69709 FISCAL POLICIES AND INSTITUTIONS FOR SHARED GROWTH IN KENYA LESSONS FROM THE GLOBAL CRISIS APRIL 2010 Note: Fiscal years, currency and equivalent units Fiscal year: 1 July-30 June Currency = Kenyan Shillings (Ksh) US$ 1.00 = Ksh 79.50 Abbrevia�ons and Acronyms EBCs Electronic Bene�t Cards Ksh Kenya Shillings OPM Office of the Prime Minister LIC Low Income Countries TJRC Truth, Jus�ce and Reconcilia�on Commission MDER Minimum Dietary Energy Requirements ASAL Arid and Semi- Arid Lands MDGs Millennium Development Goals BOP Budget Outlook Paper MNKOAL Ministry of Northern Kenya and BoT Build, Own, Operate and Transfer Other Arid Lands BPO Business Process Outsourcing MOA Ministry of Agriculture BSD Budget Supply Department MOE Ministry of Energy BSP Budget Strategy Paper MOED Ministry of Educa�on CBK Central Bank of Kenya MOF Ministry of Finance CCT Condi�onal Cash Transfer MOGC&SD Ministry of Gender and Children and CEM Country Economic Memorandum Social Development CPI Consumer Price Index MOR Ministry of Roads DFID Department for Interna�onal Development MOWI Ministry of Water and Irriga�on DOTS Direc�on of Trade Services MPER Ministerial Public Expenditure Review DPT Diphtheria, Pertussis, Tetanus MTEF Medium Term Expenditure Framework EBSC Economic and Budget Steering Commi�ee MTP Medium-Term Plan ECV European Community MVC Most Vulnerable Children EMOP Emergency Relief Opera�ons NAAIAP Na�onal Accelerated Agricultural Inputs ERD External Resources Department Access Program EWS Early Warning System NARC Na�onal Rainbow Coali�on FY Fiscal Year NCPB Na�onal Cereals and Produce Board GDP Gross Domes�c Product OVC Orphaned and Vulnerable Children GFS Government Finance Sta�s�cs PAYE Pay As You Earn GJLOS Governance, Jus�ce, Law and Order Sector PEFA Public Expenditure and Financial Assessment GoK Government of Kenya PIM Public Investment Management HIV/AIDS Human Immuno- De�ciency Virus/AIDS PIU Project Implementa�on Unit HP Hodrick-Presco� PPPs Public- Private Partnerships HSNP Hunger Safety Net Program PSNP Produc�ve Safety Net Program ICT Informa�on and Communica�on Technology SAS Seasonal Adjustments IDA Interna�onal Development Agency SMEs Small and Medium Enterprises IDA Interna�onal Development Agency SSA Sub-Saharan Africa IMF Interna�onal Monetary Fund SWGs Sector Working Groups JICA Japan Interna�onal Coopera�on Agency UNICEF United Na�ons Children’s Fund KFSG Kenya Food Security Group VAR Vector Autoregressive KIHBS Kenya Integrated Household Budget Survey VAT Value Added Tax KJAS Kenya Joint Assistance Strategy WB World Bank Km Kilometer WDI World Development Indicators KNBS Kenya Na�onal Bureau of Sta�s�cs WFP World Food Program KPIA Kenya Poverty and Inequality Assessment Table of Contents EXECUTIVE SUMMARY i A. Fiscal Policies and Ins�tu�ons for Shared Growth i B. Characterizing Kenya’s Public Investment program ii C. Automa�c Stabilizers, Counter-Cyclical Transfers to Protect the Vulnerable iv D. Conclusions vi 1. FISCAL POLICIES AND INSTITUTIONS FOR SHARED GROWTH 1 2. A GLOBALLY COMPETITIVE AND PROSPEROUS KENYA: THE ROLE OF PUBLIC INVESTMENT 12 3. PRO-POOR SPENDING IN KENYA: A REVIEW OF THE TARGETED SUBSIDY PROGRAMS 36 References 49 Appendix 50 Appendix Group A: Fiscal Policies and Ins�tu�ons for Shared Growth 51 Appendix Group B: A Globally Compe��ve and Prosperous Kenya-The Role of Public Investment 65 Appendix Group C: Pro-poor Spending in Kenya: A Review of Exis�ng Targeted Subsidy Programs 69 List of Boxes Box 1: Public investment: Does it crowd in or crowd out private investment 13 Box 2: Public investment management: A note on good prac�ce 24 Box 3: Vision 2030 and the New Coali�on Government 29 Box 4: Vision 2030, the medium term plan and spending needs 30 Box 5: Disbursement problems: The case of World Bank 33 List of Figures Figure 1: Economic performance, 2000-2008 i Figure 2: Correla�on between budget surplus and cyclical component of GDP ii Figure 3: Cyclicality of Kenya’s development budget iii Figure 4: The external sector is the main source of vola�lity in Kenya’s GDP iii Figure 5: Composi�on of development budget iv Figure 6: Kenya’s infrastructure