a cycle is apparent in looking at the recent crisis, which is much deeper, but repeats the pattern seen in the 1998 downturn. 185. Since local government expenditure has a substantial social sector element, it could be argued that it should be more protected than other forms of expenditure during economic downturns and perhaps even increased. After all, the activities financed from social sector budgets are in most cases, those that benefit households most directly. Indeed, many countries with countercyclical fiscal packages during the recent economic crisis increased social sector spending to protect the most vulnerable households. Therefore, any fiscal framework with a structural target should outline the preferred path for social sector spending, and hence local government spending, across the cycle (while keeping such targets in the budget and flexible rather than enshrined in law). Box 1 proposes using fiscal rules for social policy, rather than having the objective of smoothing government consumption across economic cycles, Engel, Nielsen, and Valdés (2010) suggest targeting the inter-temporal consumption path of the poor. 81 Box 3.1. Fiscal Rules as Social Policy Engel, Nielsen, and Valdés (2010) examine the role of fiscal policy across the cycle when there are exogenously driven and volatile fiscal revenues. They present a theoretical model in which there is a substantial welfare gain associated with running a countercyclical fiscal policy that targets increased spending to the poor. Their model is motivated by fiscal savings in Chile during the recent copper revenue boom that enabled the government to put in place a substantial and well-targeted fiscal package beginning in 2008. As part of the fiscal stimulus, the Chilean government made one-off transfers to the poorest 40 percent in March and August 2009. For the poorest decile, these transfers equaled about two months of income. Households have incomes that fluctuate across the economic cycle. The government planner in the model is focused on managing fiscal transfers so as to maximize household utility over time. Because of revenue uncertainly, precautionary savings are made in good times to finance transfers to households in bad times. Inequality in household income is critical. Equal incomes means that there is a representative agent that receives average income, whereas inequity introduces a poorer population with less stable marginal utility over time. Incorporating heterogeneity in household income introduces an important role for targeting in the context of fiscal policy. With no targeting, an optimal policy rule would be more expensive in bad times because spending would need to be higher to receive the same level of welfare compared to a situation where transfers can be targeted to the needy. Conversely, the greater the possibility to target the poor in bad times, then the lower the amount of government spending that is required. The larger and more volatile the fiscal revenues, the higher are the welfare gains associated with adopting an optimal “social needsâ€? rule over a balanced budget (BB) expenditure. Welfare gains are greater the better is targeting in bad times, but they are smaller when there is no heterogeneity between household income levels. A calibration of the model for Chile suggests that the government should spend 100 percent of revenues in very low revenue states, but less that 100 percent even in slightly below-the-mean states (due to the precautionary saving motives). Around 80 percent should be saved in the most favorable states. Source: Engel, Nielsen, and Valdés 2010. 186. Sub-national savings mechanism could be used to put away resources during good times so that procyclical spending cuts could be avoided in economic downturns. Local governments could consider setting up “Rainy Dayâ€? Funds such as those that many States have put in place in the United States. These types of funds are, however, more likely to be suitable for larger sub-national governments such as republican cities or planning regions, rather than municipalities. Local governments have the advantage that—unlike line ministries that must return unused funds—they are permitted to keep budget underruns. Effectiveness 187. There is some evidence that the intergovernmental financial system in Latvia could improve incentives for rationing services. For example, (as discussed in Chapter 4 on education) there is a lack of incentives to close under-enrolled classrooms or schools. Administrative cost data also points to a large variance in its share of overall spending across local governments. Data on individual salaries in the public sector for 2009 was leaked in 2010 and showed that salaries in local government were often 82 higher than those in the private sector and that there existed considerable differences in salaries between local governments as well as within individual local governments. 188. How can efficiency be protected? Over the past decades, numerous methods have been tried. These range from using performance data to inform budgetary decisions, and setting service standards, to adopting innovative ways to deliver services in small-scale service delivery mechanisms. Latvia is moving to further increase transparency, and accountability for the use of local budgets and local public resources. One step in this direction is the intention to publish salaries and wages of local government employees. A restriction on the holding of multiple local government salaried posts could also be considered. 189. There is no tradition in Latvia of analyzing local government budgets, to compare and explain local government budget indicators. Negotiations, discussions on local government finance issues usually are based on fragmentary budget information, not on a more complete analysis. Therefore, it would be useful to have regular monitoring and analysis of the budget performance of local governments, including designing input, intermediate and output performance indicators for budget programs and making such indicators publically available. Such a benchmarking process may be a useful tool for increasing innovation and peer learning among republican cities and municipalities with regard to service provision. 190. The Ontario Municipal Benchmarking Initiative (OMBI) is one such benchmarking exercise that resulted from a voluntary imitative established by the Ontario Ministry of Municipal Affairs and Housing (Canada), regional chief administrative officers from across the province and 15 municipalities in 2000. The aim of the initiative was for municipalities to work together to share performance information and identify operational good practice. This involved developing, collecting and making publicly available data on input and output indicators by municipality. In Box 2 below, one example of benchmarks is given for social assistance from the OMBI. The aim of this benchmarking was to assist municipalities to assess their relative performance; identify areas where service improvements or costs savings could be made; assess improvements over time; and share ideas on new processes and solutions to better use available resources. As such, the OMBI benchmarking was not designed to be a punitive process to be used to penalize low-performing municipalities, but rather one to encourage collaboration and peer learning. The idea was that this data would be a foundation for more detailed analysis of service delivery and lead to the identification of good practices that would be applicable across municipalities (OMBI, 2008). 191. The 2008 report by OMBI identifies a number of benefits that have resulted from the benchmarking process, including improvements in service performance that came following the setting of targets, the shift of focus to results and the increased accountability that municipal reporting brought, as well as the identification and study of numerous best practices in Ontario municipalities in different service areas. Of course, there are costs attached to benchmarking. The collecting and collating of data and participation in benchmarking activities across municipalities involves financial investment and staff time. This staff effort typically involves service experts, an overstretched and scarce resource for many municipalities. The strong media interest in the benchmarking results that has developed makes the process politically difficult to manage for low performers. Some participants in OMBI have viewed it as of limited use as it fosters an environment where the only factor of a service that counts is cost. The 83 concern is that this spurs on a “race to the bottomâ€? in cost terms, with little focus on service quality (Graham 2007). There have been some examples where information generated by the process is only used when it is negative (Graham 2007). Box 3.2. Ontario Municipal Benchmarking Initiative 2008: Benchmarking Social Assistance Services In measuring budget input and outputs, the process takes into account that each municipality’s results are influenced to varying degrees by a number of factors, including: • Participant readiness for work, literacy level, language skills, and lack of Canadian work experience can impact the ability to find work; • Health barriers to employment may vary across client profiles; • Client access to programs can vary due to geographical, technological, cultural or other limitations; • Differing local labor market conditions; and • Family size and caseload mix. Social Assistance Benchmarks Compared Across Municipalities 1. How long does it take to determine client eligibility? How long it takes -on average- to determine if someone is eligible for assistance after receiving their request for help, in days. These figures show whether the response time for municipalities has been improving. 2. How many households are receiving social assistance? The number of cases is one indicator of the level of service required in a municipality. It also provides an indication of the economic and social well-being of a community. Caseloads directly influence the overall cost of service delivery. 3. What percentage of clients received assistance for less than 12 months? Of course, clients with more complex needs (i.e., severe health conditions) may require social assistance for a longer period. 4. What is the average length of time that clients receive social assistance? 5. How much does it cost to provide social assistance services? Cost measured as: (a) Monthly Social Assistance Administration Cost per Case. Administration cost represents the average cost to deliver and administer the social assistance programs and services. Administration cost per case can be influenced by the caseload size and demographics, services provided and local labor costs. (b) Monthly Social Assistance Benefit Cost per Case. The benefits cost represents the average cost of benefits paid to a social assistance client. The benefit cost per case can vary based on the caseload mix (single or family) and the types of benefits required. (c) Monthly Social Assistance Total Cost per Case. The trend over time is of interest here. Source: http://www.ombi.ca/charter.asp 84 VII. Summary of Recommendations 192. There follows a summary of the recommendations on the intergovernmental system in Latvia. Municipality Size (small municipalities) ï‚· May think of further municipal mergers to creating municipalities of sustainable size/increase administrative quality. According to the Administrative Territorial Reform Law of 1998, voluntary amalgamation of local governments is permitted. However, it may be necessary to increase the incentives to overcome political opposition to further amalgamation by municipalities with low populations. Planning Regions ï‚· Unclear status and future of the five regional planning regions does not promote their effective operation. The main function of the planning regions is to ensure regional development planning, coordination and cooperation of local governments and state institutions. But to provide more effective performance of this function, a clear decision needs to be taken on the future of this sub-national level of government. Equalization Mechanism ï‚· Make clearer the objective and definition of terms used in the equalization framework. ï‚· Clarify the methodology and the order of calculations to be made. ï‚· The mechanism for redistributing revenues across local governments would benefit from a system of monitoring and evaluation. ï‚· The equalization system is procyclical in that it is fully dependent on contemporaneous government revenues. Given that the local government beneficiaries of the equalization system may be most vulnerable to economic downturns, the building up of buffer resources to be used to fund transfers to needy local governments in bad times (to finance social spending requirements) would be a valuable countercyclical tool. ï‚· There may be some room for redesigning the system to increase incentives for local governments to raise revenues. Fiscal Responsibility ï‚· Local government expenditures made up 25 percent of general government spending in 2009. Therefore, any fiscal responsibility framework would need to include them in the design or risk being weakened by municipal fiscal slippage. ï‚· Any fiscal framework with a structural target should outline the preferred path for social sector spending, and hence local government spending, across the cycle (while keeping such targets in the budget and flexible rather than enshrined in law). ï‚· Local governments could consider setting up Rainy Day Funds such as those that many States have put in place in the United States. These types of funds are, however, more likely to be 85 suitable for large sub-national governments such as cities or planning regions, rather than municipalities. Local Taxation ï‚· Could be considered. Tax pressure is relatively low in Latvia. Would create a clearer link between local spending decisions and local taxation. Budget Structure, Transparency and Control ï‚· Amalgamate the basic and special budgets to make the budget more straightforward and increase budget transparency. ï‚· Increase accountability of local companies. As in the state sector, the data on local public companies are not included in the budgets. ï‚· Increase further transparency, clearance, accountability of local budgets and local public resources. ï‚· Conduct regular monitoring and analysis of budget performance of local governments, output and input indicators. Make this publically available. See more on this under benchmarking point below. ï‚· Bring in medium-term planning (three year budget planning), an objective that was lost to the recent economic crisis. Benchmarking Municipal Spending ï‚· Benchmarking may be a useful tool for increasing competition in service provision in larger municipalities/Republic cities and in lower disparities in efficiency of service provision across all municipalities. 86 CHAPTER 4: Social Welfare in Latvia59 I. Introduction and Main Messages 193. By EU standards, Latvia spends a very modest sum on social welfare (protection), although this has risen rapidly in the last three years. Total spending in 2007 was well below spending levels in Hungary, the Czech Republic, Slovakia and Slovenia, Romania and even Estonia. Spending on social insurance pensions makes up the largest share of social welfare spending in Latvia as in other countries – a product of rapid ageing of the population. Spending on non-contributory transfers in Latvia has been comparable to that of its immediate neighbors. However, spending on targeted “exclusionâ€? benefits is – along with Estonia - the lowest among the new member states and indeed in the entire EU. 194. All forms of social welfare play a critical role in smoothing household consumption, particularly during periods of economic crisis. For this reason, the reluctance to make quick cuts in social welfare spending during the current contraction, without careful deliberation, is reasonable. However, Latvia is faced with having to make difficult fiscal choices if it wants to achieve its objective of Euro entry. And in this context, there are clear opportunities for fiscal savings in the social welfare sector, mainly from clawing back windfall spending on old age pensions. 195. Social insurance accounts for 83 percent of social welfare spending, and is the part of public expenditure where the Government is most likely to find options for immediate savings and longer-term fiscal sustainability. Due to a lagged impact of the high-growth period, the level of old-age pensions paid to those who recently retired is considerably higher than amounts paid to retirees from previous years. To illustrate, the purchasing power of a pensioner who retired in 2009 was 69 percent higher than that of a person who retired in 2005. By way of comparison and to put this increase in context, the cumulative reported real wage growth during this period was only 28 percent. Furthermore, amidst the fiscal exuberance of the high-growth years, two important changes in pension policy encouraged a rapid and unsustainable rise in benefit spending: the expansion to all retirees of what was originally a targeted supplement for low pensions; and the introduction of generous benefit indexation to the growth of salaries as well as prices. 196. Through a combination of proposed measures designed to protect the households receiving the lowest pensions – lowering the amount of pension income that is exempt from taxation, and eliminating supplements on higher pensions- the Government could save as much as LVL 205 million immediately. Replacing the earnings-related parental benefit with a far more equitable and augmented child-care benefit would save a further LVL 65 million. 59 Administrative data were provided by the Latvian Ministry of Welfare, and spending data by the Treasury Department in the Latvian Ministry of Finance. Comparative indicators of expenditure and performance were drawn from EUROSTAT and ECSHD’s social protection databases. Measures of coverage, targeting and other indicators for Latvia were calculated by ECSHD staff using the Latvian Household Budget Survey (HBS) for 2007 and 2008 collected by the Central Bureau of Statistics. The conclusions and any errors in interpreting these data are the sole responsibility of the authors. 87 197. A gradual increase of the statutory retirement age is another critical step the Government could consider to put the social insurance special budget on a more financially sustainable path. This measure would not only increase medium and long term fiscal sustainability of social insurance, but also augment the level of pension for future retirees. Strong commitment to this reform, even if implementation does not begin until 2015, could offer immediate benefits in the form of greater confidence in Latvia’s fiscal position. 198. In a period of constrained fiscal circumstances which are likely to become tighter, when considering state social benefits – the categorical, non-contributory programs administered by the national government - Latvia could achieve more progress in supporting its most vulnerable households by using resources more parsimoniously. The recommendation being made here is for re-distribution rather than reduction of what Latvia is spending on non-contributory benefits. This could be achieved with a wider use of means-testing, particularly of family benefits. New members of the EU such as Poland and the Czech Republic have followed this approach. Older members seeking to use resources for social welfare more optimally, such as the United Kingdom, are seriously considering following their example. Taking the largest state social benefit – the family state benefit - as an example, if eligibility were set at a relatively generous threshold such as the 40 percent of least well-off households about LVL 17.3 m could be saved. This estimate is based on the amount of family state benefits received by households in the 2008 Household Budget Survey. The Government can arrive at a more precise figure by cross checking administrative data on payment of sate social benefits with reported income held at the Ministry of Welfare. It is important to emphasize, that any savings from means-testing the family state benefit would be best used to fortify the targeted programs in Latvia’s social welfare system, given how meager these program are relative to the rest of the EU. 199. The remaining recomendations made in this chapter are intended primarily to improve the performance of the social welfare system, although they may deliver some fiscal savings at the margin. 200. In the branch of the social welfare system that provides employment assistance, the Government has already initiated steps to increase the relevance and flexibility of its support to job-seekers. The most important is a gradual shift away from direct provision and augmenting the supply of training programs, toward augmenting individual demand with vouchers. This shift is encouraging, as it will not only initroduce further market discipline to improve the quality of the training, but also align incentives so that people who need this form of assistance have greater choice and pursue training that is more likely to give them marketable skills and increase their employment opportunities. 201. The highly decentralized nature of social assistance in Latvia – particularly of its main targeted programs for the poorest households - may be unnecessarily impairing performance. This review motivates a careful policy discussion of the optimal role for local governments in the delivery of social assistance, particularly of programs mandated by the State, such as the GMI and the housing benefit. With devolved responsibility from the national government, local governments have a strong incentive to look for savings and to be more efficient. This can, of course, be a force for improving performance. In other parts of the region, however, this incentive has sometimes led to compromised standards and uneven delivery of entitlements to the poorest households across municipalities of different incomes. From the experience in other countries, the devolution of financial responsibility for social assistance 88 could create strong incentives for sub-national governments to cut back or ration benefits, particularly during economic downturns. This is a concern national authorities should keep in mind while monitoring the performance and standards maintained by local authorities in the current economic contraction. 202. Finally, some of the features of the social welfare system in Latvia could be modernized to deliver much needed performance gains. Firstly, the Government could consider a shift to monetization of the large share of social assistance benefits granted to household in-kind. Curiously, the share of welfare spending on in-kind benefits in Latvia has risen over the past decade while remaining flat in the other EU New Member States and in the Euro area. Secondly, there are likely to be gains from introducing systematic cross checks of administrative information already held by the authorities. Although the Ministry of Welfare has strengthened its capacity to manage administrative data, different categories of cash benefits paid by the social welfare system are administered with separate databases with no regular cross-check for inconsistencies and incompatible benefits. This is not an issue that is typically covered in depth in a PER. However, on the positive side these databases all contain a common unique identifying code (like the social security number in the United States and the national insurance number in the United Kingdom). Although income checks are conducted as part of each beneficiary’s application process in the Latvian system, the databases have never been systematically cross checked for inconsistencies and incompatible benefits. A more ambitious approach would be to cross check welfare data with administrative data held by tax authorities, the register of persons, as well as property registries maintained by local and national governments. The benefits of this cross-checking exercise often prove to be substantial even in the best administered systems, and accrue not only in the form of savings to the national budget, but also in political support for and popular confidence in the social welfare system. II. The Structure of and Spending on Social Welfare in Latvia 203. Latvia’s social welfare system provides benefits that can be categorized into three main groups: social insurance; state social benefits; and municipal social assistance (Table 4.1).60 Social insurance guarantees people who make contributions compensation for the loss of income in certain situations - such as when a person reaches the retirement age, becomes unemployed or disabled, loses their main income provider, becomes sick, has an occupational accident or disease, or takes off from work during the pre-natal and post-natal (maternity) periods. People covered by social insurance -by virtue of paying contributions from their earnings- are entitled to the pensions, benefits and compensations to ensure they can sustain their consumption when they are unable to receive an income due to any of circumstances discussed above. Social insurance is administered by the State Social Insurance Agency, which is part of the Ministry of Welfare (MOW). 204. State social benefits complement the state social insurance system and provide support in the form of cash payments to certain population groups in vulnerable situations, due to reduction or loss of income, such as: (i) people who are not covered by social insurance or whose insurance contribution 60 The PER discusses cash benefits primarily, however, a discussion of in-kind benefits is also included among the critical issues for consideration for reform. Programs that are delivered in the form of Institutional care – such as long term care for the elderly – are discussed briefly in both this chapter as well as in the health chapter of the PER. 89 record is not sufficient to receive social insurance benefits (state social maintenance benefit); and (ii) in statutory cases when income to cover additional household spending is needed but the state social insurance system does not envisage any protection (e.g., the state family benefit, the child care benefit and the child birth benefit). State social benefits are financed entirely from the national budget, and like social insurance, are administered by national agencies of the MOW. 205. In addition, and to underpin nationally financed and administered programs, municipally administered and financed social assistance programs are available to low income and needy households, as well as to other groups on whom local authorities choose to focus their resources. The housing benefit and the Guaranteed Minimum Income (GMI) transfers are the principal targeted social assistance benefits administered and financed by municipalities, although by mandate of the national government. Local authorities also provide other social assistance benefits on a voluntary basis, as well as one-off, lump sum emergency benefits to residents who have suffered shocks. Most of these voluntary benefits and the emergency benefits are not income tested. 206. Latvia’s active employment assistance is organized and administered by the State Employment Agency (SEA), a subordinate body of the MOW. All legal residents are eligible for active employment assistance, as long as they are previously registered as unemployed with SEA. A broad range of programs are provided, including vocational training, requalification and qualification improvement; paid temporary work; activities targeted to specific groups, such as youth, disabled people, older workers and mothers reentering the labor market after maternity leave; programs to help business start-ups and self-employed; and apprenticeships. Any employment provided by SEA programs usually has to conform to labor legislation and pay at least the legal minimum wage. The Government’s budget allocations are boosted with financial support from the EU’s European Social Fund (ESF). In the 2007 – 2013 EC budget planning period, the majority of ESF financing for Latvia (81.5 percent) is allocated to active employment assistance. The largest such program, by budget allocation, is “promotion of employment in the regionsâ€?, which financed public works administered by local authorities. 90 Table 4.1: Structure of Benefits Paid by Latvia’s Social Welfare System Program Type of coverage Who are eligible? Responsible Source of finance administrative agency Social Old age Contributing Ministry of Special Social Insurance insurance Disability employees, Welfare’s Budget (individual and Survivors employers and State Social employer contributions & Unemployment self employed Insurance transfers from the general Sick-leave Agency budget) Maternity/Paternity Parental State social State social maintenance Individuals Ministry of State base budget (transfers benefits benefit ineligible for SI Welfare’s from the general budget to (covering all benefits. State Social the Ministry of Welfare) contingencies above, Insurance except unemployment; All families with Agency sick – leave, children. maternity/paternity; parental) Disabled persons Family allowances Disability allowances Social Guaranteed minimum Means tested to Municipal Municipal budgets assistance income (GMI) households below governments Lump-sum emergency a per-person benefit eligibility Housing benefit threshold based on income. 207. Compared to its EU partners, Latvia spends a very modest sum on social welfare (protection), although this has risen rapidly in the last three years. In 2007, the latest year for which comparative cross-country data are available from EUROSTAT, Latvia spent about 10.7 percent of GDP on social welfare, although this has increased considerably since, largely driven by an acceleration in spending on social insurance (as discussed in detail in later sections). Total spending in 2007 was well below spending levels in Hungary, the Czech Republic, Slovakia and Slovenia, Romania and even Estonia. Spending on social insurance pensions makes up the largest share of social welfare spending in Latvia as in other countries – a product of rapid ageing of the population. Spending on non-contributory transfers in Latvia has been comparable to that of its immediate neighbors. However, spending on “exclusionâ€? benefits (which in Latvia mainly takes the form of the GMI benefit) was 0.1 percent of GDP and is – along with Estonia - the lowest among the new member states and indeed in the entire EU.61 61 Although spending housing benefits – which are typically, although not uniformly targeted across municipalities – was higher than in some of Latvia’s neighbor countries, this benefit did not become a part of mandatory targeted municipal social assistance until 2008. 91 Table 4.2: Social Welfare Spending in Selected EU Member States, as a Percentage of GDP 2007, by ESSPROS Functional Categorization Family & Total Old age Survivor Disability Sickness Unemployment child Exclusion Housing France 29.0 11.2 1.9 1.8 8.7 1.8 2.5 0.5 0.8 Sweden 29.0 11.3 0.6 4.4 7.6 1.1 3.0 0.6 0.5 Denmark 28.1 10.7 0.0 4.2 6.5 1.6 3.7 0.7 0.7 Belgium 28.0 9.9 2.8 1.8 7.4 3.3 2.0 0.7 0.1 Austria 27.1 11.3 2.0 2.2 7.1 1.4 2.8 0.3 0.1 Netherlands 26.8 9.4 1.4 2.5 8.7 1.2 1.6 1.7 0.4 Germany 26.7 9.5 2.1 2.0 8.0 1.5 2.8 0.2 0.6 Euro Area 25.8 10.0 2.0 1.8 7.6 1.5 2.1 0.4 0.4 Italy 25.5 13.1 2.5 1.5 6.7 0.5 1.2 0.1 0.0 UK 24.8 10.4 0.8 2.4 7.6 0.5 1.5 0.2 1.4 Finland 24.6 8.6 0.9 3.1 6.5 1.9 2.9 0.5 0.2 Greece 23.8 10.4 2.0 1.2 6.7 1.1 1.5 0.6 0.5 Portugal 23.4 10.1 1.7 2.3 6.6 1.2 1.2 0.3 0.0 Norway 22.3 6.8 0.3 4.2 7.3 0.3 2.8 0.6 0.1 Hungary 21.9 8.3 1.3 2.1 5.6 0.8 2.8 0.2 0.9 Slovenia 20.8 8.2 1.5 1.6 6.7 0.5 1.8 0.5 0.0 Spain 20.5 6.5 1.9 1.6 6.4 2.4 1.2 0.3 0.2 Czech Rep 18.0 7.2 0.8 1.5 6.1 0.6 1.7 0.2 0.1 Malta 17.9 7.6 1.8 1.1 5.2 0.5 1.1 0.4 0.2 Poland 17.8 8.7 2.0 1.7 3.9 0.4 0.8 0.2 0.1 Ireland 17.6 4.0 0.8 1.0 7.2 1.4 2.6 0.4 0.3 Slovakia 15.4 5.9 0.9 1.3 4.7 0.6 1.5 0.5 - Bulgaria 14.6 6.8 0.7 1.2 3.9 0.3 1.3 0.4 0.0 Lithuania 13.9 6.0 0.5 1.4 4.3 0.3 1.2 0.2 0.0 Romania 12.6 5.4 0.5 1.3 3.0 0.3 1.7 0.4 - Estonia 12.3 5.3 0.1 1.1 4.1 0.1 1.4 0.1 0.0 Latvia 10.7 4.8 0.2 0.7 3.2 0.3 1.2 0.1 0.1 Source: EUROSTAT ESSPROS 92 208. Each section of this chapter will update the spending data for the various branches of the social welfare system (social insurance, state social benefits, etc.) separately. However, there are two interesting characteristics of aggregate spending on social welfare in Latvia that are most apparent when presented in comparative context with other member states of the EU. 209. The first interesting characteristic is that as a percentage of what it Latvia spends on social welfare, what is allocated to programs that are explicitly targeted to the least well-off households is among the lowest in the EU by a substantial magnitude (Figure 4.1). This is a feature of social welfare spending in Latvia that has remained fairly constant since 1997, even as spending on targeted programs (as a share of total social welfare spending) has increased in the Euro area as a whole (Figure 4.2). Notably this increase has not been observed among most of the countries in Central and Eastern Europe that joined the EU in 2004 and 2007. However, Hungary and Poland are exceptions, with rising shares of spending on means tested programs. Figure 4.1: Spending on Means Tested Programs, as a Percentage of Total Social Welfare Spending in EU Member States, 2007 2007 Spending on means tested programs (% of social protection spending) Estonia 0.6 Lithuania 1.5 Latvia 1.8 Sweden 2.8 Luxembourg 3.0 Denmark 3.2 Czech Republic 3.3 Belgium 4.1 Finland 4.5 Norway 4.5 Italy 4.8 Bulgaria 4.9 Poland 5.1 Slovakia 5.9 Switzerland 6.6 Hungary 6.6 Austria 6.7 Romania 6.8 Greece 7.5 Slovenia 8.6 Portugal 9.3 Cyprus 10.3 Euro Area 16 10.9 Germany 12.4 Spain 12.9 Netherlands 13.6 France 14.1 United Kingdom 15.7 Iceland 17.0 Malta 17.5 Ireland 24.4 0.0 5.0 10.0 15.0 20.0 25.0 30.0 Source: EUROSTAT ESSPROS 93 Figure 4.2: Trends in Spending on Means Tested Programs in Central and Eastern European Members States of the EU, 1997 – 2007 Spending on targeted, means tested programs 15 14 Percentage of social protection spending Euro Area 16 13 12 Slovenia 11 Romania 10 9 Hungary 8 Slovakia 7 6 Poland 5 Bulgaria 4 Czech Republic 3 2 Latvia 1 Lithuania 0 Estonia 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Source: EUROSTAT ESSPROS 210. The second interesting characteristic is the rising amount that Latvia spends on social welfare programs that deliver in-kind benefits and services. This appears strange given Latvia’s level of development, as most middle and high-income countries are moving away from providing benefits in- kind, both to lower the administrative costs of program delivery and to increase transparency of how resources are allocated. Although Latvia’s spending on in-kind programs is on the lower end of the spectrum for EU countries, and actually very close to the Euro area average (Figure 4.3), in the years since 2007 it has risen substantially compared to the EU as a whole and particularly when compared to its neighbors in Central and Eastern Europe. In later sections of this chapter, possible explanations for - and implications of- this accelerated increase in in-kind programs are given. 94 Figure 4.3: Spending on In-Kind Social Welfare Programs, as a Percentage of Total Social Welfare Spending in EU Member States, of spending on social protection) 2007 Spending on in kind programs (%2007 Cyprus 22.8 Italy 27.0 Belgium 27.1 Switzerland 29.2 Malta 29.3 Austria 29.8 Bulgaria 30.1 Luxembourg 30.4 Romania 30.4 Estonia 31.0 Portugal 31.4 Latvia 31.7 Czech Republic 32.2 Slovenia 32.3 Germany 32.8 Euro Area 16 33.0 Lithuania 33.0 Hungary 34.2 Slovakia 34.4 France 35.1 Spain 35.4 Greece 35.9 Finland 37.7 Netherlands 38.7 United Kingdom 40.4 Denmark 41.1 Norway 42.3 Ireland 44.0 Sweden 44.3 Iceland 50.8 0 10 20 30 40 50 60 Figure 4.4: Trends in Spending on In-Kind Programs in Central and Eastern European Member States of the EU, 1997 – 2007 Spending on in-kind programs 50.0 Slovakia 45.0 Percentage of social protection spending Hungary 40.0 Lithuania 35.0 Euro Area 16 30.0 Slovenia 25.0 Czech 20.0 Republic Latvia 15.0 Estonia 10.0 Romania 5.0 Bulgaria 0.0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Source: EUROSTAT ESSPROS 95 III. Performance of Social Welfare 211. In this section, the performance of some of Latvia’s social welfare programs is examined using three criteria. The first is coverage of programs, which is measured using household survey data that show the benefits and services individuals and families say they have received. The second criterion is distribution, which extrapolates from reported receipt of benefits and services to calculate how total spending is spread across the poorest to the wealthiest households, as measured by household consumption. The third criterion, given the crisis and continuing economic contraction in Latvia, is the responsiveness of certain programs to the needs of households which have grown with widespread unemployment. At the close of 2009 unemployment in Latvia was higher than in any other country in the EU. The data used in the assessment of the coverage and distribution of social welfare come from Latvia’s Household Budget Survey (HBS). As with any survey, there are likely to be measurement errors arising either from low response rates, respondents’ imperfect information, or the reluctance of some individuals to respond fully.62 For these reason, the Government is encouraged to verify the results presented here with its own analysis of data from other sources, such as the EU SILC and the rich administrative databases held by the MOW. How extensive is the coverage of social welfare? 212. Table 4.3 presents coverage rates (the rate of reported receipt of benefits) calculated from the 2008 wave of Latvia’s Household Budget Survey (HBS). Coverage of social welfare is high among the population as a whole, at 81 percent, and nearly universal among the poorest 40 percent of households. This outcome is explained mainly by the high incidence of social insurance, but is also supported by extensive state social benefits. There is a sharp contrast, however, between the extensive coverage of social insurance and state social benefits, and the meager coverage of Latvia’s principal means-tested benefits (the GMI and housing benefit). 62 Although some measurement error is to be expected, the HBS succeeds in capturing well over 90 percent of spending recorded in administrative databases. 96 Table 4.3: Coverage of Social Welfare Programs in 2008 (Percentage of households reporting benefits, by quintiles of per capita consumption, net of each transfer) Total Q1 Q2 Q3 Q4 Q5 All social protection 80.8 99.6 94.0 80.8 71.8 57.8 All social insurance 48.5 92.7 57.0 42.5 26.8 23.5 Old-age pension 36.3 86.2 39.8 26.3 15.4 13.9 Disability pension 6.3 15.0 5.0 4.8 2.8 3.9 Survivor's pension 3.2 8.8 3.3 1.6 1.1 1.4 Service and special pension 0.9 1.8 1.3 0.8 0.4 0.1 Sickness + work accidents/professional disease 3.0 3.3 2.9 3.3 2.7 2.9 insurance Maternity/Paternity benefits 3.1 4.0 3.1 2.6 3.3 2.5 All labor market programs 2.9 5.6 2.4 2.4 2.5 1.8 Unemployment benefits 2.9 5.6 2.4 2.4 2.5 1.8 All social assistance 54.4 72.7 57.9 56.8 45.1 39.7 Family state benefit (including also supplement 47.9 59.7 50.4 49.1 42.6 37.5 for disabled child) Child care benefit 8.0 15.6 6.9 8.6 6.1 2.8 Guardian benefits + other state target 1.5 2.3 2.3 1.5 1.2 0.3 allowances (support for child) Birth Allowance 5.1 7.3 4.7 6.0 4.2 3.4 Social pension + transport expenses for 1.3 2.9 1.0 0.8 1.1 0.5 disabled + care allowance + food stamps Funeral Allowance (state + local) 0.7 1.6 0.2 0.5 0.1 1.0 Targeted SA: GMI + Dwelling(cash + in-kind) 2.4 5.2 3.2 2.4 1.0 0.4 Paid health services (local government) - 1.9 2.7 2.7 2.0 1.0 0.9 includes vaccination, medication, pros Local government payments (Cash + in-kind) 7.1 13.1 5.2 7.2 5.4 4.7 Scholarship for students and pupils + creativity 5.2 7.4 6.3 5.5 4.0 3.0 scholarship All remittances 36.8 41.6 32.8 36.1 36.9 36.3 Transfers from other households (in cash) 36.8 41.6 32.8 36.1 36.9 36.3 Notes: Program coverage is the portion of population in each group that receives the transfer. Specifically, coverage is: (Number of individuals in the group who live in a household where at least one member receives the transfer)/(Number of individuals in the group). Program coverage is calculated setting as expansion factor the household expansion factor multiplied by the household size. 213. The relatively low coverage rate of targeted benefits does not on its own indicate a problem, as widespread social insurance benefits – which are taken into account in the means test - may be sufficient to underpin the welfare of many households. Some households who receive social insurance may also be receiving several forms of state social benefits, most of which are not taken into account in means tests. And compared to 2007, coverage of Latvia’s main targeted programs increased in 2008, from covering just 1.8 percent of the population and a mere 4.3 percent of the poorest 20 percent of households, to covering 2.4 percent of the population and 5.2 percent of the poorest households. Although not yet available, a further expansion of coverage of these benefits is expected in the 2009 survey data, given the Government’s accelerated increase in spending on social assistance in response to the crisis, from municipal and the national budget. 97 Which households does spending on social welfare favor? 214. As explained earlier, most social welfare programs in Latvia are “universalâ€? or put more precisely are “categoricalâ€?, which means that eligibility is based on household circumstances other than low income. This is an intentional social choice to extend the safety-net beyond just those who are poor or vulnerable to poverty, and to pursue social objectives in addition to poverty relief and consumptions smoothing. Nonetheless, as the resources available for social welfare policy are constrained, it is interesting to examine how spending on different benefits is distributed across households. The indicators in Table 4.4 show how social welfare benefits are distributed, from the poorest 20 percent to the wealthiest 20 percent of households. Table 4.4: Distribution of Social Welfare Benefits (Percentage of total reported transfers received by households, by quintiles of per capita consumption, net of each transfer) Total Q1 Q2 Q3 Q4 Q5 All social protection 100.0 49.1 21.5 14.4 8.3 6.7 All social insurance 100.0 54.7 19.1 12.4 7.0 6.8 Old-age pension 100.0 60.2 17.5 10.7 5.4 6.2 Disability pension 100.0 49.5 16.8 13.7 9.8 10.2 Survivor's pension 100.0 51.6 23.5 10.4 6.4 8.0 Service and special pension 100.0 48.8 24.2 14.4 9.5 3.2 Sickness + work accidents/professional disease 100.0 28.1 23.2 10.4 18.4 20.0 insurance Maternity/Paternity benefits 100.0 23.7 23.1 12.2 16.8 24.2 All labor market programs 100.0 41.9 10.9 11.8 18.6 16.8 Unemployment benefits 100.0 41.9 10.9 11.8 18.6 16.8 All social assistance 100.0 42.8 21.0 17.9 10.9 7.3 Family state benefit (including also supplement 100.0 30.3 20.3 20.4 15.0 14.0 for disabled child) Child care benefit 100.0 50.6 18.6 20.7 8.4 1.8 Guardian benefits + other state target 100.0 40.0 24.9 16.8 13.7 4.7 allowances (support for child) Birth Allowance 100.0 26.0 18.0 23.7 18.7 13.7 Social pension + transport expenses for 100.0 59.9 9.8 8.1 14.0 8.2 disabled + care allowance + food stamps Funeral Allowance (state + local) 100.0 16.1 6.4 32.9 3.3 41.4 Targeted SA: GMI + Dwelling(cash + in-kind) 100.0 30.4 32.3 21.0 9.9 6.4 Paid health services (local government) - 100.0 35.8 28.1 16.7 10.4 9.1 includes vaccination, medication, pros Local government payments (Cash + in-kind) 100.0 49.4 12.1 14.4 11.4 12.8 Scholarship for students and pupils + creativity 100.0 25.2 24.1 20.5 21.2 8.9 scholarship All remittances 100.0 30.0 15.1 13.2 18.7 23.1 Transfers from other households (in cash) 100.0 30.0 15.1 13.2 18.7 23.1 Notes: Benefits' incidence is the transfer amount received by the group as a percent of total transfers received by the population Specifically, benefits' incidence is: (Sum of all transfers received by all individuals in the group)/(Sum of all transfers received by all individuals in the population). Aggregated transfer amounts are estimated using household size-weighted expansion factors. 215. About half of spending on reported social welfare benefits is received by the poorest 20 percent of households. Most broad categories of social welfare transfers are progressively distributed, in that the share of benefits is falling from the group in the lowest 20 percent to that in the highest 20 percent of the consumption distribution (as one moves from left to right along the rows of table 4.4). This is 98 even the case with most categories of social insurance which are not explicitly intended to benefit poorer households more. Several benefit categories are fairly evenly distributed across households in different consumption groups, notably sickness and work-accident benefits, parental (maternity/paternity) benefits, and the family state benefit, which is the largest category of state social benefit (46 percent of spending on state social benefits in 2009 and 2010). 216. However, the distribution indicators in Table 4.4 point to a problem in the targeting of the benefits that are designed to reach the poorest households, GMI and the housing benefit: over 37 percent of reported GMI and housing benefits are paid to the wealthiest 60 percent of households. While this figure shows a slight improvement in income targeting from 2007 when 39.5 percent of GMI and housing benefits where received by these households, it calls into question the capacity of municipalities to target benefits accurately and raises the concern that the national authorities are not doing as much as they could to monitor municipal performance. These concerns are taken up in later sections of this chapter. How responsive is the social welfare system to unemployment? 217. The social welfare system as a whole is likely to have played an important role in safeguarding the consumption ability of household in the wake of the economic contraction of 2009, even though many of the benefits paid are not explicitly designed to respond to this sort of shock. A specially adapted module of the continuous Labor Force Survey (LFS) collected by Latvia’s Central Statistics Bureau will soon be available that will indicate how important social welfare transfers, private transfers, and other instruments have been in helping households to cope. The crisis had a devastating impact on the Latvian labor market in 2009, which will continue to be felt in 2010. Between June 2008 and June 2009, registered unemployment increased by almost 150 percent (Figure 4.5). Unemployment rates rose sharply from 6.2 in 2007 to 20.1 at the height of the crisis in the fourth quarter of 2009. Youth (15- 24 year olds) were more severely affected than average, with their unemployment rates rising from 10.9 in 2007 to 40.1 percent in the fourth quarter of 2009. Long term unemployment also soared. In addition to significant job losses, a larger fraction of employed people are now working part-time. At the end of 2009, 7.4 percent of all employed people were working on a part time basis, compared to about 4.8 percent in 2007 and 2008. 99 Figure 4.5: Growth Rates of Registered Unemployment in European and Central Asian Countries, June 2008–June 2009, in Percent 40 35 30 25 20 Youth (15-24) 15 unemployment Unemployment (15-64) 10 5 0 2007 2008 2008 2008 2008 2009 2009 2009 2009 Q1 Q2 Q3 Q4 Q4 Q4 Q4 Q4 Source: Kuddo (2009) and staff estimates for Latvia using LFS 218. The sharpest declines in employment were in the construction, public administration, manufacturing, and utilities sectors. These four sectors accounted for almost two-thirds of the 150,000 jobs that were lost. The tough labor market conditions are apparent in the competition for vacancies. While on average 7 people competed for each vacancy in 2008, in 2009 there was a ten-fold increase with 70 people competing for a single vacancy. 219. Ajwad, Haimovich and Azam (2010) simulate the impact of the labor market contraction of poverty and income distributions in Latvia and find that the country is likely to suffer a sharp rise in poverty, a widening poverty gap, and a rise in income inequality. With the 18 percent GDP contraction in 2009 and employment contractions across the sectors, poverty (measured as total household income less that LVL 90 per person per month) may have increased from 14.4 percent to 20.2 percent nationally.63 220. To analyze where in the welfare distribution people are affected most by the economic contraction, Growth Incidence Curves (GIC) were plotted.64 Figure 4.6 plots the simulated GIC for Latvia and also disaggregates the GIC by densely (urban) and thinly (rural) populated areas. For all of Latvia, the simulation indicates that households with per capita income in the bottom 40 percent of the income distribution are likely to have been hit the hardest by the economic slowdown. This is particularly apparent in the densely populated areas of the country, which include the republican cities. However, although the simulations show that people in the bottom 40 percent of the income distribution are likely to have been hit harder by the crisis, the impact on very poor households in rural areas was 63 Ajwad, M. Ihsan, Francisco Haimovich and Mehtabul Azam (2009) “The Employment and Welfare Impact of the Financial Crisis in Latviaâ€? Eastern Europe and Central Asia Human Development Group, World Bank, Washington, DC. 64 These curves compare the growth rate in income of the pth quantile across two time periods, t−1 and t, as gt(p)=yt(p)/yt−1(p)−1. Varying p vary from 0 to 1, gt(p) traces growth incidence curve (GIC). 100 cushioned somewhat. In rural areas, it is more likely that people above the 5th percentile and below the 40th percentile were the hardest hit by the economic slowdown. Figure 4.6: Percentage change in per capita household income between 2008 and 2009 (Growth Incidence Curves) 0 Growth in per capita income -10 -20 -30 0 20 40 60 80 100 Percentile Densely populated area Thinly populated area All Source: Ajwad, et al. (2009) using base-line income estimates from the 2006 EU-SILC. 221. There are social welfare programs that are explicitly designed to act as a buffer in case of job- loss and other shocks to income. First and foremost among these is Latvia’s program of unemployment insurance, which is part of the social insurance branch of the social welfare system, financed from the social insurance special budget. Administrative data that show the share of registered unemployed who are eligible for unemployment insurance indicate how this program left many Latvians who lost their jobs unprotected. This was also the case in other countries in Central and Eastern Europe. The low levels of protection reflect eligibility criteria and pay-out periods that make it difficult for the program to act as a broad automatic stabilizer. To make up for this shortcoming, the Government eased eligibility requirements and extended the benefit period of unemployment insurance in 2009. 101 Table 4.5: Registered unemployed receiving unemployment insurance benefits (Percentage of Registered Unemployed) June 2008 June 2009 Czech Republic 32.4 45.1 Poland 15.2 22.4 Slovenia 22.7 32.7 Estonia 56.0 63.7 Latvia 56.5 54.9 Lithuania 34.6 40.4 Bulgaria 26.2 42.1 Romania 27.1 52.9 Croatia 23.4 25.9 Ukraine 72.8 76.0 Turkey 15.9 21.0 65 Source: Kuddo (2009) 222. Other social welfare instruments that helped households cope are the system’s “activeâ€? employment assistance programs. Among these, programs designed to help workers augment their skills with training and which provide job search assistance are unlikely to be very useful in the wake of a massive demand shock when jobs disappear (although participants receive accompanying stipends which may have indeed been very important). For this reason, although spending on all forms of employment assistance increased to respond to higher demand from households, the Government chose to expand more direct forms of assistance to the unemployed who do not have access to unemployment insurance in the form of public works, with the introduction of the Workplaces with Stipends (WWS) program in 2009. 65 Kuddo, A. (2009) “Labor Market Monitoring in Europe and Central Asia Countries: Recent Trends (Round Two)â€?. World Bank, Washington DC. 102 Figure 4.7: Spending on Active Employment Assistance Programs Source: Staff estimates using data from MOW 223. The public works WWS has been rolled out across Latvia by the State Employment Agency (SEA) and its regional affiliates. The program is self-targeting to needy households, with low stipends relative to the binding minimum wage and a requirement for full-time, labor intensive work. Participants in the program must be registered as unemployed, cannot be receiving unemployment insurance benefits, and are selected on a first-come, first-served basis. The availability of WWS work places depends crucially on the performance of municipal authorities in creating work sites. The most recent data (from March 2010) indicate that work places are being created in the regions of Latvia with soaring unemployment and in regions where the demand for the program, measured by the length of the waiting list, is high (Figure 4.8). This reflects very favorably on the responsiveness of national and local authorities in deploying the new social welfare program quickly and where it is needed most. In the Balvu filiÄ?le and the Ludzas filiÄ?le, 36 and 31 percent of all registered unemployed people who are not receiving unemployment benefits are enrolled in the WWS program. At the other extreme, only 6 and 10 percent of registered unemployed people in the Dobeles filiÄ?le and RÄ«gas reÄ£ionÄ?lÄ? filiÄ?le who do not receive unemployment benefits participating in the WWS program. 103 Figure 4.8: Responsiveness of WWS to Uncovered Unemployed and Waiting Lists in Latvia’s Regions Source: Staff estimates using data from MOW 224. The social assistance programs financed directly from municipal budgets can also be critical to how effectively households are able to cope with shocks to their income. Municipalities offer a wide array of social assistance benefits and services, in cash and in kind. Two programs are mandatory: the housing benefit targeted to low income households (those whose income per person is bellow LVL 150 or LVL 120 per month, depending on municipality, which can take the form of a cash payment, fuel for heating or actual housing), and the GMI which is targeted to the poorest households (presently, with household income less than LVL 40 per adult and LVL 45 per child per month). Although not mandated by national legislation, most municipalities also offer a single-payment, emergency benefit to households that seek assistance, regardless of household income. 225. Figure 4.9 provides a rudimentary assessment of the responsiveness of municipally administered and financed social assistance programs to household need, proxied by the increase in unemployment between December 2008 and December 2009 in each locality. The plots in the first column show changes in total spending, the share of spending on mandatory benefits, and the share of spending on benefits delivered in-kind, and how these changes relate to the rise of unemployment. In the second column each measure is shown in per-capita terms. 226. These data raise concerns about the responsiveness of municipal social assistance programs to household needs in the wake of the crisis. Total spending on social assistance has clearly risen in most municipalities and republican cities. In many cases, spending more than trebled. Although there is some indication of a positive correlation between these increases and greater need (shown by the upward arc in the distribution of the data to the right of the origin), there are several municipalities where despite a considerable rise in unemployment, very little change in spending on social assistance is observed. 104 Figure 4.9: Changes in Spending on Municipal Social Assistance with Changes in Unemployment, 2008 - 2009 (a) Change in Municipal Spending on Social Assistance, (b) Change in Municipal Spending Per Capita on Social 2008-2009 Assistance, 2008-2009 (Bubble Size: Total Municipal Spending) (Bubble Size: Total Municipal Spending per capita) 400% 400% Change in Total SA Spending Change in Total SA Spending 300% 300% 200% 200% 100% 100% 0% 0% -100% -100% 4 6 8 10 12 14 4 6 8 10 12 14 Percentage Point Increase in Unemployment, YE2008 -YE2009 Percentage Point Increase in Unemployment, YE2008-YE2009 Municipalities Republican Cities Municipalities Republican Cities (c) Change in Municipal Spending on Mandatory Social (d) Change in Municipal Spending Per Capita on Assistance, 2008-2009 Mandatory Social Assistance, 2008-2009 (Bubble Size: Total Municipal Spending) (Bubble Size: Total Municipal Spending per capita) Change in Share of Mandatory SA Spending 75% Change in Share of Mandatory SA Spending 75% 50% 50% Riga 25% 25% 0% 0% -25% 4 6 8 10 12 14 -25% 4 6 8 10 12 14 Percentage Point Increase in Unemployment, YE2008 -YE2009 Percentage Point Increase in Unemployment, YE2008 -YE2009 Municipalities Republican Cities Municipalities Republican Cities (e) Change in Municipal Spending on In Kind Social (f) Change in Municipal Spending Per Capita on In Assistance, 2008-2009 Kind Social Assistance, 2008-2009 (Bubble Size: Total Municipal Spending) (Bubble Size: Total Municipal Spending per capita) Change in Share of In-Kind SA Spending 100% Change in Share of In-Kind SA Spending 100% 75% 75% Riga 50% 50% 25% 25% 0% 0% -25% -25% -50% 4 6 8 10 12 14 -50% 4 6 8 10 12 14 Percentage Point Increase in Unemployment, YE2008-YE2009 Percentage Point Increase in Unemployment, YE2008 -YE2009 Municipalities Republican Cities Municipalities Republican Cities Source: Staff estimates using data from Latvian Treasury and MRDLG 227. Considering mandatory benefits on their own – those that are explicitly intended for low income households – a relationship to the local rate of unemployment is even less apparent. It is likely that the 105 nationally financed and administered programs bore a large share of the social cost of the economic contraction. But the limitations in coverage of national programs, requires that municipal social assistance play a critical safety net role, particularly for households at the lowest levels of the consumption distribution. And while there is no a priori reason to expect the share of benefits that municipalities deliver in-kind to bear a relationship to unemployment, in several cases, the rise in in-kind benefits has been substantial. Some of the likely reasons for this divergence in performance across municipalities are discussed later in this chapter. IV. Critical Issues in Social Welfare Social Insurance 228. Earlier in the decade Latvia, like Estonia and Lithuania, spent around 11 percent of its GDP on social welfare. By 2007 this ratio had reached 8.4 percent, the lowest share in the EU however rapid GDP growth over this period is the primary culprit for this trend. The comparative position of Latvia changed substantially in 2008 and 2009 when cumulative real GDP dropped by 22 percent, while social welfare spending grew by 25 percent in real terms. These developments dramatically increased the share of GDP allocated to social welfare expenditures which reached 13.4 percent in 2009. Comparative data for 2009 for other EU members are not yet available. Figure 4.10: Total Social Welfare Expenditure in EU Member States (Percentage of GDP), 2007 and Updated for Latvia 2009 Source: EUROSTAT ESSPROS 229. Figure 4.11 decomposes the growth in social welfare expenditures between 2007 and 2009 and shows the changes in the main areas of spending. For example, the share of GDP allocated to old age programs, even taking into account a 10 percent pension cut in the second half of 2009, increased from 5.0 to 7.7 percent of GDP (by 21 percent in real terms). Spending on family and child programs 106 increased from 0.9 percent to 1.7 percent of GDP (by 49 percent in real terms). Finally, sickness and disability spending increased from 1.5 to 2.7 of GDP (by 41 percent in real terms). Figure 4.11: Share of GDP Allocated for Social Welfare Functions in EU Member States (Percentage of GDP), 2007 and Updated for Latvia 2009, Cash Transfers Only Source: EUROSTAT ESSPROS 107 230. Social insurance programs in 2009 made up 83 percent of Latvia’s social welfare spending. Given that these programs are intended to be self-financing, it is important to examine both revenue and expenditure trends in real terms (Figure 4.12). Between 2005 and 2007 revenues to the social insurance special budget grew by 44 percent and then dropped by 23 percent between 2007 and 2009 in real terms, resulting in cumulative growth of just 11 percent from 2005 to 2009. Meanwhile, spending on social insurance in the same 2005-2009 period increased by a cumulative 54 percent in real terms, significantly overtaking revenues and accelerating substantially even after 2007 when the economy started to slow. Between 2005 and 2009 spending on old age benefits grew by 42 percent, while spending on disability and sickness grew by 57 percent in real terms. A new parental benefit was introduced, and financed from the social insurance special budget in 2008, but without a corresponding increase in the contribution rate owed by workers and employers. Figure 4.12: Social Insurance Special Budget Contribution Revenues and Expenditures in 2009 Real Amounts 231. The unprecedented increase in revenues to the social insurance special budget in the high- growth period up to 2007 can be better understood by decomposing the growth rate of the covered wage bill as shown in Figure 4.13. Growth in the number of contributors averaged 3.4 percent until 2007, but this trend was sharply reversed by the crisis. The recession-related decline in the number of contributors has been further exacerbated by half of the working pensioners effectively being corralled into retirement by a 70 percent cut in their pensions if they continued to work, although the cut has since been repealed.66 232. The part of growth in the covered wage bill attributable to real increases in workers’ earnings during this period was even more volatile, rising by 21 percent in 2007 and then contracting by 13 percent in 2009. High growth in registered wages prior to the crisis not only reflected growth in real wages, but also increased formalization of employment. Inflation was also unstable during this period rising to 15 percent in 2008 before moderating to 3 percent in 2009 and likely reaching negative 66 Interestingly, the 9 percent drop in the number of contributors understates the negative effect of coverage contraction on the revenues as the precipitous decrease in the number of contribution paying workers has been mitigated by the increase in other types of insured population including unemployed, sick and disabled. 