lags behind other low income countries in Africa iv Figure 7: On average, only two thirds of the development budget is u�lized each year, iv but the ra�o has been increasing Figure 8: Targeted spending accounts for about 1 per cent of GDP, mainly donor funded v Figure 9: Provincial distribu�on of service gains, 2003-2005/06 15 Figure 10: Reduced distances to various infrastructure, 1997-2007 15 Figure 11: Rise in development spending, 1999/00-2006/07 18 Figure 12: Acquisi�on of non-�nancial assets, Kenya and comparators, 1990-2007 19 Figure 13: Share of development spending by sector, per cent of total 2004/05-2008/09 20 Figure 14: Ministerial Spending 2000 Prices: Spending shi� to physical infrastructure 21 Figure 15: Total donor commitments 2008/09 22 Figure 16: Government departments with donor-�nanced projects, 2008/09 budget 23 Figure 17: Budget devia�on by ministry 28 Figure 18: Top 5 under-spenders 28 Figure 19: Sectoral budgets 2008/09 and 2009/10 31 Figure 20: Regional poverty levels (KIHBS 2005/06) 37 Figure 21: Prevalence of cri�cal food poverty (KIHBS 2006/06) 38 Figure 22: Targeted programs popula�on and shares of total expenditures 2008/09 (percent) 44 Figure 23: Spending on different categories of vulnerable groups 46 Figure 24: Regional poverty (numbers ‘000) and expenditure per capita 47 Figure 25: Poverty incidence and total targeted spending by province 47 List of Tables Table 1: Factors explaining decline in indebtedness, 1996/07-2006/07 (percent of GDP, annual average) 2 Table 2: Central government �scal outcomes as percent of GDP, 1999/00-2006/07 3 Table 3: Correla�ons among various measures of the cyclical component of GDP, 1996-2008 Q3 4 Table 4: Summary budget �gures, 1996-2008 5 Table 5: Components of public sector revenue and expenditure, 1996-2008 5 Table 6: Cyclical proper�es of public sector revenue and expenditure, 1996 Q1-2008 Q3 6 Table 7: Es�mates of revenue elas�ci�es 7 Table 8: Correla�on of the budget surplus measures with the cyclical component of GDP 8 Table 9: Variance decomposi�on of output (percent of variance of the forecast error) 11 Table 10: Public investment outcome indicators, Kenya in comparison, 2008 13 Table 11: Indicators of human capital, Kenya and SSA and LIC averages 15 Table 12: Economic composi�on of development budget, 2004/05-2008/09 as percent of total 17 Table 13: Fiscal space for increased development spending 2002/03-2008/09 18 Table 14: Kenya and comparators, development spending as percent of GDP 20 Table 15: Share of development budget by sector, 2004/05 to 2008/09 (Ksh millions) 20 Table 16: Sectoral composi�on of development expenditures (Ksh millions) 22 Table 17: Donor-�nanced development expenditures, 2002/03-2008/09, Ksh billions 22 Table 18: Development budget execu�on rates, 1999/00-2006/07 26 Table 19: Budgeted and disbursement varia�on, 2006/07 US$ millions 29 Table 20: Targeted schemes in Kenya: Popula�on coverage (numbers’000) 29 Table 21: Selected targeted programs in Kenya 40 Table 22: Programs with poten�al for scale up 42 Table 23: Expenditure levels by program, 2008/09 (Ksh million) 43 Table 24: Fiscal opera�ons (Ksh billions) 44 Table 25: Regional spending social safety nets 45 Table 26: Average government transfer in Kenya by province and poverty status 47 Acknowledgements This report was prepared by a core team led by Jane Kiringai (AFTP2), together with Cevdet Denizer (CF- PIR) and Tracey Lane (AFTP2). Important substan�ve comments were received from Jeni Klugman (AFTP2), Sanjay Dhar (AFTPM), Harold Alderman (PA9SS), Luis Serven (DECRG) and Sudharshan Canagarajah (EC- SPN). Special thanks go to the consultants Fredrick Wamalwa, Owen Nyangoro, Lydiah Ndirangu, Njeru Kirira and Yuliya Meshcheryakova for excellent research and data analysis. A larger group within the World Bank contributed valuable inputs to this report, for which the core team expresses their apprecia�on: Wolfgang Fengler (AFTP2), Louise Fox (AFTP1) and Mike Mills (AFTHE) who offered important insights and comments for the ini�al dra�s, and Carolyn Wangusi and Anne Kha�mba (AFTP2) who provided editorial and administra�ve