108 territory in 2010. Overall growth of the covered wage bill peaked at 32 percent in 2007 and contracted by 17 percent in 2009. 233. Latvia was the first country in Central and Eastern Europe to restructure its public, single-pillar pay-as-you-go (PAYGO) old-age pension plan, from a purely defined benefit (DB) structure, to a non- financial (notional) defined contribution (NDC) plan. The new public plan was implemented in January 1996. Although not funded, the benefits paid under a non-financial NDC retirement savings plan are calculated based on an affiliated worker’s accumulation of “notionalâ€? capital (the individually accounted accumulation of contributions to the plan) and a notional rate of return, which in Latvia’s case is tied to the insured wage and benefit bill which is the sum of wages of contributors and benefits of other insured including unemployed, sick, disabled and those on maternity and child care benefits.67 Simultaneously, a phased increase in the minimum retirement age for men and women was introduced, which converged at 62 in 2008. 234. In theory, one of the merits of an NDC structure – like a funded defined contribution plan – is that by tying benefits to contributions, the formula should provide workers with an incentive to work longer, since their benefits are linked to how much they “accumulateâ€? notionally, and the returns earned on their notional capital. Thus workers are able, and indeed encouraged, to work and contribute past the minimum retirement age as upon retiring, a worker’s notional annuity is calculated by dividing his notional capital by his remaining life expectancy.68 In reality, close to 100 percent of insured individuals retire before or at the statutory retirement age of 62 even though the incentive to postpone retirement has been extraordinarily high in recent years, with real notional interest rate of 21 percent and 28 percent in 2008 and 2009. It is not clear if this lack of responsiveness to incentives is attributable to the shortsightedness on the part of the insured, lack of information, pressure from employers to retire, perceived uncertainty of the benefits, lack of faith in the credibility of the plan, or other cultural factors. 235. Also as part of the structural reform of social insurance for old age, disability and survivor coverage in the 1990s, a separate pillar of private mandatory individual retirement savings accounts was introduced, and began operating in July 2001. The new funded accounts were made mandatory for any worker under 30 at that time, and all new entrants to the formal employment since. From 2001 a small share of workers’ social insurance contributions to the NDC plan (about 2 percentage points) was transferred to the individual retirement accounts. This share was increasing slowly (in 2008, it was equal to 8 percentage points), however, in the wake of the contraction in 2009 the Government decided 67 Thus, the rate of return earned during the accumulation period (when a worker is paying into the plan) reflects (a) growth in productivity and the wage rate; (b) changes in the size of working age cohorts; (c) the net migration of working age individuals; (d) the gradual formalization of the economy (measured as the increase in the “coveredâ€?, i.e., registered wage bill); (e) increases in labor force due to changes in retirement age; and (f) changes in the number of people insured by the state. 68 There are two important points to note with this estimation, which both have potential fiscal consequences. First, a unisex estimate is made which overestimates the remaining expected years of life for men and underestimates the remaining lifespan of women. Second, the life expectancy estimate is made annually for each cohort of retirees from the plan, and as such captures the longevity gains reflected in population ageing. 109 to divert back 6 percentage points of contributions to shore up the finances of the public NDC pillar, committing to restore funded pillar contribution to 6 percent by 2013. 236. Although Latvia’s pension reform was broadly considered a success, the pension plan remained vulnerable to the instability of the economic boom years. The financial stability brought about with the 1996 reform was fragile and vulnerable to even small changes in the parameters of the NDC plan (such as the contribution rate, indexation mechanism, and the minimum age of retirement) and how contributions to and rights acquired under the pre-reform plan are recognized. And the NDC formula proved to cope badly with the rapid acceleration and deceleration of economic growth from 2005 – 2009. 237. Along with the rise in contributions to the social insurance special budget prior to 2008, social insurance liabilities were also increasing. Widening surpluses in the social insurance special budget created strong pressure for additional spending, best exemplified by the introduction of the new parental benefit. The fastest growing social insurance liabilities arose from the automatic indexation of the old-age pensions to the notional interest rate, which reflected the lagged growth of the wage and benefit bill of the insured population (see Table 4.6). Even in the midst of the slow-down in growth, notional interest rates remained high until 2009 due to: (a) the lag of about 18 months between observed wage and benefit bill growth and its application to notional accounts; and (b) the indexation of benefits to the wage and benefit bill of the insured population (which includes those covered who may not be contributing, such as mothers, unemployed, disabled, etc.), which in the midst of the recession decreases more moderately than the wage bill of the contributing population. Figure 4.13: Decomposed Growth in Revenues to the Social Insurance Special Budget 238. The earliest manifestation of rapidly growing old-age pension liabilities was a sequence of high notional interest rates. Reflecting the high-growth period, the high notional interest rates steadily increased the level of pensions paid to those who recently retired, shown in Table 4.6. To illustrate, the purchasing power of a pensioner who retired in 2009 was 69 percent higher than that of a person who retired in 2005. By way of comparison and to put this increase in context, the cumulative reported wage growth during this period was only 28 percent, and it is likely that the actual income received by the 110 average contributor grew even less. The income disparity among the retiring cohorts can also be seen by comparing the average benefits of pensioners of different ages as shown in Figure 4.14: cohorts younger than 68, especially male, have benefited disproportionally from the pre-2008 high-growth boom years. Table 4.6: Real Growth in Newly Assigned Pensions 2005 2006 2007 2008 2009 Notional real interest rate 11% 11% 13% 21% 28% Average newly assigned old age pension adjusted to 2009 prices, LVL 127 146 161 180 215 Growth of newly assigned pension in real terms 100% 115% 127% 142% 169% Figure 4.14: Average Pension Benefit by Age Cohort of Current Pensioners 239. Such large differences between pensions across cohorts of the retired elderly just a few years apart are likely to have created a political climate in which it was difficult to resist proposals to introduce and maintain a more generous pension indexation scheme (a combination of adjustment to changes in prices and wage growth). These differences also are likely to have motivated the introduction of pension supplements for retirees with contribution histories prior to 1996. While newly assigned pensions are expected to remain stable in 2010 and decline by 6 percent in 2011 (due to negative notional interest rates reflecting the years of contraction in the covered wage and benefit bill), the cohorts who retired prior to 2010 have clearly benefited from a permanent increase of their benefits which could be considered a windfall from the pre-2008 bubble economy. 240. Ironically, the notional account mechanism which was designed to achieve self-balancing over the long run into the pension-portion of the social insurance special budget, has transmitted the destabilizing impact of the macroeconomic bubble into the social insurance special budget, and even exacerbated the effects of the bubble in the short term. In addition, parallel increases in the retirement age and the rapid growth in declared wages (the covered wage bill) amplified notional interest rates further, again disproportionately benefiting pre-retirement cohorts who had the most notional capital. 111 Given the sudden collapse of contribution revenues to the social insurance special budget in 2009 and the absence of adequate financial reserves, these unsustainable additional windfall increases in benefits to recent retirees may have to be scaled back, especially since the demographic and macroeconomic challenges in the near future are expected to worsen. 241. An old age pension plan’s performance is in large part determined by demographic trends, which for Latvia are presented in Figure 4.15.69 In recent years the publicly administered pillar of Latvia’s old-age pension system has functioned in a very favorable demographic environment. In the last 4 years before the fiscal crisis each cohort of 62-year olds has been smaller reflecting significant drop in birth and survival rates for people born during World War II. A gradually increasing effective retirement age accentuated this trend even further, resulting in a progressively lower numbers of new retirees. However, this trend has reversed in 2008 and now each subsequent cohort of 62 year olds will be larger than the previous for the next 15 years. Similarly, in the last 8 years prior to the crisis each subsequent cohort of 18 year olds was larger, bringing higher numbers of new contributors to the plan. Unfortunately, this trend also reversed sharply at the same time as baby boomers started retiring. Over the next decade the number of new entrants to the labor force is expected to drop to half the 2007 level. Figure 4.15: Demographic Trends in Latvia, Population Pyramids 2009 and 2060 2009 2060 90 90 75 75 60 60 45 45 30 30 15 15 0 0 Source: Staff PROST projections using EUROSTAT assumptions 242. These demographic trends combined with the recent increase in the effective retirement age, growing coverage of the working age population, and formalization of the wage bill prior to the crisis contributed to the declining old age pensioner dependency rate70 and growing surpluses until 2007. These trends have reversed sharply as early retirement has spiked and number of contributors has decreased. While some of the recent reversal is temporary, the long run trend in the dependency ratio is sharply up as shown in Figure 4.16. A more disaggregated picture of membership in the pension plan is presented in Figure 4.17. 69 Demographic projections are based on EUROSTAT no-migration demographic scenario. Life expectancy at 62 by year 2060 is expected to lengthen by 8 years for men and 7 years for women. Total fertility rate is assumed to increase from current 1.38 children per woman to 1.54 children per woman by the year 2060. 70 Dependency rate is equal to a number of old age pensioners that have to be supported by 100 contributors. 112 Figure 4.16: Simulated Projection of the Effective Dependency Rate (Number of old age pensioners per 100 contributors) Figure 4.17: Simulated Projection of Evolution in Pension Plan Membership Source: Staff PROST projections using MOW assumptions 243. Under these assumptions effective revenue from pension contributions to the social insurance special budget is projected to follow a path pictured by the lower line in Figure 4.18. This is based on the assumption that contribution rate allocated for pension insurance remains at 20 percent, with the carve- out for the funded pillar increasing from 2 percent in 2010 and 2011 to 6 percent by 2013. Recently the State Special Budget for Old Age Pensions – a line item in the social insurance special budget - has been and still is subsidized from other social insurance programs by about 2 percent of the overall wage bill. The higher revenue line in Figure 4.18 assumes that this subsidy will continue indefinitely. 113 Figure 4.18: Simulations of Projected Revenue and Expenditure for Old Age Pensions from State Social Insurance Special Budget 71 Source: Staff PROST projections using MOW assumptions 244. The lower expenditure line in Figure 4.18 represents projected growth of the notional pension pillar’s expenditures under the current law. These expenditures are expected to decline slowly before reaching the level of revenues by the end of the simulation horizon. The projected “internally subsidized deficitâ€? is presented in Table 4.7 and is calculated as a difference between projected expenditure and “subsidizedâ€? revenue lines. 245. While the system is expected to remain in deficit for most of the simulation horizon, important progress towards financial sustainability has been made with the revisions of the social pension law in 2009. Prior to 2009, the lowest pensions were indexed more generously by topping inflation indexation with additional indexation based on 50 percent of real wage growth. This top-up has been eliminated. The long term fiscal effect of this important change is illustrated by the difference between the two expenditure lines in Figure 4.18. 71 Assumptions about the future growth of the economy in the next few years are based on the projections of the Latvian Government (March 2010). Namely, GDP growth is assumed to be -4% and 2.7% in 2010 and 2011 respectively, while wage growth is assumed to be -6.3% and 2.7%. In the longer term GDP growth is projected to reach 4% by 2013 and then drop to 2.5% by 2030 before reaching a steady state growth of 1.5% by 2050. These long term assumptions are more optimistic that those used in recent modeling exercises by Latvian government and EU. Furthermore, it is assumed that after the initial post-crisis recovery covered wage bill and notional interest rate will grow at the rate of GDP growth. 114 72 Table 4.7: Simulations of the Projected “Internally Subsidized Deficitâ€? to Pensions 4 2009 2015 2020 2030 2040 2050 2060 Pension deficit 1.3% 2.1% 1.6% 1.7% 1.3% .8% -0.3% Source: Staff PROST projections 246. In addition to the reform of pension indexation, some short term savings measures have also been taken in 2009, including: (i) suspension of indexation of pension for 2009; (ii) reduction of pensions in payment by 70 percent for working pensioners and 10 percent for the other old age pensioners (later found unconstitutional by the Supreme Court); (iii) increase in early retirement penalty from 20 percent to 50 percent of the pension in payment until the pensioner reaches statutory retirement age; and (iv) temporary redirection of contributions from the funded pillar to the notional pillar. In retrospect, the increase of the penalty for early retirement proved to be especially timely as flows to early retirement have increased dramatically in 2009 and are likely to continue to rise until the end of 2011. 247. Given additional need for significant short run savings, the Government could consider additional measures focused the stock of current pensions in payment. Newly granted pensions amount to only 4 percent of total pension spending and even if reformed, significant savings from changes to the benefits of the “flowâ€? of future retirees will only be realized in the medium and long run. Importantly, of the package of pension reforms made in 2009, those that remain in place have already reduced medium and long term projected spending, while most significant measures designed to make short term savings were repealed by the Constitutional Court. For this reason, and in the spirit of burden sharing, the Government could take a fresh look at the possible savings that can be had from changes to pensions in payment. 248. The Government could tax pensions progressively. Given the accelerated increases in pensions that resulted from the booming economy prior to 2008 (driven by (i) partial wage indexation of pensions, (ii) the introduction and then extension of new supplementary benefit for pre-1996 contribution years, and (iii) windfall gains for recently retired pensioners associated with abnormally high notional interest rates), it is reasonable to attempt to claw back some of this spending. A claw-back is viable especially if it is done in a targeted manner, aimed mostly at the youngest (i.e., the most recently retired) pensioners who gained disproportionately from the pre-2008 boom, many of whom are still capable of working (although, admittedly, the differentiated nature of the benefit cuts in 2009 that penalized working pensioners adversely impacted incentives, and although repealed in December 2009, has already driven 50 percent of working pensioners out of the labor market). Figure 4.19 illustrates the effect of two different targeting measures. The lower line represents pension income tax revenue. Currently the first LVL 165 of monthly pension benefits is exempt from income tax.73 This exemption is much higher than the exemption on income from labor, and gathers LVL 34 m in annual revenues. Moving this threshold lower towards a lower bound of LVL 80 (the level of the minimum pension guarantee) can generate additional revenue of up to LVL 85 m. 72 While only 20% of wage is allocated to notional retirement accounts, around 22% of wage bill is transferred to State Special Budget for Old Age Pensions to meet current pension payment obligations. The 2% difference is subsidized from other social insurance programs. 73 Average pension, including pre-1996 service benefits stood at 168Lat in 2009. 115 249. The Government could eliminate the pension supplement for contributions prior to 1996 or return to a policy of targeting the supplement. The dotted line on the same graph represents targeted elimination of the non-contributory supplement to pensions introduced in 2006 for low income pensioners and extended to all pensioners in 2008. The benefit pays LVL 0.7 for each year of accrued service before 1996, even though actual records of contribution prior to this year are partial, at best. On average the supplement amounts to 14 percent of overall pension spending. This measure could be implemented progressively by eliminating the supplement for all retirees except the 10 percent of pensioners with the lowest benefits. Eliminating the supplement on pensions at or above LVL 140 per month would protect the incomes of the poorest pensioners, and save up to LVL 120 m annually. Figure 4.19: Range of Savings from Targeting Pension Taxation and Repeal of the Pension Supplement Source: Staff estimate based on MOW administrative data 250. The Government could index pension benefits to deflation. Another measure that could justifiably be applied to the current stock of pensioners is to index pensions to deflation. By definition, this protects the purchasing power of pensioners. Given that purchasing power of contributors has decreased on average by 13 percent in 2009 it is appropriate that the purchasing power of pensioners is at least constrained from rising. The current law indicates that pensions should be indexed to prices but cannot be reduced in nominal terms. This provision could be interpreted in two ways. Under a lenient interpretation of the law (Option 1 in Table 4.8) pensions should be frozen in times of deflation and indexed positively when the CPI is positive. Under a more stringent interpretation the downward adjustment of the pension is not allowed but upward adjustment only starts when purchasing power of the pension starts to fall behind its original level (see Option 2 in Table 4.8). Whether the more stringent application of the law is possible would have to be investigated by the authorities. However, the legality of stringent interpretation clearly runs a political risk of being challenged when inflation returns. There is also a risk that deflation could be prolonged, which would not allow the Government to claw back accumulated deflation for a long time (2019 under current assumptions). Finally, Option 3 of Table 4.8 presents the effect of automatic indexation to the CPI whether positive or negative, lagged by one year. However, this option would probably require additional legislative action. The net present value of total savings range from 3.7 percent of GDP in the case of full indexation to deflation with savings of 0.6 percent of GDP in the first 2 years of implementation. Applying the stringent interpretation of the law would save only 1.7 percent of GDP and the savings would only start accumulating in 2015- 2019. 116 Table 4.8: Simulation of Indexation Options in a Deflationary Environment 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Inflation -3.7 -2.5 0 .2 .7 1 1.5 2 2 2 Accumulated Inflation -3.7 -6.2 -6.2 -6.0 -5.3 -4.3 -2.8 -.8 1.2 3.2 1. Lax Interpretation 0 0 0 0 .2 .7 1 1.5 2 2 2. Stringent 0 0 0 0 0 0 0 0 0 1.2 Interpretation 3. Full indexation -3.7 -2.5 0 .2 .7 1 1.5 2 2 Savings of (3) over (1) 3.7 6.2 6.2 6.2 6.2 6.2 6.2 6.2 6.2 Savings of (2) over (1) 0 0 0 0 .2 .9 1.9 3.4 5.4 6.2 Source: Staff PROST projections 251. The Government could fully apply the current NDC rules leading to the decline in new pensions. The benefits paid to new retirees are expected to be hit by negative notional interest rates in 2010 and 2011 of -4 percent and -10 percent respectively. This development is already increasing the pressure to look for “solutionsâ€? both at individual and Government levels. There is evidence that people perceive retiring early in 2010 (even with the early retirement penalty) as more beneficial than waiting for the negative notional interest rate of -10 percent to be applied to their notional capital in 2011. In addition to increased early retirement, pressure on the social insurance disability program is also building, given that the negative notional interest rates are not applied to the level of disability payments, which remain a defined benefit program (Figure 4.20). The Government’s current policy to counter the decline of new pensions has been not to update the annuitization factors by increases in life expectancy, as stipulated in the NDC formula. Increasing the annuitization factor would have cut new pensions by additional 2 percent. Immediate savings can be made by increasing the annuitization factors as soon as possible and allowing the adjustment to take place automatically as intended. The expected decline in next year’s new pensions of around 10 percent could also be allowed to happen, given that it only represents a partial correction of unsustainably high notional interest rates in previous years. Figure 4.20: Increase in New Old Age and Disability Benefit Claims, January 2006 – January 2010 Disability Old Age 1100 5000 new disabled new old age 900 4000 6 month average 3000 700 2000 500 1000 300 0 Source: MOW administrative data 117 252. The Government could consider further increases in the retirement age. An increase in the statutory retirement age is also recommended, even if this measure is unlikely to generate fiscal savings in the short run, both to reduce the medium term pension deficits in the social insurance special budget and to increase replacement rates in the long run. The statutory retirement age could be safely increased by 6 month increments every year, starting in January 2015. There are several reasons to wait until then. The current early retirement option expires in 2011. Given the recent rush to early retirement and disability, this could mean very few new old age benefits are awarded in 2012-2013.74 This means that an increase of the retirement age before 2014 would not generate any significant savings, and could be counterproductive in the recessionary environment. Furthermore, the cohort turning 62 in 2014 will already have experienced a significant increase in the effective retirement age (70 percent of previous cohort will have retired early, while most representatives of this cohort will have to wait until 62), so it would be inequitable to increase their burden further. The expected savings from increasing the retirement age are presented in Figure 4.21 and can be compared with the projected deficit. Savings for the pension scheme from a retirement age increase are expected to climb to 1.5 percent of GDP by 2020 as the number of pensioners decreases and the number of contributors rises. After this, annual savings start to decline as longer careers start to translate into higher pensions. However, it is reasonable to expect additional benefits from this measure to arise elsewhere: increasing the number of taxpayers could boost revenue not only with additional contributions to the social insurance special budget as a whole, but also through the government tax base. Figure 4.21: Simulation of Projected Fiscal Savings from a Retirement Age Increase (Percentage of GDP) Source: Staff PROST projections 253. An important benefit from increasing the retirement age would be to raise replacement rates both from the notional and funded pillars. The effect on notional replacement rates is presented in Figure 4.22, which shows an increase of newly granted pension benefits of around 25 percent which will 74 It is expected that among the cohorts turning 62 in 2012 and 2013, around 70 percent will already be receiving either an old age or disability benefit. 118 be badly needed in the future when the generosity of the notional pillar is expected to decrease dramatically. A similar positive impact of retirement age increase can be expected on the replacement rates from funded pillar, to which 6 percent of affiliates’ wages will be allocated from 2013. Figure 4.22: Simulated Impact of Retirement Age Increase on Replacement Rates from the NDC Pillar Source: Staff PROST projections 254. The Government could further strengthen the funded pillar. The reduction of contributions to the funded pillar from 6 percent to 2 percent between 2009 and 2012 has reduced the fee base for the private pension fund managers. While most of these companies can survive such temporary changes, an extended period of diverted contributions could cause some fund managers to start leaving the market. This could reduce competition in fund management services, and would very likely increase account administration fees. For this reason, it is important for the authorities to adhere to the planned path for restoring contributions to the funded pillar. The sobering projection of replacement rates in Figure 4.13 underscores the crucial role that the funded pillar is designed to play – if allowed - in delivering benefits for future pensioners. 255. Other programs financed from the social insurance special budget offer opportunities for savings also. The most promising is the relatively new parental benefit. Contributory, earnings-related benefits are paid to expecting mothers for 4 months starting with the 7th month of pregnancy, at an average benefit of LVL 1,234. Subsequently, an earnings-related parental benefit is paid to parents covered by social insurance at an average amount of LVL 433 per month for the care of a child from 2 months to 12 months. In contrast, parents who are not covered by social insurance receive the flat (i.e., not related to earnings) child care benefit of LVL 50 per month (which is a component of the state social benefit branch of the welfare system, discussed in the next section). For the duration of the second year of a child’s life a flat child care benefit in the amount of LVL 30 is paid from the state social benefit budget regardless of the insurance status of the parent. 256. Until 2005, no distinction was made between “insuredâ€? (i.e., contributing) and “uninsuredâ€? (non-contributing) parents. All parents received a flat benefit of LVL 30 in today’s prices. But in 2005 the Government introduced the earnings related parental benefit for those who contribute to the social insurance special budget. However, the new benefit was financed from the state social benefits budget, 119 i.e., from general revenues and not revenues from social insurance contributions. From January 2008 the earnings-related parental benefit was increased again, and shifted to the social insurance special budget. Critically, this transfer was made without an accompanying increase in the social insurance contributions. Figure 4.23 demonstrates the inflation-adjusted growth of the total spending associated with these two benefits that share the common social objective of subsidizing the cost of child care. Figure 4.23: Spending on Child Care and Parental Benefits, Adjusted for Inflation, (LVL millions) Source: Treasury Department, MOF 257. The Government could restore the flat child care benefit. Given that no additional social insurance contributions were levied for the new earnings-related parental benefit, the recent shift to the social insurance special budget is inappropriate. This was a point made in the ruling of the Constitutional Court against the Government’s cut of pensions in 2009. Furthermore, since the parental benefit is in effect “non-contributoryâ€? the link between the level of payment and a recipient’s prior earnings introduces regressive transfers to people with higher incomes. Furthermore, having two benefits aimed at the same social objective – one earnings related and the other flat and considerably lower – is socially inequitable. The Government could correct this problem by making all parents of children aged 2 to 24 months eligible for a flat child care benefit paid from the state social benefit budget. The level of the flat child care benefit could either be: (i) maintained at the current LVL 50 and LVL 30 for the first and second year of child’s life respectively, generating savings of LVL 60 m in 2011; or (ii) the benefit for the first year could be raised to LVL 100 (which was the average benefit level in 2007) generating lower (but still significant) savings of LVL 50 m (i.e., LVL 70 m savings for the social insurance special budget, minus the additional LVL 20 m expense from the state social benefits budget). 258. The Government could consider paying the full child care benefit regardless of whether the parent is working or not. The current policy makes a parent’s decision to rejoin the labor force expensive, unnecessarily discouraging work. This is damaging for the individual and for society which would benefit from encouraging longer periods of activity in the labor market. This change is especially critical as the current pool of contributors is shrinking and would help stabilize the notional interest rates in the old age pension plan. In addition, this change in policy would also increase employability of young women. 120 259. The Government could make maternity benefits more flexible. Rather than requiring women to exit the labor force at 7 months of pregnancy, a woman could be allowed to decide to postpone the commencement of the 4 month long maternity benefit period by a few weeks or even until the time of birth. International practice suggests that many women might prefer this flexibility and choose to remain active. Each individual’s decisions would depend on a doctor’s opinion, the type of employment, and the particular economic circumstances of the family. The current policy reflects increasingly outdated attitudes toward pregnancy.75 Making maternity benefit flexible, could translate into savings for the Government in the first year of implementation. Based on a conservative assumption that women will decide to shift the date of benefit commencement by an average of 2 weeks, the savings for the social insurance budget in the first year of implementation would amount to LVL 1.4 m. In addition, increasing the flexibility of the maternal benefit would add additional taxpayers which would increase social insurance and general tax revenues, adding another 0.6 m in combined revenue, and raising total savings to LVL 2 m. Finally, the amount of maternity benefit could be calculated on at least the wages in the 2 years of prior to taking leave, rather than last 6 months, as is currently the case. This would dramatically reduce perverse incentives to manipulate reported income in the months prior to maternity leave. 260. The Government could save money from changes to how it reports and accounts for contributions. Currently, reported social insurance contributions are used to determine benefits even if contributed funds have not yet been transferred to the social insurance agency by employers. Similarly, the agency transfers 2 percent of the wage bill of participating contributors to the private pension funds even if contributions from employers has not yet arrived. Past proposals to postpone the commencement of temporary social insurance benefits - such as sickness benefit - until contributions are made in full have been struck down as illegal on the grounds that late payments from the employer are not the fault of the beneficiary. These reform proposals were valued at LVL 38 m in savings, but the delays in paying benefits they implied would have created hardship for the perspective beneficiaries. An alternative would be to base initial benefits paid only on contributions that have actually been received from employers. For example, if the contribution record of an applicant for an old age pension shows that the contributions for the last 3 months have not been received by the time of the application his pension would be calculated ignoring these last recorded contributions. Later, when the pending funds finally reach the social insurance administrator, the pension would be increased to account for the late payments but no refunds would be issued for the past differences. Similarly, money could be transferred to individual pension accounts in the funded pillar only upon the receipt of contributions from the employer. Both changes would create strong incentives for employees to put pressure on their employers, and to report negligent employers to the authorities, but minimize the negative impact on their household. This important accountability mechanism would be reinforced if the Government restored the regular statements on the status of individual pension account, which was suspended in 2009. 75 Historically, late pregnancy has been socially treated as sickness throughout Eastern and Central Europe, but this is not the case in other regions of the world where many women can and do choose to work productively in the last months of their pregnancy. 121 261. As a closing observation to this segment on social insurance, a relentless contraction of the working age population in Latvia is expected to continue in coming decades. Macroeconomic conditions are also expected to improve only very slowly in the near and medium future. Given these expected changes, it is vital to encourage labor force participation and heavily discourage exit from the labor force through early retirement or disability. Current policies discourage many pensioners and parents of young children from working. Revenue losses from taxation and social insurance contributions in these cases is likely higher than the savings from reduced benefit payments while the individuals affected by these measures also suffer a significant reduction in their income. Continued employment of younger benefit recipients also helps them to maintain their employability in the future. It is often argued that early retirement and other benefit programs that encourage withdrawal of certain people from the labor market, open employment opportunities for new entrants to the labor force and the currently unemployed. This is known as the “lump of laborâ€? argument, and has been found to be false. The number of the positions in the labor market is not fixed. Many studies suggest that increasing the proportion of working elderly does not crowd out the jobs of younger cohorts as their experience and skill sets are very different. These two groups of workers are not good substitutes. The creation of new firms often requires both skill sets making young and older workers natural complements that can increase the productivity of the economy. State Social Benefits 262. This segment focuses on state social benefits: the set of welfare transfers in Latvia for which there are no explicit contributions required from individuals and firms for eligibility, other than the taxes they pay, which pool as general revenues to finance spending out of the national budget. These benefits are universally available to individuals and households suffering losses from death, disability, etc., but who are ineligible for social insurance because they do not meet contribution criteria. Further, this category includes benefits available to all families with children, of which the flat child care benefit discussed at length in the previous segment is just one. All state social benefits are administered by de- concentrated offices of the Ministry of Welfare. 263. Latvia budgeted about LVL 97.6 m for this category of welfare programs (budget code 20.00.00) in 2010, which is a small but significant decrease in spending from 2009 (LVL 101.6 million), achieved by shortening the pay-out period of some benefits and eliminating overlapping benefits. But the small decrease (relative to health and education spending) is justified given the importance of many state social benefit programs in sustaining household consumption during the crisis. The distribution of spending has remained relatively constant. Family state benefit – a monthly payment of LVL 8 for every child to the age of 18 made to all households – is the largest program, accounting for 37 percent of planned spending in 2010. 122 Figure 4.21: Distribution of Spending on State Social Benefits, 2010 Planned Disabled transport, Child guardian Children Disabled by Funeral grant, 0.05 1.57 allowance, w/celiac, Chernobyl, Survivors of 1.15 1.00 0.26 Independence struggle, Chernobyl, 2.58 0.00 Fostering grant, 2.20 Disabled child care, 3.23 Supplemental family, 7.02 Family state benefit, 36.80 Disabled care, 11.49 Social maintenance, 12.67 Child care, 11.63 Child birth, 8.33 Source: Social Benefit Department, MOW 264. Latvia, like many of its neighbors, has chosen a relatively generous interpretation of entitlement to state social benefits, and in particular to family allowances and child benefits. As pointed out earlier, this is an intentional social choice to extend the safety-net beyond just those who are poor or vulnerable to poverty, and to pursue social objectives in addition to poverty relief such as encouraging fertility. This being said, family benefits have been shown to have little impact on fertility decisions. Where there has been a positive response, the level of benefits is typically much higher relative to average wages than the amounts paid by any of the state social benefit programs in Latvia. The MOW argues that child care and child birth benefits are both effective at promoting growth of Latvia’s birth rate (Ministry of Welfare, 2006, pp. 41-42). However, the coincidence of observed increases in births with the economic boom period would make it difficult to attribute this increase to social policy. In any case, keeping child birth, child care and other family benefits universal is a social choice, with explicit, easily quantified fiscal implications. 123 265. The Government could limit payment of the family state benefit to low income households. Assuming that any of the state social benefits is too low to make a difference to family decisions about how many children to have, the remaining justification for these benefits is to help households to maintain a minimum living standard. The main measure the Government could pursue to seek both short term and long term fiscal savings in this branch of the social welfare system is to improve distribution by means testing some categories of state social benefit. Since most of the non- contributory benefits are designed to be universally available to all who need them, limiting benefits to the poorest households would require legislative reform. Even if eligibility were set at a relatively generous threshold - such as the 40 percent of least well-off households – the distribution of the family state benefit shown earlier indicate that about LVL 17.3 m could be saved.76 266. The Government could consider even more radical action – means-testing all forms of state social benefit. However, the remaining programs in this branch of the welfare system are either very small, will gradually go out of existence (for example, the Chernobyl related benefits and those linked to the independence struggle), or socially totemic, and thus politically very difficult to reform (among these, the child birth benefit). It is important to emphasize, that any savings from means-testing state social benefits would be best used to fortify the targeted programs in Latvia’s social welfare system, given how meager these program are relative to the rest of the EU. Municipal Social Assistance 267. Social assistance, financed and administered by municipal governments, accounts for a minor but imiportant share of social welfare spending. Total annual spending on social assistance by municipalities (including republican cities) increased by 98 percent from 2006 to the end of 2009. The most accelerated increase in spending, however, took place not in response to the crisis, but from 2007 to 2008 when municipal social assistance expenditure increased by 50 percent. This increase probably reflects the decision to make housing benefits mandatory. Prior to the contraction in 2008, about 5 percent of this spending went to the targeted programs municipalities are mandated to pay (the GMI and from 2008 the housing benefit). Spending on mandatory benefits, as a share total spending, accelerated rappidly in 2008 and again at about the same rate in 2009 (indicated by the trend line on the left panel of figure 4.22). The rest was spent on benefits municipalities offer at their discretion. As mentioned earlier, while there is no apriori reason to expect the share of benefits delived in-kind to have increased during the economic contraction, in-kind benefits grew six-fold from the end of 2008 to the end of 2009. 76 About 49 percent of family state benefits are received by the wealthier 60 percent of households. The estimated savings are calculated by applying this factor to the amount that is planned for this benefit category of state social benefit in 2010, LVL 35.4 m. 124 Figure 4.22: Municipal Spending on Social Assistance, 2006 – 2009 (Real LVL) Mandatory vs. Discretionary, 2006-2009 Cash vs. In Kind, 2006-2009 80 40% 80 40% Mandatory Benefits' Share of Spending In-Kind Benefits' Share of Spending 60 30% 60 30% Million LVL Million LVL 40 20% 40 20% 20 10% 20 10% 0 0% 0 0% 2006 2007 2008 2009 2006 2007 2008 2009 Mandatory Benefits Discretionary Benefits Cash Benefits In-Kind Benefits Source: Treasury Department, MOF Figure 4.23: Program Allocation of Municipal Social Assistance Spending, 2009 GMI 8% Other programs (including meals) Housing 34% 25% Fostercare benefit Health Emergency 4% 17% 5% Transportation 3% Raising and schooling kids 4% Source: Social Assistance Department, MOW 268. As discussed earlier, when compared to its neighbors, Latvia allocates a very small share of social welfare spending to programs designed to target the poorest households. The meager spending on income-targeted benefits is apparent in the low coverage indicators, where the main benefits targeted to the poor have to be grouped in order to be captured in representative household surveys in 2007 and 2008. The relatively small volume of last-resort social assistance reaching poor households could reflect the highly decentralized character of Latvia’s social assistance programs, which are not only administered but also financed by municipal governments. This may explain the low incidence of GMI and housing benefits if poorer municipalities are less able to provide benefits. 125 269. As part of the Government’s emergency response to the crisis, a temporary line of supplemental financing (which covers 50 percent and 20 percent of what municipalities pay in GMI and housing benefits respectively) was created late in 2009 to help local authorities meet the rise in demand for the two mandatory targeted benefits. However, this is a stop-gap measure which is set to expire at the end of 2011 and which leaves an important policy question pending: How could responsibility for the different functions of delivering social assistance be assigned differently between national and local governments, and non-government service providers? Given the experience of other countries that have decentralized social assistance by varying degrees and with mixed results, this is a question for the Latvian authorities to consider carefully. This segment of the PER chapter on social welfare tries to provide some guidance to help find answers, and addresses the minor, but related issue of the extent to which social assistance is delivered in-kind rather than in cash transfers. 270. The main motivation for decentralizing responsibility for services to local governments is a real or perceived lack of efficiency, flexibility, innovation and responsiveness on the part of national government. Decentralizing responsibility for service delivery can improve outcomes by encouraging competition between local authorities and between an enlarged set of non-government service providers in a given locality. Decentralizing can also improve outcomes by opening the possibility of local innovation and differentiation of services to meet the particular needs of specific groups, which often differ widely across areas of a country. Furthermore, local authorities can bring greater knowledge to service delivery, which can be particularly useful in accurately targeting services to the households that need these most, reaching out to socially disconnected groups, and to closely monitor the welfare impact these services have. Finally, and most importantly, by placing responsibility on the shoulders of locally elected officials, decentralization can lead to improved services by increasing accountability and the pressure these households can bring to bear on local government when performance is unsatisfactory (World Bank, 2004).77 271. The literature distinguishes between three specific forms decentralization can take, which are important to consider separately when discussing how best to organize social assistance. The first, and most complete transfer of responsibility is called devolution, by which central government transfers the authority for decision making, financing and management of services to politically autonomous units of local government. The second, more limited transfer is called delegation, by which local governments are given purely administrative responsibilities for delivering public services, but are subject to the supervision of central government agencies which also retain financial responsibility. The third and most limited transfer is called de-concentration by which national governments transfer activities and responsibilities of central ministries and agencies to local offices of these agencies, but retain control over financing and the formulation of policy. 272. A seemingly obvious point underlying the theoretical arguments in favor of decentralization - although often forgotten in practice – is that the extent to which the touted benefits of decentralizing services can be had is directly determined by how quickly the capacity of local authorities can be built to manage decentralized responsibilities. However, even where responsibility is decentralized to highly 77 World Bank (2004) Making Services Work for Poor People: World Development Report, Oxford University Press, Washington DC 126 capable local governments, when considering social assistance in particular, a further fundamental distinctions has to be made: between financing and administration (or implementation). It is important to keep this distinction in mind when considering four principal economic and political arguments. 273. The first argument arises from the negative externality of poverty and the corresponding public- good nature of social assistance. Many government interventions designed to foster public goods (those that increase positive externalities and correct negative externalities) such as national defense and macroeconomic management, are only effective when provided uniformly. The public-good nature of social assistance - that is, its primary function to alleviate poverty and ensure minimum consumption and social welfare – also requires a strong degree of uniformity. When considering minimum levels of publicly-financed support, it is difficult to justify disparities across localities or discrimination in the level and quality of provision based on place of residence (Bird, 2000).78 Some authors even speculate that allowing differences in social assistance provision could lead to a “race to the bottomâ€? as local authorities seek to discourage the settlement of the poor and vulnerable by lowering cash benefits and cutting services in their area (Matkovic, 2006).79 From the standpoint of balancing overall economic efficiency and equity, the challenge of how to use public spending to achieve the highest possible rates of growth while ensuring the gains improve the welfare of all citizens, it is critical that institutions like social assistance are uniformly available across regions, particularly given the strong market forces that drive geographic concentration of economic production (World Bank, 2009).80 274. The second argument is drawn from the economics of insurance, risk management, and savings from scale. If considered “poverty insuranceâ€?, social assistance is most efficiently financed from the largest possible “risk poolâ€?. Decentralization can fragment the risk pool, if responsibility for financing benefits is devolved and the tax base on which the social assistance system is built is constrained. This creates smaller, less efficient risk pools in which there are not an actuarially sufficient number of “winnersâ€? (i.e., net contributors to the pool) supporting beneficiaries (i.e., net recipients from the pool) (Barr, 2001).81 In a related point, there are substantial savings from scale from centralizing some of the functions related to delivery of social assistance, particularly those related to managing information. In a highly decentralized welfare system, smaller municipalities can be forced to bear higher marginal costs to deliver the same minimum package of services than larger municipalities. These higher administrative costs could lead to a lower level and poorer quality of benefits and services in smaller, resource-constrained communities. 275. Thirdly there is an argument for diversifying the set of service providers and capturing the benefits of market forces. As with education and health care, gains in efficiency and quality can be had from greater competition and innovation. The education and health policy literature advocates a “splitâ€? between the purchase and provision of publicly financed services (which are typically – but not 78 Bird, R. (2000) “Intergovernmental Fiscal Relations: Universal Principles, Local Applicationsâ€?, Working Paper 00- 2, School of Policy Studies, Georgia State University 79 Matkovic, G. (2006), “Decentralization of Social Welfare in Serbiaâ€?, Center for Liberal-Democratic Studies, Belgrade, Republic of Serbia. 80 World Bank (2009) “Reshaping Economic Geographyâ€? World Development Report, Washington DC 81 Barr, N. (2001) The Welfare State as Piggy Bank, Oxford. 127 necessarily – bundled together in, large centralized systems of public service provision), that allows a larger and more diverse set of service providers to compete and for households to choose between them. Similar arguments can apply to social assistance, particularly to the provision of assistance above a socially defined minimum package. These arguments, however, are more valid with respect to social welfare services, than to cash benefits, as there is greater scope for differentiation across service providers and thus a more obvious basis for competition (Barr, 2001). 276. Finally, there are important political arguments to consider. In contrast to services that benefit the majority of households in a locality, such as education and health, social assistance is increasingly targeted to the most vulnerable groups. This is efficient from a social welfare standpoint. However, not only are these groups often a minority in any given community in terms of their relative and absolute number, they also typically are those households that are least able to exert political pressure, due either to their poverty, social disconnection, or the prejudices of the majority and local elites. Thus, with respect to benefits and services for the poorest and most vulnerable, without the right mechanisms to strengthen accountability and participation, local governments have less incentives to respond to demands for better funded and higher quality social assistance. Indeed, local authorities may have strong incentives to divert funding away from social assistance to services enjoyed by the majority or by the more powerful groups in their constituencies. The arguments for retaining financing at least for mandatory benefits at the national level of government are summarized in the Box 4.1. 128 Box 4.1 Financing Mandatory Social Assistance: Best Practice Guidance Experts agree on the advantages of central-government financing social assistance entitlements. Several arguments are presented to support this emerging consensus. First, from a legal standpoint, the national government is it is the only level of authority that can guarantee that as citizens of a country, every individual will be treated equally and have access to the protection no matter where they happen to live. Second, to act as an effective “safety netâ€? this guarantee has to function effectively as a citizen’s entitlement which implies the need for nationally uniform eligibility criteria, benefit amounts and quality guidelines, all of which determine costs and therefore are best established through nationally representative institutions. Third, from and economic perspective, the national governments is best positioned to engage in counter-cyclical spending – to redistribute resources from times of economic growth to when the economy is faltering - while in many countries local authorities are forbidden from doing so, or where they are allowed, do so far less effectively. Fourth, not only can the national government more effectively redistribute spending to counteract the economic cycle, but it is also uniquely placed economically and politically to redistribute resources from prosperous places to poor parts of a country. Fifth, financing safety net benefits and other welfare entitlements from the national budget allows people to be more mobile, and to move to improve their economic prospects, neither worrying about the entitlements they have to leave behind, nor choosing their destination according to the generosity of benefits. Sixth, taking into account the nature of risk management, the much larger risk pools offered by financing from the national budget can increase the responsiveness and efficiency of safety nets and other welfare entitlements, significantly reducing administration costs. Most countries in the EU and OECD are moving toward national financing of at least minimum entitlements. This has been the direction that welfare policy has been taking over a long period of history. Although, social welfare assistance started in many places as the responsibility of local authorities, decentralized financing has been typically put under severe strain by shocks, such as national economic crises. National governments have to move quickly in the wake of a crisis to take up financial responsibility for social welfare when constrained local authorities simply do not have the resource base to do so. 82 Source: Musgrave, (1959), Oates, (1972), Grosch, et al. (2008) 277. Taken together, these arguments make a strong case for maintaining centralized financing of at least a mandatory, uniform minimum package of cash benefits and services; while capturing the benefits of local knowledge by giving local authorities responsibility for some key aspects of administration (in 82 Musgrave, R. (1959), The Theory of Public Finance: A Study in Public Economy. New York; McGraw-Hill Oates, W. (1972), Fiscal Federalism. New York: Harcourt Brace Jovanovich Grosh, M., C. del Ninno, E. Tesliuc and A. Ouerghi, (2008), For Protection and Promotion: The Design and Implementation of Effective Safety Nets, The World Bank, Washington, DC 129 particular, outreach to excluded groups, delivery, and monitoring); responding to local needs by ensuring scope for local innovation; reaping the gains from competition by diversifying service providers to include non-government agencies; and establishing structures to ensure accountability to the poor, vulnerable and socially disconnected groups. 278. Indeed, the wide range of country experiences with decentralizing social assistance in Central and Eastern Europe generally reflect the arguments above. Moves to restructure social assistance in the region during the 1990s often took the form of decentralization of varying degrees, but particularly increasing local participation in financing and administration of benefits, as well as enlisting greater support for vulnerable groups from families and non-governmental organizations. But the experience has been mixed. This is mainly because decentralization in the region in most cases was motivated less by the primary objective of improving delivery, and more by secondary objectives. These range from coping with constrained budgets in times of economic and financial crises; devolution of fiscal responsibility and contingencies; and reactionary rejection of any form of central planning after the fall of socialism (Matkovic, 2006). 279. There are several examples of responsibility for social assistance being fully devolved with little care or thought given to adequacy or efficiency of financing. In Romania decentralization of social assistance lead to a rapid deterioration in performance. After passing a law on local public finance which devolved responsibility for social assistance to local governments, total spending on social assistance and child protection – including benefits and services for disabled and elderly people – fell by nearly 40 percentage points in real terms between 1996 (when fiscal decentralization began) and 2001. This resulted in a 16 percent decline in cash benefits from their level in 1996. A lack of financial resources at the local level forced municipalities to scale back provision, and in some cases, even to stop providing services altogether. Decentralized financing resulted in substantial inter-regional disparities in coverage, with poorer localities providing the least amount of services. In the face of resource constraints, eligibility criteria for most benefits became ad hoc, as local welfare offices used their discretion to ration scarce municipal resources. To repair the damage to poor households left in the wake of Romania’s decentralization, the country recentralized social assistance financing in 2002. 280. Even in countries where there is a mix of national and municipal funding of social assistance, there are long histories of bitter discord between the different levels of government over the matter (Ditch and Barnes, 1996).83 In Sweden, for instance, where social assistance was funded partly with block grants from national government and partly by local taxation, there was considerable political opposition to nationally imposed benefit levels. Individual municipalities could award benefits at a rate below the suggested minimum. Whilst claimants could have their right to enforce their claim to the minimum level, it emerged that only the articulate and well informed people could make use of these rights. There were similar outcomes in Finland, where the abolition, in 1991, of the normative directives obliging municipalities to pay benefits at the national rate left no redress for claimants. In this case, there was widespread recognition that the geographical variations in benefit payments were less related 83 Ditch, J, and H. Barnes (1996), “Social Assistance Coverage and Administrationâ€?, in Finding the Balance: Financing and Coverage of Social protection in Europe, eds. van Ginneken, Occasional Papers on Social Security, International Social Security Association, Geneva, July 1996 130 to the need for assistance than the ability of municipalities to provide it, and thus there was considerable pressure on the central government to increase its share of financing (Ditch and Barnes, 1996). 281. The performance outcomes of decentralizing social assistance financing can depend on whether a country is largely homogenous in social and economic terms, or one where there are substantial social and income differences across areas. In Russia, for example, soon after the breakdown of the Soviet Union, the national government’s financial role was weakened contributing to its inability to allocate adequate financial resources for social assistance. The central government had little influence over the distribution of social assistance across the autonomous republics, oblasts, or cities. As a result, social assistance became entirely a local matter: locally financed, and with benefit eligibility and levels wholly under local control. This led to a wide dispersion of the coverage and levels of social assistance in the country (Milanovic, 2000).84 282. However, there are also experiences in the region that illustrate the benefits expected from decentralization of certain functions. In Albania, for example, delegating responsibility for identifying eligible groups was found to have had a positive impact on the delivery of social assistance. Local officials in Albania had more information on households’ welfare than was available to national government authorities. The allocation of social assistance among households by local authorities was much better targeted than the allocation of social assistance funds by the central government (Alderman, 1998). 283. In Hungary, decentralization of service provision proved more successful than decentralization of cash benefits, and has led to the gains from innovation and competitions between providers, predicted in the policy literature. The Hungarian social assistance system was decentralized following the 1990 Act on Local Self-Government. The Act gave almost full control over social assistance to newly formed local councils. However, the national government remained fully responsible for financing social assistance. Whereas Romania provides an example of devolution, Hungary’s case is an example of delegation. The transfer of responsibility for providing welfare services brought greater flexibility, which spawned innovation and the introduction of new services, as well as new service providers, who have proved more responsive to households. By the end of the 1990s, with decentralization of social welfare services well entrenched, there were over 5000 non-government organizations providing services to poor and vulnerable groups under the oversight and regulation of national government authorities (Andrews and Ringold, 1999). 85 284. The Government could consider permanently re-centralizing financing for at least the GMI. There are strong economic and political arguments for retaining national financing of social assistance, and few arguments in favor of decentralizing this key function. Furthermore, there are many concrete examples to illustrate the dangers of fragmenting the larger risk pool provided by the national public finance system, and forgoing the economies of scale from many aspects of centralized provision, such as 84 Milanovic Branko (2000), “The Role of Social Assistance in Addressing Povertyâ€?, in Braithwaite, Grootaert, and Milanovic, Poverty and Social Assistance in Transition Countries, St. Martin’s Press, New York. 85 Andrews, Emily S. and Dena Ringold (1999) “Safety Nets in Transition Economies: Toward a Reform Strategyâ€?, Social Protection Unit, Human Development Network, World Bank 131 information management, benefit payment, quality assurance, and impact evaluation. A long history and rich practical experience with assigning social welfare mandates and functions to local governments in East and Central Europe, indicates that the greatest advantages from decentralization are to be had by giving municipal authorities some administrative responsibilities over delivery of cash benefits (in particular, delegating responsibility for outreach to excluded groups and other administrative functions), and much greater autonomy in the provision of welfare services. 285. It is difficult to extend the recommendation to centralize financing for the mandatory housing benefits in its current form. The benefit, although mandatory, is not uniform in the way it is provided (sometimes in cash for housing; often in cash and/or in kind subsidies for winter fuel; and in some cases in the actual form of housing), varying according to the wealth and history of a municipality. Nor is the group targeted by the national mandate the same, as the eligibility criteria – “low incomeâ€? – across Latvia’s municipalities and republican cities. 286. The Government could require mandatory benefits to be monetized and encourage a shift away from in-kind programs where possible. The quandary posed by the mandatory housing benefit reflects a concern raised earlier for the amount of social assistance spending that is delivered in-kind rather than in cash. The sharp rise in spending on in-kind programs during the contraction may be purely an artifact of the dominance of the newly mandatory housing benefit. Made obligatory in 2008, in 2009 the housing benefit accounted for 25 percent of municipal spending on social assistance, and unlike other benefits (in-kind and cash) that local authorities were able to cut the national mandate would have prevented municipalities from reducing this program. Indeed, as more people lost income, the number of households eligible for the benefit increased. This said there may be a more worrying explanation for the rise in the share of spending on in-kind programs: tactical allocation by budget-constrained municipal governments interested in reducing the resources they actually spend on mandatory targeted social assistance. Programs that provide in-kind benefits are notoriously more difficult to monitor. Data on how much is spent is driven by how local authorities assign monetary value to the in-kind goods and services they provide. There is no evidence that any municipalities are engaging in this sort of tactical allocation. However, in-kind provision makes it more difficult for the MOW to monitor whether they are or not. If the same trends in the reported data on allocation of housing and GMI from 2006-2008 (plotted in Figure 4.24) were to persist in 2009 and 2010, this would be a concern for the Government to investigate. 132 Figure 4.24: Housing and GMI Benefits, Share in Total Social Assistance and Extent of In-kind Delivery Ratio of total Housing Benefits to total Ratio of "In Kind" Housing Benefit to Municipal Social Assistance Benefit total Housing Benefit (2006-2008) 2006-2008 68.00% 40.00% 66.00% 35.00% 64.00% 30.00% 62.00% 25.00% 60.00% 20.00% 58.00% 15.00% 56.00% 10.00% 54.00% 5.00% 52.00% 0.00% 50.00% 2006 2007 2008 2006 2007 2008 Ratio of total GMI Benefits to total Ratio of "In Kind" GMI Benefit to total Social Assistance Benefit 2006-2008 GMI Benefit 2006-2008 6.00% 14.00% 5.00% 12.00% 10.00% 4.00% 8.00% 3.00% 6.00% 2.00% 4.00% 1.00% 2.00% 0.00% 0.00% 2006 2007 2008 2006 2007 2008 Source: Staff estimates using data reported to the MOW 287. In upper middle income and high income countries there is an emerging preference for paying social assistance in cash wherever there is an opportunity to do so. Not only does delivery of cash benefits lower administrative costs for local authorities, cash programs are far more transparent and easy to monitor, far less susceptible to procurement governance problems, and minimize market distortions. Cash also offers recipient households the greatest degree of flexibility to cope and overcome poverty or other difficult circumstances. Some long-standing arguments against cash benefits are frequently made in Latvia, but as explained in Box 4.2, these often break down under scrutiny. Where there are opportunities to do so, the Government could increase effectiveness and maybe even lower transactions costs by encouraging a shift away from in-kind programs. 133 Box 4.2. The Emergence of Cash Transfers as a Preferred Instrument of Social Assistance Countries of all levels of development and income have used a variety of instruments to compensate households for price changes, help them to manage economic shocks, and to alleviate the plight of poor families. In high-income countries, the use of cash has been more prevalent in temporary or permanent assistance programs and unemployment subsidies. In developing, transition and middle-income countries, cash transfers are less prevalent and price subsidies on the products and services that the poor consume and in-kind (mainly food) transfers are still common. However, the rationale for preferring these mechanisms over cash transfers are somewhat paternalistic and outdated, and rest on a fundamental distrust of household consumption decisions, particularly the capacity of the poorest household to make sound decisions that will lead to welfare improvements. For example, since cash is fungible, in-kind food transfers are believed to guarantee food consumption and to deter “frivolous spendingâ€? on goods or behavior society deems undesirable, such as alcohol, tobacco and gambling. However, none of these claims have been supported by strong evidence. In fact, quite to the contrary, the more traditional methods used by governments can lead to worse outcomes. Price subsidies are typically captured by non-poor groups and can result in regressive transfers. In kinds food assistance may compete with local food production and consumption and may have deleterious effects on local food production. For these reasons a shift in thinking about how best to protect and promote the welfare of vulnerable households has come about. In many countries policy makers and other stakeholders have begun to prefer using cash transfers—which, among other things, are cheaper to deliver—five or less percent against 30 and more percent for in-kind transfers—and present greater benefits that many of the in- kind transfers used. With cash transfers families are able to buy the foods and purchase the services they prefer, or put these resources to other welfare improving uses that can be very particular to unique household circumstances. The fundamental shift for policy makers is away from meddling in markets – whether directly in the form of price subsidies or indirectly through the distortions caused by in-kind transfers - and toward an approach that places greater trust in household decision making. This shift has been pushed by evidence showing that, although some “frivolous spendingâ€? can indeed exist, the overwhelming majority of households – even among the poor- make sensible, welfare-enhancing decisions. Source: Adapted from Castaneda and Lindert (2005) Labor Market and Employment Assistance 288. Prior to the crisis, in 2007, Latvia spent about 0.46 percent of GDP on labor market programs, which put the country in the middle ranking among the EU new member states from Central and Eastern Europe. However, this level is low relative to the EU27 and OECD averages of 1.7 and 1.32 percent of GDP. 134 Figure 4.25: Government spending on passive and active labor market measures as a percent of GDP in 2007 Source: EUROSTAT, LMP database 289. Similarly to other countries in Europe, Latvia provides a wide array of passive and active labor market and employment assistance to households. The main passive program is unemployment insurance, which has also been the primary automatic stabilizer in the wake of the global financial crisis, although with significant gaps in coverage (as shown earlier). In 2007, spending on unemployment insurance benefits and early retirement made up about 0.3 percent of GDP. Latvia’s expenditures on passive labor market programs as a percent of GDP are low relative to EU27 and OECD averages of 1.02 and 0.75 percent of GDP respectively. In 2009, spending on unemployment insurance benefits rose sharply to about 1.1 percent of GDP, as the financial crisis took its toll on the labor market. As a result, the stimulus (anti-cyclical) impact of unemployment benefits amounts to about 0.8 percent of GDP. Figure 4.26: Government Spending on Passive Labor Market Measures (Percent of GDP in 2007) Source: EUROSTAT, LMP database. Spending on active programs, relative to New Member States or compared to the EU27 was low in Latvia prior to the crisis. Among EU10 countries, Latvia’s spending on these programs of less than 0.1 135 percent was only higher than expenditures in Romania and Estonia. The EU27, on the other hand allocated about 0.47 percent of GDP to active programs, while OECD countries spent 1.3 percent of GDP in 2007.86 Low expenditures on active employment assistance in Latvia prior to the crisis suggest that coverage among unemployed people was limited and had a small impact on aggregate employability or competitiveness of the labor force. Figure 4.27: Government Spending on ALMPs as a (Percent of GDP in 2007) Source: EUROSTAT, LMP database. 290. Since 2004, Latvia’s expenditures on employment assistance have varied considerably, determined mostly by the Government’s decisions about how to use resources from the European Social Fund (ESF). In the period prior to the crisis, total spending on employment assistance varied from a low of LVL 6 million in 2004 to a high of LVL 21 million in 2006. 291. Employment assistance programs in Latvia have historically emphasized training. As discussed earlier, in 2009 the Government astutely shifted this emphasis as the magnitude of the contraction and the gaps in the effective coverage of unemployment assistance became apparent. The Government shifted ESF resources originally allocated for training to allow a large-volume deployment of emergency public works, in the form of the WWS. The deployment has been successful, and the current long waiting lists are more an indication of the weak labor market conditions than the performance of the program. Municipalities have shown tremendous creativity and efficiency in creating WWS positions within the parameters of the program, which have included several precautions to ensure that any economic distortions it may introduce are minimal. 292. While the emergency deployment of public works has been a success so far, the Government could consider preparing for a phased withdrawal of the WWS program when labor market conditions improve. Given the current high unemployment rates (22 percent) and small number of job vacancies created by the market87 withdrawal of the WWS is advisable only when signs of recovery are evident. There are indications that employment will start to increase in late 2010. If indeed the employment 86 OECD (2009) “2009 OECD Employment Outlook: Tackling the Jobs Crisisâ€? 87 The number of vacancies has decreased significantly from 17,662 in December 2007 to 3,205 in December 2008 and 1,389 in December 2009. 136 situation improves by late 2010, it would be reasonable to begin a phased withdrawal of the WWS program starting in mid-2011. Given the self-targeting features of the program, the Government could consider reducing the monthly stipend from the current LVL 100 to around LVL 80 per person for full time participants. This move will reduce the relative appeal of the program for job-seekers who would choose between participation with the reduced stipend or intensifying their search. Moves to reduce the stipend amount will no doubt be met with resistance, but in order to minimize economic distortions and increase the potential for private sector led growth, it is important that the Government continues with the withdrawal. In addition, the expected reallocation of ESF money from the WWS program into longer-term priorities will result in fiscal pressures that might be avoided by reducing the size of the WWS program. 293. In the meantime, several minor problems that have arisen during deployment of the WWS could be corrected. First, the program currently rewards early retirees with a pension, by providing them with a supplementary income. However, the intention of the program was to give people a “last resortâ€? safety net, not a supplementary income. Early retirees crowd out people with no other safety net. Given the length of the waiting list for the program, this move could help redirect resources to arguably needier participants. In addition eliminating eligibility of early retirees will help reduce incentives for early retirement at the margin. Second, the first “cohortâ€? of participants is reaching the end of their 6 months in WWS program, and there is a question about whether their terms in the WWS program should be extended. Given the long waiting list, extensions could hinder job creation for those that are still without any other assistance. The Government could consider allowing people who have completed 6 months in the WWS program to re-enlist in the waiting list such that the first-come-first-served principle is maintained. It might be good to indicate an upper bound for the wait time so that people at the bottom of the waiting list do not suspend their search for private or public sector jobs in the hopes of continuing in the WWS program. 294. As it looks to the medium and longer term, and plans its shift away from an emergency stance, the Government could consider structural reform to Latvia’s employment assistance programs. Given the relatively low amount that Latvia spends on active employment assistance, there are not any obvious areas of significant fiscal savings. However, there are problems in the design and implementation of programs that if corrected could increase their performance. The current training and retraining programs suffer from some administrative inefficiency and could place greater emphasis on personal responsibility. 295. Training and re-training opportunities of up to 6 months are offered to unemployed people and are designed to promote professional skills development and to help job seekers update or acquire knowledge in a new profession. For longer running programs, a certification of professional qualification is also issued upon completion. In response to the crisis, the Government, like many others in the region, significantly increased funding (primarily from the ESF) to training programs from LVL 3.2 m in 2008 to LVL 14.8 m in 2009. In 2010, the Government plans to allocate LVL 25.2 m to training programs. Training services are contracted out to private and public, MoES-accredited, vocational education 137 establishments that submit proposals to the SEA.88 A flat 70 LVL monthly stipend is paid to participants by the service provider during training. Recent surveys show that training institutes generally provide satisfactory services and that participants are generally satisfied with the training process. 296. On the supply side of training programs, currently, there is a centralized procurement procedure with services procured by the SEA central office for all 28 SEA regional affiliates. It is a time consuming procedure and has to be later administrated by staff in SEA affiliates when the unemployed are grouped for the different training courses. Competition among training providers is stifled because firms have no incentive to improve performance after they become accredited. However, there are attempts to change this trend. The SEA is adding criteria in order to make sure that training services procured are of better quality (minimum equipment requirements are met, preference is given to trainers able to offer training services for people with disabilities, etc.). In providing training opportunities, unused capacity in existing state-financed programs in secondary vocational education and higher education institutions could be better utilized. 297. Income support provided to participants may be insufficient for some and excessive for others, creating perverse incentives. The Government has rightly relaxed rules restricting job retraining to vocational specializations and lower-level training. However, paying people who receive unemployment insurance a stipend is likely unnecessary given fiscal constraints. The current stipend of 70 LVL per participant per month is provided to help unemployed people cope with low incomes into the household and also to compensate for costs associated with attending the training program. 298. The Government could move away from the current supply-driven system of job retraining for the unemployed, toward a more demand-driven process which is flexible to the types of training supported. The MOW has already taken an important step in this direction by proposing that vouchers be given to program participants. A voucher model is the preferred mechanism for managing training. Vouchers devolve to the individual the choice of training programs and location of training, and create an incentive for beneficiaries to choose the type of training that best prepares them for productive employment. Minimal restrictions can be instituted, such as: (i) requiring participants to choose from a positive (and regularly updated) list of licensed (by MoES) training providers; and (ii) requiring minimal educational qualifications for certain training opportunities. However, the strengths of a voucher model are flexibility and enablement of individual preferences, and hence, restrictions on participants’ choices are best kept at a minimum. Giving this freedom to participants would greatly simplify administrative and procurement requirements; engage market forces to weed out low quality providers; capture the benefits of competition among providers, particularly if encouraged by the SEA regularly conducting evaluations and publishing results of all training providers, giving potential participants the information base needed to make the choice that is right for them. 299. The Government could make better use of existing institutions to build and renew the skills of job seekers. Training vouchers could be used in special, short-duration retraining programs developed and 88 The split between private and public training institutions is an follows. For longer term programs – 49% are private institutions, 51% - public; for informal training – 68% are private, 30% - public; and for vouchers (with shorter working hours) – 70% private and 30% public. 138 offered by existing secondary vocational schools. MoES managed vocational training institutes do not participate in the request for proposals from PESs. The main reason is that MoES providers are apparently unable to perform the required quarterly financial planning forecasts. The Government could reform MoES’s financial planning procedures to allow eligible participants with vouchers into their institutions. Vocational secondary schools are centrally managed by the Ministry of Education and Science, but are scheduled to be transferred to local governments for management by September, 2015. With support from EU structural funds, all secondary vocational schools are developing restructuring plans that are designed to improve efficiency and introduce greater flexibility and responsiveness to evolving labor-market needs. In most cases, these plans include the development of modular training programs with input from local employers and greater administrative flexibility in contracting of service providers. These changes – including local management, greater administrative flexibility, and focused training modules – could be accelerated in order to speed the adoption of a new, demand-driven model for retraining unemployed adults alongside the vocational education programs for school-aged youth. A flexible voucher model would even extend the benefits of employment assistance beyond traditional target groups.89 300. The Government could carefully consider the rationale for and efficiency of income support provided to participants in training programs. The Government might consider providing more than 1 stipend level to program participants when participants have to travel to receive training. Implementing more than one stipend level can help remove some of the geographic constraints associated with the program. For example, some forms of training are not provided in some municipalities because of low demand from residents or because training providers are not willing/able to provide the training in that locale. Alternately, people pursuing training a short distance from their own homes could potentially receive a stipend smaller than the uniform 70 LVL. However, there are three potential risks. First, there might be a temptation to create a large number of stipend levels, which will unnecessarily increase the administrative requirements and discretion of the PESs. Second, participants who register for training in locations far from their residence will make requests for not only transportation but also residence. However, including residential costs will lead to a considerable increase in administrative requirements and to an increase in costs. Finally, there might be a tendency for people to register as unemployed in one region and then take up training in another location to get a bigger transportation stipend. 89 In the wake of the economic contraction, enrollments in fee-based part-time higher education programs have fallen sharply. One particular group of people who could be considered for emergency retraining support, through vouchers, are individuals who lost their jobs and had to drop out of these programs because they could no longer afford tuition fees. This training is often used by employed individuals to upgrade their skill qualifications for better jobs. 139 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Social Insurance (State Special Budget for Social Insurance) Lower the amount of pension 85 The unprecedented Given very high pension The flat cut in pensions in Immediate short term income exempt from taxation if exempt growth in revenues increases that resulted mid-2009, that was both gains that will help put the to a point between the amount is until 2007 translated from fiscally profligate regressive and penalized social insurance special current level of LVL 165 and lowered to 80 into a rapid growth in policies during the boom working pensioners, budget on long-run LVL 80 per month. LVL. wages and before 2008 (wage makes any measure that financial sustainable path, consequently of the indexation, new benefit for reduced pensions risky, as well as lead to a more 24 covered wage bill. This pre-1996 contribution particularly in the wake of progressive distribution of if exempt boom was reflected in years and windfall gains for the Constitutional Court welfare spending. amount is the notional interest younger pensioners decision of December 21, lowered to LVL rate applied to pension associated with abnormally 2009. 140 benefits from the NDC high real notional interest However, the pillar. The cohorts that rates) it is reasonable to recommended measure is retired prior to 2010 recover some of the different from the 2009 have benefited from a recent unsustainable cuts both in form and in windfall from the pre- increases . substance: 2007 bubble. Between A broader “taxation basedâ€? ï‚· The measure is 2005 and 2009 the measure may be further progressive, affecting average pension for justified (over a measure lower-income new retirees grew 69 targeting only recent pensioners relatively percent. retirees) as the increase in less. revenue to the state social ï‚· The measure does insurance special budget not distort work motivated other “high- incentives. costâ€? pension policies, during the high-growth, such as the indexation of pensions to wage growth. As the Government seeks further savings, the principle of “burden sharingâ€? should apply. Pensions have been relatively unaffected by budget cuts up to now; the adjustment and 140 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit contraction has been bourn primarily by people of working age. Pensioner households are also more likely to own their homes and to have fewer current expenditures (on child- rearing, schooling, etc). Targeted elimination of the 120 -- A non-contributory Extending the supplement Although the extension of Immediate short term pre-1996 service supplement supplement was to all pensioners was an the supplement to all gains that will help put the for pensions above LVL 140 introduced in 2006 for unsustainable policy. pensioners with a work- social insurance special per month low income pensioners. The Government could history prior to 1996 is budget on long-run This was extended to save from repealing the recent (2008), a sense of financial sustainable path, all pensioners in 2008. expansion of the benefit to “acquired rightâ€? may as well as lead to a more The benefit pays LVL all pensioners, and still already have solidified progressive distribution of 0.7 (70 centimes) for continue to supplement among retirees receiving welfare spending. each year of service the pensions of the 10 this benefit. As this is a accrued before 1996 percent of retirees with specific “constituencyâ€?, and on average lowest incomes which the group is likely to be amounts to 14 percent would place the benefit more able to organize of the overall pension elimination threshold at or resistance to the spending. above LVL 140. measure. Fully index pensions to 31 61 In 2009 and 2010 Alternatives exists for There is political risk of Combined with other changes in the CPI, both pension benefits are savings without changes in the measure being measures, CPI indexing positive and negative. (Alternatives thereon “frozenâ€? (i.e., not the law, however, these reversed when positive pensions will help put the discussed in the adjusted). alternatives would only inflation returns as well as state social insurance PER would However, in 2011, deliver savings when the a risk of prolonged budget on a long term deliver savings benefits will be indexed deflationary period ends. deflation which would not financially sustainable only in 2017) to the CPI. Under a lenient allow recovery of path. The current law interpretation the resources accumulated indicates that pensions indexation law, adjustment deflation for a long time. should be indexed to could be 0 in times of prices but cannot be deflation and positive reduced in nominal when CPI is positive. Under terms. The savings the more stringent shown here assume interpretation the 141 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit automatic indexation downward adjustment of to the CPI whether the pension is not allowed positive or negative but upward adjustment and would require only starts when the additional legislative purchasing power of the action. pension starts to fall behind its original purchasing power. Whether the more stringent application of the current law is possible would have to be investigated further. Application of true “G valuesâ€?, 2 4 At the point of retiring, New retirees are expected Application of the G For long term financial without any change in the an affiliated workers to be hit by negative values will lower benefits sustainability, the current law [6 million in accumulated “notional notional interest in 2010 awarded in 2010 and automatic parameters of 2013] capitalâ€? is divided by an and 2011 of -4 percent and 2011. the NDC formula should be estimate of life- -10 percent respectively. allowed to function. expectancy. In the NDC This is why this year formula, the life regular increase in G values expectancy parameter amounting to 2 percent is referred to as the “G decrease in pensions was valueâ€?. not applied. This is being done at the discretion of the authorities, and is not reflecting the automatic application of the NDC parameters as intended. Retirement age increase -- -- The early retirement Among the cohorts turning It is frequently argued An increase in the starting in January 2015 (Savings in option is expiring in 62 in 2012 and 2013, that raising the statutory and effective the medium 2011 – which given around 70 percent will retirement age, retirement ages is essential term extremely high “pre- already be receiving old particularly in a to the long-run financial expected.) retirement ratesâ€? age or disability benefits. recessionary period (or sustainability of the 142 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit through early This defeats the purpose of when the labor market is system, and help raise retirement and trying to increase very slack) exposes older replacement rates, as, all disability programs, retirement ages before workers to greater risk of else equal, affiliates’ would translate into 2014. Furthermore, the unemployment. notional capital will be very few old age cohort turning 62 in 2014 However, this concern larger and is spread over benefit applications in will already experience a seems misplaced. fewer years in retirement. 2012-2013. significant increase in the Workers with retirement An increase in the effective retirement age coverage who would be retirement age earlier (70 percent of previous affected by an increase in than January 2015 will cohort retired early, while the retirement age, are probably precipitate a this cohort will have to also more likely to have faster “rush to retireâ€? wait until 62), so it is not coverage of that could destabilize equitable to increase their Unemployment the state social burden further. Thus, the Insurance, as well as insurance budget. most logical time to start relatively greater job increasing retirement age security (than younger is 2015. cohorts). In the current The MoW presents a contraction, younger proposal to begin this workers have suffered increase in 2016, based on unemployment at expected unemployment relatively higher rates until then and a desire for than older workers. parity with increases in the retirement age in Estonia. Replace the earnings-related -- 65 New parental benefit Given that no additional Arguments have been Immediate short term parental benefit with an was shifted into the social insurance presented that this gains that will help put the augmented flat Childcare The law requires (Assumes social insurance special contributions were levied measure will deter social insurance special benefit of LVL 100 (paid from 306 day Childcare budget in 2008. in 2008, the recent shift of higher-income budget on long-run State basic budget: see social transition state social The parental benefit is the parental benefit households from having financial sustainable path. assistance section of this period from the benefit is earnings-related, and spending to the social children. Eliminates duplication of matrix) date of passage raised to LVL paid to those with a insurance budget is However, it is not clear benefits, and leads to a of any measure 100) history of inappropriate. from international more progressive that changes contributions. Furthermore, as with all evidence that social distribution of social benefits related However, when it was other “non contributoryâ€? welfare benefits are welfare spending. to child bearing. introduced, no increase benefits, it should not be successful in inducing in contribution revenue linked with the recipients families to have children. 143 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit to the state social earnings, which otherwise Nor is it clear why the insurance budget was introduces regressive State should subsidize made. The benefit is, redistribution to people childbearing of wealthier for all intents and with higher incomes. A flat households at a relatively purposes, a non- Childcare benefit is already more generous rate than contributory benefit, granted from the State it does to poorer reflecting the fiscal basic budget for the same households that are not profligacy of the boom social welfare objective. covered by the state period. As part of this measure, social insurance special the flat benefit could be budget. raised to its 2005 level, LVL 100. 90 Increase flexibility of 2 2 Contributory earnings The Government could -- The current structure of maternity benefit, allowing related maternity consider introducing the maternal benefit could expecting mothers to decide benefits are paid for greater flexibility to the be made more flexible to to postpone the the duration of 4 maternity benefit. Rather offer additional choices, commencement of the 4 months starting with than requiring women to and produce savings for th month benefit period by a few the 7 month of exit the labor force at 7 the Government in the first weeks or even up to the time pregnancy with months of pregnancy, a year of implementation by of birth. average benefit of LVL woman could be allowed adding additional 1,234. Subsequently, to decide to postpone the taxpayers which would an earnings-related commencement of the 4 increase social insurance parental benefit month benefit period by a and general tax revenues. (discussed above) is few weeks or even up to paid for those covered the time of birth. by social insurance. International practice suggests that many women would prefer this flexibility and choose to remain active, although individual decisions would depend on 90 This estimate is based on a conservative assumption that women will decide to shift the date of benefit commencement by an average of 2 weeks. Under that scenario, the savings for the social insurance budget in the first year of implementation would amount to LVL 1.4 million. In addition, increasing the flexibility of the maternal benefit would add additional taxpayers which would increase social insurance and general tax revenues, adding another 0.6million in combined revenue raising total savings to LVL 2 million. 144 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit doctor’s opinion, the type of employment, and particular economic circumstances of the family. State Social Benefits Target Family State Benefits -- -- Family state benefits FSBs have a small impact Some political resistance A stronger, more robust (FSB) to poorer households constitute about 37 on non-poor households is to be expected. and responsive safety net with children. The Not clear if the Given how percent of state social and are more symbolic However, anecdote for the poorest Government could achieve a 306-day fair little Latvia benefits. FSBs are than welfare enhancing. suggests a many households, particularly higher social return from its notice spends on monthly payments of For example, households households in the those with children. social welfare spending, by requirement social LVL 8 to all families in quintiles 4 and 5 spent wealthier segments of Keeping family benefits reducing coverage of FSB for which applies to assistance, with a child between about 2,300 and 4,200 LVL society do not bother to universal is a social choice non-poor households, but benefits related any, and all the ages of 1 up to 18 per person per year in collect the FSB they are with explicit fiscal maintain assistance for poor to child birth, is potential years of age. 2008 and hence the added entitled to. implications. The current households. relevant – this savings from 72 LVL per child per year transfer amount has a Means-testing all state social requires restricting has little impact on minimal impact on non- benefits (i.e., those financed consultation FSB– which household welfare. In poor households The 91 from the national basic budget with legal 17.3 million quintile 3, whose yearly government needs to for the Welfare sector) would counsel.) – should be per capita household weigh the benefits of be advisable – with the re-distributed consumption is almost LVL universality, with the fiscal exception of benefits related to strengthen 2,000. However, the fiscal implications of providing a to disability. targeted implications of the benefit to households FSB could be the best benefit social program are significant. whose welfare is hardly to start with, given its currently assistance improved by the transfer. wide coverage; large volume programs. (and thus potential savings); and that the public is “sensitizedâ€? to reform of this benefit (earlier public discussion of possible reform). 91 This estimate is made by estimating the amount of FSB being received by 60% wealthiest households (consumptions quintiles 3-5) in the 2008 HBS. The SILC survey may provide a more accurate estimate of the benefits received by households. However, any survey is only a reported “reflectionâ€? of what is actually being paid. The Government can arrive at a far more precise figure by cross checking administrative data on payment of FSB with reported income held at the Ministry of Welfare, or even at the State Revenue Service. 145 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Restore Childcare Benefit to a -- -- Currently the The Parental Benefit is No anticipated risks of Reverting to a single, flat flat amount of LVL 100 (see (see Parental coexistence of the essentially non- raising Child benefit, if Childcare benefit will lead above recommendation on Benefit earnings-related contributory. Once it is Parental Benefit is considerable fiscal Parental Benefit) segment on Parental Benefit transferred to the State eliminated. savings, and a more State Social alongside the flat basic budget, it is Political resistance to be progressive distribution of Insurance) Childcare Benefit inequitable and unjust to anticipated from social welfare spending. augments household maintain an earnings- households covered by income and welfare related benefit in parallel social insurance, as the inequality, and imposes to a flat benefit that have earnings related Parental a high fiscal cost. the same social welfare Benefit is considerably objective. higher than the flat Child benefit, even with an increase to 100 LVL. Municipal Social Assistance Increase co-financing of GMI -- -- GMI and Housing The co-financing The budget “reprieveâ€? on Raising co-financing and and Housing benefit: Keeping benefits are increasing introduced in the ESSNS municipalities from eventually recentralizing the amounts of mandatory [14 million in importance as (50 percent of spending on greater financing of financing for mandatory targeted assistance benefits increase, safety-net instruments, GMI and 20 percent of mandatory benefits from targeted benefits from the constant, increase co-financing based on and appear to be spending on housing national budget, could State budget, will ensure a share from State budget for anticipated reasonably well benefit) helps mitigate the also lead to perverse robust, uniform safety net GMI and Housing benefit to 75 spending in targeted to the lowest problem of perverse outcomes: for the lowest-income, percent. 2010, income groups. incentives, particularly in (i) moral hazard needy households financed However, there are still the face of higher demand (municipalities awarding wherever they happen to from improvements in for benefits from the crisis. GMI and housing benefit live. targeting FSB, targeting accuracy that Savings from targeting with less care for means as above] could be achieved. Family State Benefits to testing); The mismatch between households in quintiles 1 (ii) diverting social a national mandate and 2 (as above) could be assistance spending to and local financing used to increase the share benefits that don’t benefit causes perverse of co-financing of the poorest households. incentives at the local mandatory benefits from These risks can be level, and disparities in the State budget, and still mitigated by instituting provision of these leave room for overall more rigorous monitoring mandatory benefits fiscal savings. of municipal compliance 146 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit across wealthy and As the Latvian Welfare with GMI and housing poor municipalities. system does not yet have benefit mandates; For this reason few EU the information targeting; and quality and OECD countries management tools to requirements (something retain decentralized monitor targeting that the MoW is planning financing of mandatory performance of to do anyway). targeted benefits. municipalities, less-than- full State financing may still be appropriate. Employment Assistance Shift training to a demand- -- -- There are three main Vouchers devolve to the Lack of competition Vouchers will have several based model. The Government Efficiency gains Efficiency inefficiencies of the individual the choice of between providers, and benefits over the status could continue its shift away at the margin. gains at the current training training programs and lack of information quo: from the current supply-driven margin. program: location of training, and among voucher recipients (i) Administrative system of job retraining for the (i) Administrative create an incentive for can reduce the benefits of requirements stemming unemployed, toward a more requirements on the beneficiaries to choose the this shift. This risk can be from procurement of demand-driven process by State Employment type of training that best mitigated by measures to training services for each expanding voucher use by Agency (SEA) to prepares them for improve competitions course by the SEA can be program participants. procure training productive employment. It among providers and greatly simplified. services is high; allows participants to information of those (ii) Market forces could (ii) Failure to give choose the training seeking training. weed out low quality freedom to discipline, the firm to licensed training providers participants to choose provide the training, the through low demand for their training courses location of the training their services. This is can lead to sub-optimal institution, etc. Minimal important because under personal choices; restrictions can be placed, the current regulations, (iii) Competition among such as: (i) requiring the SEA is not able exclude training providers is participants to choose low quality training stifled because firms from a positive (and providers who have been have no incentive to regularly updated) list of licensed by MoES even if improve performance licensed (by MoES) training additional quality criteria after they become providers; (ii) requiring are set during the accredited. minimal educational procurement process. qualifications for certain (iii) Greater competition 147 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit training opportunities. among training providers However, the strength of could be fostered if the vouchers is flexibility and SEA publishes evaluation hence, restrictions on results of all training participants should be providers. This publication minimized. can also give potential participants the information base needed to make career choices. Allow vouchers to be used for -- -- MoES managed It is not cost-effective for -- Besides reducing unused all training options. Training Efficiency gains Efficiency vocational training the Government to finance capacity in MoES managed vouchers should also be usable at the margin gains at the institutes do not new, tailored programs for vocational training in special, short-duration margin participate in the the unemployed when institutions, including retraining programs offered by request for proposals there is significant unused these institutions will lead existing secondary vocational from PESs. The main capacity in some existing to increased competition schools, including those reason is that MoES state-financed programs in among service providers managed by MoES. providers are not able secondary vocational and hence increase to perform the education and higher training quality and required quarterly education programs that relevance. financial planning are explicitly or implicitly forecasts. ignored. Although the recent financing reforms reduced excess capacity, considerable excess capacity remains. Eventually reduce stipends paid -- -- Income support Lowering the stipend could Withdrawing stipends too Lowering the stipend to training recipients. Efficiency gains Efficiency provided to those who ensure that training soon (before the labor amount will lead to a 148 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Although appropriate during and budget gains and participate in training program participants are market starts to recover) reduction in the waiting the current contraction, savings budget programs may lead motivated by human could increase the list for training programs government should consider savings people to request the capital accumulation number of needy job- and will lead to more take reducing the stipend paid to program for the objectives rather than the seekers waiting for up of training programs by training participants to reduce financial rather than safety net objective of limited WWS workplaces. people interested in the take up by those who want a the human capital training programs. The courses rather than people safety net rather than a human accumulation motive. Government might taking the course with the capital investment. consider providing more least transaction costs than 1 stipend level to associated with the LVL 70 program participants when stipend. participants have to travel to receive training. Maintain Workplace With -- -- The WWS program If labor market conditions Withdrawing the WWS Phasing out the WWS Stipends (WWS) until labor formed a crucial part of begin to improve in mid- too soon (before the program will be important market recovers. The the emergency social 2011 as forecasted, it labor market starts to to minimize economic Government could begin safety net to mitigate would be reasonable to recover) will increase the distortions and increase preparations for a phased the impact of the crisis begin withdrawing the number of people seeking the potential for private withdrawal of the WWS on households. The WWS program in the third GMI and other forms of sector led growth. In program when labor market design features of the quarter of 2011. A minimal social assistance. addition, the expected conditions improve. program meet best- credible exit strategy for reallocation of ESF money practice conditions for the government can be away from WWS into public works programs through stipend reduction longer-term priorities will around the world. The (to around LVL 80 per result in fiscal pressures program was rolled out person per month for full that might be avoided by in record time and time participants). This minimizing the size of the targeted individuals move will reduce the WWS program. who do not receive relative attractiveness of unemployment the program for insurance benefits. unemployed people who In 2009, LVL 8 million would be faced with the were allocated to the option of choosing the WWS program. program with the reduced Continuing soaring stipend or intensifying unemployment their job search. 149 Matrix 2. Options for the Government to Consider in the Social Welfare Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit numbers have forced the government to allocate LVL 26.7 m to the WWS program in 2010. Restrict entry of early retirees -- -- The program currently Early retirees have made a In addition to the benefits in Workplace With Stipends rewards early retirees, choice to retire and should of reallocating resources (WWS). The Government who receive a pension, not have access (and away from people who are could exclude early retirees by providing them with thereby crowd out people already receiving pension who are receiving a pension a supplementary with no safety net) to the to people with no safety from the WWS program. income. However, the WWS program. Given the net, eliminating eligibility intension of the length of the waiting list of early retirees will help program was to give for the program, this move reduce the incidence of people a “last resortâ€? will help redirect resources early retirement. safety net, not a to more deserving supplementary income. participants. 150 Chapter 5: Education I. Introduction and Overview 301. Up to the implementation of structural reforms in 2009, Latvia’s education system was characterized by high coverage but low efficiency, caused by demographic change since the start of the transition. Decentralization and diversification of financing had led to greater inequality in learning achievement across municipalities and schools. Competition in the global economy has changed skill requirements in ways that make many education programs less relevant to today’s labor-market needs. The main driver of inefficient spending was the cost of teaching and non-teaching staff. 302. A previous PER, conducted in June, 2007, documented the rapid growth of budget expenditures for education programs and the problem of overstaffing in schools, and recommended a number of fundamental, structural reforms to help contain budget expenditures and improve efficiency of education provision. The key reform measures recommended to the Government in 2007 were to: ï‚· Create the means for improved efficiency in school management by relaxing central norms on class size and allowing school directors and local authorities flexibility in managing education resources, ï‚· Create the incentive for improved efficiency and quality by revising the basis of budget financing of primary and general secondary education, so that resources finance outputs (students) rather than inputs (teachers and schools), and ï‚· Adopt similar reforms -- flexible management and per student financing– to improve efficiency and responsiveness of vocational secondary education and higher education to changing skill needs in the labor market. As documented earlier in this report, in 2008 the global financial crisis hit Latvia, reversing a decade of double-digit economic growth and creating severe strains on government budgets. This event added urgency to the containment of public spending in the education sector, and the Government moved decisively to implement cost-containment measures. A plan for per-student financing of primary and general secondary schools was developed in 2008 and early 2009, and put into effect in September, 2009 at the start of the 2009/2010 school year. Implementation of per-student financing is being managed by the 118 new units of local government that were established in July, 2009 as part of a difficult reform of territorial administration.92 Cost-reduction measures were also adopted in vocational secondary and higher education. These are major accomplishments, which are leading to improved efficiency in management of education resources, and will yield further efficiency gains in the future. 92 Local government under the administrative reorganization of July 1, 2009 comprises 9 republican cities and 109 municipalities (novads). For further details on the administrative reorganization, see Inga Vilka, “Local Governments and their Finances in Latviaâ€?, World Bank consultant report, April 19, 2010. 151 These structural reforms are already showing promise of delivering efficiency gains, and the likelihood of improving performance, although it is yet too early to observe measureable improvements. But given the extent and speed of reductions in financing to the education sector, no further substantive savings can be realistically expected in the immediate term. The structural reforms that have been initiated need time to take their full effect. For this reason, no further major reductions in the budget to the education sector are advised at this juncture. 303. However, there are some further reforms that would help full implementation of the structural reforms initiated in 2009, and in doing so, deliver minor but significant budget savings in 2011 and 2012. The Government could continue further its progress in implementing the new financing model by providing more guidance to municipal authorities and to school administrators on options for application of per-student financing, based on the lessons of implementation during the first year of the new financing scheme. Many novads still are not aware of all the options available to them to achieve better efficiency – such as transporting teachers (rather than students) between schools. With more guidance from MOES (and knowledge-sharing between municipalities organized by the Ministry) local authorities are likely to be able to exploit more fully the flexibility availability under per-student financing. 304. The per-capita formula itself is coming under the scrutiny of academic specialists and experts interested in seeing further efficiency improvements. Even after just one year of implementation, already the data are available for researchers to evaluate the impact of the structural reform of education financing, to identify problems and simulate changes to the parameters of the formula. This is a welcome development. As in other upper middle- and high-income countries with plenty of local expertise, the Government of Latvia could benefit from encouraging a continuous transparent discussion of how the formula can be improved. 