assistance. The team also expresses their apprecia�on to the Targeted Food Subsidy Scheme Task Force for the invalu- able data used for the analysis. We are also grateful for the contribu�ons provided by the Office of the Prime Minister, Ministry of Gender and Children Affairs, Ministry of Educa�on, Ministry of Agriculture, and World Food Program. Overall guidance was provided by Kathie Krumm (Sector Manager AFTP2). EXECUTIVE SUMMARY This report brings together three budget notes that A. Fiscal Policies and Ins�tu�ons for Shared assess Kenya’s �scal capacity to respond to global Growth crisis and deliver shared growth without compro- mising macroeconomic stability. The key messages Introduc�on from this analysis point to areas where Kenya’s �s- Kenya’s strong economic performance between cal policy requires strengthening. This Execu�ve 2002 and 2007 has been partly a�ributed to mac- Summary summarizes messages from three budget roeconomic stability and strong �scal consolida- notes. The �rst note, Fiscal Policies and Ins�tu�ons �on. A�er two decades of sluggish performance, for Shared Growth, is an assessment of Kenya’s �scal economic growth resumed in 2002 and steadily in- stance and suggests an appropriate �scal anchor for creased from 0.5 per cent to 7 per cent in 2007 (see Kenya. The second note, A Globally Compe��ve and Figure 1 below). During this period, the govern- Prosperous Kenya: The Role of Public Investment, ment re�red debt and started crea�ng �scal space reviews the status of public investment in Kenya to fund essen�al infrastructure. The ra�o of debt and suggests the reforms required to improve pub- to Gross Domes�c Product (GDP) declined from 60 lic investment planning and implementa�on. The per cent in 2000 to 40 per cent in 2008. Fiscal space subject of the third note is Pro-poor Spending in was achieved through a strong revenue effort and Kenya: A Review of the Targeted Subsidy Programs. stringent �scal management. The budget de�cit av- This study suggests that the targeted cash transfers, eraged about 2 per cent during this period. As a re- unemployment bene�ts, and workfare programs sult of the strong growth performance, income per provide automa�c stabilizers for �scal policy during capita increased and poverty declined from 56 per crisis. This Execu�ve Summary ends with some con- cent in 2000 to 47 per cent in 2005/06. cluding recommenda�ons on the opportuni�es for strengthening �scal policy. Figure 1: Economic performance, 2000-2008 Public Debt% of GDP 8 7 6 GDP gr owth r ate 5 4 3 2 1 0 Source: Budget Speech 2009 i Execu�ve Summary This remarkable growth was disrupted by three Characterizing Kenya’s Fiscal Policy shocks, star�ng with the poli�cal crisis experienced Surprisingly, unlike most developing countries, in the �rst quarter of 2008, the global �nancial cri- Kenya’s �scal policy is acyclical and does not ‘lean sis, and the prolonged drought running through to with the wind’. This �nding is signi�cant because the third quarter of 2009. The combined effect of most developing countries adopt a pro-cyclical �s- these shocks is manifested through sluggish eco- cal policy, contribu�ng to macroeconomic vola�l- nomic performance, with GDP growth at 1.7 per ity. Figure 2 below shows the correla�on between cent in 2008, and about 2.5-3.0 per cent in 2009. In- budget surplus and the cyclical components of GDP come per capita contracted by 1.1 per cent in 2008, between two periods, 1996-2008 and 2003-2008. and it is es�mated that close to 6 million Kenyans The primary balance is posi�vely correlated with required food assistance as a result of the prolonged GDP (0.08), thus it is higher during booms and low- drought and high food prices. Although the govern- er during recessions, sugges�ng counter-cyclicity. ment had created �scal space to �nance essen�al However, for the period 2003-2008, the coefficient infrastructure and social spending, the addi�onal dropped to 0.02, sugges�ng that �scal policy had �nancing to counter the effects of the drought and become acyclical. global crisis will impart signi�cant pressure on the available resources. Since the current government