305. To support the further implementation of the new per-capita financing model, it is important to underscore that although low teacher salaries have been a threat to education quality in the past, teacher earnings are now at a level that is generally comparable to earnings in other sectors of public service, although this important balance needs to be closely monitored. Historically, teachers in Latvia have earned less than workers with equivalent qualifications in other occupations, raising concerns for the quality of teachers that can be retained at relatively low earnings. The gap between reported average earnings in the education sector and comparator sectors narrowed in 2009, reflecting the combined effect of salary increases adopted in 2008 and the first of the salary reductions initiated in 2009. As a consequence of the major reduction in the earmarked transfer for primary and general secondary education, there were cuts in all components of teachers’ earnings during the last half of the 2008/2009 school year and the beginning of the 2009/2010 school year. Out of concern that the speed and depth of these cuts could further threaten the quality of teachers, in January, 2010, central budget transfers to local governments for teacher salaries were increased by 37 percent, with no additional teaching requirement. This increase was intended to finance the restoration of payments to teachers for teaching-related tasks such as class preparation and follow-up and correction of exams. As of February, 2010, average monthly earnings of employees of state budget institutions were LVL 500 and falling, whereas in April, 2010 average monthly earnings of teachers were LVL 476 and rising. 152 306. In order to preserve a positive incentive to enter the teaching profession, the Government could maintain the January, 2010 salary increase for primary and general secondary education. This would help maintain the recently attained balance in earnings between teachers in primary and general secondary schools and people with similar qualifications who are working in other sectors. The Government would have to closely monitor teacher earnings to maintain general equivalence with other professionals working in the public sector, which is a task made very difficult by the antiquated “piece workâ€? model of teacher remuneration still in place. If further salary increases for teachers in the immediate future are being considered, moderating these increases to maintain this earnings comparability across sectors could yield significant savings. 307. Eventually, MOES could move toward a more modern, full-time concept of remuneration in the teaching profession based on learning outcomes rather than time inputs for specific tasks. In such a model teaching responsibilities would be defined to include a broad array of teaching-related tasks that support student learning, including lesson preparation and follow up, correction of exams, collegial support, and meeting with students and parents. This approach would be more consistent with the output-oriented financing reforms, and would provide more explicit incentives for teachers to focus on learning outcomes. Adoption of a new, full-time concept of the teaching profession based on learning outcomes will require preparation of the methodology and further strengthening of assessment instruments. It would be easier to apply this concept once the economic recovery has allowed the restoration of more adequate financing for base salaries and other school inputs for effective teaching and learning. 308. This chapter of the PER also includes a review of vocational secondary education and higher education. For these segments of the sector, although there are no obvious areas for budget savings to be found in the immediate term, several suggestions are made to improve efficiency and performance. II. Public Expenditure on Education 309. In 2007, just prior to the financial and economic crisis, Government expenditures on education accounted for 10.6 percent of total government expenditures and 4.4 percent of GDP, as shown in the first column of figures in Table 5.1. Both figures were well below the EU-19 averages of 12.2 percent of total government expenditures and 5.4 percent of GDP, as well as the OECD averages of 13.3 percent of total government expenditures and 5.3 percent of GDP and (Figures 5.1 and 5.2). Low teacher salaries were one of the reasons for the modest level of education expenditures. 153 Table 5.1: Evolution of Public Expenditures on Education 2007–2010 by Expenditure Components, in millions of current LVL % % % 2007 2008 2009 2010 Change Change Change Central Government education expenditures, all ministries, 524.8 23.7% 649.3 -14.4% 556.0 -17.2% 460.1 including earmarked subsidies to local governments Earmarked education transfers to 251.0 27.6% 320.2 -16.6% 267.1 -18.9% 216.6 local governments Central Government education expenditures, all ministries, not 273.8 20.2% 329.1 -12.2% 289.0 -15.7% 243.5 including earmarked subsidies to local governments Ministry of Education and Science 307.9 14.8% 353.7 -35.0% 230.0 -15.6% 194.2 budget Local government expenditures on education from all sources, 371.0 21.4% 450.4 -15.3% 381.5 NA NA including earmarked subsidies to local governments Local government expenditures on 120.0 8.5% 130.2 -12.1% 114.4 NA NA education from own sources Total Government education 644.8 20.9% 779.5 -14.0% 670.4 NA NA expenditures Total government expenditures, 6,091.7 20.9% 7,363.6 -10.4% 6,599.6 NA NA all sectors GDP 14,779.8 10.1% 16,274.5 -18.6% 13,244.3 -8.7% 12,089.0 Government education expenditures as % of total 10.6% 0.0% 10.6% 134.8% 10.2% NA NA government expenditures Government education 4.4% 9.8% 4.8% 75.1% 5.1% NA NA expenditures as % of GDP Source: GDP figures from IMF World Economic Outlook database, including projected GDP for 2010. Data for local government education expenditures for 2008 and 2009 from MOES school-level data on actual expenditures. (Excel file “ Local Government Education Expenditures by School and Source, 2007 and 2008.â€?) MOES budget data from MOES Budget Department. All other data from Treasury Data System. (Tab 5, Excel file “Components of Education Expenditure, 2007-2010â€?) 310. In late 2007, the Government began preparation of a new per-student financing mechanism that was designed to bring about more efficient staffing of schools. The Government also made a commitment to increase teacher salaries starting in 2008. Education budgets also rose in other expenditure areas in 2008. Government expenditures for education grew by 20.9 percent in nominal terms in 2008 -- the same as the average increase in government expenditures in other sectors – and 4.8 percent in real terms (paragraph 14 and Table 5.1). As a result, the share of education expenditures in total government expenditures remained constant, at 10.6 percent. Government expenditures on education increased as a percentage of GDP, from 4.4 percent in 2007 to 4.8 percent in 2008. 154 Government Education Expenditure as % of Total Government Figure 5.1: Public Spending on Education as percent of Total Public Expenditures, Expenditure Latvia and OECD Comparators 20% 15% 10% 5% us ) un 9) Es CD Fi ium B - 19 La Rep ny ak o r k A 008 (2 c 7 , ce lic he om an S and EU d d nd Ja ly en a te or a U I nia ep y ra n ( 2 Fr a y ze e n N ing al Po ds K s O n ov N ar ia li R wa D e ni P tri n r e H 00 de de C G pa tio Ita K g a 00 n ga or tv ub te la ub at n la m E to et d d tu ch rm g 2 nl Fe we rla ni r e St el ia tv si Sl U us La R Source: Data for Latvia from Table 5.1. Data for other countries from Table B.4.1, Education at a Glance, 2009, OECD. Reference years for Latvia as shown; 2006 for OECD countries. Tab 3, Excel file “Government Expenditures by Levelâ€? 311. The first three columns of Table 5.1 summarize the financing components of the 2008 expenditure increase. This rise in education expenditures was financed from three sources: (1) a 20.2 percent increase in central budget allocations for education activities – largely, but not exclusively for activities managed by the Ministry of Education and Science (MOES); (2) a 27.6 percent increase in earmarked transfers to local governments for education; and (3) an 8.5 percent increase in the expenditures of local governments for education from their own resources (not including earmarked transfers to local governments for education). 312. The fact that the 2008 teacher salary increases came while the new per-student financing measures were still in preparation meant that the salary increases were applied to a still-overstaffed network of primary and general secondary schools. Indeed, as described below, the number of teachers in primary and general secondary schools actually increased between the 2007/2008 and 2008/2009 school years, despite declining enrollments.93 The number of budget-financed places in higher education also increased despite declining total enrollments. Second, and more importantly, it coincided with the end of Latvia’s decade-long economic boom. The increases in teacher salaries and other areas of education expenditure appeared affordable amidst double-digit economic growth. But 2008 turned out to be a critical watershed – the arrival in Latvia of the global financial crisis and the abrupt end of the economic boom. The increase in teacher salaries in 2008 was short lived: by 2009, restoration of fiscal stability and expanded safety-net measures became the overriding budget priorities. 93 Significant efficiency improvements began in September, 2009 as central constraints on staffing were eliminated and as local governments assumed responsibility for school management under the new per-student financing scheme, described in later sections of this chapter. 155 313. As shown in Table 5.1 and Figure 5.2 (and further described below), central government expenditure on education fell by 14.4 percent in nominal terms in 2009, reflecting the first wave of budget cuts that were made in the effort to restore fiscal balance. Local government expenditures on education fell less steeply, helping to buffer the effect of falling central government outlays on education. Government expenditures on education fell by 14.0 percent in 2009 – less than the decline (of -18.6 percent) in GDP and more than the decline (of -10.4 percent) in total government expenditures. As a result, government expenditure on education declined as a share of total government expenditure from 10.6 percent in 2008 to 10.2 percent in 2009, and increased as a share of GDP from 4.8 percent to 5.1 percent. 314. Table 5.1 and Figure 5.2 summarize the changes in components of education spending in the final year leading up to the crisis, and in the subsequent two years: ï‚· Central ministry expenditures for education (net of earmarked transfers), which include financing for secondary vocational and higher education, increased by 20.2 percent in 2008, but fell by 12.2 percent in 2009 and by another 15.7 percent in 2010.94 ï‚· Earmarked education transfers to local governments, which cover the core financing for teacher salaries in preschool, primary, and general secondary education, increased by 27.6 percent in 2008 but fell by 16.6 percent in 2009 and a further 18.9 percent in 2010, necessitating a major rollback in teacher salaries (discussed further below). ï‚· Education expenditures from local governments’ own resources rose by 8.5 percent in 2008, but fell by 12.1 percent in 2009. Figure 5.2: Evolution of Components of Public Financing of Education, 2007-2010 in Millions of Current LVL 600 500 400 300 200 100 2007 2008 2009 2010 Education expenditures by Central Ministries Earmarked education transfers to local governments Local government expenditures from own sources MOES budget net of transfers Source: Figures from the MOES (“MOES Budget 2003-2010â€?) and text Table 5.1. 94 For higher education, Figure 4.2 somewhat overstates the extent of the financing decline in 2009 because the income of autonomous (but still publicly funded) higher education institutions revenues from paid services, facilities rental, and other activities was included in budget allocations up to and including 2008, but not since then. This change was introduced in part to address a government concern that the prior prevailing practice of reducing budgets of higher education institutions by the entire amount of their own income undermined the motivation to generate additional income from service provision. 156 315. Thus, all central-government components of education spending expanded at about the same rate in 2008 and contracted at about the same rate in 2009 and 2010. Local-government expenditures played a moderating role – rising less than central-government expenditures in 2008, and falling less than central-government expenditures in 2009. The budget of the Ministry of Education and Science (MOES) rose less in 2008 (by 14.8 percent) and fell more steeply in 2009 (by -35.0 percent) than central government education expenditures and earmarked education transfers, and declined by another 15.6 percent in 2010 (although education expenditures by the MOES are themselves a component of education expenditures by central ministries, its budget is shown separately in Figure 5.2). The MOES budget includes financing for vocational secondary schools (which it manages directly) as well as financing for higher education institutions (which are autonomously managed). MOES expenditures can exceed total central ministry education expenditures (as they did in 2007 and 2008) because the MOES budget includes several activities, such as scientific research and sports that are not classified as education in the chart of public accounts. Other central ministry education expenditures support the cost of running the specialized schools which are managed by other sector ministries – including the nursing schools under the Ministry of Health, the agriculture schools under the Ministry of Agriculture, and the arts, music, and dance academies under the Ministry of Culture. MOES expenditures have averaged about 80 percent of central ministry expenditures on education over the past five years, and comprise 85.7 percent of 2010 planned expenditures.95 95 Information provided by MOES Budget Department. 157 Table 5.2: Basic Budget Evolution for Ministry of Education and Science Programs, 2007-2010 (current LVL) Program Description 2007 2007 2008 2008 2009 2009 2010 plan amendments plan amendments plan amendments plan II General education 5,170,555 5,365,939 5,990,488 6,160,861 5,757,331 3,433,362 1,693,459 Vocational education 53,262,678 56,410,509 59,644,931 61,417,175 64,662,488 48,583,355 41,737,547 Higher education 104,973,123 112,906,369 121,877,262 127,073,144 87,486,340 53,922,894 43,367,773 National Language Policy and Administration 2,109,247 2,304,823 1,976,397 1,946,243 961,707 674,976 466,280 Science 46,040,140 39,073,806 43,923,464 42,531,592 37,267,560 21,988,797 16,532,657 Sector Management 6,713,867 6,880,132 7,076,540 7,896,982 5,190,194 3,657,614 2,579,343 IT development and maintenance, training, etc. 1,362,853 879,004 377,564 International Youth Program Agency 2,171,508 Sports 23,528,726 26,867,452 27,563,159 27,585,643 27,905,963 17,777,177 13,999,825 Free School Meals Grant 1,648,869 2,264,018 Higher Education Council 112,419 112,419 120,453 113,382 103,027 80,587 58,173 General education quality 1,244,250 1,313,989 1,313,031 1,339,191 2,112,188 Increasing teacher salaries 26,000,000 0 33,500,000 0 Interest education 602,059 602,059 646,397 654,226 562,748 Participation in EU Structural Funds projects 15,623,052 29,017,249 9,774,894 30,332,830 5,581,052 11,332,500 Academic Program Agency 15,480,708 8,020,708 10,664,708 10,622,469 11,482,098 11,361,486 National Education Development Agency 15,935,394 18,372,287 22,044,128 27,273,673 8,628,176 24,689,280 Support for education 6,806,000 7,121,842 4,414,800 3,531,840 European Economic Area Financial Mechanism 141,777 694,624 710,665 1,260,681 and Norwegian Financial Mechanism State agencies and their activities 2,668,766 4,085,117 Other 684,577 306,960 66,719,861 MOES Total 316,796,218 307,932,318 353,063,629 352,763,877 264,189,190 209,969,656 194,202,617 Source: Treasury data system. Excel file “Budgets for Education, 2007-2010â€? 158 316. The reductions of the MOES budget in 2009 and 2010 occurred in stages (Table 5.2). The 2009 total plan budget for the Ministry was 25 percent lower than the prior year’s amended budget. This budget was itself amended later in the year to a level 40 percent below the amended 2008 budget. The 2010 plan budget made further reductions, bringing the total MOES basic budget to a level 45 percent below the amended 2008 MOES basic total. Some education subsectors experienced even greater reductions – particularly in higher education, where the amended 2009 budget was 58 percent below the amended 2008 budget, and the 2010 budget fell even further to just one-third of the amended 2008 budget.96 317. The contraction of government expenditures for education in 2009 further lowered the share of education spending in total government expenditure in Latvia by comparison to other countries: Government spending on education as a percentage of total government spending was 10.2 percent in 2009, versus a mean of 12.2 percent for the EU-19 for the most recently available year and 13.3 percent for the OECD.97 Because of the precipitous fall in GDP in Latvia in 2009, government expenditure for education increased slightly as a percentage of GDP in 2000, from 4.8 percent to 5.1 percent. Even so, government spending for education in Latvia, expressed as a percentage of GDP, remains below the average expenditure levels for the OECD and the EU-19 for the latest available year, where government education expenditures averaged 5.3 percent and 5.4 percent of GDP, respectively (Figure 5.2). 318. Because of price changes, nominal expenditures (in current LVL) do not accurately reflect the magnitude of expenditure changes in terms of purchasing power. Prices increased by an average of 15.3 percent in 2008, and fell by an average of 1 percent in 2009 and 5 percent (projected) in 2010.98 When prices are rising, nominal expenditure comparisons overstate the impact of rising expenditures on purchasing power, and understate the effect of falling expenditures on purchasing power. When prices are falling, nominal expenditure comparisons understate the impact of rising expenditures on purchasing power, and overstate the effect of falling expenditures on purchasing power. Table 5.3 and Figure 5.3 illustrate the effect of changing prices on components of education expenditure. Real government expenditures on education, expressed in constant-price 2003 LVL, increased less in 2008 (when nominal expenditures and prices were both increasing) and decreased less since 2008 (when nominal expenditures and prices were both declining) than shown in Table 5.1 and Figure 5.2. In terms of purchasing power, government expenditures on education increased by 4.8 percent in 2008 (versus the 20.9 percent increase in nominal expenditures), and declined by 13.3 percent in 2009 (versus the nominal decline of 14.0 percent). The MOES budget actually declined slightly in real terms in 2008, and fell by 34.5 percent in 2009. 96 The MOES budget figures shown in Table 3 and reflected in Figure 2 somewhat overstate the actual decline in government expenditures on education in 2009 because higher education institutions’ own earnings from service provision, facilities rental and other activities were included in the higher education budget allocations up to and including 2008, but not thereafter. Own earnings of higher education institutions amounted to 40.1 million LVL in 2008. If this amount is subtracted from the MOES budget allocation for 2008, it implies a budget reduction of 26.7 percent (rather than 35.0 percent) in 2009. 97 Same sources as for Figures 4.1 and 4.2. 98 Figures from IMF World Economic Outlook database, April, 2010. 159 Table 5.3: Evolution of Public Expenditures on Education 2007–2010 by Expenditure Components, in Millions of Constant-Price 2003 LVL % % % 2007 2008 2009 2010 Change Change Change Central Government education expenditures, all ministries, 336.8 7.3% 361.2 -13.7% 311.6 -12.9% 271.3 including earmarked subsidies to local governments Earmarked education transfers to 161.1 10.6% 178.1 -16.0% 149.7 -14.7% 127.7 local governments Central Government education expenditures, all ministries, not 175.7 4.2% 183.1 -11.5% 161.9 -11.3% 143.6 including earmarked subsidies to local governments Ministry of Education and Science 197.6 -0.4% 196.7 -34.5% 128.9 -11.1% 114.5 budget Local government expenditures on education from all sources, 238.1 5.2% 250.5 -14.7% 213.8 NA NA including earmarked subsidies to local governments Local government expenditures on 77.0 -6.0% 72.4 -11.5% 64.1 NA NA education from own sources Total Government education 413.8 4.8% 433.6 -13.3% 375.7 NA NA expenditures Total government expenditures, 3,909.1 4.8% 4,096.0 -9.7% 3,698.3 NA NA all sectors GDP 9,484.4 -4.6% 9,052.8 -18.0% 7,421.8 -4.0% 7,128.5 Source: Table 5.1 data, plus GDP deflator from IMF World Economic Outlook database, April, 2010. (Tab 5, Excel file “Components of Education Expenditure, 2007-2010â€?) Figure 5.3: Evolution of Components of Public Financing of Education, 2007-2010 in Millions of Constant-Price (2003) LVL 250 200 150 100 50 0 2007 2008 2009 2010 Education expenditures by Central Ministries Earmarked education transfers to local governments Local government expenditures from own sources MOES budget net of transfers Source: Same as for Table 5.2, plus GDP deflator from IMF World Economic Outlook database, April, 2010. 160 319. Local governments manage the earmarked education transfers which finance salaries and social insurance contributions for teachers in preschool, primary, and general secondary schools. These earmarked education transfers accounted for 39.8 percent of total government expenditure for education in 2009 (Table 5.4). As shown in Table 5.4, the transfer is subdivided into specified portions for education for special-needs students (23 percent of the earmarked transfer for 2010), primary and general secondary schools (72 percent of the transfer), and preschool education for 5-and 6-year olds (5 percent of the transfer). The amount of the transfer to each local government is determined by the MOES Budget Department in consultation with the MOF Budget Department. For preschool and primary and general secondary education, the transfer for each local government is determined by a spreadsheet model which applies financing coefficients to students in each of several cost categories, based on grade level, urban/rural location, special-purpose schools (including state gymnasia, music, dance, and art schools, and boarding schools), and the number of special-needs students. Transfers are provided on the basis of these ex-ante calculations, and adjusted later in the school year when the results of the September and January school censuses are tabulated. 320. Earmarked education transfers declined by 16.6 percent in 2009 and a further 18.9 percent in 2010, reducing the overall amount of the transfers to 67.6 percent of its nominal 2008 level (and 71.3 percent of its constant-price 2008 level). Of the three components of the earmarked education transfer, the transfer for salaries of teachers in primary and general secondary schools experienced the largest absolute and relative budget reduction – a nominal decline of 103 million LVL since 2008, or 32.3 percent of its 2008 level (Table 5.4). In real terms, the cumulative decline amounted to 28.3 percent of its 2008 level. 321. Local governments in Latvia are responsible for financing all non-salary costs of preschools, primary schools, and general secondary schools from their own resources, (including the block grants which they receive under the Municipal Equalization Fund). Local governments’ own resources are intended to cover non-salary education costs for textbooks, library books, and other teaching and learning materials, as well as student transportation, teacher training, utilities, and the physical maintenance of schools. Information from local governments shows that their own resources are used to meet a broad array of education costs -- including the above, but also salary supplements to teachers (Figure 5.4), hiring of additional teachers, librarians, and other school staff, and transport allowances for teachers. 161 Figure 5.4.Supplementary Financing of Teachers’ Salaries by Local Governments, September, 2009 and April, 2010 Share of local funding 16% 600 thous. LVL/month 14% Local funding, 500 12% 10% 400 8% 300 6% 200 4% 2% 100 0% 0 Sep-09 Apr-10 Sep-09 Apr-10 Sep-09 Apr-10 Sep-09 Apr-10 Sep-09 Apr-10 Riga Other rep. cities Other cities Rural Total Share in total compensation (left axis) Amount, thous. LVL/month (right axis) Source: Analysis of school survey data by Professor Mihails Hazans, Consultant for the World Bank team 162 Table 5.4. Latvia Budget Plan and Amendments, 2008-2010 Earmarked Education Subsidies For Local Governments (Ministry Code 62), By Program Ministry Program Program 2008 2008 2009 2009 2010 Code Code Description plan amendments plan amendments plan II 62 01.00.00 Earmarked subsidy for special education 56,223,498 59,040,638 64,935,246 57,701,223 48,615,467 62 02.00.00 Earmarked subsidy for local folk art 904,169 904,169 982,792 622,434 collectives, covering salaries and state social security contributions 62 05.00.00 Earmarked subsidy for primary and general 214,496,736 230,191,525 262,602,459 197,687,081 155,665,155 secondary education – salaries of teachers and local education authorities, and the corresponding social security contributions 62 10.00.00 Earmarked subsidy for preschool education 13,820,957 15,152,104 17,882,603 11,687,704 12,324,000 for 5 and 6-year old children – staff salaries and corresponding social security contributions 62 12.00.00 Earmarked subsidy for local activities 2,158,968 2,158,968 1,477,340 400,000 0 62 14.00.00 Earmarked subsidy for local nursery schools - 11,847,834 15,842,663 22,847,604 0 0 staff salaries and state social security contributions 62 Total 299,452,162 323,290,067 370,728,044 268,098,442 216,604,622 Source: Data provided by Ministry of Finance - Excel file “Budgets for Education, 2008-2010â€? 163 322. There is considerable variation across local governments in the level of financing per student. At the school level, per-student costs of primary and general secondary education as reported by local governments in December, 2009 ranged from LVL 142 per year to LVL 2,302.99 The extreme variation in school-specific costs as reported by local governments reflects the peculiarities of individual schools, including capital investments and repayment of loans for earlier school improvements. A more indicative measure is the variation across local governments in average expenditures per student across the various schools in each jurisdiction. Average per-capita education expenditures of local governments provide an approximation for the variation in average per-student expenditures of local governments for primary and general secondary education. 323. In 2009, average per-capita education expenditures of local governments varied from 100 LVL to almost 700 LVL (Figure 5.5). Yet despite the dispersion of expenditures per student that this implies, the OECD’s PISA international student assessment for the latest available year100 found that between-school differences in learning outcomes of 15-year-old students in Latvia were among the lowest in Europe (Figure 5.6). This low between-school variation in learning achievement is a very positive accomplishment of Latvia’s education system, matched only by the education systems of the Scandinavian countries, Spain, and Poland.101 Figure 5.5 Per-Capita Education Expenditures of Local Governments (LVL), 2009 Per-Capita Education Expenditure of Local Governments, 2009 (LVL) 800.0 700.0 600.0 500.0 400.0 300.0 200.0 100.0 0.0 Source: Analysis by Dr. Inga Vilka, Consultant for the World Bank team Legend and source for figure on next page: 99 Information provided by MOES Budget Department. (Word file, “Per-Student Costsâ€?) 100 OECD’s Programme for International Student Assessment, 2006. PISA measures student achievement of 15- year-old students in participating countries using an internationally comparable survey instrument. Low between- school variance in assessment results means that the differences in community resources and other differences across schools have little effect on measured learning achievement of students. Latvia’s mean science score on the 2006 assessment was 490 – slightly below the OECD mean, but above the United States mean. Among Latvia’s th neighbors, Estonia was conspicuously high in the ranking, with a mean score of 531 – the 5 highest score of the 57 participating countries. 101 Although these results are from the 2006 PISA survey, they should change only slowly over time. The results of the May, 2009 PISA survey are scheduled to be released in December, 2010. 166 Figure 5.6 Variance in 15-Year-Old Student Performance Between Schools and Within Schools on the PISA 2006 Science Scale Total between-school variance Total within-school variance Between-school variance explained by Within-school variance explained by the the PISA index of economic, social and PISA index of economic, social and cultural status of students and schools cultural status of students and schools Between-school variance Within-school variance 100 80 60 40 20 0 20 40 60 80 100 120 Bulgaria Germany S lovenia Czech Republic Austria Hungary Netherlands Belgium Argentina Chile Japan I taly Greece Chinese Taipei I srael Brazil Slovak Republic Turkey Uruguay Switzerland Romania Serbia Hong Kong China Croatia Tunisia Korea Kyrgyzstan Luxembourg United S tat es P ortugal Thailand Mexico Lithuania Colombia Russian Federation United K ingdom Montenegro New Zealand Australia Jordan I ndonesia Macao China A zerbaijan Canada Ireland E stonia Denmark Latvia Spain P oland S weden Norway I celand Finland Source: Table 5.1, PISA 2006: Science Competencies for Tomorrow’s World – Volume 1: Analysis, OECD, Paris, 2007. 167 Table 5.5 Total and Per-Student Government Expenditures by Level of Education 2008 2009 2010 Primary and General Secondary Education Local government education expenditures from all sources (including 418,035,814 354,066,330 NA earmarked transfers) Local government education expenditures on primary and general secondary education = Total government expenditures on primary and 297,641,499 260,946,885 NA general secondary education of which, from state budget 230,191,525 197,687,081 155,665,155 of which from own budget 67,449,974 63,259,804 NA Total Government expenditures 7,363,564,100 6,503,997,107 NA Expenditures on primary and general secondary as % of total government 4.0% 4.0% expenditures Total Government education expenditures 1,144,144,239 980,000,356 NA Expenditures on primary and general secondary as % of government 26.0% 26.6% NA education expenditures Total Government expenditures 7,363,564,100 6,503,997,107 NA Enrollments (school year (t-1)/t) 263,140 248,647 239,139 Government expenditure per student 1,131 1,049 NA Vocational Secondary Education MOES vocational education budget 61,417,175 48,583,355 41,737,547 Enrollments, total 37,667 38,819 36,660 Of which, in MOES institutions 34,389 34,134 32,296 Government expenditure per student (MOES) 1,786 1,423 1,292 Imputed budgets for vocational education programs managed by other 5,854,357 6,668,220 5,639,790 ministries (at MOES unit cost) Government expenditure for vocational education 67,271,532 55,251,575 47,377,337 Primary and Secondary (general plus vocational) Government expenditure on primary and secondary education 364,913,032 316,198,460 NA GDP (millions of current LVL) 16,274 13,244 12,089 Government expenditures on primary and secondary education as % of 2.2% 2.4% NA GDP Government expenditure on primary and secondary as % of total 5.0% 4.9% NA government expenditures Government expenditure per student (all primary and secondary students) 1,213 1,100 NA Per-capita GDP (current LVL) 7,167 5,857 5,365 Government expenditure per student/per-capita GDP 16.9% 18.8% NA Higher education MOES higher education budget 127,073,144 53,922,894 43,367,773 Higher education budgets, other ministries 35,100,000 25,100,000 20,400,000 Government expenditure for higher education 162,173,144 79,022,894 63,767,773 Government expenditure on higher education as % of GDP 1.0% 0.6% 0.5% Government expenditure on higher education as % of total government 2.2% 1.2% NA expenditure Enrollments in public institutions, year (t-1)/t 87,017 84,389 77,693 Government expenditure per student (students in public institutions) 1,864 936 821 Source: GDP data from IMF World Economic Outlook database, April, 2010. Other data provided by MOES and Treasury data system. Excel file, “Government Expenditures by Levelâ€? 168 324. A further perspective for assessing the adequacy of education financing is the amount of government expenditures by level of education, expressed as a percentage of GDP. Table 5.5 presents estimates of total and per-student government expenditures on education by level (primary, secondary, etc.), in current LVL and expressed as a percentage of GDP and total government expenditures for 2008, 2009 and 2010.102 Figure 5.7 compares the level of government expenditures for primary and secondary education and higher education as a percentage of GDP in Latvia and countries in the EU and the OECD. 325. As shown in Figure 5.7, the level of government expenditure for primary and secondary education in Latvia, expressed as a percentage of GDP, was the lowest in Europe in 2008 and 2009. Government expenditures on primary and secondary education comprised 2.4 percent of GDP in Latvia in 2009, versus an average of 3.5 percent in the EU-19 countries and 3.7 percent in the OECD countries for the latest available year. The level of government expenditure for higher education in Latvia was the second lowest in Europe in 2008 (at 1.0 percent of GDP), but the budget cuts of 2009 and 2010 brought it to by far the lowest level in Europe (at 0.6 percent and 0.5 percent of GDP, respectively). Figure 5.7 Government Expenditure on Primary,-Secondary and Higher Education, as a Percentage of GDP, Latvia and Comparator Countries 5.0% 4.5% 4.0% Primary and Secondary Higher 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% Note: Secondary education figures include secondary vocational education. Reference years for Latvia as shown; for comparator countries, 2006. Source: Latvia figures from Table 5.7, above. Figures for other countries from Table B2.3 Education at a Glance, 2009, OECD. Tab 3, Excel file “Government Expenditures by Levelâ€?. 326. These aggregate comparisons do not address the level of expenditures on secondary vocational education (which are subsumed under primary and secondary education). Internationally comparable 102 The total estimate of government expenditures by level in Table 6 amounts to 11.9percent of total government expenditures. This is slightly less than the 13.1percent figure shown in Table 1 for the share of education in total government expenditures for higher education because it excludes preschool education and education activities not specific to grade level. 169 data on vocational education expenditures are more limited than they are for general education. In any case, shares of vocational enrollments at the secondary level vary widely across countries, so a comparison of expenditure levels for vocational education would not provide many insights on the appropriateness or adequacy of the level of financing for vocational education. What is more relevant here is a comparison of levels of per-student financing in vocational and general education. This information is useful because it tells us what average level of resources is provided for each student in secondary vocational education by comparison to students in general secondary education, independently of what percentage of students are enrolled in vocational education. 327. Figure 5.8 presents a ranking of the countries for which per-student expenditure data are available for vocational secondary and general secondary education. The estimates of government expenditure per student in Latvia shown in Table 5.5, and Figure 5.8 show expenditure levels for secondary vocational education that are 58 percent higher than for primary and general secondary education in 2008 and 36 percent higher in 2009.103 The latter figure appears to be well within the range of reasonable comparators. In view of the financing needs for specialized equipment and expendable materials in vocational education, vocational education costs per student should be significantly higher than for secondary general education. The lower expenditure levels reported in Figure 5.8 for secondary vocational education in Estonia, the Netherlands, and Slovakia are an anomaly that suggests either reporting errors or budget deprivation. (In Austria, this finding reflects the fact that employers participating in its dual system of vocational education pay some of the costs of education provision.)104 In Germany, which is renowned for its highly effective dual-system apprenticeship training program at the secondary level,105 government expenditures on vocational education are more than twice as high as they are for general secondary education (which includes the highly-evaluated network of gymnasia). Expenditures on vocational education in Hungary, where vocational school restructuring with major investments in the late 1990s led to improved labor-market relevance,106 are 42 percent higher than they are for general education. 103 The available information on class size and student/teacher ratios in Latvia (discussed below) suggests that government expenditures per student may be slightly lower for upper secondary general education than they are for all of primary and general secondary education, in which case the Table 5 expenditure estimates may understate the actual differences. 104 Michael Hörtnagel, “The Financing of Vocational Education and Training in Austria,â€? Institute of Industrial Science, Vienna (CEDEFOP, Thessaloniki, 1999). 105 The German dual system has had a history of providing high-quality training with close involvement of employers and very positive labor-market experiences for graduates. But the system is not without its problems. See, for example, Diane-Gabrielle Tremblay and Irène Le Bot, “The German Dual Apprenticeship System: Analysis of its Evolution and Present Challenges,â€? Research Note Mo 2003-4A, Université du Québec, February, 2003. 106 András Benedek, “Vocational Education and Training Reform in Hungaryâ€?, Background Paper, Budapest, 2008. 170 Figure 5.8: Government Expenditure Per Student in Vocational Secondary Education as a Percentage of Government Expenditure Per Student in General Secondary Education, Latvia and European Comparators 250% 200% 150% 100% 50% 0% en nd ia lic a ry d y s a n 9) 8) nd an tri an ni ea ak ed la ga ub 00 00 to us m nl rla m er ov Sw un ep Es er A (2 (2 Fi itz he D Sl H R G EC Sw ia ia et h tv tv N ec O La La Cz Source, Data for Latvia refer to the years indicated. Data for other countries are from Table C1.2 of Education at a Glance: OECD Indicators, OECD, Paris, 2007, and refer to 2004. III. Improvements in the Education Sector’s Efficiency 328. The introduction of per-student financing for primary and general secondary education and the transfer of management responsibility to local governments in 2009, combined with the budget reductions described above, motivated improved efficiency in delivery of primary and general secondary education. As shown in Table 5.6, the primary and general secondary education system was already moving towards reduced numbers of schools and classes even before the introduction of per-student financing in September, 2009. The number of schools, classes, and class sets107 declined between the 2007/2008 and 2008/2009 school years (although not by as much as total enrollments). But the most budget-sensitive parameters were moving in the wrong direction – from low efficiency towards even lower efficiency. Although enrollments declined by 5.5 percent between September, 2007 and September, 2008, the number of teaching workloads actually increased by 1.4 percent. This led to a further decline in the already-low net student/teacher ratio108 from 8.0 to 7. The average number of 107 A class is a group of students in the same grade that are taught by a single teacher in a single classroom. A class set is a group of students from one or more grades that are taught by a single teacher in a single classroom. Joint classes – a subset of class sets -- are groups of students from a particular grade that are taught in a multi- grade setting, with students from more than one grade being taught in a single classroom by a single teacher. For example, a situation in which a single teacher teaches five students in grade one, five students from grade two and five students from grade three together in the same classroom would comprise one multi-grade class, one class set, and three joint classes. Joint classes are often an efficient mode of education delivery when the number of students per grade is too small to justify a full-time teacher for each grade. 108 The gross student/teacher ratio is defined as the number of students divided by the total number of classroom teachers, regardless of the number of hours they spend teaching. The net student/teacher ratio is defined as the number of students divided by the number of teaching workloads taught by classroom teachers. This indicator is based on full-time teacher equivalents, and is a better reflection of actual teacher cost because teacher 171 paid workloads per teacher increased by 0.8 percent putting further strains on the budget. These perverse staffing changes, combined with the large salary increase for teachers that was granted in 2008, help explain the 20.9 percent increase in government education expenditure in 2008 (Table 5.1). Table 5.6: Evolution of Efficiency Parameters in Primary and General Secondary Education, 2007-2009 % 2007/2008 % Change 2008/2009 Change 2009/2010 a Students 263,140 -5.5% 248,647 -3.8% 239,139 Schools 992 -1.0% 982 -10.7% 877 b Classes 15,431 -3.0% 14,973 -7.4% 13,858 b Joint classes 14,399 -3.0% 13,971 -7.8% 12,887 Total workloads 45,139 1.4% 45,783 -28.9% 32,564 Teaching workloads 32,928 1.4% 33,419 -24.9% 25,099 Share of non-teaching workloads 27.1% -0.1% 27.0% -15.2% 22.9% c Total school staff (number of posts) 33,605 -0.8% 33,321 -15.5% 28,153 c Teaching staff (number of posts) 26,537 0.6% 26,695 -12.1% 23,469 Total school staff (persons) NA NA 28,144 -14.4% 24,091 Teaching staff (persons) NA NA 22,657 -10.7% 20,228 Workloads: total/teaching 1.37 0.0% 1.37 -5.3% 1.30 Teaching workloads per teacher post 1.24 0.8% 1.25 -14.5% 1.07 Total workloads per school staff post 1.34 2.2% 1.37 -15.8% 1.16 Students/teaching staff (gross S/T) 9.9 -6.1% 9.3 9.4% 10.2 Students/teaching workloads (net S/T) 8.0 -6.8% 7.4 27.9% 9.5 d Average class size 19.0 -2.8% 18.5 4.2% 19.3 a Notes: Number of students is slightly smaller than in official statistics because it excludes general education b students in 10 vocational secondary schools. Joint classes refer to either multiple-grade classes or single-grade classes that were formed from merged classes and taught by one teacher. Calculation of classes and joint classes .c d includes distance education students. Each staff member counts once in each school at which he/she works. According to OECD and Eurostat methodology, calculation of class size excludes special schools and classes. Source: Calculation with annual School Census data provided by MOES. 329. After the management and financing reforms that were introduced during the summer of 2009, the improvement in network rationalization that was already under way accelerated substantially, according to most indicators. One hundred and five schools either closed or were consolidated with neighboring schools; the number of classes and class sets declined at more than twice the pace of the previous year; and average class size increased. The number of joint classes declined slightly. This is consistent with improved efficiency, since many former joint classes were absorbed into larger classes, as classes and schools were consolidated. compensation is based on the actual number of hours that a teacher teaches. It normalizes across the many teachers who teach more than a full workload or less than a full workload. 172 330. The crucial difference from the changes in the previous year is that after the adoption of the 2009 reforms, the number of teaching workloads declined by about 25 percent and the number of occupied school staff positions declined by over 5,000, including more than 3,200 teaching positions (Table 5.6). The actual number of staff employed in schools declined by over 4,000, including more than 2,400 teachers. The share of paid workloads for non-teaching tasks also declined from 27.0 percent to 22.9 percent. These very substantial improvements in efficiency reversed the earlier expansion of staff and budgets, and raised average class size to 18.5 and the net student/teacher ratio to 9.5. IV. Teacher Salaries and Teacher Earnings 331. Historically, teachers in Latvia have earned less than workers with equivalent qualifications in other occupations. The results of the household labor-force survey provide an indication of the magnitude of those differences. As shown in Figure 5.9, average reported earnings in the education sector since 2006 have been lower than earnings in the public sector as a whole, and lower still by comparison to earnings in public administration and defense. The gap between reported average earnings in the education sector and comparator sectors narrowed in 2009, reflecting the combined effect of the salary increases adopted in 2008 and the first of the salary reductions initiated in 2009. 332. As a consequence of the major reduction in the earmarked transfer for primary and general secondary education, there were cuts in all components of teachers’ earnings during the last half of the 2008/2009 school year and the beginning of the 2009/2010 school year: ï‚· Central budget financing for teachers’ base salary was reduced from LVL 345 to LVL 250 per month; ï‚· Financing for non-teaching tasks and overtime teaching was sharply reduced, and ï‚· financing for teaching-related tasks (for which teachers had been compensated on an hourly basis) was suspended. 333. Based on its planning of education transfers for the 2009/2010 school year, the MOES projected that these measures would lead to a 50 percent decline in average teacher earnings -- from LVL 552 per month during the 2008/2009 school year to LVL 276 per month in the 2009/2010 school year (Table 5.7). 173 Figure 5.9: Average Monthly Earnings (in LVL per month) Education and Comparator Sectors, 2006-2009 700 650 average monthly earnings (LVL) 600 550 Education 500 Public sector 450 400 Public administration and 350 defense 300 250 200 2006 2007 2008 2009 st Note: 1 Quarter data for respective years. Source: Enterprise administrative data Table 5.8: Projected Change in Components of Teacher Earnings in Primary and General Secondary Education, September 2008 – September, 2009 (Current LVL per month) September 1, 2008 September 1, 2009 Base salary for 21-period workload 345 250 Payment for teaching-related tasks 100 0 Payment for extra teaching hours (beyond basic 107 26 workload) Total 552 276 Source: Information provided by the MOES Budget Department. 334. One of the limitations of the sectoral data from the household labor-force survey data is their level of aggregation: they include all employed individuals who are identified as working in the education sector – school administration and school maintenance workers as well as teaching staff, and private as well as public educational institutions. Payroll data provide a far more accurate view of how the recent budget reductions have affected teacher earnings in primary and general secondary schools. Actual payroll records of primary and general secondary schools indicate that in September, 2009, the typical base salary for a 21-lesson workload was, as planned, 250 LVL in all schools, small and large, with few exceptions. Maximum base salary was 255 LVL in almost all schools with fewer than 150 students and in half of larger schools. But some school administrators were able to raise teachers’ base salary by supplementing their earmarked education transfer with savings from school and class consolidation or income from other sources, such that teacher base salaries ranged from 255 to 280 LVL in half of schools 174 with 150 to 499 students and from 280 to 315 LVL in one out of four of schools with 500+ students (Figure 5.10).109 Figure 5.10 Salary Range per Workload for Teachers in Primary and General Secondary Schools, in LVL per month, September, 2009 Teachers' wage rate variation within schools, Sept 2009 by school size 350 300 250 200 <90 90-149 150-299 300-499 500+ (total) Lowest wage rate in the school Most frequent wage rate Highest wage rate in the school Excludes compensation for non-teaching functions. Excludes special, board, and private schools Exclude outside (few atypical) values Source: Calculation with payroll data provided by MoES Wage rate refers to one teaching workload Source: School payroll data. 335. In addition to the supplementation of base salary by local governments, teachers were able to increase their earnings substantially above the Ministry’s projections at the start of the 2009/2010 school year through paid overtime teaching and various non-teaching tasks. In Latvia, the 21 lesson-per- week teaching requirement under a normal teaching workload is meant to be accompanied by class preparation and follow-up activities amounting up to a 40-hour work week. Yet teachers are permitted to be paid for work up to 1.5 teaching loads, and in addition to be paid for supplementary tasks such as lesson preparation.110 Payroll data for individual schools show that staff in primary and general secondary schools worked an average of 1.23 paid workloads in September, 2009 (Table 5.8); teaching staff111 worked an average of 1.08 teaching workloads. 336. These figures understate the actual workloads and earnings of teachers, because many teachers augment their income by teaching in more than one school. Matching school-specific payroll information for individual teachers allows a calculation of actual average workloads for individual school staff which takes into account their work in multiple schools. As shown in Table 5.8, staff in primary and 109 The number of schools with base salaries above 255 LVL was so small compared to the total, that these schools are considered ‘outliers’ and not shown in the ‘total’ panel of the graph. 110 In prior school years, teachers were permitted to be paid for up to 2 full workloads. 111 Teaching staff are defined here as school staff whose largest number of compensated hours is for teaching. 175 general secondary schools worked an average of 1.33 paid workloads in September, 2009, taking into account their work in more than one school; teaching staff worked an average of 1.19 teaching workloads. Table 5.8 Average Number of Paid Workloads per Teacher and School Staff in Primary and General Secondary Schools, September, 2009 and April, 2010 September 2009 % Change April, 2010 Teaching workloads per teacher 1.09 6.9% 1.16 Total workloads per staff post 1.22 7.9% 1.32 Teaching workloads (in all schools) per teacher 1.19 8.4% 1.29 Total workloads (in all schools) per person 1.33 9.0% 1.45 Source: Calculation with MOES school-level payroll data. Excel file “Students, Schools, and Teachers 2007- 2010 Comparing Payroll and School Report Dataâ€? 337. The 2009 budget cuts led to a sharp reduction in the number of workloads for non-teaching tasks which were financed under the earmarked education transfers. As of September, 2009, central budget financing for non-teaching tasks was reduced, providing only a 15 percent complement to calculated teaching workloads for school administration (principals and deputy principals).112 Many local governments, however, provided financing from other sources to maintain payments for these activities. As shown in Table 5.9, there were more than 7,000 paid workloads for non-teaching activities in September, 2009 – almost twice the number financed by the 15 percent provision in the earmarked transfers to local governments. 112 I.e., for each teaching workload, an additional 15percent is provided under the earmarked education transfer for administration and other non-teaching tasks (15 ÷ 1.15 = 13.0percent). The actual percentage of non-teaching workloads shown for 2009/2010 in Table 7 (23.2percent) is larger than 13 percent because local governments financed additional workloads from their own resources and/or budget savings from school consolidation. 176 Table 5.9: Paid Workloads for Teaching and Non-Teaching Tasks in Primary and General Secondary Schools, 2008 - 2010 Of which Sept. % Sept. % April in boarding & 2008 Change 2009 Change 2010 special schools TOTAL 44,457 -29.0% 31,583 12.1% 35,400 4,317 12.2% teaching grades 1-12 32,331 -24.0% 24,574 11.2% 27,332 2,445 8.9% other tasks, of which: 12,126 -42.2% 7,009 15.1% 8,068 1,872 23.2% school principals and deputy principals 2,941 -28.1% 2,113 7.2% 2,266 271 12.0% boarding-school staff 1,675 -31.4% 1,148 6.1% 1,219 1,168 95.8% interest group teachers 1,504 -34.8% 981 22.3% 1,200 89 7.4% pre-school education teachers 1,184 -23.0% 912 14.8% 1,047 24 2.3% teachers’ work in 'extended day' groups 1,501 -73.7% 395 58.9% 628 48 7.6% librarians 1,172 -58.3% 489 15.1% 563 55 9.8% education psychologists 482 -50.0% 241 17.4% 283 51 18.2% social educators 316 -32.3% 214 11.7% 239 49 20.7% speech therapists 317 -38.8% 194 15.3% 224 29 13.0% other educators 575 -84.1% 91 46.6% 134 16 11.9% sports organizers 147 -34.7% 96 6.2% 102 7 7.3% special educators 197 -59.2% 80 14.9% 92 9 9.2% teachers' assistants 115 -52.8% 54 32.0% 72 55 77.0% Source: Data from school payroll database. Word file “Mihails June 14â€? 338. In January, 2010, central budget transfers to local governments for teacher salaries were increased by 37 percent, with no additional teaching requirement. This increase was intended to finance the restoration of payments to teachers for teaching-related tasks such as class preparation and follow-up and correction of exams. Following the January2010 increase in earmarked transfers to local governments for teacher salaries, the minimum base salary of teachers remained unchanged, but the number of paid workloads for primary and general secondary school staff increased substantially. The average number of paid workloads in the principal school of employment increased from 1.22 to 1.32 for all tasks, and from 1.09 to 1.16 for teaching tasks. Paid workloads across all schools of employment increased from 1.33 to 1.45 for all tasks and from 1.19 to 1.29 for teaching tasks (Table 5.8). At the same time, the infusion of new financing in January, 2010 led to a large increase in compensation to teachers and other school staff for all categories of non-teaching tasks following the substantial reduction that occurred between September, 2008 and September, 2009). 339. Information from the school payroll database for September, 2009 and April, 2010 provides a breakdown of teachers by the number of paid tasks performed and the number of schools in which they work. As shown in Table 5.10, 1,948 primary and general secondary school teachers worked in more than one school in September, 2009. That figure increased to 2,250 by April, 2010. The number of teachers working at more than one type of task (for which they were separately compensated) increased from 10,862 in September, 2009 to 12,656 in April, 2010. 177 Table 5.10 Primary and General Secondary Teachers with Multiple Jobs, September, 2009 and April, 2010 September, 2009 April, 2010 Number of schools in Number Number of schools in Number which the teacher of which the teacher of works Teachers works Teachers 1 2 ≥3 1 2 ≥3 Number of 1 12,828 0 0 12,828 11,641 0 0 11,641 types of 2 7,153 998 0 8,151 7,981 996 0 8,977 compensated 3 1,478 530 91 2,100 2,006 666 94 2,766 tasks (including teaching) ≥4 283 245 84 611 419 375 119 913 Number of Teachers Total 21,742 1,773 175 23,690 22,047 2,037 213 24,297 Source: Calculation with MOES school-level payroll data. Excel files “Teacher Salaries and Teacher Deployment, September, 2009 and April, 2010 – Mihails Hazans, May 3â€? and “Teachers with Multiple Jobsâ€? 340. Table 5.11 summarizes the effect of these changes on teachers’ earnings, and compares them to recent changes in earnings of employees in state budget institutions. As of September, 2009, average monthly earnings of teachers (LVL 378) were much lower than average monthly earnings of employees of state budget institutions (LVL 508). Since September, 2009, however, average monthly earnings have risen for teachers and fallen for employees of state budget institutions, such that the two are now roughly equal. As of February, 2010, average monthly earnings of employees of state budget institutions were LVL 500 and falling, whereas in April, 2010 average monthly earnings of teachers were LVL 476 and rising (not including any non-school income earned by teachers). 178 Table 5.11 Earnings Comparison for Teachers and Employees of State Budget Institutions Average monthly earnings Source (LVL) At the main Across all job jobs Primary and General Secondary Teachers September, 2008 552 - MOES Budget Department. September, 2009 - projected MOES Budget Department. "Teacher (LVL 255 average salary per 281 - Workloads & Salaries by Novad and School workload X 1.1 workloads) Type). MOES school-level payroll data. Excel files September, 2009 343 378 “Teacher Salaries and Teacher Deployment, September, 2009 and April, 2010 MOES school-level payroll data. Excel files April, 2010 382 476 “Teacher Salaries and Teacher Deployment, September, 2009 and April, 2010 Employees of State Budget Institutions Ministry of Finance, March 1, 2010 PPT: " MOF January-December, 2009 484 566 PPT on Public-Sector Salary Changes - 2009 to average salaries 2010" Ministry of Finance, March 1, 2010 PPT: " MOF September, 2009 average 432 508 PPT on Public-Sector Salary Changes - 2009 to salaries 2010" Column 2 figure from Ministry of Finance, March 1, 2010 PPT: “MOF PPT on Public-Sector February, 2010 average Salary Changes - 2009 to 2010â€? The figure in salary and estimated total 422 500 column 3 is from SRS (VID) data compiled by earnings Statistics Latvia for February, 2010. Word file “Earnings of Employees of State Budget Institutionsâ€? Sources as indicated in rightmost column IV. Interpreting the Expenditure Evidence & Implications for Education Policy 341. Two sets of conclusions emerge from the evidence on education expenditures and teacher earnings presented in the previous section. These have both short-term and long-term implications. In the short term, the major reforms in management and financing of primary and general secondary education that were introduced in 2009 are having the intended effect in bringing about improved efficiency. These reforms will take time to yield their full benefits. There will be further efficiency gains as local governments become more adept at applying the new procedures and opportunities for any discretionary redistribution of earmarked funds are minimized. The MOES could accelerate this process by disseminating lessons of successful implementation of per-student financing. The capitation formula 179 has come under the critical scrutiny of local specialists and academic experts, who are engaged in a public debate on how it can be further refined to increase efficiency. 342. Government expenditures on education in Latvia, whether expressed as a share of GDP or of total government expenditures, are low in relation to European comparators. Other comparisons of education expenditure levels as a percentage of per-capita GDP tell a similar story: of significant underfunding of education in Latvia at all levels -- particularly in higher education. Continuation of this low level of support for education could risk putting Latvia at an educational disadvantage vis-à-vis its neighbors. The fact that Estonia spends twice as much (as a share of GDP) on higher education as Latvia and one-third more on primary and secondary education may have something to do with the fact that Estonia’s mean science score on the most recent PISA assessment was 531 – the fifth highest among all participating countries and the second highest in Europe, while Latvia’s was 490 – 10 points below the mean.113 343. Earlier concerns over low teacher salaries relative to other sectors, and the threat that they pose for education quality have been alleviated to an extent by the Government's action in January 2010 to restore payment to teachers for teaching-related tasks that increased the amount of per-student financing. Recent data from schools indicate that average teacher earnings for the largest group of teachers – primary and general secondary school teachers -- are now roughly at parity with earnings of people working in budget-supported institutions. This outcome may reduce the urgency of further salary increases for teachers in the immediate future, although when compared to other countries, teachers in Latvia still appear relatively underpaid. Figure 5.11 compares statutory salary (annualized wage rate per one workload) of Latvian teachers in school years 2005/2006 to 2009/2010, expressed as percentage of the GDP per capita114 to similar indicators for OECD member and EU partner states (as of 2007). The OECD average (as well as the average for the 19 EU/OECD members) is more than two times higher than the Latvian figure. Only 5 out of 33 comparison countries pay their teachers significantly less than 100 percent of GDP per capita; these are Hungary and Iceland (about 80 percent at primary and lower secondary level, but close to 100 percent at upper secondary); Norway and Israel (slightly more than two thirds); and Estonia, the only country where the indicator is below Latvian level. 344. Actual teachers’ earnings differ from the statutory salary if teachers’ workloads fall short of or exceeds a normal teaching load, and also due to various bonuses and supplements. Depending on the design of the compensation system, the difference might be small in some countries and substantial in others. In Latvia, the average classroom teacher’s actual earnings are 525 LVL if earnings in all schools where a teacher works are taken into account. In any case, the difference between earnings and statutory salary in Latvia is larger than in most comparators. To avoid any confusion in international comparisons, Figure 5.12 presents annual average earnings (rather than statutory salaries) per employee of public school teachers in Latvia, the United States, and European countries. 113 Figure 2.11c, PISA 2006: Science Competencies for Tomorrow’s World – Volume 1: Analysis, OECD, Paris, 2007. 114 Average salary in school year t/(t+1) is compared to GDP per capita in year t. 180 Figure 5.11. Annual statutory teachers’ salaries per work load after 15 years of experience, as per cent of GDP per capita Annual statutory salaries af ter 15 years of experience, per cent of GDP per capita 200 190 180 Lower secondary Primary Upper secondary 170 160 150 140 130 120 110 100 90 80 70 60 50 40 30 20 10 0 Switzerland Slovenia Finland Australia Poland England France Estonia Luxembourg Hungary Czech Republic Chile Iceland Ireland Denmark Greece Netherlands New Zealand Portugal Germany Japan Austria Latvia 2009 Latvia 2008 Spain Latvia 2010b Latvia 2010a Latvia 2010 Latvia 2007 Latvia 2006 Norway United States Israel Mexico Turkey Sweden Italy Belgium OECD average EU 19 average Notes: OECD countries’ data (with few exceptions) are for school year 2006/2007. Both for OECD countries and for Latvia, statutory salary in school year t/(t + 1) is related to GDP per capita in calendar year t. Latvia 2010a: September- December 2009; Latvia 2010b: January-August 2010; Latvia 2010: average for the school year 2009/2010. Sources: Prepared by Professor M. Hazans, Consultant to the World Bank team. All countries except Poland and Latvia: calculation based on OECD (2009, Tables D3.1 and X2.2), using GDP of 2006 (rather than 2007) as a base. Poland (2006): calculation based on ILO LABORSTA data; Latvia 2006-2009: calculation based on government regulations for teacher salaries. Latvia 2010: calculation based on school payroll data provided by MoES 181 Figure 5.12: Annual average gross teacher earnings in public schools, per cent of GDP per capita), by level of education, in Latvia, selected countries Belarus 2007 Russia 2007 Primary Secondary Slovakia 2007 Estonia 2009 Czech R. 2007 Lithuania 2007 Norway 2007 Latvia 2009/2010 (1) Latvia 2008/2009 (1) Latvia 2009/2010 (2) Hungary 2007 US 2006 Finland 2006 Romania 2007 Poland 2006 UK 2007 Germany 2007 0 25 50 75 100 125 150 175 Notes: Entries ‘Latvia 2008/2009 (1)’ and ‘Latvia 2009/2010 (1)’ refer to all classroom teachers, while ‘Latvia 2009/2010, (2)’ and other countries’ data refer only to teachers with full-time workload. Source: Prepared by Professor Mihails Hazans, Consultant to World Bank team. Calculated with teacher earnings data (Latvia – administrative data; Estonia – LFS data, other countries - ILO LABORSTA data), GDP (in national currency, current prices) and population data of Statistics Latvia, Statistics Estonia and UNESCO Institute of Statistics. 345. In order to preserve a positive incentive to enter the teaching profession, the Government could maintain the January, 2010 salary increase for primary and general secondary education. This would help maintain the recently attained balance in earnings between teachers in primary and general secondary schools and people with similar qualifications who are working in other sectors. The Government would have to closely monitor teacher earnings to maintain general equivalence with other professionals working in the public sector, which is a task made very difficult by the antiquated “piece workâ€? model of teacher remuneration still in place. Eventually, MOES could move toward a more modern, full-time concept of remuneration in the teaching profession based on learning outcomes rather than time inputs for specific tasks. In such a model teaching responsibilities would be defined to include a broad array of teaching-related tasks that support student learning, including lesson preparation and follow up, correction of exams, collegial support, and meeting with students and parents. This approach would be more consistent with the output-oriented financing reforms, and would provide more explicit incentives for teachers to focus on learning outcomes. 346. In the longer term, as resources permit, it would be desirable to increase budget support for education – particularly for higher education –in order to bring Latvia closer to international comparators and ensure that Latvia’s education system is playing an appropriate role in maintaining the 182 competitiveness of Latvia’s economy. Furthermore, the current situation of rough parity in earnings of teachers and workers with similar qualifications will need to be sustained in order to safeguard education quality. It would be beneficial if future increases in teacher compensation maintain that balance. Finally, education sector outcomes could benefit by designing any future education budget increases to build in stronger incentives for improved quality and efficiency. Using Budget Policy to Reinforce Positive Incentives 347. The recent per-student financing reform in primary and general secondary education is a financing measure that incorporates powerful incentives for improved efficiency and quality. As we have seen, these incentives are already having a measurable effect on improved efficiency, and should yield further benefits for efficiency and education quality in the longer term. The January 2010 increase in earmarked transfers for teacher salaries provides a contrary illustration. The increase in teacher earnings under the increase in budget transfers was desirable in that it brought about a desirable rise in teacher earnings. But the manner in which the increase was implemented -- linking increased teacher compensation to performance of additional, separately compensated tasks – creates incentives for teachers and other school staff to focus on additional paid hours of school work rather than focusing on activities that will improve student learning outcomes.115 348. This input-based approach to teacher compensation is outdated and inconsistent with the new output-based approach of per-student financing. Moreover, it appears to be inconsistent with the requirement that teachers work a full 40-hour week on teaching and teaching-related tasks in order to qualify for payment for a 21-lesson teaching workload. Additional payment for teaching-related tasks is not warranted unless it can be demonstrated that actual hours worked exceed the level of teaching- related tasks associated with a full teaching workload. In the medium term, educational outcomes could benefit by moving from this hourly-based compensation approach to a more modern, full-time concept of the teaching profession which focuses on student learning outcomes and explicitly recognizes the important teaching-related tasks that teachers perform in addition to actual lesson delivery. 349. Another sense in which the current approach to teacher compensation embodies perverse incentives is the practice of paid overtime teaching hours, in which teachers work up to and often beyond the maximum permitted number of overtime hours in order to earn extra income. As the average number of workloads per teacher (1.45) implies -and the data from the combined school payroll registry confirm- some teachers work and are paid for more hours than the 1.5 workload limit.116 This 115 Some, but not all paid, non-teaching activities contribute to better learning outcomes. More modern approaches to teacher professional development explicitly recognize teachers’ contribution to better learning outcomes. Moreover, this approach treats teachers unequally across disciplines and grade levels. For example, math teaching offers more opportunities than science teaching for extra payment for correction of homework. But effective science teaching has its own requirements for teaching-related activities (such as keeping up-to-date on new scientific developments) that are not recognized in the hourly compensation formula. 116 In April, 2010, teachers who worked in two schools were paid for an average of 1.82 workloads. Teachers who worked in three or more schools were paid for an average of 1.98 workloads. (Excel file “Teacher Salaries and Teacher Deployment, September, 2009 and April, 2010â€?) 183 may raise risks to education quality because at this level of overtime work it is difficult for teachers to spend the class preparation and follow-up time that the curriculum assumes for a normal 21-lesson-per- week workload. Here, too, an outcome-based approach to teacher compensation could provide more positive incentives for teachers to focus on learning outcomes and still permit them to earn salaries that are competitive with work in other sectors. V. Education Sector Policy Options for Government Consideration 350. The following measures are proposed as options for government consideration to address the issues raised in the preceding paragraphs. Preschool, Primary and General Secondary Education 351. The Government could help to ensure the best, equitable start for all students by maintaining budget financing of preschool education for 5 and 6 year old children. Budget financing for 5 and 6 year old children in preschool education was restored under the World Bank Special Development Policy Loan. Budget financing of preschool education is consistent with the compulsory nature of preschool education and the demonstrated benefits for subsequent school performance. Preschool education plays a particularly important role in providing educationally-at-risk children with the foundation skills that are critical to adequate learning progress in primary and secondary schooling. 352. The Government could maintain the January, 2010 salary increase for primary and general secondary education. This would help maintain the recently attained balance in earnings between teachers in primary and general secondary schools and people with similar qualifications who are working in other sectors. 353. The Government could continue further its progress in implementing the new financing model by providing more guidance to novads and schools on options for application of per-student financing, based on the lessons of implementation during the first year of the new financing scheme. Many novads still are not aware of all the options available to them to achieve better efficiency – such as transporting teachers (rather than students) between schools. 354. In the medium term the authorities could monitor the evolving situation of teacher earnings in relation to salaries of employees in budget institutions, and reflect this information in consideration of any future salary increases in order to maintain comparability in earnings of primary and general secondary school teachers and people with similar qualifications working in other sectors. 355. Eventually, MOES could move toward a more modern, full-time concept of the teaching profession based on learning outcomes rather than time inputs for specific tasks. In such a system, teaching responsibilities would be defined to include a broad array of teaching-related tasks that support student learning, including lesson preparation and follow up, correction of exams, collegial support, and meeting with students and parents. This approach would be more consistent with the 184 output-oriented approach of the recent financing reforms, and would provide more explicit incentives for teachers to focus on learning outcomes. Adoption of a new, full-time concept of the teaching profession based on learning outcomes will require preparation of the methodology and further strengthening of assessment instruments. It would be easier to apply this concept once the economic recovery has allowed the restoration of more adequate financing for base salaries and other school inputs for effective teaching and learning. 356. The cost-effectiveness of special education could be scrutinized, in order to take advantage of beneficial opportunities for “mainstreamingâ€? students with special needs. . An expenditure item in the MOES budget which appears particularly large in relation to other programs is the earmarked transfer to local governments “for educational activities.â€? This allocation – amounting to LVL 48.6 million in the 2010 plan budget, larger than the entire budget for higher education (Table 5.4) – is provided for the support of the special education institutions which are managed by local governments for special-needs students. It is possible that other approaches to education provision for some of these special-needs students – including mainstreaming in general education schools -- could lead to more cost-effective education provision. In cases of severe disability, which is not the comparative advantage of the Ministry of Education, the Government could consider transferring principal financing and management responsibility to the Welfare sector. For students with lesser disabilities, the Government could consider financing under a more comprehensive, per-student financing approach which provides an incentive for local governments to seek more efficient models for educating special-needs students, including moving students to mainstream schools where feasible. Mainstreaming special-needs students whose conditions are mild and needs can be met in mainstream schools can improve their opportunities, as well as sensitize other students to the challenges faced by students with learning disabilities. Vocational Education 357. The Government could consider financing vocational education places more adequately as vocational schools are upgraded and transferred to local government management. Per-student expenditures on vocational secondary education in Latvia are below international comparators with high-quality vocational education programs. With major support from the EU, Latvia is in the process of upgrading and restructuring secondary vocational schools and transferring the improved schools to local government management. More adequate recurrent financing of the upgraded vocational schools would help exploit the important potential benefits of this investment. There is a legitimate question about the appropriate scale of vocational education (in terms of percentage of secondary students enrolled in vocational programs). Streaming students into vocational education at early ages could risk locking them into lower-lifelong earnings, and close off other opportunities. This risk could be mitigated by ensuring a “level playing fieldâ€? between vocational and general secondary, and keeping opportunities open for movement from one stream to the other. To the extent that vocational education is justified by skill needs in the labor market, it should be funded adequately to provide high-quality and relevant skill training. 185 358. To improve household choices about secondary education, stipends for vocational education students could be eliminated, beginning with incoming students in September, 2010. All vocational secondary students receive stipends of between 1 and 70 LVL per month. A small part of this amount is designed to offset the child allowances paid to parents whose children attend general secondary schools, to which families with students in vocational secondary schools are not entitled. During the 2009/2010 school year, the budget-financed portion of the stipend amounted to LVL 3.1 million for the 32,296 students in MOES VET institutions and c. 0.4 million for the 4,364 students in VET programs under other ministries – an average of LVL 96 per student per year. The remaining portion of the VET stipend is paid by the EU structural funds. Stipends for vocational-school students are often presented as necessary to attract students to attend vocational schools. But if vocational education provides valuable skills for young people, students should not need to be paid as an inducement to attend vocational schools, particularly, with the enhanced VET programs which are being developed under EU financing. Stipends could act as an unnecessary inducement for households to stream their children into VET, particularly for lower income households. Suspension of student stipends in vocational education would help students and parents make an unbiased choice between vocational education and general education, and would help to emphasize that vocational education is an investment that should be fully justified by its impact on lifetime earnings relative to other education options. 359. Public vocational schools could be allowed to compete to deliver adult retraining programs with other providers. In order to do so, budget regulations could be changed to allow vocational schools to participate in MOW-financed training programs. Elimination of stipends for vocational education students would be complemented by giving vocational schools a fair opportunity to compete for contracts to provide training and retraining services to unemployed people participating in Latvia's programs of employment assistance administered by the State Employment Agency under the MOW. Most secondary vocational schools have the expertise and capacity to offer adult retraining. Yet most of this retraining is contracted with private providers. Providing some of this training instead through public vocational schools, by allowing them to compete would help finance some of the schools’ fixed costs, and thus could yield budget savings. But vocational schools report that they cannot participate in competitive tenders for adult retraining because current regulations for budget institutions require them to plan and get prior approval for all expenditures, whereas the income from competitive tenders is uncertain and thus cannot be planned with any certainty. 360. The transfer of vocational schools to novads could be accelerated. With major EU support, all secondary vocational schools are preparing and implementing restructuring plans. All of the restructured vocational schools are scheduled to be transferred to local governments by 2015. Local management and financing of vocational schools would help ensure that vocational programs are cost- effective (in comparison to general secondary education), and provide good value in terms of improved job prospects for graduates. 186 Higher Education and Science 361. For higher education, in the medium term, it is desirable to provide a higher level of budget support comparable to other European countries, while moving towards greater reliance on private sources of financing.. As a short-term phasing measure to increased overall financing for higher education, the Government might consider providing more adequate financing for each budget-financed student place in higher education by reducing the number of budget-financed student places, financing these at a more adequate level, and prioritizing them for the poorest applicants (including those interested in a part-time course that would allow them to retain employment during their study), while keeping the total "envelope" of public financing for higher-education constant in the next two years. 362. Government expenditure on higher education is very low in Latvia by comparison to other European countries. Budget-financed places are rationed on the basis of student performance: The state budget finances the tuition of students with the highest scores on entry examinations for particular programs. Currently, budget financing covers half of full-time students in higher education. Other students pay full cost-recovery fees, which vary by specialization. This selective use of budget financing in higher education allows education authorities to maintain adequate levels of budget support per student in a situation of overall budget contraction by reducing the number of students who receive budget support. But the recent budget reductions in higher education may have taken the level of financing per student too low because the number of budget-financed students did not shrink as budget resources contracted. 363. As shown in Table 5.5, the number of budget-financed places increased by 7 percent between 2008 and 2010 even as higher education budgets overall fell by over 60 percent. Maintaining the number of budget-financed places in higher education under progressively reduced financing will erode and eventually destroy the quality and credibility of higher education for new graduates and could even undermine public perception of the qualifications of earlier graduates. Although there is clearly political pressure to maintain or even to increase the number of budget-financed places in higher education, doing so in a situation of shrinking overall financing imposes a severe cost on all students by degrading teaching and learning conditions. 364. A preferable approach is to establish the resource requirements by field of specialization, and to maintain that level of budget support by varying the number of budget-financed places depending on the availability of budget resources. A short term option to consider is reducing the number of budget- financed student places, financing them at a more adequate level, and prioritizing them for the poorest applicants (including those interested in a part-time course that would allow them to retain employment during their study), while keeping the total "envelope" of public financing for higher- education constant in the next two years. 365. As currently used, budget financing of higher education is inequitable because without means testing, it is more likely to disproportionately benefit students from higher-income families. Available 187 household surveys do not capture budget-financed university places. However, survey evidence does show that 65 percent of university students come from the wealthiest 50 percent of Latvian households (Figure 5.13). Greater reliance on fee financing could help to shift the public perception of higher education, to view it as an individual human capital investment decision. Current students may have acquired rights to continued financing under the arrangements that were in place when they entered higher education. Thus, any move to greater reliance on fee financing of higher education may need to be introduced gradually – for example, by applying it to newly registered students in September, 2010 and September, 2011. Figure 5.13 Distribution of Total University Student Enrollment, by Household Income Decile, 2007 20 18 Percentage of Total university 16 14 enrollment 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 Decile by Household Income Source: Staff estimates, using Latvia SILC 2007 366. There could be some savings from suspending higher education student stipends. Currently, 3,379 higher education students receive stipends of LVL 70 per month. These are granted to reward high performance. To make them more equitable, targeting criteria were introduced in 2009, and slightly over half of stipends are now awarded on the basis of need as well as performance. Student stipends are inefficient because high performance reaps its own lifelong rewards through higher earnings and professional mobility. The returns to higher education in Latvia are among the highest in the EU. Student stipends are inequitable because they are only awarded to budget-financed students, who tend to be from higher-income families. Because abrupt suspension of student stipends would adversely impact the personal finances of students already enrolled, the “acquired rightâ€? principle may also need to be applied. Suspension of student stipends would further help to shift the public perception of higher education, to view it as an individual human capital investment decision rather than a universal entitlement. 367. The Government could also consider discontinuing the subsidies for interest on student loans. The Government currently pays accrued interest on student loans during the duration of studies and for one year after completion. It also guarantees repayment and pays any excess of borrowing costs above 188 5 percent. This arrangement is unnecessarily generous to students and entails serious contingent liabilities for the Treasury. The cost of the interest subsidy has risen sharply – from LVL 3.3 million in the 2008 budget to LVL 8.2 million in the 2010 budget.117 But recent declines in interest rates mean that interest payments on student loans in CY 2010 are likely to be about LVL 1.6 million below the budgeted amount. Interest subsidies on student loans distort students’ choices about how much higher education to seek. The current structures by which the Government negotiates interest margins and guarantees repayment (in addition to individual guarantees) offer sufficiently attractive borrowing costs to students that ensure equitable access to credit. Actual interest rate subsidies beyond these measures are neither necessary nor desirable. Maintaining interest subsidies risks inducing students who do not need loans to borrow money and earn interest on the funds during the duration of their studies. This risk could rise sharply with higher interest rates, presenting an unnecessary and unwarranted contingent liability for the budget. 368. Finally, the Government could place greater emphasis on competitive financing of research and development. The Government could promote improved quality of R&D by moving towards a heavier reliance on competitive financing of research outputs rather than financing inputs. This could involve shifting some of the financing of salaries and facilities of public R&D institutes to competitive, international peer-reviewed programs, in which public and private universities, as well as public research institutes, can participate. The largest element of budget financing for science and research (amounting to slightly over LVL 7 million) finances inputs in the form of salaries and facilities for state science and research institutions. Greater reliance on competitive financing of scientific research would encourage higher productivity and fuller participation by universities. 117 Information provided by the MOES Department of Higher Education. 189 Matrix 1.3 Options for the Government to Consider in the Education Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Education, Preschool, Primary and General Secondary Maintain budget financing -- -- Budget financing for 5 Budget financing of -- -- preschool for 5 and 6 year old and 6 year old children preschool education is children in preschool in preschool education consistent with the education. was restored under the compulsory nature of World Bank Special preschool education and Development Policy the demonstrated benefits Loan. for subsequent school performance. Maintain salaries of general -- -- The January, 2010 Maintaining teacher Reversion to former low -- primary and secondary salary increase brought salaries at a comparable salaries for teachers teachers. Maintain January, average teacher level with employees of would endanger 2010 salary increase for earnings (including state budget institutions education quality by primary and general overtime) to parity with will help attract and retain encouraging departure of secondary education. employees of state competent teachers. the most competent budget institutions. teachers. Maintain implementation of -- -- Many novads are not Many novads still do not -- Adoption of a new, full- “funds follow the studentâ€? yet getting the full understand the options time concept of the financing model potential benefit of available to them to teaching profession, based --Exploit more fully the per-student financing – achieve better efficiency – on learning results, would flexibility availability under in terms of either such as transporting require preparation of the per-student efficiency or quality. teachers (rather than methodology and further financing: students) between schools. strengthening of --MOES to provide more The determination of assessment instruments. It guidance to novads and teachers’ earnings under would be easier to apply schools on options for per-student financing is this concept once the application of per-student still based on the old, economic recovery has financing input-based approach of allowed the restoration of --Move towards full-time hours spent on more adequate financing concept of teachers’ reimbursable tasks, rather for teacher salaries and profession, based on than on what their other school inputs for learning outcomes rather than students are learning. effective teaching and time inputs. learning. 190 Matrix 1.3 Options for the Government to Consider in the Education Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Review cost-effectiveness of -- -- Central financing of The approach of Care needs to be taken to Mainstreaming special 118 budget for special education. each institution sequestering special-needs preserve special needs students whose In cases of severe disability, provides no incentive students in separate education facilities for conditions are mild and transfer principal financing and for local governments institutions is inconsistent those students with whose needs can be met in management responsibility to to seek more efficient with the global trend severe conditions. mainstream schools, can the Welfare sector. service delivery towards mainstreaming improve their For students with lesser models. special-needs students. opportunities, as well as disabilities, consider financing Many students in special sensitize other students to under a more comprehensive, needs facilities could the challenges faced by per-student financing approach function well in students with learning which provides an incentive for mainstream schools, to disabilities. local governments to seek their benefit, as well as more efficient models for that of classmates. educating special-needs students, including transferring students, where appropriate, to mainstream schools. Vocational Secondary Education Finance vocational education -- -- The level of per- With major support from There is a legitimate To the extent that places more adequately as student expenditures the EU, Latvia is in the question about the vocational education is schools are upgraded and by government on process of upgrading and appropriate scale of justified by skill needs in transferred to local vocational secondary restructuring secondary vocational education (in the labor market, it should government management. education in Latvia is vocational schools and terms of percentage of be funded adequately to below those of transferring the improved secondary students provide high-quality and international schools to local enrolled in vocational relevant skill training. comparators with high- government management. programs). quality vocational More adequate recurrent Streaming students into education programs. financing of the upgraded vocational education at vocational schools would early ages could risk help exploit the important locking them into lower- potential benefits of this lifelong earnings, and investment for the growth close off other of the economy. opportunities. This risk 118 An expenditure item in the MOES budget which appears particularly large (in relation to other programs) is the earmarked transfer to local governments “for educational activities.â€? This allocation – amounting to LVL 48.6 million in the 2010 plan budget– is provided for the support of the special education institutions which are managed by local governments for special-needs students. This amount appears to be possibly larger than necessary – even larger than the budget for all of higher education. 191 Matrix 1.3 Options for the Government to Consider in the Education Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit can be mitigated by ensuring a “level playing fieldâ€? between vocational and general secondary, and keeping opportunities open for movement from one stream to the other. 119 120 Suspend financing of stipends 1.2 2.3 All vocational Stipends for vocational- To the extent lower Suspension of student for students in vocational secondary students school students are often income households are stipends in vocational education and training (VET), receive stipends of presented as necessary to more likely to have education would help to beginning with incoming between 1 and 70 LVL attract students to attend children enrolled in emphasize to parents, students in September, 2010. per month. A small vocational schools. But if vocational schools, students, and teachers that Family State Benefits should part of this amount is vocational education immediate repeal of vocational education is an be paid to parents of children designed to offset the provides valuable skills for stipends could impact investment that should be in vocational schools under child allowances paid young people, students negatively on their fully justified by increased the same regime as applies to to parents whose should not need to be paid welfare. Measures lifetime earnings. parents of children in general children attend general as an inducement to recommended to secondary.. secondary schools. attend vocational schools, strengthen the social During the current particularly, for the welfare system could help (2009/2010) school enhanced VET programs mitigate this impact in the year, the budget- which are being developed short term. financed portion of the with EU financing. stipend amounted to The state budget already LVL 3.1 million for the finances free room and 32,296 students in board for boarding MOES VET institutions students in vocational and c. 0.4 million for schools. the 4,364 students in Stipends could act as an VET programs under unnecessary inducement other ministries – an for households to stream average of LVL 96 per their children into VET, 119 This is estimated at 96 LVL per student per year (equal to current average budget-financed stipend for students in MOES VET schools), times the number of incoming VET students in September, 2010 (estimated as one-third of current total VET enrollments of 36,660). This savings estimate does not include the cost of replacing VET student stipends with targeted family allowances. 120 This is estimated through same procedure, applied to incoming class and 2nd-year class of September, 2012, estimated as 12,220 for each class. This savings estimate does not include the cost of replacing VET student stipends with targeted family allowances. 192 Matrix 1.3 Options for the Government to Consider in the Education Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit student per year. The particularly for lower remaining portion of income households. the VET stipend is paid by the EU structural funds. Parents of students in VET are not paid Family State Benefit (FSB). Enable VET school to compete -- -- Most secondary Most of the budget- Allowing public VET In the medium and long for adult retraining. Revise vocational schools have financed retraining which schools to compete with term, adult education budget regulations to allow the expertise and is provided under the private training providers programs in public vocational schools to capacity to offer adult MOW’s employment could introduce political vocational schools should participate in MoW-financed retraining. Vocational assistance program is obstacles to full be fully self-financed retraining projects, schools report that contracted with private competition. This risk through fees, as they are administered by the State they cannot participate providers. Allowing VET could be mitigated now. But under the Employment Agency (SEA). in competitive tenders schools to compete to through close monitoring current crisis, public for adult retraining offer this training, would and adherence to vocational schools should because current help finance some of the competitive procurement be able to compete for regulations for budget schools’ fixed costs, and procedures. contracts to deliver adult institutions require thus yield budget savings. retraining. them to plan and get prior approval for all expenditures, whereas the income from competitive tenders is uncertain and thus cannot be planned with any certainty. Accelerate transfer of -- -- With major EU support, Local management and In the short run, poorer Financing and managing vocational schools to novads. all secondary financing of vocational municipalities could secondary vocational vocational schools are schools would help ensure struggle with the education under the same preparing and that vocational programs additional burden of process which is used for implementing are cost-effective (in vocational schools. general secondary schools restructuring plans. All comparison to general would increase the of the restructured secondary education), and likelihood of unbiased vocational schools are provide good value in decisions about the scheduled to be terms of improved job appropriate scale and 193 Matrix 1.3 Options for the Government to Consider in the Education Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit transferred to local prospects for graduates. content of general governments by 2015. secondary and vocational secondary education. Higher Education & Science Finance higher education Government As currently used, budget Current students may In the medium term, it is places more adequately, even if expenditure on higher financing of higher have acquired rights to desirable to provide a it requires reducing the education is very low in education is inequitable continued financing higher level of budget number of budget-financed Latvia by comparison to because it under the arrangements support for higher places. most countries. disproportionately benefits that were in place when education at a level Budget-financed places students from higher- they entered higher comparable to other are rationed, solely on income families. education. Thus, this European countries. the basis of student Greater reliance on fee measure might need to performance. Half of financing would help to be introduced gradually, full-time students pay shift the public perception by applying it to newly full cost-recovery fees, of higher education, to registered students in which vary by view it as an individual September 2010 and specialization. The human capital investment September 2011. other half pays no fees; decision rather than a their places are universal entitlement. It financed by the budget. would improve equity and fiscal sustainability of higher education, and promote quality by encouraging competition based upon relevance to changing skill needs in the labor market as reflected by changing patterns of earnings. 194 Matrix 1.3 Options for the Government to Consider in the Education Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit To reduce the number of The number of budget- budget-financed places, the financed places in Government could target higher education budget financing to students increased over the past from households with three years, despite insufficient means to attend large budget reductions university. in higher education – The Government could reduce implying a budget-financed student deterioration of places, financing these at a teaching and learning more adequate level, and conditions. prioritizing them for the poorest applicants (including Maintaining the those interested in a part-time number of budget- course that would allow them financed places in to retain employment during higher education under their study), while keeping the progressively reduced total "envelope" of public financing will erode financing for higher-education and eventually destroy constant in the next two years. the quality and credibility of higher education for new graduates and could even undermine public perception of the qualifications of earlier graduates. 195 Matrix 1.3 Options for the Government to Consider in the Education Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit 121 122 Suspend higher education 0.6 1.2 Currently, 3,379 higher Student stipends are Abrupt suspension of Suspension of student student stipends education students inefficient because high student stipends could stipends would help to receive stipends of LVL performance reaps its own adversely impact the shift the public perception 70 per month. These lifelong rewards through personal finances of of higher education, to are granted to reward higher earnings and students already enrolled. view it as an individual high performance. To professional mobility. The The “acquired rightâ€? human capital investment make them more returns to higher education principle may also need decision rather than a equitable, targeting in Latvia are among the to be applied, with universal entitlement. criteria were highest in the EU. Student gradual implementation introduced under the stipends are inequitable of this measure. SDPL, and slightly over (despite the SDPL- half of stipends are supported targeting action) now awarded on the because they are only basis of need as well as awarded to budget- performance. financed students, who tend to be from higher- income families. 123 124 Discontinue budget subsidies 0.03 0.03 The Government Interest subsidies on In the short term, some Individuals will make for interest on student loans. currently pays accrued student loans contaminate students could be decisions about how much interest on student students’ choices about dissuaded from taking up higher education to invest loans for the duration how much higher university studies by in, based on earnings of studies and for one education to seek. The repeal of the subsidy. prospects. year after completion. current practice under However, the high rates It also guarantees which the Government of return from investment repayment and pays negotiates interest margins in higher education any excess of and guarantees repayment should eventually prevail borrowing costs above (in addition to individual on individual decisions. 5 percent. guarantees) offers This arrangement is sufficiently attractive unnecessarily generous borrowing costs to 121 This is estimated as LVL 700 per student, applied to new students in September, 2010 (estimated as one-quarter of current enrollments). 122 This is estimated on the same basis (footnote above), but applied to the incoming classes on 2010 and 2011. 123 This is an estimate based on current interest rates, or LVL 0.1 million at 5 % interest rate, as calculated by the MOES, based on the number of students currently receiving interest subsidies on student loans. This amount could increase significantly if the loan uptake rate or loan default rate were to increase significantly. 124 As above, based on current interest rates, or 0.1 at 5 % interest rate LVL 0.03 million at current interest rates, or 0.1 at 5 % interest rate. This amount could increase significantly if the loan uptake rate or loan default rate were to increase significantly. 196 Matrix 1.3 Options for the Government to Consider in the Education Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit to students and entails students. Actual interest potentially serious rate subsidies beyond that contingent liabilities for are neither necessary nor the Treasury. desirable. Due to falling interest Maintaining interest rates in CY 2010, subsidies risks inducing interest payments on students who do not need student loans in CY loans to borrow money in 2010 are likely to be order to earn interest on about 1.6 million Lats the funds during the below the budgeted duration of their studies. amount. This risk would rise sharply with higher interest rates. Place greater emphasis on -- -- The largest element of More competitive -- Likely to increase competitive financing of budget financing for financing would encourage performance and quality of research and development. science and research higher productivity and budget supported research The Government could (amounting to slightly fuller participation by and development. promote improved quality of over LVL 7 million) universities. R&D by moving towards a finances inputs in the heavier reliance on form of salaries and competitive financing of facilities for state research outputs rather than science and research financing inputs. Shift some institutions. of the financing of salaries and facilities of public R&D institutes to competitive, international peer-reviewed programs, in which public and private universities, as well as public research institutes, can participate. 197 CHAPTER 6: Health 369. This chapter covers health conditions in Latvia in the first section, then in the second section it considers how Latvia compares to other similar countries and states in terms of health resources, utilization, and expenditures. The third section reviews trends and issues in public expenditures for health, mainly from 2004 through 2010. The fourth section considers the private health insurance market. The fifth section, the conclusion, summarizes the findings and the options that Latvia might consider to improve the financing and operation of the health system. 370. In examining how Latvia spends on health, we assume that the purpose of this public investment is to make the Latvian population healthier and to improve equity of access to health services – goals fully consistent with each other – not necessarily to finance the infrastructure and services that exist today and were for the most part inherited from another era. I. Health Conditions Overview Figure 6.1: Life Expectancy at Birth for Both Sexes in Latvia, Estonia, and Lithuania (Three-Year Moving Average) 371. Latvia, along with Estonia Life expectancy at birth, in years and Lithuania, has been slow 75 relative to other European Union 74 accession countries to shake off the inheritance of poor health 73 outcomes from the Soviet era. 72 Within the World Health 71 Organization’s classification, these Estonia Latvia three countries join Belarus, 70 Lithuania Hungary, Kazakhstan, Moldova, 69 Russia, and Ukraine as “low child 68 and high adult mortalityâ€? countries. High adult mortality is 67 especially problematic for men 66 and imposes a high economic cost. 65 1970 1980 1990 2000 372. In terms of overall life Source: WHO/Europe Health for All Database January 2010 expectancy, shown since 1970 for all three countries in Figure 6.1 Latvia and Lithuania by 2006 had only recovered their peaks in the 1980s, although Estonia had passed its peak and seemed poised for consistent improvement. Estonia’s improvement reflects a widening gulf between men and women, with women now 12 years ahead of men, at 79 years. In Latvia in 2007, life expectancy at birth was 66 years for men and 76 years for women. 198 Latvia and the 2004/2007 EU Figure 6.2: Life Expectancy at Birth in Latvia and the EU member States Accession Countries before 2004, and the Other EU Accession Countries 373. EU Accession Countries have diverged in their performance Life expectancy at birth, in years 81 on health outcomes since 1990. 80 Figure 6.2 shows Latvia (green, at 79 the bottom) plus the EU States 78 77 before 2004, (dark red, at the top). 76 Other accession countries are 75 Czech Republic Latvia Poland included as gray lines, but the 74 Slovenia EU members before May 2004 three top performers are 73 highlighted: Slovenia (yellow), 72 71 Czech Republic (red), and Poland 70 (blue). All of the accession 69 countries were tightly packed at a 68 life expectancy of about 70 years in 67 1970. There was little 66 improvement for the next 20 years 65 1970 1980 1990 2000 2010 (in fact, it had been virtually stagnant since 1960), then the Source: WHO/Europe Health for All Database January 2010 divergence began in earnest in 1990. For Latvia, which lagged the EU by roughly 2 years (72 versus 70) in 1970, the unfortunate fact is that by 2007 the difference had expanded to 9 years, 80 versus 71. 374. The question is what it would take for Latvia to shift from stasis to rapid gains in life expectancy so it can close the gap more rapidly than looks possible today. Next we look at specific causes of death that are at the root of the gap. The panels in Figure 6.3 show Latvia’s performance relative to the accession countries and the EU15 on standardized death rates for the most common causes of death (note that the scales are different for each cause of death). We focus only on deaths in the age group 0- 64, which are termed “premature deathâ€? in the literature. Latvia ranks among the top three in death rates for each of these diseases, using data mostly for 2007 and 2008. Most of these are adult deaths in the productive working years of life. Latvia ranks first by a wide margin on deaths due to coronary artery disease and motor vehicle accidents. It is second or third – again by a wide margin relative to the fourth position – for deaths related to strokes and injuries. In each case the difference between Latvia and the pre-2004 EU countries is measured in large multiples, not increments. 199 Figure 6.3: Standardized Death Rates (SDR) in Latvia and Other EU Accession Countrie s SDR, ischaemic heart disease, 0-64 per 100000, Last available Latvia 2007 Lithuania 2008 Hungary 2008 Romania 2008 Estonia 2008 Bulgaria 2006 Slovakia 2005 Poland 2007 Czech Republic 2008 Slovenia 2007 EU members before May 2004 2007 0 50 100 SDR, cerebrovascular diseases, 0-64 per 100000, Last available Bulgaria 2006 Romania 2008 Latvia 2007 Lithuania 2008 Hungary 2008 Poland 2007 Estonia 2008 Slovakia 2005 Czech Republic 2008 Slovenia 2007 EU members before May 2004 2007 0 50 SDR, external cause injury and poison, 0-64 per 100000, Last available Lithuania 2008 Latvia 2007 Estonia 2008 Romania 2008 Poland 2007 Slovakia 2005 Slovenia 2007 Hungary 2008 Bulgaria 2006 Czech Republic 2008 EU members before May 2004 2007 0 20 40 60 80 100 120 140 SDR, motor vehicle traffic accidents, 0-64 per 100000, Last available Latvia 2007 Lithuania 2008 Romania 2008 Slovenia 2007 Poland 2007 Slovakia 2005 Estonia 2008 Hungary 2008 Czech Republic 2008 Bulgaria 2006 EU members before May 2004 2007 0 10 20 Source: WHO/Europe Health for All Database January 2010 Notes: Ischemic heart disease is basically heart failure due to coronary artery disease. Cerebrovascular diseases tend to be related to hypertension, resulting in strokes. 200 Aspirational Comparison of Latvia with the EU15 Countries 375. If we consider the difference between Latvia’s standardized death rate and the average of the EU15 to be Latvia’s excess mortality by cause relative to where it aspires to be, it starts to become clear Figure 6.4: Comparison of Standardized Death Rates (All Ages and Sexes) for Latvia, Lithuania, Estonia, and the European Union, 2007 Latvia Lithuania Estonia European Union Pre-2004(15 countries) European Union Today (27 countries) 0.0 200.0 400.0 600.0 800.0 1000.0 1200.0 Age Standardized Death Rate per 100,000 Population, 2007 Circulatory System Cancers External Causes Other Digestive System Respiratiory System Nervous System Endocrine Infectious and Parasitic Genito-Urinary System Mental Perinatal Congenetal Abnormalities Musculoskeletal System Skin Blood Pregnancy and Childbirth Source: Eurostat, April 2010 just how challenging Latvia’s health situation is. Figure 6.4 shows the major causes of death for all of the Baltic countries and both sets of EU countries. It becomes glaringly obvious, despite the detail, that the major sources of differences are two disease categories, circulatory system and external causes. These two account for 86 percent of the difference in deaths between Latvia and the EU15. For this reason we look only at these two causes, plus cancers, and put everything else in the “otherâ€? category. Latvia is only 15 percent higher than the EU15 in overall deaths due to cancer, but it is an important and expensive cause of death. The overriding policy question is how Latvia should spend its health Lat to improve these outcomes. If Latvia in 2007 had the EU15 death rates just for circulatory system diseases and external causes, about 6,800 people would not have died (about 20 percent of total deaths). 201 376. Figure 6.5 illustrates another problem common to CIS countries and Latvia: premature deaths of adult males. The top section shows standardized death rates for women in the EU15 and Latvia for all ages and for below 65. The bottom section shows the same for men. Comparing the top bar for Latvian men below 65 with the bottom bar for all Latvian men, 48 percent of all male deaths occur before age 65, compared to 28 percent for women. In the EU15, male deaths below 65 are only 29 percent of the total, and even more surprising, it is easy to see in the chart that death rates for males in Latvia below Figure 6.5: Comparison of Standardized Death Rates by Age and Sex for Latvia and the EU15, 2007 Age <65 EU 15 All Ages Women Age <65 Latvia All Ages Age <65 EU 15 All Ages Men Age <65 Latvia All Ages 0 200 400 600 800 1000 1200 1400 1600 Age Standardized Death Rate per 100,000 Population, 2007 Circulatory Cancers External Causes All Other Causes Source: Eurostat, April 2010 65 are almost as high as for the EU15 for all ages. 377. Figure 6.6 shows that circulatory system deaths are primarily attributable to heart attacks and strokes. Moreover, Latvia exceeds the EU15 for men and women at all ages by a factor of about 4 overall, but for men below 65, it is by a factor of nearly 6. 378. Figure 6.7 disaggregates deaths due to external causes. By international standards, male Latvians (the pair of bars at the bottom of the chart) lead extraordinarily dangerous lives. Almost all of the deaths due to external causes in Latvia are premature, 94 percent for men and 86 percent for women. This ratio is also high for the EU15, but not nearly as extreme, 78 percent for men and 59 percent for women. Accidents, especially highway accidents, falls, poisoning and exposure to noxious substances, self-harm, and assaults are shown by the EU15 numbers to be highly preventable but are extraordinarily common in Latvia among prime-aged men. 202 Figure 6.6: Composition of Cardiovascular Deaths by Age and Sex for Latvia and the EU15, 2007 Age <65 EU 15 All Ages Women Age <65 Latvia All Ages Age <65 EU 15 All Ages Men Age <65 Latvia All Ages 0 100 200 300 400 500 600 700 800 Age Standardized Death Rate per 100,000 Population, 2007 Other Heart Diseases Ischaemic Heart Diseases (Heart Failure) Cerebrovascular Diseases (Stroke) Source: Eurostat, April 2010 Figure 6.7: Composition of Deaths Due to External Causes by Age and Sex for Latvia and the EU15, 2007 Age <65 Latvia EU 15 Latvia EU 15 Women All Ages Age <65 All Ages Age <65 All Ages Men Age <65 All Ages 0 20 40 60 80 100 120 140 160 180 200 Age Standardized Death Rate per 100,000 Population, 2007 Transport Accidents Falls Accidental Poisoning/Noxious Substances Intentional Self-Harm Assault Event of Undetermined Intent Other Accidents Source: Eurostat, April 2010 203 Figure 6.8: Comparison of Health-Related Behaviors in the FINBALT Surveys, 2006 FINBALT Assessment of Health Behavior 2006 Daily Smokers 379. The FINBALT Health Monitor’s most recent nationally Exposed to Tobacco Latvian Men representative survey of 15-64 Smoke at Work 1-5+ year-olds’ health behavior Hours Per Day Lithuanian Men suggests that there is much to be Estonian Men done to improve the underlying Exposed to Tobacco Smoke at Home Finland Men causes of the bad health outcomes just observed. FINBALT 0 20 40 60 uses an almost identical survey Percent instrument and sampling strategy are used in Finland, Estonia, Latvia, and Lithuania at least every other year. Use Seat Belts in Rear Seat Nearly 380. Forty-seven percent of Always Latvian Men Latvian men and 18 percent of Lithuanian Men women smoke cigarettes daily. Use Seat Belts in Exposure to passive tobacco Front Seat Nearly Estonian Men smoke at work has dropped from Always Finland Men 36 percent for men in 1998 to 26 0 50 100 150 percent in 2006 (and to 10 percent for women), but it has Percent showed no improvement after 2004. Over 50 percent are exposed to passive smoke at 30 Minutes Exercise home because of the high rate of at least 2-3 Times a… daily smoking in the population. Not Visited a Doctor Last Year Latvian Men The top panel in Figure 6.8 shows Body Mass Index > 25 Lithuanian Men that smoking is a bigger problem Kg/m2 in Latvia than for its neighbors. Estonian Men Physical condition Although it will be a long and Poor or Very Poor Finland Men arduous task to get smoking down to Finland’s level – about 0 20 40 60 80 half of Latvia’s – reducing Percent exposure to smokers in the workplace, in public places, and Source: FINBALT Reports for Estonia, Finland, Latvia, Lithuania, 2006 Surveys. See at home are public health Annex Table 10. problems that can be addressed directly with other instruments (such as regulations) even if smoking remains prevalent. 204 381. Seat belt usage is definitely moving in the right direction. Use of front seat belts “almost alwaysâ€? increased from 64 percent among men in 2004 to 84 percent in 2006, with use going from 54 percent to 80 percent among men 15-24 years of age. Use of seat belts in the rear seat jumped from 6 to 21 percent during those two years. Finland has succeeded in getting rear seat occupants to wear seat belts 75 percent of the time. Latvia has 263 people killed or injured in road accidents per 100,000; Finland’s rate is 178. Although the gap will not be closed by seat belt usage alone, trying to close the seat belt gap with Finland would not be a costly intervention (see the second panel in Figure 6.8). 382. Traffic deaths are encouraged by drunk driving. Nearly 60 percent of Latvian men and 50 percent of women from age 15 to 34 know of someone who has driven under the influence of alcohol during the past year. The proportion starts declining for women after age 34, but it stays at nearly 60 percent for men until age 44. More generally, use of alcohol in the previous year was reported by 88 percent of the respondents. About 40 percent of men and 20 percent of women had consumed spirits in the previous week, and binge drinking of 6 or more portions of alcohol during 1 session at least once a week was common for 24 percent of men (but only for 6 percent of women). Nearly 60 percent of men and 47 percent of women know of one or more persons who have used narcotics in the previous year. Lowering the use of drugs and alcohol especially among the young would be beneficial; keeping them from behind the steering wheel of a car is a separate and possibly easier task to accomplish. 