took office in 2003, �scal policy has been ‘�me consistent’. There have The global �nancial crisis and the nega�ve domes- been no major differences between policy an- �c shocks present challenges and opportuni�es nouncement percentents and outcomes, and the as well. For Kenya, the introduc�on of a targeted overall macroeconomic framework has remained cash transfer program and a youth workfare pro- broadly consistent with targets. This posi�ve per- gram are good examples of the opportuni�es. If formance can be a�ributed to the strong �scal ad- well designed, these social programs will provide justment between 2002 and 2007, with remarkable in-built automa�c stabilizers for Kenya’s �scal sys- reduc�on in debt as a share of GDP and a strong rev- tem, which could be scaled up during crisis. In ad- enue effort. Public borrowing was limited to 1.8 per di�on to the social programs, a �scal s�mulus pack- cent of GDP. These efforts have paid off, and Kenya age equivalent to 1 per cent of GDP is also being can now issue debt at single digit interest rates. implemented in response to the crisis. Figure 2: Correla�on between budget surplus and cyclical component of GDP - - Overall Primary Adjusted Adjusted - Balance Balance Overall Primary Balance Balance Source: World Bank staff es�mates ii Execu�ve Summary Public investment, other recurrent spending, and Figure 4: The external sector is the main source corporate taxes exhibit counter-cyclical proper- of vola�lity in Kenya’s GDP �es, but development spending is highly vola�le. Figure 3 below shows the correla�on between �scal variables and the cyclical component of GDP for two periods, 1996-2002 and 2003-2008. The low corre- la�on coefficients for most variables suggest that �scal variables are neutral to swings in the business cycle. During the �rst period, 1996-2002, pension payments exhibited counter-cyclical proper�es but, since 2003, when the Na�onal Rainbow Coali�on (NARC) government took office, a new set of vari- ables, development and other recurrent spending Source: World Bank staff es�mates have become more counter-cyclical. Development spending remains extremely vola�le at 50 per cent In summary, the analysis suggests that Kenya’s �s- variability compared to 8 per cent for other recur- cal policy is not the cause of vola�lity in GDP and, rent spending. therefore, does not lower growth. As men�oned earlier, vola�lity in Kenya’s output emanates from Figure 3: Cyclicality of Kenya’s development budget monetary policy, and notably from the exchange - rate. Since the NARC government took office, devel- opment spending has become appropriately more counter-cyclical, but remains vola�le. This is cer- tainly a posi�ve performance rela�ve to developing countries whose �scal policy responses have been pro-cyclical. However, there is scope for further re- forms that would make �scal policy counter-cyclical to complement the posi�ve macroeconomic devel- opments. - - B. Characterizing Kenya’s Public Investment program Source: World Bank staff es�mates Key: Dev-Development, Rec-recurrent, Inc. Tax Corp-Corporate Income Although total government expenditure accounts Tax, Ims-Imports, T-Total, VAT_L-Vat Local, Ex-Excise, Int-Interest, for about 27 per cent of GDP, development spend- ing has averaged 4 per cent of GDP, and only 2 per It is Kenya’s monetary policy, and not its �scal poli- cent of GDP is used for the core assets. On a posi- cy, that has a signi�cant impact on GDP’s vola�lity. �ve note, development spending has been rising in The main sources of vola�lity are in oil price shocks, recent years; in 2008/09 it increased to 6 per cent of and other external shocks transmi�ed through the GDP. However, a closer analysis of this expenditure exchange rate. Figure 4 below shows that �scal GDP (see Figure 5) shows that, on average, 48 per cent accounts for 13 per cent of vola�lity in output, while of development spending is used for the acquisi�on the federal funds rate accounts for 70 per cent of of assets, with the rest used for recurrent type ex- vola�lity. In this regard, the federal funds rate is penditures. This implies that acquisi�on of non-�- used as a proxy of the global market. Overall, �scal nancial public assets has been about 2 per cent of shocks have limited impact on GDP vola�lity; a 10 GDP. However, it is notable that a signi�cant share per cent change in cyclically adjusted primary bal- of development spending is off-budget, especially ance leads to a 1 per cent change in GDP. through state-owned enterprises.