383. The bottom panel of 6.8 shows a collection of other general health statistics. Except for Finland, 40 percent or less of the population exercises regularly during leisure time in any of the countries. A more disconcerting issue is the high rate of failure to visit a doctor in Latvia. About 32 percent of Latvian men and 20 percent of women had not visited a doctor during the previous year, which is highest among these four countries. The fail-to-visit rate is 36 percent for men aged 45 to 54 (compared to 19 percent in Finland). An even higher 42 percent of Latvian men had not visited their general practitioner during the past year. This is quite a stunning statistic given the high morbidity and mortality Latvians face. On the positive side, though, Latvian men are the least likely of the group to be overweight or obese, and their self-reported health assessment is similar to that of their neighbors.125 384. There are apparent shortcomings in how Latvia addresses its high levels of morbidity and mortality connected to strokes and heart failure. Apart from excessive exposure to tobacco products, which is a risk factor for both problems, it appears that the health system is not monitoring and treating men effectively. For example, only 60 percent of men had their blood pressure checked (Figure 6.9) during the previous year. For older adult men, who in Latvia’s health environment should probably be receiving annual physical exams, the proportion having their blood pressure checked increases but never exceeds 72 percent. 125 The graphs in this section only include men to simplify and focus on the major at-risk group. Data for both men and women are included in Annex Table 10. 205 385. The second panel in Figure 6.9 shows the proportion of men diagnosed or treated for high blood pressure in 2006. Only about 10 percent were diagnosed or treated, less than half the proportion in Lithuania. The largest share of Figure 6.9: Comparison of Hypertension-Related Treatment in the FINBALT Surveys, the shortfall is accounted by a 2006 lack of diagnoses among the 45-54 age group, but it is also Blood Pressure low among the oldest group. Checked Last Year… The causes may be failure to Blood Pressure Latvian Men visit the doctor, a failure of Checked Last Year… Lithuanian Men procedures when a man visits Blood Pressure Estonian Men his doctor, or myriad other Checked Last Year… mistakes. Latvia almost Finland Men certainly does not have a 0 20 40 60 80 100 significantly lower incidence Percent of high blood pressure than its neighbors, given the mortality patterns already discussed. If Diagnosed or Treated anything it is likely to be for High Blood… higher, so these are actually Diagnosed or Treated Latvian Men missed diagnoses of an for High Blood… Lithuanian Men invisible killer. Diagnosed or Treated Estonian Men 386. The third panel shows for High Blood… Finland Men whether medication was used 0 10 20 30 40 50 the previous week to manage Percent high blood pressure. Again, Latvia ranks lowest by far in using this method to manage high blood pressure. For the Used Medicine for 55-64 age group, the High Blood Pressure… percentage using medicine is Used Medicine for Latvian Men higher than the percentage High Blood Pressure… Lithuanian Men newly diagnosed or treated Used Medicine for Estonian Men shown in the previous panel High Blood Pressure… (25.2 percent versus 23 Finland Men percent). Unless this is a 0 10 20 30 40 50 statistical fluke, it could be an Percent indicator of inadequately supervised treatment. Source: FINBALT Reports for Estonia, Finland, Latvia, Lithuania, 2006 Surveys. See Annex Table 10. 387. In Figure 6.10 we have a similar display of data 206 for cholesterol screening. Only 20 percent of Latvian men had their cholesterol measured during the year prior to the survey. 24 percent of men did not know if their cholesterol had ever been checked (top panel in Figure 6.10). 388. Whereas blood Figure 6.10: Comparison of Cholesterol-Related Treatment in the FINBALT Surveys, 2006 pressure screening was more even across countries, Blood Cholesterol cholesterol screening is a not Checked Last Year Age as common a practice, and 55-64 Finland tends to lead the way Blood Cholesterol Latvian Men in implementing it. For men Checked Last Year Age 45-54 Lithuanian Men 45-54, Latvians are much less Blood Cholesterol Estonian Men likely to have their cholesterol Checked Last Year checked than their neighbors. Finland Men Total In fact, Latvia almost does not 0 20 40 60 show up on the graph (middle Percent panel in Figure 6.10). 389. However, it seems to perform about the same as Diagnosed or Treated the other Baltic States in using for High Blood… medicine to manage high Diagnosed or Treated Latvian Men cholesterol, but this therapy is for High Blood… Lithuanian Men five times more common in Diagnosed or Treated Estonian Men Finland. Males in Finland from for High Blood… 0-64 experience a Finland Men standardized death rate of 44 0 10 20 30 40 per 100,000 from ischemic Percent heart disease, compared to 157 for Latvia. Yet 24 percent of men in Finland at age 55-64 Used Medicine for High are taking cholesterol- Blood Cholesterol Age lowering drugs, compared to 7 55-64 Last Week percent of Latvian men. And Used Medicine for High Latvian Men Blood Cholesterol Age again, we have the 45-54 Last Week Lithuanian Men unmistakable statistical Used Medicine for High Estonian Men conundrum of only 1.2 Blood Cholesterol Last Week Finland Men percent of Latvian men in this age group being diagnosed or 0 10 20 30 treated for the condition in Percent the previous year while 6.7 percent are taking drugs. This Source: FINBALT Reports for Estonia, Finland, Latvia, Lithuania, 2006 Surveys. Annex Table 10. 207 could be a statistical Figure 6.11: Percent of Men and Women Having Blood Cholesterol Measured During Previous 5 problem but it has Years, Latvia FINBALT Survey 2002-2006 happened for Latvia, for this age group, for both 70 hypertension and 60 cholesterol. If it reflects confusion by respondents, 50 it could mean a lack of Percent 40 knowledge about an important health risk for 30 Women 55- them. If it reflects 64 20 Men 55-64 something about medical practice in Latvia, it could 10 Women 45- suggest poor continuity of 54 0 care for these men and 2000 2002 2004 2006 poor handoff and monitoring of them from a Source: Eurostat, April 2010 specialist to the GP. Something is awry. Nothing about blood pressure or cholesterol screening requires a hospital bed and may ultimately prevent people from ending up in a hospital bed with a serious illness. 390. As with seatbelt use, the trends in cholesterol Table 6.1: Percent Having Heard about Public screening are more or less in the right direction. Figure 6.11 Health Programs, Latvia FINBALT Survey 2006 shows a steady increase in the percentage of older Latvians, Public Health especially older men, screened for high cholesterol. But a Campaign Heard About It dramatic improvement for everyone 45 to 64 is still in the Anti-Smoking 52% future.126 Passive Smoking 28% 30% Quit and Win 391. Two types of behaviors are important for success in (9% Participated) reducing Latvia’s high burden of morbidity and mortality. Healthy Heart 31% Individual Latvians can improve the healthfulness of their own Healthy Food 30% behavior, and there are distinct changes that can be made, Iodized Salt 20% from the simple act of wearing a seatbelt to the difficult but Family Health Week 17% important act of not smoking. The government, through tax Source: Latvijas JedzÄ«votÄ?ju VeselÄ«bu JetekmÄ“joÅ¡o Paradumu PÄ“tÄ«jums, 2006 policy, public information, laws, the efforts and examples of its (Health Behavior among Latvian Adult leaders, and many other low-cost interventions, can help to alter these behaviors (especially among the Population, 2006) young). Latvia is trying. However, Table 6.1 shows the percent of respondents in the FINBALT survey 126 In the United States, recommendations for men are to “…have your cholesterol checked regularly starting at age 35. If you are younger than 35, talk to your doctor about whether to have your cholesterol checked if you have diabetes, you have high blood pressure, heart disease runs in your family, and/or you smoke.â€? http://www.ahrq.gov/ppip/healthymen.htm 208 who have heard about specific public health campaigns. Obviously it is challenging to get people to pay attention, so efforts need to be coordinated across multiple fronts. 392. A second type of behavior is the conduct of the medical system. We have seen that despite extremely high mortality from cardiovascular failure, the health system is not consistently monitoring well the key at-risk men from age 45 to 64. Whatever the reason, this is a behavior that can be fixed with an effort by the MOH, specialists who treat these diseases, the family doctors tasked with monitoring chronically ill patients, and the payments system that creates the incentive structure under which providers work. The men have to begin experiencing higher and more consistent contact rates with the health system, they need to understand the importance of monitoring the risk factors for cardiovascular disease, and case management by the health system needs to be up to international standards. Improving Latvia’s performance on these measures will help to determine how fast it converges in health outcomes to the EU15. Summary – Health Conditions 393. The population of Latvia is not particularly healthy, and it lags well behind the EU15 and even the other accession countries that joined the EU in 2004 and 2007. In 2007, a male child born in Latvia could expect to live about 66 years. Roughly 51 of those years, or 77 percent, would be healthy. A male child born in Sweden could expect to live 79 years. Roughly 68 of these years, or 80 percent, would be healthy. That is 17 more years of healthy life than in Latvia. 394. Latvia is less different from Estonia and Lithuania, although Estonia is starting to move ahead. From 2000 to 2007, all three countries succeeded in reducing infant, child, and adult mortality, but Estonia has done significantly better. As a result, in 2007, 18.6 percent of Estonia’s adults between 15 and 60 years of age would die; in Latvia, 21.3 percent would die; and in Lithuania, 23.1 percent would die, while in Germany, for example, only 8 percent would die.127 395. The large gap in mortality – especially premature mortality – between Latvia and the EU15 has been persistent for years, but most of it is attributable to a handful of causes, namely the high probability that Latvians face of dying from heart attack, stroke, or external causes. This is especially a problem for men, but the better numbers for women are still multiples worse than for women in the EU15. 396. Preventing these deaths and the associated morbidity, apart from the impact on the welfare of those living healthier and longer lives, would have a positive economic impact on Latvia. It would reduce the loss of working adults, reduce levels of disability, increase labor force participation, and raise the incentive of people to invest in themselves (raising skill levels). Correcting the problem may seem like a long-term proposition, but the speed of improvement depends heavily on government spending decisions in health. In less than 20 years, the Czech Republic added 6 years to overall life expectancy. Latvia needs to add about 9 years to catch up to the EU15. 127 World Health Organization, World Health Statistics 2009; the latest data are available at www.who.int/whosis. 209 397. Often it is assumed that improved health outcomes, because they depend in large part on individual behavior, take a very long time. However, the FINBALT data suggest that there may be some straightforward improvements in medical practice that could directly affect mortality from cardiovascular disease. And there are now many examples of public health strategies that have been effective in causing substantial improvements in personal behavior. II. Cross Country Comparisons of Health Resources, Utilization, and Expenditure Overview 398. Table 6.2 shows a list of countries that will be used as comparators for Latvia’s availability of resources in health. The American state of West Virginia is also included, to provide a comparator from outside of Europe. It performs poorly relative to other states in the United States. Neither a comparison Table 6.2: Comparators for Latvia in Health Sector Resources, Most Recent Year Square Life Expectancy % Urban Population per Kilometers Population at Birth GDP/N in PPP Population km2 Ireland 70,273 4,422,077 80 $40,823 60.5 60.34 Lithuania 65,300 3,358,114 72 $15,739 66.6 52.06 Latvia 64,559 2,266,094 71 $15,389 67.8 35.42 West Virginia, USA 62,755 1,819,777 75 $30,831 36.1 29.06 Croatia 56,594 4,434,508 76 $14,309 56.5 78.53 Estonia 45,227 1,340,675 74 $19,155 69.1 29.79 Denmark 43,094 5,489,022 79 $35,125 85.6 126.12 Source: Europe from WHO/Europe HFA Database January 2010; West Virginia Wikipedia like this nor a regression-based comparison is perfect, but this approach allows us to see some of the variation that actually exists across countries. Because an important element of the debate over health resources in Latvia revolves around its physical size and a claimed high share of rural people, these two factors are emphasized in the choice of comparators in Table 6.2 which is ordered by geographical area (square kilometers). Latvia, Estonia, and West Virginia are the least densely populated, but over two- thirds of the population of Latvia and Estonia are urban. West Virginia is the least urban by far, with just over a third of the population in cities. There is a wide range in population and incomes in the table. Personnel 399. Despite the differences in incomes and size of country, except for some outliers, there is a fair amount of consistency in availability of human resources. Croatia stands out for low numbers of physicians and nurses overall, even though it maintains good coverage by general practitioners. West Virginia more or less doubles the coverage of general practitioners/primary care physicians relative to the others, and Ireland is an extraordinarily heavy user of nurses (Table 6.3). These two, plus the other OECD country, Denmark, are the top three in availability of nurses. 210 400. Latvia has the Table 6.3: Human Resources in the Health Sector per 100,000 Population, Most Recent Year lowest availability of General general practitioners Physicians Practitioners Dentists Nurses per 100,000 Croatia 259 65 72 524 population, even Denmark 317 77 79 953 though it is close to Estonia 328 63 87 655 the middle in terms of Ireland 303 72 58 1550 physicians overall, and Latvia 305 55 68 548 is second lowest on Lithuania 407 77 70 735 nurses. Are 1,250 West Virginia, USA 300 130 60 932 general practitioners Source: Europe, WHO/Europe European HFA Database January 2010; West Virginia, enough? Are Latvia’s statehealthfacts.org Note: Physicians are all practicing physicians; General Practitioners are the subset of Primary Care 12,472 nurses enough Physicians; Dentists are practicing dentists, and Nurses are practicing nurses. Depending on the (of whom nearly 77 country, the data are from 2006, 2007, or 2008, but these numbers change little during a three- percent work in year period. 128 hospitals)? Unknown; however, the result is heavily influenced by budget decisions both in health and education. The most market driven health sector of the comparators, West Virginia, has 136 percent greater density of general practitioners and over 70 percent greater density of nurses than Latvia, the least expensive professionals in the health system. Hospitals129 401. There is wider variation in the availability of hospital resources, ranging from a total of 349 beds Table 1: Hospitals per 100,000 Population, Most Recent Year Total Acute Care Acute Care Number of Total Number Percent Average Psychiatric Hospital Beds Hospitals per Hospital Beds Acute Care of Acute Care Privately Acute Care Hospital Beds per 100,000 100,000 per 100,000 Hospitals Hospital Beds Owned Hospital Size per 100,000 Croatia 535 1.14 352 52 16,053 <1 308 94 Denmark 349 1.16 310 62 16.779 2 271 60 Estonia 557 2.68 380 36 5,101 10 142 56 Ireland 534 1.15 274 50 11,627 233 85 Latvia 757 3.08 523 70 11,910 6 170 137 Lithuania 814 2.40 509 81 17,193 <1 212 102 West Virginia, 73 nonprofit 410 3.08 410 56 7,476 133 71.2 USA 17 for profit Source: Europe, WHO/Europe European HFA Database January 2010; West Virginia, statehealthfacts.org. “Psychiatric hospital bedsâ€? is in italics for West Virginia because it is the US average (available at http://www.cdc.gov/nchs/data/hus/hus09.pdf#099). Note: The data are for 2007 except Ireland is 2006. 128 These are the number of GPs, nurses, and proportion of nurses in hospitals reported in 2007 to WHO. Source: Health for All Database (http://www.euro.who.int/HFADB) 129 Note that we compare hospital resources but not primary care or outpatient resources. That number is not consistently reported by countries, and what is included is too variable to compare. Probably the best guide to outpatient services for this sample is the number of general practitioners, discussed under Personnel. 211 per 100,000 people in Denmark to more than double that, 814 beds, in Lithuania (Table 6.4). Latvia rivals Lithuania in density of bed coverage, at 757 beds per 100,000 in 2007. 402. Just focusing on acute care, at 3.08 acute care beds and 523 acute beds per 100,000, Latvia has more than 290 percent the coverage of hospitals, and from 150 to 190 percent the number of acute care beds per 100,000 in Croatia, Denmark, or Ireland. Latvia leads all of the countries in density of acute care hospitals, acute care beds, and psychiatric beds per 100,000 population. The average acute hospital size, at 170 beds, is in the middle of the range but about a third smaller than in Denmark or Ireland. The EU15 averaged 3.09 hospitals and 554 beds per 100,000 in 2006 (but only 375 acute care beds per 100,000), the most recent year for which the data are available. 403. Ownership of hospitals is almost entirely by governments except in West Virginia. In that state, government-owned hospitals account for only 10 percent of the total, while 73 percent are independently owned nonprofit hospitals and 17 percent are for-profit. 404. There is little question that Latvia still has an over-capacity of acute care hospitals and beds, and that it could continue to reduce their number. It almost certainly has excessive psychiatric beds, exceeding the density of Croatia, for example, by over 40 percent. Reducing total hospital beds per 100,000 to the EU15 average would mean closing 4,600 beds. Halving the density of psychiatric beds, part of the overall total, would mean closing nearly 1,600 beds, leaving 3,000 acute care beds to be closed. This change would simply put Latvia even with Estonia’s progress in reducing beds. However, Estonia has not been equally successful in reducing the number of hospitals, so its average hospital has 131 beds and it continues to have 1 more hospital per 100,000 people than the average in the EU15. For Latvia, to reach the EU15 average for both hospitals and beds, it would need to close 26 hospitals. The result would be fewer hospitals of higher capacity, both physically and medically. 405. Simply closing hospital beds but keeping the buildings in operation – unless the buildings are given a new, valuable economic purpose – does not remove the hospitals from the public budget. Almost all of Latvia’s hospitals are owned by the State or by local governments, so keeping them open to provide another type of medical or social service would simply mean shifting them off of the inpatient medical care budget and onto another part of the budget. They would become less costly, of course, because they would no longer be treating inpatients; yet the important, sustainable step for Latvia to take is to transform unneeded medical resources into economically valuable activities. 406. In addition, closing hospitals would free nurses and staff physicians to move to other hospitals and to other endeavors. These professionals could be redeployed to higher impact purposes in the health sector, such as increased home care, and strengthening primary care. 407. It is important to put the EU15 indicators into perspective. In a similarly large market, the United States, the number of staffed acute care beds declined from 282 to 241 per 100,000 population from 1996 to 2006, a 13 percent decrease (compared to 375 acute care beds per 100,000 in the EU15 in 2006). In many markets of the US, particularly in the west and northeast, the density was substantially 212 lower, closer to 150 beds per 100,000. Therefore, even the EU15 averages may not reflect the true potential for Latvia to reduce hospital capacity.130 408. In the United States, the reduction in hospital beds has been accompanied by an increase in hospital-based employment, which rose by 6.2 percent over ten years to 2006 – nurses increased by 14 percent – as hospitals shifted their product mix to diagnostic services and short-stay or outpatient based services. 131 This could be a future path for Latvia. Choosing how fast to converge to the EU15 or to move beyond it depends almost completely on public expenditure choices because its hospitals are financed by the public budget. Use of Services 409. Latvia is just behind Table 6.5: Use of Services, Most Recent Year Acute Care Average Length Outpatient Lithuania in the Admissions Bed Occupancy of Stay Acute Contacts per rate of per 100 Rate Care Hospitals Person admissions to Croatia 15.4 85 7.3 6.0 acute care Denmark 17.8 84 3.5 4.2 hospitals – Estonia 16.8 70 5.8 7.4 about 20 per Ireland 14.1 87 6.1 6.6 100 population Latvia 20.0 76 7.2 6.0 were admitted Lithuania 21.5 73 6.3 7.0 to an acute West Virginia, USA 15.8 67 4.8 4.1 care hospital in Source: Europe, WHO/Europe European HFA Database January 2010; West Virginia, 2008, statehealthfacts.org. West Virginia numbers are incomplete. Italicized numbers are for the US as a whole. Depending on the country, the data are from 2006, 2007, or 2008. compared to the low of 14 in Ireland in this group. Estonia has reached about 17, which should also be feasible for Latvia (Table 6.5). 410. Average length of stay in Latvia, at 7.2 days, is among the highest for this group and could decline by at least by 1 or 2 days without beginning to approach the lower bound, Denmark, which has an average of 3.5 days per stay, less than half Latvia’s. In contrast to these measures of inpatient care, Latvia is well below Estonia and Lithuania in the use of outpatient care but is much higher than Denmark and the U.S. Latvia’s bed occupancy rate, 76 percent, is right in the middle. 130 We realize that in 2009 and 2010, Latvia has reduced the numbers of hospitals and beds contracted with the Payments Center, as detailed later in this chapter. Of course international comparison lag behind the current reality, but we should not assume that the other countries we are using for comparison have been idle in cutting capacity over the same period. 131 David C. Goodman, Elliott S. Fisher, Kristen K. Bronner (2009) “Hospital and Physician Capacity Update A Brief Report from the Dartmouth Atlas of Health Care Atlas Projectâ€? Available at http://www.dartmouthatlas.org/atlases/Capacity_Report_2009.pdf 213 Figure 6.12: Trends in Use of Health Services, 2000 and Most Recent Year Acute care hospital admissions per 100 Lithuania 2008 Latvia 2008 Estonia 2008 Denmark 2001 Ireland 2007 0 5 10 15 20 25 2000 Last available Average length of stay, acute care hospitals only Latvia 2008 Lithuania 2008 Estonia 2008 Ireland 2007 Denmark 2005 0 5 10 2000 Last available Outpatient contacts per person per year Estonia 2008 Lithuania 2008 Latvia 2008 Denmark 2007 Ireland 1988 0 5 10 2000 Last available Source: WHO/Europe Health for All Database January 2010 411. The panels in Figure 6.12 illustrate changes since 2000 in these measures to the most recent year, which is 2008 for Latvia. Latvia’s admission rate is unchanged; length of stay has dropped by about 1.5 days, and outpatient contacts have increased more than in any of the other countries. Although bed occupancy rate is not included, it changed little in any of the countries except for increases in Estonia and Ireland. Expenditures 412. Data on total health care expenditure are quite old (2005), but only the absolute amounts will have changed significantly (Table 6.6). The ratios do not move much. Estonia, Latvia and Lithuania tend 214 to spend relatively less of GDP on health, although Latvia is high in this group, and particularly Latvia and Lithuania have relatively small shares accounted for by the public sector, in Latvia’s case only 61 Table 6.6: Expenditures for Health Care, Most Recent Year Private Out of Total Health Total Health Pocket Expenditure Expenditure Per Public Sector as Inpatient as Expenditure as in GDP Capita PPP Percent of Total Percent of Total Percent of Total Croatia 7.4 $1,001 81 18 Denmark 9.4 $3,169 84 36 15 Estonia 5.0 $846 77 33 21 Ireland 8.2 $3,125 80 59 12 Latvia 6.4 $860 61 52 39 Lithuania 5.9 $862 67 32 West Virginia, USA 15.0 $6,714 46 31 14 Source: Europe, WHO/Europe European HFA Database January 2010; West Virginia, statehealthfacts.org. West Virginia numbers are incomplete. Italicized numbers are for the US as a whole. The most recent data are generally for 2005, except US is 2007. Note: PPP is purchasing power parity. It adjusts for differences in prices across countries to attempt to equalize what $1 will buy across countries. Therefore it will be different from the number obtained by using the exchange rate. percent. This leaves 39 percent of spending in 2005 in Latvia coming from households through out-of- pocket payments, employer contributions, and insurance. Spending on inpatient care is a relatively high share of the total, over half. This share in Denmark, Estonia, and the U.S. is 36 percent or less. Summary – Cross Country Comparisons 413. This section presents comparisons with other countries or states that are about the geographic size of Latvia, although population and other characteristics differ. These comparisons show Latvia viable alternatives to its approach in the health sector. In short, the comparisons suggest: ï‚· There are no big differences across countries in availability of human resources in health, although Latvia ranks relatively low in general practitioners and nurses. ï‚· Closing 1600, or half, of its psychiatric beds and 3,000 acute care beds would allow Latvia to match Estonia’s progress in reducing beds and to reach approximately the EU15 average. It probably needs to do better than Estonia in closing hospitals; however, and closing 26 hospitals would get it to the EU15 average. With technology changing the role of hospitals quickly, just reaching the EU15 average might be less than Latvia should aspire to achieve. ï‚· It is essential also to achieve other improvements in efficiency of hospitals, including reducing the rate of hospital admission and average length of stay. ï‚· Reducing hospital infrastructure and moving patients to other care settings would presumably reduce the share of spending on hospitals and allow the government to better offset the high out-of-pocket spending of Latvian households ï‚· There is room to increase spending on health as a percent of GDP, and as the economy recovers, pressures to do so will increase. However, spending either the current public budget or future increments on the existing stock of resources (such as Latvia’s stock of hospitals and beds) – 215 would probably be wasteful and do little to improve health. Certainly past spending on these resources has achieved little in terms of improving measurable health outcomes for the population. ï‚· There is a need to better understand the connection between how health sector resources are used today and how they could be better deployed to improve life expectancy and years of healthy life quickly. Latvia has room to deploy its health care resources more effectively, but its medical and scientific establishment will need to focus on how best to do this. 216 Box 6.1: A “Base-Closingâ€? Strategy for Hospitals In all of the countries of Eastern Europe and the former Soviet Union, the inherited and excess hospital infrastructure was devolved 20 years ago to local and regional governments, and to universities, with the State maintaining ownership of some of the main national or university-based hospitals. In most cases they operate as independent organizations under a state-owned enterprise law. Because all are backed by governments, they have been nearly impossible to close or to shrink significantly in any country. Often they are a major economic entity and employer in the locality where they exist. Politicians consider it political suicide to be associated with the closing of such a resource in their community. Exactly the same could be said of military bases in the United States. There are often too many of them, bases are owned by the federal government, they are important contributors to local economies, they are important employers of civilians, and any congressman would consider it political suicide to be associated with the closing of a base in his or her district. After reaching an impasse many times over the problem of closing bases, Congress created a Base Realignment and Closure (BRAC) procedure to remove politics from the process. BRAC rounds were completed in 1988, 1991, 1993, 1995, and 2005, closing or realigning almost 400 bases. These repeated tests of the procedure have shown that it is possible to make sensible technical and economic choices despite the enormous political hazards of the resulting decisions. The key to the procedure is that after Congress is presented a list of bases to be closed, its only option is to reject the entire list or to do nothing and allow it to go forward. It has always done nothing. Could the same procedure be used in Latvia to close hospitals? Here is how the base closing approach might be adapted: Step Action Possible Timeline 1. Prime Minister Appoints Hospital Closing Commission (HCC) (the BRAC Commission has 9 November 1, 2010 members) 2. Ministry of Health prepares a report and proposal of hospital closings and realignments December 1, 2010 consistent with its strategy and estimated needs through 2025. [If slated for closure, a hospital would lose its license to operate and end medical operations completely; it would not just lose its inpatient contract with the payments department of the MoH.] 3. HCC conducts hearings and visits the hospitals to be closed or realigned. It may add or March 31, 2011 subtract hospitals from the list. HCC submits its final report to the Prime Minister. 4. Prime Minister accepts or rejects the report in its entirety. April 15, 2011 5. If Prime Minister does not reject the report, the Saeima then has 45 days to reject the June1, 2011 report in its entirety by passing a resolution to that effect. If no action is taken, the Ministry of Health is automatically required to implement the recommendations. A special process would probably be needed for psychiatric beds. One option would be to have a technical panel evaluate every patient currently occupying an inpatient psychiatric bed and recommend whether that patient could be managed on an outpatient basis. Based on this review and international practices, guidelines would be drawn up for a stringent process governing future admissions for inpatient care. Current patients and their families would be given the option to follow the new guidelines on a voluntary basis, but no current occupants would be ejected involuntarily, and they would be welcomed back if they could not make the transition to outpatient care. However, future admissions would be governed strictly by the new guidelines, and the MoH would make recommendations on inpatient psychiatric infrastructure based on the expected impact of these guidelines and the gradual attrition of current residents. Finally, perhaps this “base closingâ€? procedure could be used to help determine appropriate status and ownership of hospitals. Should they be owned by government entities, shifted to public trusts (as in the UK), shifted to corporate entities (as is happening to a limited extent in Poland), to charitable foundations (common in the US), or sold? 217 III. Recent Trends in Latvia’s Health Policy and Expenditures Overview of Public Health Expenditures to 2010 414. Despite its low public fiscal effort in health relative to other advanced countries, Latvia consistently raised real per capita government spending on health care from 2000 to 2007 (the bars in Figure 6.13), Figure 6.1: Government Spending on Health as a Percent of GDP and in Inflation-Adjusted Lats, Latvia, by about 2000-2010 139 percent Per in real 160 3.7 Capita terms. (MoH) 140 Percent Real Lats (Base = 2000) However, 3.5 of GDP 120 Percent of GDP with the 3.3 (MoH) 100 beginning of 3.1 economic 80 2.9 60 problems in 2008, 2.7 40 spending 2.5 20 began falling 2.3 0 and by 2010 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 will drop back nearly Source: Health Compulsory Insurance State Agency News No. 16, page 7, for 2000 to 2008 with revisions to its 2005 from the MoH. GDP is from the Central Bureau of Statistics. Population from Eurostat. Staff calculations for level. Still, it 2009 and 2010 health budgets based on MoH data. This figure uses 2000 as the base; 2003 is the base for all subsequent real or deflated numbers. remains 74 percent higher than in 2000. The rise in per capita spending is assisted by a 5.5 percent decline in population over this period. 415. Health spending by government increased to 3.8 percent of GDP in 2007 (the blue line in Figure 6.13) and has fluctuated around that level since then. After 2008 both spending and GDP dropped, and the Government appears likely to keep spending at around 3.6 percent of GDP even if that means a decline in the absolute amount. 416. The State budget finances about 82 percent of the activity in contracted health providers. “Paid health servicesâ€? from private patients finances another 9 percent, uncompensated “patient depositsâ€? account for 3.5 percent (these are copayments connected with use of Compulsory State Insurance benefits), and municipalities account for less than 1 percent. Facilities gain another 5 percent from EU funds and their own non-health services (such as cafeterias). 417. This section of the chapter focuses on the 85.5 percent that comes from the State and patient copayments (excluding the 9.5 percent from private paid services, the 1 percent from municipalities, and the 5 percent from EU or own resources). Latvia does not have a significant, purely private health 218 sector. Almost all formal establishments (hospitals and clinics) are owned as limited liability companies by the State or by local Table 6.7: Expenditures for Health Care, 2008 through 2010, Constant 2003 Lats governments. General Percent Percent of the of the Percentage Percentage practitioners are in Budget Budget change 2009 change 2010 private practice; some 2008 2010 vs. 2008 vs. 2008 specialists are staff in 01.00.00 Sector Management 1% < 1% -33% -51% clinics and hospitals, and 02.00.00 Medical Education 4% 4% -35% -33% others are in private 06.02.00 Medical History Museum < 1% < 1% -30% -58% practice. Pharmacies are 33.00.00 Health Care Provision 77% 79% -27% -22% private. Medical Treatment 64% 55% -31% -34% Medicines and Material 12% 17% -10% 1% The Crisis in 2009 and Central Procurement Vaccines and 1% 2% 10% 46% 2010 and How the Dialysis Population Genome Database Project < 1% < 1% -36% -50% Health Sector Coped Education VSIA Children Clinical < 1% < 1% -13% -33% University Hospital Expenditures Rare Disease Medical Treatment of < 1% < 1% Children (new) 418. Government Social Safety Net (new) < 1% 6%% discussions of the 37.00.00 International Commitments 1% 1%% -2% -10% budgets are based on 39.00.00 Specialized Health Care Provision 12% 12% -17% -26% nominal amounts. With Communicable Disease 4% 4% -21% -27% inflation of 3.5 percent in Sports Medicine < 1% < 1% -29% -35% 2009 and -3.7 percent Blood Supply 1% 1% -21% -25% expected in 2010, this Emergency Medical Assistance 5% 6% -5% -10% may not cause great Disaster Medicine 1% < 1% -41% -100% harm. However, from this Forensic < 1% < 1% -33% -43% point forward we will use 40.00.00 E-Health Implementation < 1% < 1% -42% -61% constant 2003 Lats.132 45.00.00 Medical Administration of Funding 1% 1% -35% -63% The two columns of 46.00.00 Health Care Inspectorate 1% 1% -30% -25% numbers in Table 6.7 47.00.00 Health Economic Evaluation < 1% < 1% 21% 60% show the change from Medicines and Medical Devices 48.00.00 < 1% 1% -9% 56% Evaluation and Registration 2008 to 2009 in actual European Regional Development 62.00.00 1% 1% 419% -34% spending and from 2008 Fund 63.00.00 European Social Fund < 1% 1% 671% 955% to the 2010 budget. 70.00.00 Other European Union < 1% < 1% -42% -34% EEA and Norwegian Financial 419. The budget was a 71.00.00 < 1% < 1% 17% -27% Mechanisms moving target for all of Eliminated institutions and funds 2% 0% 221% -100% during 2009 2009. In nominal terms, Total Overall -17% -24% the MoH started the year National Budget Only (no EU) -21% -25% with LVL 533 million but Source: Ministry of Health for the nominal amounts. Deflated to 2003. Emergency Medical in the June supplemental Assistance was included in Health Care provision in 2008, in 2009 it started shifting to the MoH. For consistency, in this table it is treated as if it is always in the MoH. See Annex Tables 5 and 6. 132 Note, however, that figure 9 has a base year of 2000, not 2003, but from this point forward 2003 is the base. 219 budget, its allocation dropped by 15 percent to LVL 452 million, LVL 81 million below the original amount. The final out-turn, after additions in the second half of the year, was LVL 493 million, only 7.5 percent below the original estimate. Nevertheless, this was nominally LVL 70 million below 2008. In real terms, the MoH experienced the 17 percent cut shown in the second-to-bottom line in Table 6.7. According to initial budget estimates, it faces another 7 percent decline in the 2010 budget, bringing it 24 percent below the 2008 amount. 420. The ministry responded in 2009 with budget cuts that prioritized the maintenance of service levels in the following areas (in a relative sense, as they generally were also cut): primary care (general practitioners), services for children and pregnant mothers, emergency medical assistance (ambulances and emergency care at hospitals), and subsidies for covered pharmaceuticals. By October it added emergency safety net provisions, which supplemented financial protection for needy and low income households. The cuts are briefly described below, and we examine in more detail later exactly what happened. 421. Streamlined MoH: The Government reorganized the Ministry and its affiliated agencies, and cut employment. Sector Management, the first line of Table 6.7, was cut by 33 percent (with another cut of 18 percent planned for 2010). However, it accounts for only 1 percent or less of the budget so had little impact on the total amount saved. A number of organizations were eliminated or merged by the ministry.133 422. Reductions in Health Care Provision: Health care provision accounted for 64 percent of spending in 2008; necessarily that is where large amounts of money could be saved. The Ministry targeted this part of the budget for a 28 percent cut in 2009 while trying to maintain the services listed above. Thus large cuts were taken from hospitals and secondary ambulatory services. 423. Reductions in Specialized Health Care Provisions: The other largest single part of the budget, accounting for about 12 percent of spending, includes infectious disease control, treatment of communicable diseases (e.g. tuberculosis and HIV/AIDS patients), and the other items in Table 6.7. All of these elements except emergency medical assistance experienced large cuts. Employment and Pay in the Health Sector 424. Table 6.8 shows, for approximately the same line items as Table 6.7, changes in posts and remuneration from mid-2009 to what is anticipated in the 2010 budget. Items that are really procurement of services in Table 6.7, such as Health Care Provision, do not have a separate entry for staff. In that case the staff who manage it are under “Medical Administration of Funding.â€? 133 Those cuts for the most part show up in the line for “Disaster Medicineâ€? and “Eliminated institutions and funds during 2009â€? near the bottom of the table. These allocations may not have entirely disappeared because they reflect both the discontinuance of some activities and the merging of others. 220 Table 6.8: Changes in Posts, Consultants, and Remuneration from mid-2009 to the 2010 Plan Change in Change in Total Expenditures on Posts Change in Remuneration Remuneration Freelancers 2010 Posts per Post 2010 Change Sector Management -28% 3% 85 -27% 13,586 -1% Medical History Museum -25% 87 -16% 4,170 -11% Communicable Disease -4% 1,311 -3% 6,591 -1% Sports Medicine -17% -100% 74 0% 6,872 -17% Blood Supply -17% 5% 343 0% 6,579 -17% Disaster Medicine -100% - -100% Forensic -17% 4% 206 0% 7,095 -17% E-Health Implementation -48% -100% 3 -57% 11,979 21% Medical Administration of Funding -49% -34% 139 -38% 9,287 -17% Health Care Inspectorate 33% 38% 252 0% 8,849 33% Health Economic Evaluation 76% -100% 68 10% 11,476 61% Medicines and Medical Devices -16% -57% 156 0% 12,161 -16% Evaluation and Registration European Social Fund 13% - -100% Other European Union -51% -54% 22 -58% 18,048 15% EEA and Norwegian Financial 26% 1 0% 11,872 26% Mechanisms Eliminated institutions and funds -100% -100% - -100% during 2009** Total -33% -17% 2,747 -23% 7,057 -13% Source: Ministry of Health. Emergency Medical Assistance is eliminated from the count because it was being absorbed into the MoH gradually during this time period. With it, total posts in 2010 are 5,449, compared to 5,364 in 2009. We only account for the 2,747 non-EMA posts in this table. Remuneration per post in 2010 is in nominal 2010 Lats, but all of the percent change columns referring to Lats show the real change in constant 2003 Lats. See Annex Tables 7 and 8. 425. Already some changes to employment and remuneration had taken place before mid-2009, but they are not accounted for here. Even without data Figure 6.14: Employment in Human Health Activities in Latvia, 2005 - 2009 from the beginning of 2009, 50000 41,552 42,202 43,669 41,466 the changes are substantial. 41,038 The total expenditure on 40000 32,470 33,126 32,726 33,577 30,894 staff (remuneration in the 30000 first column) will drop by 33 20000 percent in 2010, 8,568 8,426 9,476 10,092 10,572 expenditures on consultants 10000 (freelancers) will fall by 17 0 percent, the number of 2005 2006 2007 2008 2009 posts will drop to 2,747 (by Total Public Private 23 percent), and Source: Central Statistics Bureau remuneration per post will fall by 13 percent to LVL 221 7,057 (LVL 588 per month). 426. Figure 6.14 shows that employment in the sector dropped from 2008 to 2009 by 2,683 in the public sector. It was offset by a gain of 480 in private employment, for a net loss of 2,203 in the sector. This was a 5 percent drop from 2008. Hours worked between the end of 2008 and the end of 2009 dropped by 17 percent, including an 18 percent decline in the public sector and a 10 percent decline in the private sector. Nominal wages of health workers fell from LVL 526 to LVL 475 per month from 2008 to 2009. In constant Lats, the drop was 13 percent. For those working in local government institutions – many of the caregivers – the drop in wages was 17 percent. 427. As the details above clearly show, the health sector shrank substantially in 2009; a rough estimate based on these data would be that earnings declined by about LVL 40 million in 2009. The MoH data suggest a substantial further shrinkage in 2010. The change in 2009 mainly took the form of a reduction in hours worked and less in the number of jobs. Latvia’s Purchase-Provider Split 428. Latvia finances health care through general taxation, but there is a purchaser-provider split. The Payments Center in the Ministry of Health behaves like the payments side of an insurance system. This approach makes good sense in that it raises the flexibility of the MoH to manage its affairs; it generates an enormous amount of centrally-held and processed information about the functioning of the system, but it allows virtually all of the daily operational decisions to be made at the facility or provider level. Hospitals are paid a combination of fixed payments and fee-for-service, specialists are paid fee-for- service, and general practitioners are paid for the most part on a capitation basis. 429. Unlike in a direct Figure 6.15: Overall Health Budget and Payments System Share of the Spending, delivery system, this split gives Latvia, 2003-2010 a service entitlement to citizens. Virtually 100 percent 450,000,000 of Latvians are members of the 400,000,000 Real Lats (Base = 2003) system and are registered with 350,000,000 a General Practitioner (GP), 300,000,000 who acts as the gatekeeper for 250,000,000 most secondary outpatient 200,000,000 services and hospital 150,000,000 resources. 100,000,000 50,000,000 430. Figure 6.15 shows that - up to 2007, almost all of the 2003 2004 2005 2006 2007 2008 2009 2010 funding was channeled Overall Health Budget Payments System for Medical Care through the payments system. The wedge grew after 2007 due to MoH administered Source: Health Compulsory Insurance State Agency News for various years and 2010 health budget. 222 funds for Medical Education, European Regional Development Funds, and Specialized Health Care Provision (as shown in Table 6.7, this category includes Communicable Disease, Sports Medicine, Blood Supply, Emergency Medical Assistance, Disaster Medicine, and Forensic Medicine). Of these, Specialized Health Provision could certainly be financed through the payments system on a contractual basis so that the contract-based purchaser/provider split is maintained for as many service provision agencies as possible. 431. The change in status Figure 6.16: Distribution of Medical Care Spending 2003-2010 (in Constant 2003 Lats) for Emergency Medicine from the contract system to 350,000,000 Patient direct provision is explicit – Pharmaceuticals these teams have now been 300,000,000 Emergency Medical brought into the Ministry of 250,000,000 Teams Health and are counted Dentistry 200,000,000 there as employees. Previously they would have 150,000,000 General Practitioners been part of an institution 100,000,000 Secondary Outpatient contracted through the 50,000,000 payments system. The Inpatient Care purpose of the change is to - 2004 2005 2006 2007 2008 2009 2010 unify the system under one authority and to rationalize what was an overly Source: Health Compulsory Insurance State Agency News for various years and the decentralized patchwork of 2010 health budget. municipal providers with redundancies and operational bottlenecks. However, once it is rationalized, there is no reason to keep it within the civil service. In fact, in reforms announced in February 2009, the MoH plans to shift most of these institutions to contracts, Table 6.9: Distribution of Medical Care Spending in Percent expecting to do so in 2010. 2005 2008 2009 2010 Impact of the Crisis on the Inpatient 48 50 45 35 Composition of Government Secondary Ambulatory 14 18 18 18 Expenditures General Practitioners 11 9 11 14 Dentists 2 1 1 2 432. Figure 6.16 illustrates how Emergency Medical 6 6 6 7 payment system spending has been Assistance distributed over the years.134 In the Patient Pharmaceuticals 15 15 16 22 Centrally Procured pre-crisis years of 2005 through 2008, 4 1 1 2 Pharmaceuticals Latvia focused additional resources Settlements with the EU 0 0 1 1 Source Health Compulsory Insurance State Agency News for various years 134 account for almost all spending, 97 percent in 2007, 98 percent in Medical care and patient pharmaceuticals and its 2009 operational report. 2008, and 91 percent in 2009. There are three other spending categories: “other programs,â€? “reserve fund,â€? and central procurement of medicines (such as vaccines and chemicals for dialysis). They constitute a small but consistent fraction of the budget. These other spending categories are not included Figure 6.16. 223 on inpatient care, secondary ambulatory services, and patient pharmaceuticals. Inpatient expenditures rose by 79 percent in real terms from 2005 to 2008. Secondary outpatient ambulatory payments rose by 121 percent. General practitioners saw an increase over the period of 45 percent. Payments to dentists increased 28 percent. Real spending on pharmaceuticals rose by 73 percent. 433. Table 6.9 shows the impact. By 2008 inpatient and secondary outpatient spending had reached 68 percent of the total, up from 62 percent in 2005. Everything went up in absolute amount, but the share of basic outpatient services (general practitioners and dentists) fell. 434. Did productivity go up? If we measure it in the simplest possible way – outputs – the answer is “not much.â€? Quality of care may have changed, amenities may have changed, but volumes were much less elastic than expenditures. The first two columns in Table 6.10 show the increase in expenditures from 2005 to 2008 and the corresponding change in volume. Table 6.10: Response of the Health System to Budget Increases and Budget Cuts 2005 to 2009 (based on constant 2003 Lats) Change in Change in Expenditures Change in Volume Expenditures Change in Volume 2005 to 2008 2005 to 2008 2008-2009 2008 to 2009 5% (unique inpatients) -21% (unique inpatients) Inpatient 79% -27% 4% (bed-days) -27% (bed-days) Secondary Outpatient 121% 16% -27% -9% General Practitioners 45% 18% +4% -1% Dentists 28% ? -11% ? 43% (patients) +1% (patients) Patient Pharmaceuticals 73% -9% 59% (prescriptions) -1% (prescriptions) Source Health Compulsory Insurance State Agency News for various years and its 2009 operational report. 435. On a 79 percent increase in expenditures, hospitals produced a 5 percent increase in inpatients and a 4 percent increase in the number of bed-days. Virtually all of the increased cost was consequently greater intensity of care, or higher cost per patient. For an even larger increase in expenditures on secondary outpatient services, they saw only 16 percent more patients. As fee-for-service is an important element of reimbursement at these two levels, they can increase revenues by doing more for each patient. The experience for hospitals can be seen in Figure 6.17. Budget and volume estimates for 2010 foresee another drop in inpatient volume and a rise in cost per patient. 436. Spending for General Practitioners rose 45 percent and volume increased by 18 percent. As their compensation is for the most part through capitation, the increase in use – close to the change in volume for secondary outpatient services – is probably a reflection of greater demand for services by the population rather than efforts by the GPs to increase patient visits. In contrast, the high additional expenditures per patient by hospitals and secondary outpatient services would reflect decisions of the providers to a large degree. They have a financial incentive to do more while general practitioners have a financial incentive to do less. 224 437. For dentists, we Figure 6.17: Cost Per Inpatient Episode (in Constant 2003 Lats) and Changes in Volume do not have a direct measure of output, but 400 600,000 from 2005 to 2008, the 350 published cost per child 500,000 treated rose in real 300 Constant 2003 Lats Number of Cases terms by 32 percent, 400,000 250 suggesting a small reduction in volume as 200 300,000 the number of children 150 200,000 declined. It is easy to see 100 the relative costs and 100,000 volumes in the top and 50 bottom panels of Figure 0 - 6.18. Like hospitals, 2005 2006 2007 2008 2009 2010 emergency medical Cost per Inpatient Episode Inpatient Cases (est.) assistance also steadily has increased the cost Source: Health Compulsory Insurance State Agency News for various years and the 2010 per call over this period health budget. on a fairly constant volume (Figure 6.18). 438. The prescription drug benefit was expanded significantly in 2005. The number of patients receiving medicines increased by 43 percent by 2008, and the number of prescriptions rose by 59 percent. Increased pharmaceutical spending was a real added benefit that broadened coverage of patients and prescriptions. The cost per patient receiving support rose 30 percent over the period, but the cost per prescription rose a much smaller 18 percent. 439. In 2009 inpatient care resources suddenly dropped 27 percent in one year (column 3 in Table 6.10). Hospitals responded with a huge reduction in volume (the last column in Table 6.10 and clearly visible in the bottom panel of Figure 6.18). Hospital admissions fell 21 percent and bed-days fell 27 percent. In fact, admissions and bed days fell 17 percent and 24 percent, respectively, relative to 2005, even though the budget even in 2009 was 31 percent higher in real terms. As hospitals gained resources, they spent more per patient; as they lost resources, they cut patients. 440. Expenditure on secondary outpatient services dropped by 22 percent in 2009, to below 2007 levels (but still 72 percent higher than in 2005). Patient volume dropped 9 percent. Budgets of general practitioners were prioritized in 2009, rising 4 percent year-on-year and reaching a 50 percent increase over 2005. Their patient numbers dropped by only 1 percent. Dentists ended up with a cut of 11 percent in their budgets, leaving them 15 percent above 2005. 225 441. Like General Practitioners, Figure 6.18: Cost Per Patient Episode (in Constant 2003 Lats) and Changes in Volume pharmaceuticals were 45.00 prioritized during the 2009 40.00 contraction. Expenditures fell 35.00 only 9 percent in 2009, and 30.00 they remained 56 percent 25.00 above the 2005 level. The 20.00 cost of a prescription fell 9 15.00 percent, and the cost per 10.00 patient fell 11 percent, which 5.00 allowed the overall number of 0.00 visits and prescriptions to stay 2005 2006 2007 2008 2009 2010 (est.) flat despite the 9 percent Cost per Prescription Secondary Outpatient Primary Outpatient drop in expenditures. As can Inpatient Bed Day EMA per Call be seen from Table 6.9, in 2009 inpatient spending fell to 45 percent of the total, 8,000,000 Primary Outpatient which funded the protected 7,000,000 categories of general 6,000,000 Prescriptions practitioners and 5,000,000 pharmaceuticals, and they 4,000,000 Secondary expanded their shares. 3,000,000 Outpatient 2,000,000 1,000,000 Inpatient Bed Days 442. The extraordinary - run-up in medical costs does 2005 2006 2007 2008 2009 2010 EMA Calls not seem to correspond to a (est) large increase in employment. During the period from 2005 Source: Health Compulsory Insurance State Agency News for various years and the 2010 to 2008, employment in health budget. “human health activitiesâ€? rose by 2,631 jobs or 6.4 percent. Salaries rose substantially, as shown in Table 6.11 – on average 21 percent at the outpatient level and 28 percent at inpatient institutions. Note that these numbers do not include the vast majority of family doctors and specialists in private practice; only employees of institutions are included. Nevertheless, if these salary changes reflect market conditions, we could infer that wage changes in private practice would be similar. 226 443. The result for hospitals, for example, is that revenues increased 79 percent, wages went up 28 percent, employment probably did not go up more than 6 percent, and Table 6.11: Increase in Salaries for Medical Staff between 2003 and 2008 there was almost no change in (based on constant 2003 Lats) output. Many hospitals also have Type of Staff Outpatient Inpatient Physicians dentists, and specialists specialist clinics, so they also related to them 12% 15% participated in the rise in spending Paramedical staff 32% 32% on secondary ambulatory services. Nurses of dentistry, sanitarians, and 23% 20% Specialists admit inpatients, so these assistants to nurses two parts of the system are symbiotic Other personnel 16% 44% Source: Health Compulsory Insurance State Agency News for various years. in some ways or can be substitutes in others. In short, hospitals and secondary outpatient services consumed an incredible increment of resources between 2005 and 2008 with little change in output. The only way to explain the high additional spending for inpatient and secondary outpatient care is a revolution in medicine in Latvia – either costly new procedures that had never been financed before or an almost overnight increase in the use of technology and diagnostics for care. These changes persisted during the downturn, when the cost per patient did not fall in proportion to the cut in funding. Yet as we have seen in the FINBALT data, this expansion of activity had not, at least by 2006, bought significantly improved screening of at-risk men or management of their cardiovascular risks, an example of one set of improvements. Changes to Medical Services 444. One of the first acts in 2009 was to reduce all reimbursements from the Payments Center by 20 percent of salary costs, as the government attempted to reduce wages by 20 percent across all of its operations as part of the internal adjustment to make the economy more competitive externally (and to reduce the budget deficit). 445. The effort to protect outpatient, emergency, reimbursed medicines and devices resulted in the large reduction in spending on inpatient care Table 6.12: Target Share of Health Spending in 2009 reviewed above. This change is also Type of Care 2009 2010 part of an effort by the MoH to reduce Outpatient (Primary permanently Latvia’s excessive dependence At least 32% At least 38% and Secondary) on inpatient care by reducing the number of Not more than Not more than Inpatient 60.6% 52% hospitals and beds, encouraging shorter Emergency Medical lengths of stay, and shifting patients to day At least 6.4% Less than 9% Care surgeries and home care, with the short-term Source: MoH. targets shown in Table 6.12. 446. To encourage greater use of “day hospitalsâ€? the day-bed payment to hospitals was increased by over 500 percent to LVL 7.43, resulting in a 64 percent increase in day patients from March to November 2009 relative to the same period in 2008. Surgical services increased by 221 percent to 6,074 unique 227 patients. In July 2009, chemotherapy and radiotherapy also became outpatient services; by November these services had reached 466 and 64 unique patients respectively, up from 0 the previous year. For the first time, beginning in 2009, home care for chronically ill patients became a reimbursable service. The number of patients increased from 252 to 1646 between January and November, with the average number of visits per patient per month also growing from 5.5 to about 8. In all of these cases, the reduction in inpatient stays may not have been 1-for-1, but they show considerable potential for reducing the use of inpatient facilities. Similarly, beginning in January 2009, the scope for home psychiatric care and care at day centers was expanded. In July, 273 inpatient psychiatric cases that were on waiting lists to shift to social care institutions were shifted off of the medical care budget and on to the social care budget, paying for them at the lower rate that would be received by a social care institution. This change has created an incentive for the psychiatric hospitals to move the patients or internally to reorganize care for them along the lines of a social care institution. All of these initiatives received a boost under the emergency safety net program in that it finances any copayments for needy households who require these new services. 