� � The expenditure analysis here is limited to Central government spending and excludes expenditure by state-owned enterprises, which will be covered in the more detailed 2010 Public Expenditure Review. iii Execu�ve Summary Figure 5: Composi�on of development budget enue from 21 per cent of GDP in 2003/04 to 24.4 per cent of GDP in 2008/09, and restraint on the growth of recurrent expenditures (maintained at an average of 20percent of GDP over the same peri- od). However, the increase in infrastructure spend- ing also reflects a trade-off largely at the expense of the budget shares to the health, educa�on and governance sectors. The execu�on rate of the development budget has been improving from around 40 per cent in 2001/ Source: Ministry of Finance, Notes constant 2004 prices 02 to over 70 per cent by 2006/07, but it s�ll lags behind total budget execu�on. On average, one The challenges in Kenya’s public investment pro- third of development budget is not spent each year gram are reflected in the state of infrastructure. (see Figure 7 below). Donor �nancing contributes Kenya’s infrastructure development lags behind to poor execu�on of development budget through other low income countries in sub-Sahara Africa delays in disbursements, but the problem can also (SSA) (see Figure 6 below). The most signi�cant dif- be traced to over-budge�ng and weaknesses in re- ference is in the paved roads density, where Kenya por�ng. As the government mobilizes domes�c re- has a density of 16km compared to a SSA average sources, the share of donor-�nanced projects has of 31km, and an average of 134km for other Low In- been declining and, between 2002 and 2008, the come Countries (LIC). Electricity coverage also lags, share of donor-�nancing declined from 73 to 41 per with a popula�on coverage of 18 per cent in Kenya cent. compared to 72 per cent for middle income coun- Figure 7: On average, only two thirds of the tries. Even the country’s 2011 target coverage of 33 development budget is u�lized each year, but the per cent will be below the current average for LICs. ra�o has been increasing Poor infrastructure, notably in transport and elec- tricity, constrains economic growth and has been iden��ed as a major performance challenge under the Doing Business Indicators. Figure 6: Kenya’s infrastructure lags behind other low income countries in Africa Source: Ministry of Finance (as reported in the Concept Note), 0 2007/08 World Bank staff es�mates Weak budget execu�on increases the lead �me to project comple�on, leading to escala�ng costs. In addi�on to low execu�on rates, weaknesses persist in the project planning process. For instance, there Source: Yepes et al. (2008), WDI (2008 ) are no procedures or evalua�on criteria for apprais- al and screening to ensure that high-return projects The good news is that the addi�onal �scal space are priori�zed. As a result, some of the projects are created through debt re�rement has been u�lized poorly designed, which delays implementa�on. Fur- to increase infrastructure spending. This increase thermore, future recurrent cost implica�ons of such in development spending has been brought about projects are not taken into account, leading to inad- by the �scal space created from an increase in rev- equate provision for opera�ons and maintenance. iv Execu�ve Summary Improvement in development budget execu�on of poverty, such as hunger and malnutri�on, which is key to improving the ability of �scal policy to are especially damaging to children. The global es- respond to exogenous shocks. By crea�ng �scal �mates indicate that spending on social safety nets space for infrastructure �nancing, Kenya has made varies substan�ally by country, but averages about remarkable progress in reversing the trend of low 1.9 per cent of GDP for both cyclical and counter- levels of public investment. Kenya is implemen�ng cyclical transfers. The current review shows that a �scal s�mulus package equivalent to 1 per cent Kenya is currently spending about 1.0 per cent of of GDP in response to the global �nancial crisis. GDP, including external funding (see Figure 8 be- The s�mulus package has a strong focus on rural low), but it is mainly ad hoc and not counter-cyclical infrastructure and social spending. However, weak by design. implementa�on capacity and low absorp�on of de- Figure 8: Targeted spending accounts for about velopment funds could neutralize the poten�al im- 1 per cent of GDP, mainly donor funded pact of the counter-cyclical s�mulus. Accelera�ng investment budget execu�on and implementa�on, and in par�cular on projects with aid �nancing com- mitments, can complement the s�mulus package. This approach would also have bene�cial impact on the balance of payments. For instance, increasing development budget execu�on from 50 to 80 per cent in �scal year (FY) 2009/10 would increase total spending by an addi�onal US$ 400 million, without addi�onal debt commitments. Source: World Bank staff es�mates C. Automa�c Stabilizers, Counter-Cyclical Kenya has a signi�cant number of programs fund- Transfers to Protect the Vulnerable ed and managed by different agencies, and mainly donor supported, but with limited coordina�on. Kenya has a high number of poor and vulnerable Consequently, there is no program that the govern- individuals and households, but social safety nets ment could easily and quickly scale up to cushion to cushion the poor are limited and fragmented. the vulnerable against crisis. Kenya has more than Es�mates based on the 2005/06 integrated house- 14 targeted subsidy programs, but the geographic hold budget survey indicated that about 47 per cent coverage is limited and skewed in favor of drought- of Kenyans live below the poverty line. Further, es�- prone rural geographic areas, with limited a�en�on mates indicate that about 15 per cent of the urban to other vulnerable groups, par�cularly the urban popula�on and 22 per cent of the rural popula�on poor. live below the cri�cal food poverty line. These es�- mates were based on a good year when the Kenyan The Government introduced a generalized maize economy was growing at about 6 per cent, and in- subsidy scheme in November 2008. This approach comes per capita expanding at 3.0 per cent. How- was expensive and had to be suspended. The Gov- ever, a�er the three shocks hit the country, incomes ernment is now in the process of designing a new per capita contracted by 1.1 per cent, meaning that cash transfer program, and made a provision of Ksh the number of poor and vulnerable Kenyans will in- 7.6 billion (US$ 100 million) in the �scal year 2009/ crease un�l economic growth exceeds popula�on 10 budget for social safety nets and a youth work- growth on a sustained basis. fare program. Global experience shows that the design of a comprehensive and effec�ve targe�ng The impact of the crisis poses signi�cant challeng- system takes at least nine months, some�mes even es towards the achievement of the Millennium Development Goals (MDGs) and, if not addressed, longer, to put in place. Furthermore, even in coun- can reverse the gains made so far. Global experi- tries that have counter-cyclical safety nets in place, ences show that well designed, targeted counter- scaling up in �mes of crisis can pose signi�cant chal- cyclical safety nets prevent long term consequences lenges, especially when �scal space is limited. v Execu�ve Summary With the high number of poor people in Kenya, it recovers, adherence to a clear exit strategy for the is unlikely that a subsidy scheme can comprehen- �scal s�mulus program would ensure that Kenya sively cover all the poor and vulnerable within the quickly reverts to a credible and sustainable �scal limited �scal space, and without compromising the stance, and the economy is once again on a sustain- hard earned �scal space. In addi�on, the coverage able growth path. and design of the scheme may require