447. Those are the Table 6.13: Changes in Contracted Inpatient Facilities 2008 to 2017 “carrotsâ€? or positive Contracted 2008 2009 2010 2017 adjustment incentives. Hospitals 78 72 41 ? The main “stickâ€? used to 12,929 prod adjustment in the Beds (end of the year) 17,001 15,121 10,550 (start of year) sector has been lower Beds per 100,000 749 669 575 478 payments and a shift in Source: MoH. the character of provider contracts with the Payments Center. The Payments Center, for example, on April 1, 2009 ended inpatient contracts with 13 small hospitals with 287 beds (22 beds each on average). Another 273 beds were removed in this manner on September 1, 2009. Riga Hospital #1, a 600-bed institution, was shifted from an inpatient to an outpatient institution on January 1, 2010. Its patients could be absorbed by other large hospitals in Riga, yet it still maintains a large first- and second-level outpatient clinic, and it offers day surgery. Patients can still get inpatient care but without financing from the Payments Center – they pay the full cost. Riga City finances social care beds at the hospital. Thus the hospital and the beds disappear from the set of contracted services available with State subsidy, but the capacity does not disappear physically, or disappears slowly as other funding sources fail to support the level of plant, equipment, and professionals who are present. Table 6.13 shows the number of hospitals with inpatient contracts declining from 78 in 2008 to 41 in 2010. The number of contracted beds fell from 17,001 to an estimated 12,929 at the beginning of 2010. This is an impressive reduction that moves the key indicator, contracted beds per 100,000 population, down to 575. However, that number is still higher than Estonia and above the EU average of 554 beds per 100,000 in 2007. By 2017 the stated goal is to reduce this ratio to 478 beds/100,000. 448. Table 6.14 illustrates how the reduction is being done. By mid-2010, there will be 22 emergency hospitals in Latvia’s new unified emergency medical system, down from 34 at the beginning of 2009. These emergency hospitals include the major university, regional, and multi-profile hospitals. 228 449. A new type of hospital, Table 6.14: Composition of the Hospital Sector 200 called a “careâ€? hospital, is created Contract Type 2009 2010 Change that will now supply 1,129 beds for Emergency hospital 34 22 -12 day hospital care or patient-paid Care hospital 0 7 7 care (about 160 beds per hospital; Specialized hospital 22 12 -10 these are sizable institutions). Other hospital 16 0 -16 Specialized hospitals (including Outpatient institution 0 30 30 psychiatric, orthopedics, addiction, Merged 0 1 1 maternity, and rehabilitation Total 72 72 0 hospitals) are reduced primarily by Hospitals 72 41 -31 shifting rehabilitation and other Source: MoH. small specialized hospitals to outpatient status. The large hospitals remain. 450. Despite the reduction in beds in 2009, the occupancy of beds hovered at about 70 percent throughout the year. This reflects a 25 percent decline in hospitalizations in 2009, from 16 per 100 population in 2008 to 12 per 100 from March to November in 2009. It is too early to say anything about the possible health consequences of the change in organization of the health sector in 2009, but initial figures from the Central Statistics Bureau indicate that the crude death rate in 2009 was about three percent below 2008, and that the crude death rate has dropped further in the first three months of 2010. 451. One thing is certain. The Figure 6.19: Inpatient Expenditures Per Bed in Constant 2003 Lats hospital sector cannot function as it has become accustomed, at 10,000 the level of resources it will 9,000 Lats per Bed in Constant 2003 receive in the 2010 budget. 8,000 Figure 6.19 shows inpatient 7,000 expenditure per bed in constant 6,000 Prices 2003 Lats. The number of beds 5,000 decreased by just 1,000 during 4,000 the period 2004 to 2008, a 3,000 missed opportunity to reform 2,000 the health sector in a time of 1,000 rising budgets. As we have seen, - the result was continuing heavy 2004 2005 2006 2007 2008 2009 2010 use of hospitals – but no increase in output – with no Source: Health Compulsory Insurance State Agency News for various years and MoH efficiency improvement and much higher spending per patient. The remaining problem is that once again in 2010, hospitals will have to reduce patients to maintain the current high level of resources per patient. Already there is enormous 229 pressure to increase spending on hospitals in the 2010 budget, including suggestions to reallocate the funds set aside for the emergency safety net to hospitals. It is easy to imagine a number of scenarios that might unfold: i. Increase spending to maintain the 2009 rate per bed, which would require an additional 36 million Lats in the 2010 hospital budget (a 31 percent increase). A solution like this would certainly be preferred by the hospitals. ii. Achieve the 2009 rate per bed by cutting another 3,091 beds, which would put Latvia at 437 beds per 100,000. A possible mechanism for doing this would simply be to cut the number of beds contracted but put no restrictions on hospitals to finance services from other sources of funds, but it is clear from Latvia’s experience that unless hospital buildings are closed, the expected savings will not be realized. iii. Reduce current global hospital budgets by a specific amount per bed, which would create an incentive for them to reduce the number of beds. For example, cutting budgets by 50 percent per existing bed on July 1 could result in a 50 percent reduction in inpatients (meaning occupancy would fall by 50 percent), spending half as much on the same number of inpatients with the same average length of stay (reducing ancillary costs), reducing average length of stay by half, or some combination of these measures.135 Changes in Copayment Policies and Consumer Spending Copayments 452. Latvia has a system of copayments, patient contributions that are required for most services. One of the first Table 6.15 Basic and Maximum Copayments 2008 to 2010 (Nominal Lats) and Percent Changes actions taken in the (based on constant 2003 Lats) crisis of 2009 was Percentage to raise the copay Change 2008 to Type of Staff 2008 2009 2010 2010 (Real) for a number of Outpatient Visit to GP 0.5 1.0 1.0 101% services, on the Outpatient Visit to Specialist 2.0 5.0 3.0 50% assumption that it Outpatient Visit to Hospital 2.0 5.0 3.0 50% would reduce use. Outpatient Surgery in a Hospital 0.5 5.0 5.0 903% The MoH was less 9.5 interested in Daily Inpatient Charge in Hospitals 5.0 12.0 (starting 91% additional revenue day 2) Maximum Patient Contribution for from patients than 80.0 250.0 250.0 214% One Hospital Episode in reducing use, Maximum Patient Contribution for 150.0 400.0 400.0 168% because the copays One Year Source: MoH and vnc.gov.lv. 135 Some may argue that the reimbursement system is based on services, not on the cost per bed. However, the same people tend to argue that with the shift to global budgets in 2010, service volume is now less important. One advantage of using the cost per bed as a metric under a global budget is that it creates incentives for hospitals to use their beds more efficiently within the global budget constraint, thus requiring them to economize on a resource they have in excess. The goal would be to give them an incentive to reduce both beds and square meters per bed – all of the costs associated with beds. 230 were a small fraction of the total cost of the services being consumed. Fees had not been changed since 2005, so this was considered a reasonable action given the substantial inflation that had occurred in the interim. 453. Table 6.15Table gives the basic copayment categories for recent years and the percentage change from 2008 to 2010. There are additional charges for diagnostic tests and certain procedures. In a policy change, prices were adjusted downward in 2010 for secondary outpatient services and for the daily charge to stay in a hospital. If a household is certified as needy, it would have paid half of these fees in 2009; now the safety net pays it for them. 454. Copayments were raised substantially in 2009, doubling the cost of a visit to a GP and more than doubling the costs of a visit to a specialist at his or her office or in a hospital setting. However, these copayments were reduced in 2010. By far the largest increase brought outpatient surgery more in line with its relative cost (see Table 6.15). 455. Today it costs an individual LVL 9.5 per day to stay in a hospital, and there is a LVL 30 payment for surgery. In addition to these charges, there are fees for some diagnostic tests while an inpatient, although most of them are free once a person is admitted to the hospital.136 This difference between inpatients and outpatients creates a potential incentive for admittance. Such a problem is not unique to Latvia but is characteristic of insurance plans that cover inpatient services, which are considered catastrophic, at a higher percentage than for outpatient services. A question is whether in the end the insurer would save money by rerating outpatient and inpatient tests the same, and whether the health outcomes would be as good or better. 456. The maximum copayments were also increased significantly in 2009, from LVL 80 to LVL 250 for one hospital episode, and from LVL 150 to LVL 400 as the maximum out-of-pocket cost for one year. The LVL 400 total copay in 2009 was 6.8 percent of per capita GDP (LVL 5,874). The best evidence we have (from the Rand insurance experiment in the U.S. although it is now quite old) is that copayments do not reduce needed care, although more recent evidence suggests that this may not be the case for people suffering from chronic diseases. 457. Latvia’s coverage of pharmaceuticals has been governed since the early 1990s by a positive list that evolved from covering drugs for specific chronic conditions (such as diabetes) to also covering specific groups of patients (e.g. children) to a more extensive subsidy program for a broader range of diagnoses and medicines (with subsidy levels keyed to the severity of illness). Starting in 2005, Latvia shifted to a reference price system that provides subsidies for prescribed drugs based on the least cost alternative drug. There are three compensated drug lists. List A is the reference price list. The reference price is for the base drug that the government subsidizes 100 percent. If a replacement drug is chosen, the patient pays the difference in cost between it and the reference price (the difference is called the 136 In 2009, several expensive diagnostic tests, including computed tomography, nuclear magnetic resonance imaging, and coronary angiography, which previously had copayments if done on an outpatient basis but not on an inpatient basis, were assessed the same copayment if done on an inpatient basis. 231 reference supplement). The government compensates drugs at a rate of 100, 75, or 50 percent, based on the diagnosis, so the reference list shows each diagnosis with its compensation rate, then the list of medicines and/or medical devices eligible for that diagnosis code (and who can prescribe it). 458. List B is the same type of list, but it is populated with non-replaceable drugs, so there is no choice over which one the government will compensate. List C was introduced in 2007 for expensive individual treatment cases when the patient needs medicines to sustain his or her life functions and the maximum annual treatment cost is over LVL 3,000 (it covers primarily oncological, hematological, and rheumatology drugs). 459. Coverage is improved by adding new drugs (and medical devices) to the reimbursable lists, which has been done every year. Even in the current difficult environment, 83 new medicines and medical devices were added to the compensated list in 2010. The Health Economics Center includes new drugs if the manufacturer submits them and if they pass a test of relative effectiveness and price. 460. Among European Figure 6.20: Share of Generics in the Pharmaceutical Market 2006 countries, Latvia has succeeded in shifting the largest volume of pharmaceutical purchases to generics (nearly 80 percent), and it trails only Poland in the percentage of value accounted for by generics. In Figure 6.20, Latvia is second from the right. The blue bar on the left is the share of generics in the total value of drugs sold, and the yellow bar on the right is the generic Source: European Generic Medicines Association share of volume, in 2006. 461. During the economic crisis, Latvia made some adjustments to the lists of reimbursable drugs. Those diseases for which medicines were 100 percent reimbursed remained so (for example, insulin- dependent diabetes, glaucoma, heart failure, oncological diseases, and more). The 90 percent reimbursable category was eliminated, and these diseases were moved to the 75 percent category (asthma, Parkinson’s disease, and others), and a number of 75 percent reimbursable diagnoses were moved to 50 percent (e.g. osteoporosis, hypertension). No diseases were removed completely, but in general copayments rose sharply. 232 Exemptions from Copayments (or deposits) 462. Latvia has a long list of exemptions that are shown in Table 6.16. They combine many different criteria, ranging from medical, to social, to poverty, to insurance caps. The total amount spent for this purpose – limiting the burden of financial costs of health care on households – in 2008 was about LVL 12 million; however it went up to LVL 27 million in 2009 because of the large increase in copayments enacted that year. In 2009 the increase in “compensated copaymentsâ€? added 12 percent additional public resources above the budgeted amount for inpatient spending, 8 percent for secondary outpatient services, and 6 percent for general practitioners. 463. Fully 75 percent of the spending in 2009 is accounted for by just four groups: children, the mentally ill, pregnant women, and oncology patients. In addition to the funding in this table, children receive free dentistry; the mentally ill and cancer patients also qualify for subsidized purchases of medicines. 233 Table 6.16: Exemptions from Patient Deposits in 2008 and 2009 for Inpatient and Outpatient Care Total Total Expenditure Expenditure Percent in Nominal in Nominal Category of Patient LVL 2008 Percent LVL 2009 Preventive All patients undergoing prophylactic exams (special list of 661,241 5.7% 951,764 3.5% Care exams) Patients receiving vaccination or passive immunotherapy 37,622 0.3% 232,338 0.9% within the State immunization program Infectious TB patients and those under examination for TB 279,012 2.4% 373,875 1.4% Diseases Patients with specific infectious diseases (list of 43 102,758 0.9% 274,278 1.0% diseases) Chronic and Mentally ill patients under treatment 2,270,032 19.4% 7,162,983 26.4% mental Chronically ill patients receiving dialysis, 36,946 0.3% 213,941 0.8% illnesses hemodiafiltration, and peritoneal dialysis Disabled person 1 0.0% 793,179 2.9% Social Victims of political repression and participants of the 373,449 3.2% 774,682 2.9% Exemptions national resistance movement Persons at the age over 80 years 1 0.0% 0.0% Children under the age 18 4,480,610 38.3% 9,642,826 35.5% Victims of Chernobyl nuclear reactor accident 83,492 0.7% 157,658 0.6% Poor Poor people officially recognized according to the 285,304 2.4% 600,279 2.2% legislation Patients whose monthly pension is less than Ls 60 22 0.0% 0.0% Persons under the care of state specialized social care 120,295 1.0% 587,249 2.2% centers and municipal pensions (centers) Payment Caps Patients whose total patient deposits within one calendar 14,652 0.1% 23,854 0.1% year exceed 150 lat (400 in 2009) Patients, whose total amount of patient’s deposit per one stay in one and the same hospital has exceeded 80Ls (250 394,949 3.4% 545,498 2.0% in 2009) Specific Deceased 368,324 3.1% 865,907 3.2% Procedures or Draftees 16 0.0% 0.0% Conditions Pregnant women up to 42 days after childbirth 973,977 8.3% 2,372,268 8.7% Surgical treatment in day-hospital 16 0.0% 2,635 0.0% In-patient oncologic treatment for adults 735,054 6.3% 1,093,104 4.0% In-patient oncohematologic treatment for adults 79,964 0.7% 150,182 0.6% In-patient treatment for adults within the framework of 128,001 1.1% 236,524 0.9% alcoholism restriction program In-patient treatment for adults within the framework of 32,048 0.3% 92,726 0.3% drug addiction restriction program Treatment in the second stage of rehabilitation 237,389 2.0% Total 11,695,170 100% 27,147,744 100% Source: Health Compulsory Insurance State Agency News #17 2009. 464. On October 1, 2009, the government implemented its emergency social safety net strategy, through which the MoH originally proposed to eliminate all of these categories and simply subsidize families falling below the “needyâ€? line – households with per capita income 50 percent or less of the minimum wage (which is less than LVL 90 per month under current law). The proposal was rejected, and the safety net now simply clarifies who is considered poor for that category in the middle of Table 6.16. Note that only 2.2 percent of subsidies went to this group in 2009. 234 465. The safety net introduced important exemptions for the needy. In October 2009, when it was first introduced, the safety net would finance 100 percent of the cost of copayments for the needy at every level of the system. To encourage less use of inpatient care, the poor also became eligible for free overnight “hotelâ€? stays in hospitals (in connection with travel for day surgeries, chemotherapy, etc.), for home care of the chronically ill, and for day care of the mentally ill. In this way the MoH could begin to institutionalize alternatives to inpatient stays by subsidizing them for the poor, who depended more on inpatient accommodations to gain an entitlement to free or low-cost services. 466. By February 2010, when it Table 6.17: Safety Net Coverage for the Poor 2009 and 2010 became clear that initial take-up of October February the new benefits was slow enough Benefit 2009 2010 that the MoH could safely expand the Copayment at GP, Specialist, Hospital, “hotelâ€? stays in hospitals, home care, safety net within the original budget, day care for mentally ill patients it modified the cutoffs by shifting Income < LVL90 100% eligibility for a 100 percent subsidy to Income < LVL120 100% Income > LVL120 and < LVL150 50% LVL 120 and adding a second level for 100% reimbursement for the cost of incomes between LVL 120 and LVL prescriptions after LVL50 deductible is 150 that is eligible for a 50 percent reached Income < LVL90 100% subsidy. In addition, under the 100% reimbursement for the cost of original safety net, needy patients prescriptions with no deductible became eligible for a 100 percent Income < LVL120 100% Source: MoH subsidy for prescriptions after they met a deductible of LVL 50 for the year. By February the new income ceilings were also adopted for this benefit, and the deductible was eliminated (Table 6.17). Over LVL 24 million has been allocated for the safety net measures in 2010. 467. There are many good reasons for the government to reconsider and proceed with the MoH’s original proposal to shift the entire exemption system to the new safety net criteria. Doing so would simplify the process of determining eligibility, which could then be handled by municipalities, and would perhaps improve the targeting of subsidies. 468. Moreover, the health safety net improves the incentives at the municipality level, where previously a person would have to apply for social assistance from the municipality to cover health co- payments. Assistance with the co-payments was not a mandatory form of municipal social assistance, like the GMI and the housing benefit. Municipalities could provide it on a voluntary basis and at a level according to their wealth and commitment. The municipality, bound by its own budget constraint, therefore had an incentive not to provide the assistance, and the individual could not get the help unless he or she had already paid the fee. Anyone can ask for help, so these subsidies were probably not as well targeted as one connected to the “needyâ€? line. 469. Under the MoH system, if the individual meets the “needyâ€? or “near-needyâ€? criteria, no cash changes hands because the provider claims the copayment directly from the MoH. The municipality has an incentive to correctly identify the poor to make sure they are eligible for the State-paid benefit. 235 470. Under a new safety net based system, it would be important to maintain the “payment capsâ€? section in Table 6.16 to protect patients from catastrophic expenses. It is interesting to note that in 2008 only 3.5 percent of the subsidy was accounted for by these caps. It dropped in 2009 even with higher deductible limits, which suggests that out-of-pocket spending on covered services is not a large source of household spending. 471. Of course under the new system, many of the individuals now meeting the criteria in Table 6.16 would continue to receive subsidies, but not all. Improving the exemption system in health would be one aspect of an overall improvement in Latvia’s system of subsidies, and the MoH has already taken a big step in the right direction. Household Expenditures 472. We can estimate household health expenditures from the annual Household Budget Survey, which is available for 2003 through 2008. Spending on health care is income elastic, so as incomes rise, health spending Figure 6.21: Government Spending on Health as a Percent of GDP and in Inflation-Adjusted Lats, Latvia, tends to rise 2000-2010 faster. Figure 6.21 adds 7 250 Insurance Premiums household 6 Per Capita Per Capita Spending in Lats spending to 200 Household Spending Figure 6.13 for 5 Per Capita Percent of GDP the years that it (Base=2000) 150 Government 4 is available. Spending Per Capita Overall 3 Total Spending in 100 GDP spending by 2 Government in GDP government and households 50 1 HH Spending in GDP peaked at nearly 6 percent 0 0 Insurance Premiums of GDP in 2007, 2003 2004 2005 2006 2007 2008 Percent of GDP with household spending Source: Health Compulsory Insurance State Agency News, various years. GDP is from the Central Bureau of Statistics. Population is from Eurostat. Staff estimates for household spending based on the Household accounting for Budget Survey. Insurance premiums are from the Financial and Capital Market Commission. See Annex 2.13 percent of Table 3. Data are not yet available for 2009 on household expenditures, so the graph stops at 2008. GDP that year. It has hovered around this level of 2 percent of GDP, but because of the expansion of GDP, real per capita out-of-pocket spending increased by 102 percent from 2003 to 2007, while the government’s per capita spending increased by 82 percent and GDP per capita rose by 51 percent. The result is that by 236 2008, government was spending LVL 131 while households were spending LVL 80 per capita, about 38 percent of the total. In 2003 households accounted for 34 percent.137 Household Expenditure Patterns 473. With household spending on health rising so significantly from 2003 to 2008, health spending became a larger part of their budgets, rising from 3.6 Table 6.18: Expenditures for Health Care Per Month Per Household Member by Quintile in Latvia, 2003 and 2008 percent in 2003 to 4.8 Expenditure Quintile 1 2 3 4 5 Total percent in 2008, with little Health 2003 1.53 3.11 4.06 3.99 6.18 3.63 difference between rural and Health 2008 in Constant 2003 Lats 3.85 7.17 7.46 7.13 11.36 7.21 urban households (although Percent of Expenditures 2003 2.9 4.5 4.8 3.6 3.1 3.6 Percent of Expenditures 2008 4.8 6.6 5.4 4.1 4.2 4.8 because rural households are Percentage Increase Health 152% 131% 84% 79% 84% 99% poorer, their absolute Percentage Increase Total 56% 57% 63% 58% 36% 46% expenditures were about Source: Central Statistics Bureau two-thirds of urban expenditures in both years). 474. Table 6.18 shows household spending by quintile for 2003 and 2008. Total expenditures for the whole population increased by 46 percent over the period (bottom row), while health care expenditures increased 99 percent. The expansion in consumption during these boom years helped the bottom two quintiles increase their overall spending by 56 and 57 percent, respectively, but their spending on health rose by 152 and 131 percent. Health as a percent of all expenditures expanded more rapidly among the bottom three Figure 6.22: Distribution of Per Capita Household Spending on Health and Health Insurance quintiles than for 2003 and 2008 (in Constant 2003 Lats) the top two, and by 2008 the poorer 90 three quintiles were 80 Hospital services spending more of 70 Lats (Constant 2003) their household Services of medical 60 budgets on health auxiliaries 50 care than were the Dental services 40 richer 40 percent. This suggests that 30 Medical services financial protection 20 for the poor could 10 Appliances, Other be improved. In 0 Poland, for example, 2003 2008 despite rapidly rising incomes Source: Household Consumption Survey 137 Figure 10 does not include spending on health care by businesses, which would probably add about one percent of GDP, but we do not have estimates for it after 2006. 237 during the same period, health expenditure across the board by households fell significantly both in absolute amount and as a percent of income because the government budget picked up a larger share of the cost. 475. Figure 6.22 shows the relative size and composition of household spending on health in 2003 and 2008 in constant Lats.138 All elements of spending rose, and the composition of spending changed little. The proportion devoted to dental services rose by 3 percentage points to 16 percent, and this was accommodated by small declines in the share spent on pharmaceuticals, appliances, outpatient medical services, and hospitals. A reasonable goal for a health insurance system is to make sure the poorer quintiles spend a smaller percentage of their expenditures (or income) on health than do the upper quintiles. If that is not possible, at a minimum, policy should aim for proportional spending across quintiles. 476. Pharmaceuticals accounted for nearly 60 percent of household spending in both years. Industry data for 2009 through the third quarter indicate that 28 percent of wholesale drug sales were over-the- counter drugs and 72 percent were prescription drugs. Of the prescription drugs, 69 percent were reimbursed and 31 percent were not. If we assume that out-of-pocket spending on pharmaceuticals follows the same patterns, 50 percent of out-of-pocket spending would be on over-the-counter or non reimbursed prescriptions. The other 50 percent would be copayments on reimbursed drugs. The government can therefore affect about half of this spending by changing its current reimbursable drug policies and another 22 percent by covering prescriptions that currently are not covered.139 Therefore, a priority to reduce out-of-pocket spending on drugs would be to reduce existing copayments rather than include more drugs. Overall Health Expenditures Figure 6.23: Combined Government and Household Spending on Health 2008 (in Constant 2003 Lats) 477. Figure 6.23 combines 200.00 Household… (Constant 2003) Millions of Lats 150.00 Government household with 100.00 government 50.00 spending on health - services.140 It reveals that public spending in Latvia essentially finances a catastrophic insurance system, Source: Household Consumption Survey and Health Compulsory Insurance State Agency News financing 95 138 The figure adds one more expenditure item relative to table 6 – insurance connected to health – so the totals are slightly higher. 139 Data from the MoH. 140 As before, it does not include spending by non-household entities, like businesses. 238 percent of inpatient care and emergency services in 2008. It finances a high percentage of general and secondary ambulatory services (76 percent), only 28 percent of pharmaceuticals and appliances, and 11 percent of dentistry (most State expenditure for dentistry is for children only). Pharmaceuticals remains an area where more could be done to reduce household costs, and this has been done in part through the safety net. But the impact on household spending of recent policy changes needs to be monitored, and drug policy reimbursement policies adjusted, to help households – especially poorer and older ones – to manage this expense. Consumer Views 478. In 2008, the Payments Center commissioned a survey to assess Latvians’ views on receiving State-paid health care services.141 It provides some additional population-based data to complement the claims data used so far. It is a nationally representative survey of 18 to 74 year-olds conducted in November 2008, requesting for the most part recall during the past year. Ninety-five percent of the respondents knew who their family doctor was and less than 2 percent did not have one, suggesting that the first level of care works extremely well in covering the population. 479. About 73 percent of the population had seen their doctor at least once in the previous year, 73 percent had seen a staff specialist in a clinic, and 30 percent had seen a specialist in private practice.142 When asked if they needed to visit a doctor but did not last year, 19 percent responded yes. Of that group, 46 percent cited “not enough money,â€? 40 percent responded “not enough time,â€? and 22 percent cited long queues (more than one answer was allowed). Figure 6.24: Patient-Reported Waiting Times in 2008 480. About 14 percent of the Waited more than a month for sample had an elective hospital treatment inpatient stay, and Waited more than a week for a this varied private practice specialist … predictably by age. Waited more than a week for a staff specialist appointment Below age 45, 10 Waited more than a week for percent or fewer family doctor appointment had a hospital stay, 0% 20% 40% 60% 80% 100% but from 45 to 60, 15 percent had an No Always or Sometimes encounter, and for those over 60, Source: Health Compulsory Insurance State Agency nearly 30 percent had been hospitalized. The challenge of reducing hospitalizations is concentrated among the old. 141 http://www.vnc.gov.lv/files/VOAVA_petijums_2008.pdf 142 Note that the FINBALT data cited in Figure 6.8 of a 32 percent failure rate for men to see a doctor and 20 percent for women is consistent with the statistic of 73 percent seeing a doctor for the population as a whole. The concern from the FINBALT survey is that at-risk men were among the least likely to see a doctor and perhaps most importantly, their family physician, in the past year (42 percent did not). 239 481. Satisfaction with the health system appears to be reasonably good for family doctors and less so with the system as a whole. Seventy-seven percent were either completely or partially satisfied with their family doctor and only 16 percent were completely or partially dissatisfied.143 However, positive responses to a more general question about the possibility to gain access to State-funded care were much rarer, with only 50 percent saying it was good or somewhat good, and 36 percent saying it was somewhat or completely bad.144 The answers to these two questions were strongly and positively correlated – if a respondent was satisfied with his or her family doctor, he or she was more likely to assess access to care positively, which is consistent with the key role of the family doctor in the system. 482. If a respondent had visited a specific provider, he or she was asked about waits to get an appointment (Figure 6.24). About 60 percent reported waiting at least a month for an elective hospital procedure. In contrast, about half of the GPs and private specialists were accessible within a week. Staff specialists in clinics were much less available. Some of the constraints in the system come through clearly: among specialists, waits – whether for staff or private specialists – were most common for dentists, gynecologists, Figure 6.25: Patient-Reported Expenditures in 2008 ophthalmologists, and surgeons. 5% 483. Finally, the responses on 17% More than LVL150 14% health expenditures are consistent with the conclusion in From LVL51 to 150 the preceding section that Latvians seem to be protected From LVL10 to 50 reasonably well financially. Although 17 percent did not Below LVL 10 27% respond, 63 percent reported 36% costs below LVL 50, and only 5 Hard to Say or percent reported costs higher Refuse than LVL 150, with the remaining 14 percent falling in between Source: Health Compulsory Insurance State Agency (Figure 6.25). IV. The Private Health Insurance Market in Latvia The Market 484. The general insurance market (life, casualty, auto, health, etc.) in Latvia is well developed and profitable. At the end of 2009, there were 14 insurance companies and 11 branches of foreign companies operating in Latvia. As shown in Table 6.1, in 2008 both gross premiums paid peaked and claims peaked at about LVL 337 million and LVL 183 million, respectively, leaving a gross margin of LVL 154 million (which of course ignores all other costs of doing business). The industry had a terrible 2009, 143 7 percent did not answer. 144 14 percent did not answer. 240 Table 6.19: Financial Data on General Insurance and Health Insurance in Latvia Percentage Change 2007 Percent 2008 Percent 2009 Percent 2008 to Amount in Lats of Total Amount in Lats of Total Amount in Lats of Total 2009 Total Premiums 306,771,000 337,388,000 223,304,000 -34% Total Claims Paid 130,501,000 183,141,000 153,870,000 -16% Total Health Premiums 42,524,000 13.9% 53,871,000 16.0% 39,228,000 17.6% -27% Total Health Claims Paid 23,774,000 18.2% 37,703,000 20.6% 46,762,000 30.4% 24% Source: Latvia Financial and Capital Market Commission (www.fktk.lv) when gross premiums paid dropped by 34 percent but claims dropped by only 16 percent, leaving a gross margin of LVL 69 million. Several insurers experienced losses in 2009. 485. Health insurance premiums dropped by 27 percent from 2008 to 2009, but claims actually increased by 24 percent, resulting in an overall loss on health insurance for the industry in 2009 (again, ignoring all other costs, which would have increased the loss). Nevertheless, premiums and claims for health insurance steadily rose from 2007 through 2009 as a percentage of total premiums and total claims. Health is a relatively more risky line of business for insurers, as health claims each year accounted for a larger fraction of total claims than of premiums, peaking at 30.4 percent of claims in 2009 (compared to only 17.6 percent of premiums). Health premiums paid to private insurers were 9, 10, and 11 percent of the total government budget for the public insurance system in 2007, 2008, and 2009, respectively. 486. Figure 6.26 Figure 6.26: Coverage of Accident and Health Insurance in Latvia, 2005 through shows that the 2009 number of people with health insurance rose 500000 25% by 100,000 from 2005 450000 Percent of Percent of Populatin Insured to 2008, then dropped 400000 20% Population Covered to 262,345 in 2009 – by Accident Number of Insured 350000 Insurance back to the 2005 level 300000 15% almost exactly. Percent of 250000 Coverage of the Population Covered 200000 10% by Health Insurance population by private 150000 insurance peaked at 100000 5% Number Coverd by 16 percent in 2008, dropping back to 12 percent in 2009. Coverage by accident insurance continued to rise in 2009, suggesting that some employers may have substituted accident insurance for health insurance during the economic crisis to maintain at least minimal coverage. The rise in claims shown in Table 6.19 for 2009 is likely connected to the loss of insurance for so many Latvians, as people typically take advantage Source: Latvia Financial health services or anticipate future of their health insurance to purchase deferredand Capital Market Commission (www.fktk.lv)needs prior to losing their coverage. 241 Types of Health Insurance Policies 487. In 2010, the public insurance system paid approximately LVL 166 per person.145 For an additional payment of LVL 119 per insured (see Table 6.20) an employer could purchase basic private coverage that would pay all patient fees in the public system and provide limited subsidies for outpatient and inpatient care from private providers (allowing the insured to jump queues and/or get specialist care without a referral from the family doctor). Mid-level coverage in the table, at about double the basic price, would allow many patients to continue to use family doctors but to opt out of the public system for most secondary outpatient and inpatient care. If they use the insurer’s network for this care, they would typically experience no out-of-pocket payments, and reimbursement would be directly to the provider at a contracted rate. 1 Table 6.20: Types of Private Insurance Coverage Available in Latvia 2010 Payment Limits in Lats Approximate Annual Patient Fees Privately Paid Privately Paid Premium 2010 in Public Outpatient Inpatient 2 Policy in Lats System Treatment Treatment Payment Method Basic Coverage 119 400 300 300 Reimburse patient who files claim Mid-Level 237 400 700 1,000 total As above or directly to Coverage and 500 per contracted institution in the illness network High-Level 319-346 2,500 to 3,500 total As above Coverage Notes: 1. Additional services can be purchased, such as coverage of medication copayments, dental care, and fitness activities. Other variations include different copayment rates. 2. The insurance pays all patient fees assessed by the public system including outpatient visits to family doctors, specialists, diagnostics, outpatient surgery, and home visits for persons over 80; all inpatient fees connected to unplanned treatments and to planned treatments if prescribed by the family doctor. The LVL400 limit is the maximum out-of-pocket annual payment required in the public insurance system (the stop-loss amount). Source: Interviews with private insurers. 488. High-level coverage, costing LVL 319-346 annually, creates an alternative to the public system for care from a network of providers contracted by the insurer. Patients with this coverage could still use the public system – and they might fall back into it if their costs exceed the yearly limits – but the point of this insurance is to provide access to private providers for all services, including ambulances, without recourse to the publicly financed system. 489. The table oversimplifies the types of coverage available, but most insurers provide policies roughly along these lines, including copayment options and optional services noted in the table. Private insurers have developed networks of contracted providers all over the country, and they have years of experience now administering claims. They market insurance primarily to groups through employers rather than to individuals. Some Policy Issues from the Private Insurer Perspective 145 In 2008, at the peak, the government spent about LVL226 per person in nominal terms. In 2009 the amount was LVL170 per person. 242 490. Clearly there is a mixed market for insurance in Latvia. As the constraints on the public budget continue to tighten even more in the near future, it is likely that employers will seek alternatives for their workers, and this market is likely to grow. Private insurers work in the shadow of the pubic purchaser and work with the same providers. The result is that they have a well informed perspective on the system that is worth considering. In limited discussions with insurers, the author gleaned the following insights from them: a. Private insurers organize tenders to develop their networks. In their dealings with physicians and institutions, they see a wide variance in the capabilities of providers to manage their businesses, to account for claims they make, and to improve and apply standards of care systematically. The State currently does not tender for services and is a passive payer with little capability to verify claims it pays. The forms providers use to request payment from the public system do not require that the patient verify having received the services. There is no centralized database that allows different providers or the payment system to easily track the tests and services a patient is receiving. The result is not only opportunities for abuse but also poor continuity of care, repetition of the same tests by different providers, and little ability to implement and enforce standards of treatment. Improving purchasing in the public sector is a high priority and could start by developing tenders to create a competitive procurement process for provision of State-financed care. Beyond that, systems to improve the validity of claims and to follow patients through the system need to be developed. b. Physicians and institutions are free to contract with any payer, including the public payment system and private insurers. The result is that suppliers exert considerable control over the system despite the single payer model in the public sector. There is little prevention systematically pursued in the system because of the prevalence of fee- for-service payments for so much of the work. One way to assert more power in the market would be to limit the ability of contractors to sell services to other insurers, to better specify standards of service and care, and to build these elements into competitive tenders so that providers could participate voluntarily and bid prices would reflect the exclusive nature of the contract and the standards of care that will be delivered. c. Because private insurers for the most part provide insurance that fills in gaps in public financing (often called “medigapâ€? insurance), the private market is susceptible to and hindered by a poorly defined public role. This problem reveals itself in a number of ways. First, the State more or less claims to cover everything except what is on the negative list. However, time is implicitly on the negative list because for many conditions, financing is rationed through waiting times. There are no standards or guarantees on the time element of care. It would be far more predictable – allowing people to pursue alternatives – if coverage were more narrowly defined so that time in the queue could be eliminated or shortened to a standard that can the State can guarantee. Even an appointment with a General Practitioner may take two weeks in the current system. Second, the scope of coverage by the state is not well defined (“to cover 243 everything is meaninglessâ€?) so it is constantly in flux, as illustrated by the large changes that took place in coverage of services during 2009. The result is that private insurers may face yearly changes on the public side that make risk management in their business more difficult. To be effective in filling “medigaps,â€? private insurers need more stability and predictability in their business environment, which is largely defined by the public purchaser. d. What might be a reasonable package for the State to finance that would be predictable for it and would eliminate some of today’s problems? It could include the family doctor and gatekeeper, paid largely through capitation, which would be a predictable expense that the State could guarantee. In fact, during the economic crisis the State continued to guarantee this coverage while withdrawing in other areas. Financing emergency and catastrophic care is to a large degree predicable as well. So the State could guarantee financing of these services either directly or through insurers. In fact, it also attempted to protect financing for these services during the crisis. Most everything else has a high discretionary component that makes management, financing, and administration time intensive and difficult. The State could make a contribution for those elements based on poverty and other criteria, but let those who can pay finance a larger share and turn these elements over to private insurers to manage. Later in this section we take a look at how such an approach might be financed and what it would cost. Do the Numbers Tell Us Anything about Future Options for Insurance in Latvia? 491. In considering Table 6.21: Spending by Government on Medical Care in Latvia and a Hypothetical Reallocation to Use Private Insurers, 2009 reforms to the Indirect current health Direct Purchases insurance Purchases by through the Private system, some of Total Per Capita Government Insurers the discussion Inpatient 169,246,517 74.84 74.84 considers shifting Emergency Medical Teams 26,410,502 11.68 11.68 completely to a Outpatient 113,100,372 50.02 voluntary Dentistry 5,086,566 2.25 2.25 General Practitioner insurance system Practices 40,677,685 17.99 17.99 or excluding Secondary Outpatient 67,336,121 29.78 29.78 some members Medicines 71,811,842 31.76 31.76 of society who do Settlements with EU 2,809,721 1.24 1.24 Medicare Total 383,378,954 169.54 30.91 138.63 not pay income Source: Staff Calculations based on MoH numbers taxes, or other measures that would lead to differential coverage. Our view is that any reform needs to follow the principle of 100 percent coverage of the population by a comprehensive plan that provides focuses on supporting people to improve their health and to prevent financial catastrophe associated with illness. This approach necessarily requires mandatory coverage of everyone and subsidies for those who cannot 244 afford the full cost of this protection. Therefore this section assumes full coverage of the population and that private insurer are not allowed to pick and choose risks. 492. Currently the only private plan that substitutes for much of the existing public package is the High-Level Coverage plan in Table 6.20. In 2009 the Government of Latvia financed medical services at the level of LVL 170 per insured, which includes about 2 percent for administrative costs. High-Level Coverage in the private sector costs LVL 346 per insured for a high-limit plan, which includes administrative costs of roughly 20 percent and probably a considerable margin for profit and risk management in the limited medigap market that exists.146 Let’s assume that if the market for private insurers expanded to the whole population, the cost for this package of services could be reduced roughly to LVL 300, just for the sake of some back-of-the-envelope calculations.147 493. Table 6.21 shows how the system might be divided between direct public purchasing of services and indirect purchasing through private insurers. Following the suggestion in point 4 of the previous section, let us assume that the government continues to finance directly from the budget the following services, as shown in the fourth column of Table 6.21: Emergency Medical Teams, General Practitioner Practices, and Settlements with the EU. The rest of the elements of the package it starts to purchase Table 6.22: Possible Government Subsidy Distribution for Use of Private Insurers in Latvia, 2009 Distribution of Subsidies for… Poorest Quintile 2 Quintile 3 Quintile 4 Richest Total Inpatient 30 25 20 15 10 100 Dentistry 40 30 20 10 0 100 Secondary Outpatient 30 25 20 15 10 100 Medicines 30 25 20 15 10 100 Household Spending on Health 2008 per 70.92 132.00 137.28 131.28 209.04 132.73 person annually Source: Staff Suggestion and CSB from private insurers. Just as citizens choose their family doctors today, in this example they would continue to do that, but they would also choose a private insurer for the rest of the services, which are shown in the last column. They include inpatient care, dentistry (for children), secondary outpatient services, and medicines. The public sector would provide subsidies that would go to the household’s choice of insurer for these services. 146 The surplus of revenues over claims for health insurance in Table 1 was 44 percent in 2007, 30 percent in 2008, and -19 percent in 2009. In a normal year, these gross margins would be adequate to finance the administrative costs and a substantial profit. 147 There is no expectation that administrative costs would decline, although they would likely do so, but that the selection of bad risks and the high cost of managing that problem considerably raises the costs for private insurers today in the marginal market they are in. Solving this problem for them by including the whole population should also allow them to both cut risk margins that are built into their pricing and to raise the benefits limits on their plans. However, the system would also have to provide for reinsurance beyond these limits, which would likely be inexpensive and is ignored for the moment. 245 494. Not all households would be equally able to pay for insurance, so an allocation of subsidies by income quintile must be decided. One option is shown in Table 6.22, which reallocates from the top two quintiles to the bottom two quintiles. Each cell contains the percent of the relevant subsidy that would go to each quintile. The bottom row shows out-of-pocket spending on health services per person by quintiles in 2008, in nominal 2008 Lats. 495. Table 6.23 shows the full set of subsidies under these assumptions. The first three rows show the assumed distribution of government subsidies for the services it will continue to finance directly. We Table 6.23: Resulting Subsidy Pattern for Use of Private Insurers in Latvia, 2009 Distribution of Subsidies for… Poorest Quintile 2 Quintile 3 Quintile 4 Richest General Practitioner (Government) 17.99 17.99 17.99 17.99 17.99 Emergency Medical Teams 11.68 11.68 11.68 11.68 11.68 (Government) Settlements with EU (Government) 1.24 1.24 1.24 1.24 1.24 Inpatient (Private Insurers) 112.27 93.56 74.84 56.13 37.42 Dentistry (Private Insurers) 4.50 3.37 2.25 1.12 0 Secondary Outpatient (Private Ins.) 44.67 37.22 29.78 22.33 14.89 Medicines (Private Ins.) 47.64 39.70 31.76 23.82 15.88 Total Government Subsidy 239.98 204.76 169.54 134.32 99.10 Source: Staff Calculation assume the subsidies are provided equally across income quintiles but of course the actual distribution will depend on use patterns. Subsidies to private insurers depend on the quintile of the household, following the patterns laid out in Table 6.22. The last row shows the total subsidy assumed to be provided to each quintile. There is no increase in the amount of budget; this table simply distributes existing government spending in 2009 in this manner. 496. The final step is to see how the result might be financed. Table 6.24 shows in the first row the amount of the subsidy retained by government for the services it will finance directly, and the second Table 6.24: Financing the Private Insurance Package in Latvia, 2009 Poorest Quintile 2 Quintile 3 Quintile 4 Richest 1. Direct Government Purchases 30.91 30.91 30.91 30.91 30.91 2. Indirect Purchases through 209.07 173.85 138.63 103.41 68.19 Private Insurers 3. Net Cost of Private Policy 269.09 269.09 269.09 269.09 269.09 4. Addition for Medicines 31.76 31.76 31.76 31.76 31.76 5. Total Cost of Private Policy 300.85 300.85 300.85 300.85 300.85 6. Amount to be Financed (row 5-3) 91.78 127.00 162.22 197.44 232.66 7. Amount to be Financed Beyond Current HH Spending (row 6- 20.86 (5.00) 24.94 66.16 23.62 bottom row of Table 6.22) Source: Staff Calculation row shows how much will go annually for each person to the private insurer of choice. This part declines as household income increases. The third line is the total cost of the private policy, which is the LVL 300 246 assumed cost, less the funding retained by government for services the package will not finance (principally GPs and EMTs), or LVL 269.09 for everyone.148 “Addition for Medicinesâ€? adds back the current per capita cost of medicines because it is not included in the private policy we are using as our model. The row labeled “Amount to be Financedâ€? is the difference between the government subsidy and the total cost of the package for each quintile. The final row compares this amount to existing household out-of-pocket spending (from Table 6.22) on the assumption that a comprehensive package should be able to reduce existing household spending. For example, for the poorest quintile, LVL 91.78 per person would need to be financed for this approach to work. If somehow the existing out-of-pocket spending of LVL 70.92 could be captured to finance the package, it would still fall short by LVL 20.86, which would have to be financed through other means. We fall short of current spending for every quintile but Quintile 2. In practice, it would be almost impossible to make up the difference without spending more tax money on health care, but a dedicated reformer would want to come up with a variety of economically sensible options for closing the gap in costs. 497. Financing the gap and doing so in part by capturing current household spending would require some combination of additional taxes, copayments, and deductibles. Perhaps some of these costs could be eliminated through further savings by insurers. 498. The point of this experiment is simple. We took the most expensive private insurance package available today in Latvia because it covers more or less the same services as the current government plan, although probably without the waiting times. We examined how it might be financed for the whole population through a combination of targeted government subsidies and additional funds, either from the insured or the tax system. Surprisingly, the numbers in this example come close to adding up. In this example, we come up about LVL 60 million short to finance the whole package. Developing other options and much more accurate estimates of costs would be necessary to take this thinking anywhere close to implementation. Policy Implications 499. Shifting to the type of subsidized multi-insurer approach outlined here would be a major change and would have to be done with considerable care. The example is grounded in existing offerings in the market, and logically it is a simple extension of the existing “choice-of-family-doctor modelâ€? that Latvians are familiar with. 500. We do not advocate this solution. Rather, instead of worrying about the philosophy of what to do, we take as a given the stated desire of many policymakers to consider multi-insurer options. We start with the numbers and ask whether it might be possible. We do not advocate this approach. We are only suggesting that any consideration of alternatives needs to look carefully at the numbers, how to continue to cover the entire population, how to distribute public subsidies to do that, and how a reform would be implemented within the institutional and resource environment of Latvia today. 148 The High-Level Coverage option includes emergency services, so it is legitimate to remove direct government funding for that element. We also remove the small reimbursement for the EU. 247 501. Getting to such a large reform would require considerable planning and careful preparation of the public, providers, and insurers. It would require extremely active purchasing on the part of the MoH because the MoH would have to design requests for proposals from insurers based on what it would like to accomplish. Beginning to tender now for providers and possibly piloting the tendering of insurers could help the government learn how to proceed with confidence that it would not do harm and would actually improve the current system. The data and analytical requirements for monitoring such a system and trying to make it focus on health results would be enormous. V. Conclusion 502. Here are the main points made in this chapter: a. Health Status. The health of Latvians, especially of men, is well below where it could be. From the evidence presented, despite rapidly increasing spending on health care after 2004 until the economic crisis hit, it appears that inadequate attention and resources were given to reverse this situation through better primary care and prevention. Instead, rather resources were spent to improve acute care upon occurrence of a health event. Without a national focus on the crisis in health, with all of its resources aligned behind overcoming the problem in a cost-effective way, Latvia has little prospect of converging on health with the original EU States within an imaginable time line. However, more can be done. Nevertheless, the health data point to obvious interventions at the GP and secondary specialist levels and in general public health interventions for the major killers (heart attack, stroke, and external causes). We recommend holding the health system and providers more accountable for health outcomes. This can be accomplished by adjusting GP and secondary specialist compensation to require improved monitoring, prevention, treatment, and coordinated care related to cardiovascular health. It also means redoubling efforts to reduce deaths due to external causes. These changes would be part of a commitment by the MoH to the Cabinet of Ministers to use all of its instruments to achieve goals in these areas, which would help Latvia catch up to the EU. b. Population-Based Monitoring Data. The FINBALT survey provides important evidence of health outcomes and behavior that is an essential complement to the claims-based data that the government generally uses to monitor the health system. We would recommend making the survey a much more robust tool to measure Latvians’ health and health-related behaviors as part of the health convergence campaign and as part of health policy-making. c. Hospital Capacity. The 2010 budget cannot finance the inpatient care infrastructure that exists. Nor in our view is that infrastructure needed. The overall health budget in 2010 is 24 percent below 2008 in real terms, but the inpatient care budget is 51 percent lower. While aggressive steps to reduce hospital beds have been made, cutting them from about 17,000 in 2008 to 13,000 in 2010, this 24 percent reduction does not begin to keep up with the cut in spending. So Latvia still has difficult choices to make on the 248 2010, 2011, and 2012 health budgets: either to raise inpatient spending to keep the hospitals running or reduce hospital beds to put them in line with the available budget. We estimate that to keep the same level of spending per bed as in 2009 (which was 18 percent below 2008) would require an increase of about LVL 36 million in the inpatient care budget in 2010. If, on the other hand, Latvia chooses to keep the total inpatient budget where it is and force the physical resources to adjust, we estimate it would need to have cut approximately 3,000 more beds by January 1, 2010, and of course more beds the later in the year that the decision is made.149 The fact that so much inpatient volume could be cut in 2009 and that there was such rapid uptake of day services in their first year shows that the system can adjust rapidly while still providing the needed care. Therefore we would recommend shrinking the hospital infrastructure to fit within the budget rather than the other way around but recognize this cannot be done overnight. The Government could consider a 5,000 bed cut that takes place at a measured pace from 1 July 2010 to 31 December 2013. That would leave Latvia with about 350 beds per 100,000, at about the level of Denmark in 2007.150 The additional beds cut in 2010 would not fully solve this year’s problem, and it is quite likely that half or more of the shortfall of LVL 36 million would need to be restored to the inpatient care budget – an addition to the budget deficit. Some system, such as that proposed in Box 6.1 or an alternative, is urgently needed to right-size the hospital sector for Latvia’s budget today and its needs over the ensuing decade. d. Efficiency of Service Provision by Hospitals and Secondary Outpatient Services. Along with a reduction in the capacity of the inpatient sector, there is a need to shift hospital compensation toward prospective payment mechanisms such as diagnosis-related groups (DRGs). In addition, DRGs for secondary ambulatory services could help to integrate compensation of secondary outpatient services and hospitals to improve incentives for efficiency and continuity of care.151 Inefficiencies abound in hospitals. First there is way too much capital in the form of buildings, beds, and the accompanying equipment. Second, it is used inefficiently. Between 2005 and 2009, spending on inpatient care rose 31 percent while the number of unique patient admissions in 2009 fell 17 percent from the 2005 level, and total bed days dropped 24 percent. Nevertheless, Latvia’s long average length of stay hardly changed. During the same period, expenditures for secondary outpatient services increased 72 percent in real 149 Such a cut in many countries would not be feasible, but in Latvia, at least nominally, it is a matter of dropping or reducing inpatient contracts with selected hospitals. However the fact that the contract is changed does not physically eliminate the beds, and they may come roaring back onto the budget in the future if hospitals succeed in exerting political pressure to be reinstated. 150 We mix acute and long-term stay beds because Latvia has too many of both. The Denmark number is acute care beds only. 151 Latvia currently uses a mix of prospective payment and fee-for-service in hospitals, paid within a global budget. Specialists are compensated on a fee-for-service basis. 