pu�ng into considera�on spa�al differences in inequality and The development budget’s planning and its im- poverty, which has become a source of tension and plementa�on is the Achilles’ heel in Kenya’s �scal fragility. Overall, the crisis has presented opportuni- management. In the short term, increasing the rate �es and challenges and, in this regard, has opened of budget execu�on remains the priority. However, up policy dialogue, providing a good opportunity for the medium to long term, a government system for a more coordinated approach between govern- for project evalua�on and selec�on criteria will be ment and donors in the design of social safety nets. necessary to help priori�ze and sufficiently fund Preparing the groundwork for such programs now those projects with the highest rates of return and, could help ensure that they can be scaled up later thus, ensure that they also play a cataly�c role in to respond to the next crisis or nega�ve shocks. Kenya’s growth. D. Conclusions The global crisis has revealed that in addi�on to ad- dressing the needs of the vulnerable, social safety Kenya has made good progress in macroeconomic nets can be scaled up and will enhance the coun- management, which is benchmarked on a debt to ter-cyclical role of the budget. However, to ensure GDP ra�o as the �scal anchor. This approach has �scal sustainability, a degree of modera�on is war- paid off; Kenya is able to �nance a counter-cyclical ranted for the schemes and, in par�cular, stronger �scal s�mulus from the domes�c market at single alignments with complementary long term devel- digit interest rate without compromising macroeco- opment strategies for poverty reduc�on is crucial. nomic stability. Going forward and as the economy vi 1. FISCAL POLICIES AND INSTITUTIONS FOR SHARED GROWTH Introduc�on The government is well aware that the current glo- The post-elec�on violence in early 2008 has shown bal �nancial crisis puts a premium on sound �scal that when poverty and inequality provide a plat- policy management and, in this regard, there are sig- form for poli�cal contest, the risks emana�ng from ni�cant cross-cu�ng �scal issues from the budget, ethnic and social tensions are indeed real. As a re- the public investment plans, and the poverty points sult of the violence and security-related issues, all of view. Sound �scal policy has two major compli- sectors suffered declines in value added, and GDP ca�ons for government. First, is that although the growth was li�le over 2 per cent in 2008. The cri- government has adopted debt to GDP ra�o as a �s- sis added to government expenditure, and early cal anchor, the indicator does not factor in the cycli- es�mates indicate that the budget de�cit for the cality of GDP. This suggests that a cyclically adjusted year 2008/09 was around 5 per cent of GDP, almost de�cit would be a useful tool for government and, double the amount of previous year. The turmoil in fact, would help Kenya manage its na�onal debt in interna�onal markets, and the rising energy and in a more appropriate way. The second complica- commodity prices brought new challenges to policy �on is the potency of �scal expansion to boost eco- management. Infla�on, mostly due to external fac- nomic growth, and the ra�onale of a �scal s�mulus. tors, has increased and the exchange rate has come While this issue has been under discussion in Kenya under some pressure, though recent data suggests since early 2009, there has been no formal analysis that there is a degree of stability in foreign exchange of the ma�er. One key objec�ve of this note is to markets in Kenya now. While the impact of the on- provide such an analysis to be�er inform the ongo- going interna�onal credit crunch may not have hit ing debate. Kenya in a pronounced and direct way, indirect im- pacts have manifested themselves in the form of re- Given the government’s interest in issuing a Eu- duced capital flows, slowdown in exports, migrant robond (though it did not take place in 2009) to �- remi�ances, reduced access to interna�onal credit nance large infrastructure needs, the Interna�onal markets for Kenyan �rms and government, and