249 terms on a 16 percent increase in the volume of patients.152 General practitioners (GPs), who are paid almost entirely on a capitation basis, saw a spending increase of 50 percent with a 16 percent increase in patient visits. The rise in costs for lower production in hospitals and no improvement in productivity among secondary outpatient services relative to GPs is likely a result of the weight given to fee-for-service payments for hospitals and secondary outpatient services and an expansion of technology for diagnosis and treatment of acute cases. It also suggests scope for improving productivity in these two areas with a change in how they are compensated. This would complement efforts to reduce hospital over-capacity. In 2010, Latvia changed how it pays hospitals to a global budget system, although within that “black boxâ€? they still account for their volume on a fee-for-service basis. None of the options suggested here would immediately solve the problems but would lay the groundwork for a more flexible system in the future that would allow the government to better define the limits of what it can finance and encourage greater provider accountability for performance of the system. i. One option it to leave everything as it is but to reform the payment system for inpatient and specialist care. In this scenario, Latvia would adopt a DRG payment system as soon as possible and start piloting it as a parallel system in its major cost centers no later than 2012. If the results of this pilot are promising, full implementation could begin in 2013 and 2014. ii. A second option, complementary to the first, is to begin contracting competitively for services. The current system of annual contracts could be replaced incrementally with competitive contracts to deliver a specified mix of services to a defined population. Latvia could thus begin defining the quantity, continuity, and quality of care it would like to purchase and encourage providers to collaborate to deliver those services – and most importantly to begin bidding competitively to provide them. It could also put in place rules that providers – such as specialists – who are parties to these bids provide 100 percent of their efforts to government-financed patients and meet performance targets for quality, results, and satisfaction. The current system exposes patients to efforts to switch them to private practices and encourages specialists to work in multiple full time jobs simultaneously. iii. A third option that would be much more costly and difficult to implement, with uncertain results, would be to begin offering a voucher for individuals to purchase insurance on the open market that would provide the same level of coverage that exists today, as shown broadly in Figure 6.1: catastrophic protection for emergency and inpatient care, general practitioner services, coverage of essential drugs, dental care for children, and specialist referrals. The voucher would be fully subsidized for the needy and perhaps require a 152 Although secondary outpatient services include tests and other services, the number of secondary outpatients is a reasonable measure of the quantity of services provided because all of the tests and other activities are presumably for these patients. 250 contribution on a sliding scale from those in higher income groups. Coverage beyond this basic level would be allowed, but at an extra out-of-pocket cost. This change would be a complex undertaking and would require a careful design, but it would allow the government to open the market for care, define the rules of competition, define a basic benefit package, set performance standards, and encourage payers to achieve greater continuity of care. It would be a revolutionary change from the current approach. It is not recommended by us but is certainly an option. In short, within the current financing system, there is a need to move away from fee-for-service payments as soon as possible and shift to DRGs for care at the specialist and inpatient levels. More radical changes to encourage greater competition (and in fact a greater level of control over results for the MoH while reducing its direct role in contracting) would be to shift to competitive procurement for providers (almost certainly with longer term contracts than is the practice today). Other more radical options can also be considered. e. Efficiency of Administration. We recommend that service delivery activities that remain within the MoH budget be shifted to the contracting system managed by the Payments Center and that these care-giving organizations be compensated according to DRGs or capitation, as appropriate. Once the EMT consolidation is completed, it could follow the same path back to the purchaser-provider split managed by the Payments Center. f. Emergency Safety Net. The ministry’s design of the health part of the emergency social safety net deserves plaudits on many fronts. It provides an incentive to local authorities to identify beneficiaries correctly and make sure they receive benefits; it treats everyone the same nationally; the payment mechanism is administratively simple and creates no burden on the client; and revisions in early 2010 removed the 100 percent tax on benefits that originally occurred at the “needyâ€? line. It protects the needy from copayments and encourages providers to deliver traditional inpatient services on a day or outpatient basis. We recommend that the Government shift fully to the safety net approach while keeping the payment caps that are in place to provide the whole population with financial protection. g. Insurance. Many people in Latvia advocate additional insurance coverage for the population that would supplement the insurance provided through the public budget, allowing Latvia to mobilize additional resources for health outside of the tax system. After reviewing government and household expenditures, as summarized in Figure 6.1, the PER concludes that Latvia now has a reasonable system to protect its population against catastrophic financial risks in health. It also has a functioning secondary private market for insurance that fills in gaps in public coverage for just over ten percent of the population. There is no reason to stop the development of private insurance, which is likely to continue to evolve as a gap-filling mechanism (and the gaps are growing as public budgets decline). The secondary market would benefit from a stable, clear 251 definition of what the public budget will finance. The current system of a negative list that excludes few services but rations care through waiting lists does not create clear rules of the game that would allow private insurers to better define their offerings. If Latvia would like to move to a system where households choose their insurer as they choose their family doctor today, then specification of the benefit package financed by the government will be essential. Under any reforms, insurance would need to maintain the principle of 100 percent coverage of the population, financial protection for everyone, subsidies for the poor to reduce further barriers to use for them, and a long list of highly probable benefits relative to the current approach that could offset the considerable risks that radical reforms would create. h. Health Care Provision as a Business. Although it is not directly addressed in this chapter, only limited elements of the health system have been privatized, namely General Practitioners, Specialists in private practice, and pharmacies. For the most part, even these elements of the system are on contract to the State. Capital-intensive hospitals and most of the infrastructure are owned by the State and by local governments. They are run – to a limited extent and in varying degrees – like businesses but also have political overseers and protectors. If they have contracts with the Payments Center, most of their revenue comes from the State. European Structural Funds support investment in the infrastructure, even though much of it is not economically viable and may not be suitable for its purpose. i. We urge Latvia to review this situation and consider whether the delivery of medical care can be moved into the market to a greater degree. The public sector can continue to have a primary role in the financing of care, but the market will only develop fully if the delivery of care has a greater element of private participation. Advocates of a competitive insurance market in fact tend to argue for it in part as a means to encourage development of health as a sector outside of government control. It would probably be better to do this directly by separating existing institutions fully from government ownership. Poland, for example, is advancing a program to “corporatizeâ€? its hospitals by shifting them from government-owned limited liability companies to joint stock companies that would face bankruptcy if they cannot pay their bills. Shares at first would be owned by the same governments that own the hospitals today, but if they become economically viable they could be sold. They will be governed by the same laws as private businesses, and there will be more of an arm’s length relationship with government. Another alternative is to turn hospitals into nonprofit companies owned by foundations that are managed by independent boards but have no connection to a government. This is an economic develop agenda as much as a health agenda because the purpose would be to spur greater efficiency and entrepreneurial activity in the health sector. 503. In short, the pressing issue in Latvia is the one highlighted by MoH decisions to prioritize GPs, maintain emergency services, build a safety net for the poor, maintain services for mothers and children, and maintain access to medicines. That has made inpatient care the residual category in the 252 current budget environment. In the short term – 2010 through 2012 – shrinking inpatient capacity in line with available resources is the fundamental policy problem. There is so much overcapacity and so many preferred alternatives to inpatient care that further downsizing it is unlikely to affect health results negatively and may in the longer term contribute to improvements. The money freed up to fund the government’s priorities may contribute to better health, although it would take years to know the result. The MoH made many smart choices in 2009 as it adjusted to the economic crisis. It probably has helped the health system morph into one that is more appropriate to Latvia’s needs. The longer term challenge is how to marshal the resources at hand to boost the health and life expectancy of the Latvian people. This agenda needs all the attention it can get as well as leadership and commitment by the Government of Latvia. VI. Recommendations 504. The PER matrix contains details of the recommendations, but they are summarized here: a. As a result of budget cuts and priorities set in 2009, Latvia now prioritizes adequate funding of general practitioners, emergency care, health services for mothers and children, dental services for children, covered prescriptions, referred secondary outpatient services, a safety net for the needy, and unplanned hospitalizations. Hospitals have become a residual in the budget, which provides funding for continued catastrophic protection at that level. However, the hospital infrastructure has – not surprisingly – not adjusted in one year to substantially lower funding. Therefore we recommend restoring some of the funds for hospitals in 2010 if the funding is available but only with the proviso of an agreed technical plan for closure of hospital beds and buildings – a plan that will be implemented from 2010 through 2013. The Government should consider a 5000 bed cut that takes place at a measured pace from 30 September 2010 to 31 December 2013. By 2013 the goal would be to achieve an overall ratio of 350 beds per 100,000 population. Although there would be some additional spending required over this period, by 2013 the goal would be to realize spending for inpatient hospital services at 35 percent or less of total medical care expenditure. Hospitals must be closed or they will come back on the budget in the future, and as long as buildings are kept open much of the expected savings will not materialize. b. Reduce average length of stay by 1.2 days in 2010 (to 7.5), 0.5 days in 2011 (to 7.0), and 0.5 days in 2012 (to 6.5). Latvia’s length of stay has been stuck at a high level for many years; cutting it is essential if fewer beds are to be adequate. The high length of stay may reflect reimbursement policies and could be tackled with conversion of secondary outpatient services and inpatient care reimbursed based on diagnostically related groups. But it will take time to change reimbursement policies and for providers to adjust; in the meantime, direct action will to be taken to reduce lengths of stay. c. Reduce MoH specialized health care provision through marginal improvements in productivity. These services provide care and should be financed through the Payments Center. 253 d. Review copayments. Copayments are a variable that the MoH has control over. Copayment policies can be adjusted further so that those who can pay, pay more, while the poor are better protected. Close attention needs to be paid to annual data on the percent of income devoted to health care by quintile in the household consumption data. In particular, pharmaceutical reimbursement policies should be adjusted as needed to reduce the burden of these expenses on the poor and chronically ill. A reasonable goal is to maintain in the household consumption data health care as a flat or increasing function of expenditure per capita (so the poorer quintiles are devoting a smaller proportion of their expenditures to health care than are the rich). e. Fully means test exemptions. Shift targeting of exemptions from social groups to completely means testing. f. Pilot competitive contracting, which would develop several instruments missing in the system: public competitive procurement of medical services based on tenders transparently defining the services to be provided; tenders encouraging providers to cooperate to integrate care and compete for patients on that basis; contracts exceeding a year; much clearer rules of the game for providers and other insurers on what the government will cover, on service levels, quality of care, and so on. g. Clarify the guaranteed package of health coverage and the rules of the game. There is no compelling argument for structural reform of health finance at this juncture, such as a shift to a social insurance model, private insurance, or a mix of public and private insurance beyond what exists today. However, to encourage public debate, we have provided analysis of one of many possible options for development of the insurance market. We do argue for development of competitive contracting, which would allow the State, as it gains experience, to extend this to the competitive contracting of insurers if so desired. We do not advocate it. A clearer definition of what is the core level of services the State will finance (and limits possibly on providers’ ability to game multiple payers) would encourage the development of a more robust private supplementary insurance system that would open possibilities for the future. 505. The most important opportunities created by the economic crisis have been taken advantage of by the Ministry of Health: prioritizing the right services (in our view) and creating a means tested safety net that did not exist previously. The path ahead demands a much stronger effort to rationalize the hospital sector, to improve the prevention of Latvia’s main killers, to reform the reimbursement system to improve incentives for secondary outpatient services and hospitals, and to encourage the use of market forces to make the health sector more dynamic as a contributor to Latvia’s economy and not just a ward of the State and local governments. 254 Chapter 7: Long-Term Care I. Introduction 506. Latvia, like other European countries, faces a significant aging of its population over the coming decades. Although Latvia is relatively early in its demographic transition when compared to Western European countries, the elder population above the age of 65 is nevertheless projected to increase considerably over the next 50 years. This increase in the elder population will have serious consequences for the demand of certain publicly financed services, like for example health care services. Although there is some uncertainty about the exact impact of aging on health care expenditures, there is consensus in the literature that the demand for long-term care (LTC) services is bound to increase strongly. The number of dependent people who are in need of LTC services will undoubtedly increase in the future, while at the same time the number of potential care givers in the rest of the population will decrease. This raises the question of who will take care of the elderly in the future. At the same time, the population in working age will also decrease significantly, which raises the question of who will pay for future LTC needs? 507. This section will explore some of these issues by looking at the demographic transition in Latvia and drawing some conclusions on the long-term impact of aging on the dependent population. The section also explores current public expenditures on LTC in Latvia, and discusses important potential efficiency gains in the health sector by shifting services to the LTC sector. The main finding is that Latvia has to prepare for the demographic changes ahead by adjusting its health and social policy towards more community-based LTC services and by strengthening the linkages and the coordination between the health and the social sector. Latvia currently spends relatively little on LTC, so in the short term there seems to be no room for decreasing spending, to the contrary. In the medium term, though, increased investments in community-based LTC services could lead to important efficiency gains in the health sector. II. Aging and Its Long-term Impact on the Dependent Population 508. Over the next 50 years, the Latvian population will age considerably. Latvia is at a relatively early stage of its demographic transition when compared with Western European countries. Its largest cohort—in a Western European context called Generation X—is currently 26 years old.153 The children of this bulge generation, which will be born over the next 10 years, will help to keep the growth of the youngest age group positive for some time, but will be negative after 2020 (see Figure 6.1). The all important working age group—aged 15 to 64—is projected to constantly decrease over the next 50 years. The older age groups, in contrast, are projected to grow strongly. As the parent generation of Generation X—the so-called baby boomers—start to retire over the next 20 years, the annual growth of 153 In France, in comparison, the largest cohort is currently around 35, in Italy around 40, and in Germany and the United Kingdom around 45. 255 the 65 to 74 age group will increase to up to 3 percent in 2022. This is followed by a significant and constant expansion of the very old—the population aged 75 and older. This age group is also expanding strongly between now and 2018. Figure 7.1: Projected annual population growth rates for Latvia by age group (2010-2060) Source: Eurostat (2010) and author’s calculations. 509. The reason why the growth rates of the old and very old are important is that they are the main groups demanding LTC services. People who are in need for LTC services have restrictions in performing activities of daily living (ADLs). In other words, they are hampered in their mobility and thus in cooking, cleaning, personal hygiene, shopping, eating, and so on. They are therefore dependent on others to support them. The reason for their dependency could be a mental or physical disability—in particular dementia—but in the case of elderly people it is not necessarily disability, but frailty in general. In any case, the share of the dependent population increases strongly with age. In Latvia, only 1.4 percent of the population aged 15 to 24 is considered severely dependent (or severely hampered), while the same share is 27.2 percent for the population aged 75 to 84, and 41 percent for the population aged 85 and older (see Table 7.1). These shares are more or less in tune with other EU countries. The share of the population with dementia also increases strongly with age (see Table 7.2). Table 7.1: Share of severely hampered people by age group in 2008 (percent) 15-24 25-34 35-44 45-54 55-64 65-74 75-84 85+ Latvia 1.4 1.9 4.0 5.9 11.2 15.9 27.2 41.0 EU-average 1.5 2.3 3.8 6.9 10.9 14.6 25.0 39.5 Source: Eurostat (2010). 256 Table 7.2: Share of people with dementia by age group in Latvia (percent, 2005) 60-64 65-69 70-74 75-79 80+ Women 0.5% 1.1% 3.9% 6.7% 17.7% Men 1.6% 2.2% 4.7% 4.9% 15.7% Source: Alzheimer Europe (2006) 510. The increase in the older age groups will result in a strong growth of the severely hampered population. Combining the age-specific shares of the hampered population with population projections allows for an estimation of how the dependent population will develop over the next 50 years.154 For Latvia, the severely hampered population is projected to constantly grow over the next 40 years, of up to 0.5 percent annually (see Figure 7.2). At the same time, the healthy, non-hampered population will strongly decrease for the next 50 years to come, by sometimes as much as 1.1 percent annually. This means that the aging of the Latvian population will also mean a shift to a much more dependent population. Figure 7.2: Projected annual population growth rates by level of restriction in activities of daily living (2010- 2060) Source: Eurostat (2010) and author’s calculations. 154 The age-specific shares of the hampered population used in the projections are an average of the shares between 2004 and 2008. The projections do not take into account any eventual improvement in the health status of elder populations over time. The limited data available from between 2004 and 2008, though, seems to suggest an increase in the share of the severely hampered population, and therefore a deterioration and not an improvement in the health status of elder populations. This is not only true in Latvia, but also in other European countries. 257 511. These profound changes in Latvian society are also reflected in the projected development of the inverse old-age and care dependency ratios. While today, there are about 7 non-hampered Latvians for one severely hampered Latvian; by 2060 it will be only 4. In other words, while today there are 7 potential care givers for one dependent person, in 2060 there will be only 4 potential care givers. At the same time, there are 4 people in working age per person in retirement age today in Latvia, while by 2060 it will less than 2. Or, in other words, while today there are 4 persons who can potential pay for the retirement—and the care needs—of the elder population, in 2060 there will be less than 2. Figure 7.3: Projected inverse dependency ratios for Latvia (2010-2060) Source: Eurostat (2010) and author’s calculations. 512. For these reasons, the questions at the forefront of the minds of policy makers are who will care of the elderly population of Latvia in the future, and who will pay for this care. The next section will start investigating these questions by looking at current public expenditures on LTC in Latvia. III. Current Public Expenditures on LTC 513. Unfortunately, there are few data available on current public expenditures—let alone total expenditures—on LTC. This is not only a problem in Latvia, but also in most other countries, and is a general problem when investigating issues of LTC. This makes it particularly difficult to compare LTC expenditures across countries. It mainly stems for the undefined position of LTC between the health and the social welfare sector, which makes it difficult to accurately collect data. 514. According to the System of Health Accounts (SHA) methodology developed by the Organization for Economic Development and Co-operation (OECD), LTC comprises the following categories of services: (i) in the health sector, palliative care, long-term nursing care (including accommodation in nursing homes), personal care services to assist with ADLs, and services and financing in support of 258 informal (family) care; and (ii) in the social sector, home help and care assistance, residential care services other than nursing homes, and other services like daycare and transportation.155 515. The actual data on LTC, though, that are available in the SHA database varies strongly across countries. This makes the data hardly comparable across counties. In the case of Latvia, expenditure data on LTC is only available for two years, 2005 and 2006. According to these data, Latvia spent 1.4 percent of GDP on LTC in 2005, and only 0.2 percent of GDP in 2006.156 This large variation already shows that even for the same country, the data seem not comparable across time. For Latvia, for example, the large variation is explained by recorded expenditures on LTC of LVL 111.04 million in the social welfare sector in 2005, while in 2006 this expenditure category is missing. Overall, the SHA data on LTC have a strong bias towards health sector data, with data on expenditures in the social welfare sector largely absent, especially for EU10 countries. 516. Nevertheless, the limited data from the SHA database that is available suggests that Latvia— along with other EU10 countries—spends relatively little on LTC. Table 7.3 lists total expenditures on LTC (as a percentage of GDP) only for countries where expenditures are relatively constant over time, which suggest at least consistency of data collection within countries. The data imply that high-income countries currently spend as much as 3.7 percent of GDP on LTC (Sweden). The EU10 countries, in contrast, all report much lower numbers, generally spending less than 1 percent of GDP on LTC. Ignoring the outlier for 2005, the data suggest that Latvia spends as little as 0.2 percent of GDP on LTC, which would certainly put it at the bottom of the comparator countries presented. 155 See OECD (2008). 156 These numbers refer to total expenditures, that is, including public and private expenditures. 259 Table 7.3: Total LTC expenditures as a share of GDP in select countries (according to system of health accounts, 2003-2007) Average 2003 2004 2005 2006 2007 2003-2007 Sweden 3.8 3.8 3.7 3.7 3.6 3.7 Norway 2.4 2.3 2.2 2.1 -- 2.3 Switzerland 2.2 2.2 2.2 2.1 2.1 2.1 Finland 2.0 2.1 2.2 2.1 2.1 2.1 Denmark 2.0 2.0 2.1 2.1 2.0 2.0 Germany 2.0 2.0 2.0 1.9 1.9 1.9 Iceland 1.9 1.9 1.9 1.9 -- 1.9 Japan 1.6 1.7 1.7 -- -- 1.7 Belgium -- -- 1.6 1.6 1.7 1.6 France 1.4 1.5 1.5 1.6 1.6 1.5 Austria -- 1.3 1.3 1.3 -- 1.3 Netherlands 1.3 1.3 1.3 1.3 1.2 1.3 Slovenia 1.1 1.1 1.2 1.2 1.1 1.1 United States 1.0 1.0 1.0 1.0 -- 1.0 Spain 0.7 0.7 0.7 0.7 0.8 0.7 Lithuania -- 0.4 0.4 0.4 0.5 0.4 Poland -- 0.4 0.4 0.4 0.4 0.4 1 Hungary 0.3 0.3 -- 0.2 0.2 0.3 Czech Republic 0.3 0.2 0.2 0.3 0.3 0.3 Cyprus 0.2 0.2 0.2 0.2 0.2 0.2 Estonia 0.1 0.1 0.2 0.2 0.2 0.2 Portugal 0.1 0.1 0.1 0.1 -- 0.1 Bulgaria 0.0 0.1 0.2 0.0 0.0 0.1 Romania 0.0 0.0 0.0 0.0 0.0 0.0 Average 1.2 1.2 1.2 1.1 1.1 1.2 Latvia -- -- 1.4 0.2 -- 0.8 Note: 1. The 2005 value for Hungary (8.23 percent of GDP) was dismissed as an outlier. Source: Eurostat (2010). 517. An alternative—more limited but also more reliable approach—is to focus on public expenditures instead of total expenditures on LTC. As already mentioned, the main gap in the SHA data is expenditures in the social welfare sector. In most countries, the largest share of public expenditures for LTC in the social welfare sector—in particular for the elderly—occurs on local government level. This decentralized financing of LTC services makes it difficult to obtain consolidated expenditure data for a whole country. A recent initiative by the World Bank, though, aims at developing consolidated government expenditures databases that include expenditures at all government levels. This allows also for the first time to obtain detailed public expenditure data on LTC at the local level of administration. A first detailed analysis of this type of data was recently conducted in Poland, and resulted in estimated 260 public expenditures on LTC in the health and social sector of 1.0 percent of GDP in 2007 (as opposed to estimated total expenditures, including private expenditures, of 0.4 percent according to SHA). 518. A similar analysis in Latvia results in estimated public expenditures on LTC in the social welfare sector of about 0.44 percent of GDP in 2008 (see Table 7.4). Expenditures increased strongly from LVL 59 million in 2007 (0.4 percent of GDP) to LVL 72 million in 2008. Note that this excludes any spending on LTC in the health sector or any private spending on LTC. Rather, these are expenditures in the social sector for incapacity, temporary incapacity, disability, and support for the elderly. It excludes any cash benefits or social security benefits that are not specifically labeled to support care or care givers. Most importantly, it excludes any old-age or disability pensions. Rather, these expenditures truly focus on LTC services provided by the social sector—that is, in-kind LTC benefits and some limited cash benefits for care. Accordingly, the main expenditure items are remuneration, goods and services, subsidies and grants (mainly to NGOs, public providers, and charity organizations), and capital investments. Table 7.4: Total government expenditure on LTC according to government expenditure data (LVL, 2007/08) Code Description 2007 2008 10.100 Social protection in case of incapacity 2,985,554 3,894,635 Central government 255,944 3,618 Local government 2,729,610 3,891,017 10.110 Social protection in case of temporary incapacity 0 83,828 Central government 0 83,828 Local government 0 0 10.120 Social protection in case of disability 34,790,221 40,951,782 Central government 34,790,221 40,951,782 Local government 0 0 10.200 Support for the elderly 21,304,872 27,343,030 Central government 0 0 Local government 21,304,872 27,343,030 Total 59,080,647 72,273,275 as share of GDP 0.40% 0.44% Central government 35,046,165 41,039,228 as share of GDP 0.24% 0.25% Local government 24,034,482 31,234,047 as share of GDP 0.16% 0.19% Source: Latvia Ministry of Finance. 519. At the central government level, expenditures on LTC increased strongly during 2008 (+17.1 percent nominally) and contracted significantly during the crisis year of 2009 (-16.8 percent). In nominal terms, expenditures were below 2007 levels in 2009 (LVL 34 million versus LVL 35 million). The central government’s responsibility in providing LTC is strictly limited to the population with disabilities. The strongest increases in spending during 2008 and decreases during 2009 were recorded for expenditures 261 on remuneration and the purchase of goods and services. The only exception is fixed capital formation, which increased significantly during 2009 (+117 percent). This suggests that sending cuts focused on salaries while investments in infrastructure were expanded. 520. On local government level, unfortunately no data are available for 2009. Local governments’ responsibilities focus on support for the elderly and the population unable to work. During 2008, expenditures increased at a much higher rate (+30 percent) than on central government level, from LVL 24 million to LVL 31 million. As was the case at the central government level, this increase was driven by increases in remuneration and purchases of goods and services. Interestingly, expenditures on subsidies and grants went down considerably, while fixed capital formation increased strongly. This could suggest a shift from buying LTC services from other organizations (like NGOs and charities) to the provision of LTC services by municipalities themselves. This raises some concerns about crowding out private and non-profit sector provision and could be further investigated by the authorities. IV. Future Policy Directions 521. LTC is a crucial policy area, with important fiscal implications in the medium and long-term. It is important that Latvia improves its data collection and recording practices on LTC services in order to develop good and fiscally sustainable policies in this area. The OECD SHA methodology gives good guidance on how to record public (and private) expenditures, and it is important that Latvia strengthens its efforts along these lines, also at the level of local government. 522. What can be said with certainty at this point is that Latvia spends relatively little on LTC and that increased investments in LTC services could lead to important efficiency gains in the health sector. Due to the overcapacities in the hospital sector it is conceivable that a large part of LTC is currently provided in hospitals, but in the guise of in-patient services. This is both expensive and to the determent of the wellbeing older people, who receive the wrong type of care and are exposed to infection risks in hospitals. 523. The relatively high average length of stay in hospitals (presented in the previous chapter) suggests that improved patient discharge management could decrease inpatient expenditures. As discussed at length in Chapter 6, patient discharge management aims at reducing length of stay in hospitals by passing on patients to more cost-efficient outpatient services like post-surgical treatments and rehabilitation. The fact that Latvia currently spends only 0.8 percent of total health expenditures on rehabilitation hints at a lack of rehabilitative services. In contrast, for those countries where such data are available, expenditures on rehabilitative services as a share of total health expenditures is around 2.9 percent (including many EU10 countries). 524. In the case of older patients, patient discharge management also involves social workers who can help organizing temporary (or permanent) assistance with ADLs after discharge. For older patients who live on their own and lack family support, even a minor injury—like a broken leg—can leave them temporarily dependent on outside support. Given the overcapacities in hospital beds, hospital management might currently be more than happy to keep such patients hospitalized until they are fully 262 recovered, although such patients are not in need of medical, but social welfare services. A more cost- efficient solution would be to discharge the patient when no more medical treatment is necessary and provide the social support through community-based LTC services. This could either happen through temporary home-based or even residential LTC services. The city of Riga has already successfully implemented this approach. A careful evaluation of the experience in Riga could form the basis of a countrywide rollout of discharge management. 525. Any publicly funded investments in building up more LTC (and also rehabilitative) service facilities, though, would need to be to be carefully evaluated against other policy objectives. Various potential policy objectives come to mind. First, reforms should be cost-effective and have the wellbeing of patients in mind, which implies a focus on community-based services like home care and daycare. Institutional care is also an important component of any LTC system, but it results in higher-intensity care, is therefore more expensive, and is not the preferred form of care by patients. Second, cost shifting between the health and the social welfare sector should be avoided as much as possible. This implies that coordination of care services has to focus on the needs of patients rather than cost implication for either the health or the social welfare sector. The current LTC needs assessment in Latvia involves both a social worker and a general practitioner, which is an excellent starting point from which to build a patient-focused care coordination system. Third, provider payment rules and capital investments have to ensure a leveled playing field between the public and the private sector. If public providers are subsidized—either directly from budget through staff salaries or through capital investment subsidies—private sector providers are unfairly disadvantaged and discourage the interest of the private sector in providing LTC services. 526. The conversion of small municipal hospitals into institutional LTC facilities is an example of how LTC reform can run counter to some of these policy objectives. International experience shows that many countries at some point in their history have converted redundant municipal hospitals into LTC institutions. The risk that is created by such reforms is that it introduces a bias towards expensive institutional care in a country’s LTC system. Poland provides an example of how this shift can become costly. Some years ago, Poland started to convert small hospitals into medical nursing homes, run and financed by the health sector, that were intended to provide post-surgical treatment at lower costs than in regular hospitals. The unintended consequences were that—given the general shortage of LTC services in Poland—patients and their families continued to use the medical sector as a substitute for social LTC services, at a much higher price than this could be done in the social welfare sector. The medical nursing homes were largely financed from the health insurance fund and came at a much lower price for patients than private or social welfare sector LTC services. The private sector was discouraged from entering the market because for-profit and even non-profit organizations could not compete with the lower user fees in medical nursing homes. Municipalities, who finance most of the social LTC services, also found it cheaper to shift patients to the health sector than provide their own social services. 527. A better approach could be to convert redundant municipal hospitals into community centers— privately or publicly owned—that provide a whole range of LTC and rehabilitative services. Such 263 community centers could be at the center of care coordination for patients. They could house daycare centers for elderly and disabled people (or even childcare), but also outpatient services like physical therapy. They could also be the hosting facilities for home-based services like care assistants or community nurses who support dependent people in their homes. To the extent necessary, they could of course also provide limited facilities for residential care, in particular for limited periods of time and respite care. The exact setup of such community centers would depend on local needs and circumstances and also have to be balanced against transportation costs in sparsely populated areas in rural Latvia. High transportation costs could in fact justify more residential care facilities with low levels of care intensity. 264 Matrix 4. Options for the Government to Consider in the Health Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Inpatient Prepare and implement a 10 10 -15 In 2010, Latvia has The MoH has cut planned The MoH’s priorities have De-hospitalization of technical plan for closure of achieved about 575 inpatient spending to LVL been to maintain health care for a more hospital beds and buildings. [An additional [10 -15 beds per 100,000, just 106 million, plus an expenditures for primary effective and efficient The Government should 25 to 36 million million 2013] above the average for additional LVL 8 million for care, for medicines, for system. Hospitals would consider a 5,000 bed cut that is likely to be Poland, Estonia, and day surgery. As a result, emergency care, for need to be closed and takes place at a measured pace needed in 2010, Europe as a whole in while the overall health mothers and children, permanently removed from 1 October 2010 to 31 which could 2007. This number budget in 2010 is 24 and for the safety net. from the health system for December 2013. By 2013 the then be includes psychiatric percent below 2008 in real These are sound priorities this strategy to succeed. goal would be to achieve an recovered from beds, which Latvia has terms, the inpatient care from population health Otherwise they may return overall ratio of 350 beds per savings to in in abundance relative (plus day surgery) budget and equity standpoints. in the future as a cost. 100,000 population. spending from to its neighbors and is 51 percent lower. While These core expenditures Some sort of privatization bed/hospital Europe. By cutting aggressive steps to reduce mean that inpatient care program or bankruptcy closures in 2011 3,000 beds, Latvia contracted hospital beds is funded by the residual program would be and 2012 would reach about 440 have been made, cutting that is left. appropriate to remove budgets.] beds per 100,000; a them from about 17,000 in While the expenditure them as wards of 5,000 bed cut would 2008 to 13,000 in 2010, priorities are sensible municipal governments allow it to reach 350 this 24 percent reduction from a population seeking support from the beds per 100,000. does not begin to keep up standpoint, the resulting State. Both beds and with the planned cut in limitation on inpatient If given the power and hospitals need to be spending. And the cuts in care poses risks to political support to do so, closed, not just beds beds have not actually individuals if they fall ill or the MoH could rapidly eliminated while the eliminated the hospitals, if they must delay develop and begin buildings are which for the most part planned procedures. This executing a strategic plan maintained. were converted to another risk can be reduced by to reduce the hospital The larger decrease type of care institution. moving deliberately sector so that it consumes would put Latvia at Consequently, in 2010, we rather than precipitously no more than 35 percent about the same density estimate that the budget to further cut hospitals; of acute care expenditures of beds as Denmark in for inpatient care is LVL 25- they should be reduced - about average for the 2007. 36 million below the likely but in a measured and OECD. cost of maintaining the transparent way. current infrastructure and Latvia could benefit from expected utilization. further cuts in hospitals, But it is dangerous to but the Government suddenly shrink inpatient would need to carefully 265 Matrix 4. Options for the Government to Consider in the Health Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit care to fit the budget. The manage the risks that go Government could instead with it for individual implement a strategic patients during the reduction beginning in transition. 2010 but with the major reductions coming in 2011, 2012, and 2013, with the MoH given an objective of achieving LVL 30 million in permanent annual savings on inpatient care by 2013, with agreed benchmarks on reductions in infrastructure to reach that point. Reduce average length of stay -- -- This would accelerate Greater use of modern Very little risk if doctors De-hospitalization of by 1.2 days in 2010 (to 7.5), 0.5 the improvement in outpatient-based are given incentives to health care. See above. days in 2011 (to 7.5), and 0.5 [Savings would [Savings 2009, when average medicine, management of manage patient stays days in 2012 (to 7.0) be part of the would be part length of stay dropped diseases through the better and if safe reductions for of the from 9.5 to 8.7 days, primary care system, day alternatives are available inpatient care reductions and would take surgery, home care. for psychiatric patients. included above] for inpatient advantage of changes Shifting to a prospective Much safer to handle care included encouraged by MoH to payment system for through incentives to above] use day surgery and hospital services, such as doctors and hospitals home care. DRGs, would create rather than through If Latvia moves to 350 incentives for hospitals to edicts from the central beds per 100,000, reduce length of stay. administration. average length of stay Currently, the combination must fall to use the of case-based plus fee-for- remaining beds more service payment both for efficiently. specialists and hospitals Elimination of long provide incentives to term care beds for increase the cost per psychiatry is included patient. It is unclear that 266 Matrix 4. Options for the Government to Consider in the Health Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit in this estimate. the global budget system Because those stays put in place in 2010 will are extremely long, change current incentives they distort the totals. for hospitals to keep patients in beds. 157 Reduce MoH specialized health 2.5 2.5 This area of the MoH Request the MoH to If part of the savings is Unification of the financing care provision through budget includes manage a 5 percent cut in due to deflation in the of care-giving medical marginal improvements in communicable this area through increases economy, this change can institutions. productivity. diseases, sports in efficiency in each fiscal be risk free. As these are medicine, blood year. The logic is simply service providers, it also supply, emergency that small additional cuts makes sense to put them medical assistance, even in areas of high in the same “arm’s disaster medicine, and importance may be lengthâ€? bargaining forensic labs. possible without resulting position (i.e., contracts in major losses of output. with the Health Payment Center) as other service providers are today with the Payments Center. Pilot competitive contracting -- -- MoH could begin DRGs or prospective Minimal for a piloting We are suggesting only piloting requests for payment systems will take exercise. Risks would that this be started or proposals to provide time to develop. The manifest during the pilot experimented with. All services to supplement current system seems and Government would the benefits would be or replace the current locked into a high level of learn how to manage longer term if it becomes a system of renewing unit costs that increased these if decision to go for practical way to do contracts and rapidly between 2005 and a full roll-out of the policy business. Competitive negotiating with 2008. There are clear is taken. contracting is not a providers on quantities signals that the quality of panacea but provides and costs. This would care would benefit from another instrument to make the system more greater integration of manage services financed transparent and give services. by the State. It is a logical 157 To arrive at possible savings, we assume that part of the reduction will be due to deflation and part will be actual savings. These services are currently funded directly by the MoH, and we further assume that they will be put on contract like other service providers. 267 Matrix 4. Options for the Government to Consider in the Health Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit the MoH another tool Competitive contracting extension of Latvia’s to manage costs, would also give the MoH purchaser-provider split. quality, and the another tool to reduce integration of care. hospitals and beds. Competitive contracting would add another instrument to allow the MoH to influence behavior and encourage providers to improve their business acumen. In addition, with competitive contracting, the MoH could require participants to exclusively serve State financed services, reducing the “gamingâ€? that currently is possible for providers under the current mixed system. Copayments and Targeted Exemptions from Health Charges Review copayments. -- -- Copayments were We are not arguing that Latvia does not maintain Fine-tuning co-payments Copayments are a variable that raised in 2009 for the copayments be changed. sufficient household can improve the the MoH has control over. [Variable [Variable first time in five years, We include the option in survey data to distribution of government Copayment policies can be impact] impact] and then reduced in this table because it is a understand the impact of subsidies in health and adjusted further so that those 2010. way to manage demand copayment policies and help manage demand. who can pay pay more, while while also reducing the does not undertake the poor are better protected. amount of subsidy that is sufficient analysis to necessary. It is a tool to understand how best to use if additional revenue is manage copayments. needed. 268 Matrix 4. Options for the Government to Consider in the Health Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Fully means test exemptions. 1 1 The Emergency Social Consistent with other The shift to means testing Shift targeting of exemptions Safety Net is already a suggestions (for the exemptions would from social groups to step in this direction. Welfare and Education improve the transparency completely means testing. sector), targeting can and consistency of public improve the fairness and subsidies in health as well impact of public subsidies. as contribute better to a In this case they are national safety net. The unlikely to save a change is also consistent substantial amount but with fundamental would be consistent with a principles of public government-wide switch economics. The measure to targeting based on saves money and improves means. the equity impact of public Using means-testing to subsidies. target subsidies would also allow the Government to take efficiency measures with greater confidence that the impact of these measures on low-income groups is mitigated. Health Finance Clarify the guaranteed package -- -- Latvia has a mix of Attract additional sources Latvia today has a health “Zero sum gameâ€?. In the of health coverage. There is no public insurance taking of finance to the health system with very low end, Latvia will have to compelling argument for the first level of risks sector, where demand administrative costs decide how much to spend structural reform of health and private insurance seems to surpass the because of its single on health care no matter finance at this juncture, such filling in gaps (provided ability of the State to payer model. A private what the source of as a shift to a social insurance by employers as a finance services. insurance market exists at funding. It can be on model, private insurance, or a benefit to their Coupled with independent the margin. budget or off budget (i.e., mix of public and private employees). hospitals run as via private premiums or insurance. A first priority is to businesses, insurance earmarked payroll taxes), clarify exactly what the could contribute to a more but households will still be public sector will pay dynamic health sector, financing it. It is possible for under current which is now to a large to affect the mix of 269 Matrix 4. Options for the Government to Consider in the Health Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit conditions and degree a ward of the State. payments: increase budgetary restrictions. administrative costs, This has become decrease payments to evident in the MoH’s providers, and change budgetary priorities. copayments depending on But the rules may not the approach followed, but yet be clear to it is unlikely that the total providers, clients, and cost to society will change insurers so that they very much in the next 5 may adjust to the new years. reality. If the “rules of the gameâ€? were set out publicly and pursued consistently, the current public/private provision of insurance would work much better. Private insurance can provide the State the ability to cap public spending for a core package of services for everyone plus subsidies to the needy for additional services. Beyond this set of State-financed benefits, additional insurance would be shifted to individuals (either via premiums or payroll taxes) Long Term Care 270 Matrix 4. Options for the Government to Consider in the Health Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit Improve discharge -- -- Due to the Patient discharge A more cost-efficient management, especially of the overcapacities in the management aims at solution would be to elderly, and eliminate biases in hospital sector it is reducing length of stay in discharge the patient current financing model that conceivable that a hospitals by passing on when no more medical encourage practitioners and large part of LTC is patients to more cost- treatment is necessary and currently provided in efficient outpatient provide the social support households to prescribe hospitals, but in the services like post-surgical through community-based medical in-patient rather than guise of in-patient treatments and LTC services. long term care. services. This is both rehabilitation. The fact expensive and to the that Latvia currently determent of the spends only 0.8 percent of wellbeing older people, total health expenditures who receive the wrong on rehabilitation hints at a type of care and are lack of rehabilitative exposed to infection services. In contrast, for risks in hospitals. those countries where such data are available, expenditures on rehabilitative services as a share of total health expenditures is around 2.9 percent (including many EU10 countries). Convert redundant municipal -- -- Many countries have Sustainable LTC requires The exact setup of such A new “Community hospitals into LTC Community converted redundant models that change the community centers would centersâ€? LTC model could Centers—privately or publicly municipal hospitals bias to community and depend on local needs be at the center of care owned—that provide a whole into residential LTC home care, while still and circumstances and coordination for patients. range of LTC and rehabilitative institutions. This providing specialist and also have to be balanced They could house daycare services. introduces a costly bias some residential care against transportation centers for elderly and towards expensive when it is truly necessary. costs in sparsely disabled people (or even institutional care in a populated areas in rural childcare), but also country’s LTC system Latvia. High outpatient services like transportation costs physical therapy. They could in fact justify more could also be the hosting 271 Matrix 4. Options for the Government to Consider in the Health Sector Savings 2011 Savings 2012 Options for policy action (LVL million) (LVL million) Background Rationale Risk Longer-Term Benefit residential care facilities facilities for home-based with low levels of care services like care assistants intensity. or community nurses who support dependent people in their homes. 272 Annex 1. Health Sector Detailed Data Table A2.1: Details of Government Health Expenditure 2003-2010 in Lats (Nominal) 1 2003 2004 2005 2006 2007 2008 2009 2010 2 Health Overall Budget (millions) 211.5 245.2 296.8 414.2 518.1 575.6 503.7 432.8 3 Insurance System Budget 193,517,885 230,981,297 276,195,340 385,859,105 471,321,668 514,267,447 359,350,384 373,991,728 4 Medicine 17,307,067 20,009,209 30,413,006 42,689,612 61,325,383 71,074,046 71,811,842 71,621,763 5 Medicare 150,822,909 171,375,603 179,072,120 226,230,684 368,366,346 427,578,256 311,567,112 270,337,727 6 Central Procurement of Medicine 7,045,705 9,721,693 7,382,437 3,397,680 4,589,507 4,736,944 5,593,402 6,934,147 7 Reserve Fund 3,031,187 2,629,113 2,650,100 4,916,742 1,408,612 287,982 49,995 8 Other Programs 15,311,017 27,245,679 56,677,677 108,624,387 35,631,820 10,590,219 8,259,885 25,098,091 9 Participants in the Insurance System 2,339,858 2,339,566 2,307,807 2,290,554 2,280,385 2,271,266 2,261,266 2,251,266 10 Medicare (Line 5) Broken into Components 11 Inpatient 105,404,861 99,565,957 143,892,767 201,980,292 241,811,897 169,246,517 105,992,907 12 Inpatient Care 105,404,861 99,565,957 143,892,767 194,754,561 234,333,874 152,812,335 95,049,612 13 Compensation of Copayments 7,225,731 7,478,023 16,434,182 10,943,295 14 Outpatient 54,749,528 55,226,783 69,530,380 120,717,938 138,083,190 113,100,372 121,298,249 15 Secondary Outpatient 23,920,819 29,723,462 36,260,607 69,116,815 88,987,920 67,336,121 75,352,400 16 Secondary Outpatient 23,920,819 29,723,462 36,260,607 69,116,815 86,238,373 63,038,109 69,982,865 17 Compensation of Copayments 2,749,547 4,298,012 5,369,535 18 General Practitioner Practices 26,891,786 21,823,136 29,086,460 45,803,589 42,708,200 40,677,685 40,247,331 19 General Practitioner Practices 26,891,786 21,823,136 25,926,513 41,771,379 41,260,501 38,379,780 37,639,056 20 Compensation of Copayments 3,159,948 4,032,210 1,447,700 2,297,905 2,608,275 21 Dentistry 3,936,923 3,680,185 4,183,313 5,797,534 6,387,070 5,086,566 5,698,518 22 Settlements with EU 1,010,472 1,132,977 1,897,666 2,809,721 2,591,522 23 Emergency Medical Teams 11,221,228 13,336,722 18,127,317 23,175,979 26,842,253 26,410,502 32,226,238 24 Medicare Total 171,375,617 168,129,462 232,560,936 347,007,186 408,635,007 311,567,112 262,108,916 Source: Ministry of Health and the annual reports of the compulsory insurance systems at http://www.vnc.gov.lv/eng/health/publichealthanalysis/ and http://www.vnc.gov.lv/lat/publikacijas/publiskie_gada_parskati/. Participants estimated from 2008 on assuming a loss of 10,000 per year. 273 Note: Line 24 and Line 5 should be equal but they are often not. 2010 is planned expenditure in the budget. All other numbers are actual expenditures. Table A2.2: Details of Government Health Expenditure 2003-2010 in Lats (Real) 1 2003 2004 2005 2006 2007 2008 2009 2010 2 Health Overall Budget (millions) 211,500,000 230,885,122 261,923,799 343,219,181 389,930,969 375,395,544 317,394,889 283,197,192 3 Insurance System Budget 193,517,885 217,496,513 243,740,339 319,735,022 354,724,792 335,395,601 226,436,322 244,716,745 4 Medicine 17,307,067 18,841,063 26,839,252 35,373,959 46,154,538 46,353,162 45,250,569 46,864,793 5 Medicare 150,822,909 161,370,624 158,029,818 187,461,878 277,238,846 278,858,534 196,326,800 176,892,064 6 Central Procurement of Medicine 7,045,705 9,154,137 6,514,946 2,815,425 3,454,142 3,089,346 3,524,553 4,537,271 7 Reserve Fund 3,031,187 2,475,624 2,338,694 4,074,167 1,060,146 187,816 31,503 8 Other Programs 15,311,017 25,655,065 50,017,630 90,009,592 26,817,120 6,906,743 5,204,775 16,422,618 9 Participants in the Insurance System 2,339,858 2,339,566 2,307,807 2,290,554 2,280,385 2,271,266 2,261,266 2,251,266 10 Medicare (Line 5) Broken into Components 11 Inpatient 99,251,282 87,866,218 119,234,084 152,013,841 157,705,192 106,646,773 69,355,115 12 Inpatient Care 99,251,282 87,866,218 119,234,084 146,575,632 152,828,166 96,291,154 62,194,508 13 Compensation of Copayments 5,438,209 4,877,027 10,355,619 7,160,606 14 Outpatient 51,553,228 48,737,226 57,615,065 90,854,396 90,055,272 71,267,580 79,369,971 15 Secondary Outpatient 22,524,312 26,230,735 30,046,682 52,018,504 58,036,255 42,430,297 49,305,888 16 Secondary Outpatient 22,524,312 26,230,735 30,046,682 52,018,504 56,243,052 39,722,004 45,792,400 17 Compensation of Copayments 1,793,203 2,708,293 3,513,487 18 General Practitioner Practices 25,321,833 19,258,756 24,101,958 34,472,568 27,853,489 25,632,101 26,335,331 19 General Practitioner Practices 25,321,833 19,258,756 21,483,526 31,437,858 26,909,326 24,184,129 24,628,639 20 Compensation of Copayments 2,618,432 3,034,711 944,163 1,447,972 1,706,692 21 Dentistry 3,707,084 3,247,736 3,466,425 4,363,324 4,165,528 3,205,182 3,728,753 22 Settlements with EU 837,309 852,698 1,237,622 1,770,481 1,695,730 23 Emergency Medical Teams 10,566,128 11,769,558 15,020,866 17,442,640 17,506,015 16,641,966 21,086,830 24 Medicare Total 161,370,638 148,373,003 192,707,325 261,163,575 266,504,101 196,326,800 171,507,646 25 CPI for Total Goods and Services 2.9 6.2 6.7 6.5 10.1 15.4 3.5 -3.7 26 Re-Index with 2003=100 100.00 106.20 113.32 120.68 132.87 153.33 158.70 152.83 274 27 Value of 1 Lat relative to 2003 1.00 0.94 0.88 0.83 0.75 0.65 0.63 0.65 Source: Annex Table 1 deflated according to Line 25 using the Consumer Price Index (CPI) from “PC02. Consumer Price Changes by Commodity Groups,â€? www.csb.gov.lv. The 2010 estimate is from the MoF. In Line 26 the CPI is used to construct an index with 2003=100. Line 27 uses the index to calculate what 1 Lat is worth in each year relative to 2003, and that number is multiplied by each entry in the corresponding column of Annex Table 1 to generate Annex Table 2. The underlying calculations are available on request. 275