low- Monetary Fund (IMF) has raised the issue of estab- er tourism revenue. Collec�vely, these show that lishing a ceiling on both domes�c and external bor- the favorable external environment Kenya enjoyed rowing as the new �scal anchor. In agreement, the un�l late 2007 has been affected, and is no longer a government has since adopted a total debt to GDP posi�ve factor suppor�ng economic growth. as the �scal anchor, with a medium term target of 40 percent of GDP. This budget note is the �rst in a 1 Fiscal Policies And Ins�tu�ons For Shared Growth series of three notes and it analyzes Kenya’s �scal invest in produc�ve ac�vi�es. The end result has stance, and suggests an appropriate �scal anchor in been the steady rate of growth of Kenya’s economy the context of a more challenging post-crisis envi- during the period 2002-2007 which, however, could ronment. The note starts with an overview of the not be sustained in 2008 due to the combined ef- �scal situa�on in Kenya and then uses two econo- fects of violence that followed the December 2007 metric approaches to gauge Kenya’s �scal stance. In elec�ons, and the ongoing global �nancial crisis. conclusion, the note observes that unlike most de- veloping countries, Kenya’s �scal policy is acyclical A closer look at factors that explain the decline in Kenya’s debt to GDP ra�o is provided in Table 1 and does not lean with the wind, which is a posi�ve above. By and large, all the �ve factors worked to- outcome as a result of �scal retrenchment during gether in most of the three periods, and total debt the last decade. stock rela�ve to GDP declined. The most signi�cant Key Fiscal Policy and Growth Issues factors have been real GDP growth and real ap- precia�on of the exchange rate. What then can be Overview of Fiscal Situa�on in Kenya expected within the next few years? Star�ng from Kenya’s debt to GDP ra�o fell by over 30 points dur- the current year, 2009, and making some assump- ing the period 1995/96 to 2006/07. It is es�mated �ons, it is possible to deduce some forward looking that the current level of total debt stock is around es�mates. As noted already, economic growth has 40 per cent of Kenya’s GDP, with the share of for- slowed down and, given the current situa�on in in- eign exchange denominated debt being at 60 per terna�onal markets, it is reasonable to assume that economic growth rates will not be as high as in the cent. This is a remarkable �scal adjustment, which 2002-2007 period. One of the perceived risks at the had the effect of reducing country risk, which in turn beginning of the global �nancial crisis was the po- led to declines in real interest rates. This situa�on, ten�al deprecia�on of the real exchange rate. How- combined with sustained revenue effort, enabled ever, Kenya has weathered the global �nancial crisis Kenya to lay the founda�on for the solvency of the rather well and, as a result of adop�ng a new meth- public sector. As a result, the country has been able odology for measuring infla�on, real interest rates to issue debt at single digit interest rates, a situa�on have been posi�ve since 2009 and the exchange not enjoyed by larger emerging market economies. rate has stabilized against the dollar. This policy credibility, which paid par�cular a�en- �on to the level of domes�c debt−a net domes�c Table 1 shows the evolu�on of overall �scal out- �nancing posi�on−has been the main �scal indica- comes. Primary balance was in surplus during most tor for the government. It has allowed the private of the past seven years, though its size has been de- sector to expand its horizons, and encouraged it to clining. Table 1: Factors explaining decline in indebtedness, 1996/07-2006/07 (percent of GDP, annual average) 1996/07– 2003/04– 1996/07– 2002/03 2006/07 2006/07 Change in public sector debt -1.8 -4.9 -2.7 Contribu�on from: 1. Primary de�cit (- surplus) -2.5 -0.1 -2.2 2. Real GDP growth -1.4 -2.6 -1.8 3. Real interest rate 2.1 0.5 1.6 4. Real exchange rate (- apprecia�on) 0.6 -2.4 0.0 5. Other factors -0.6 0.8 -0.2 Source: CEM: Accelera�ng and Sustaining Inclusive Growth, July 2008 2 Fiscal Policies And Ins�tu�ons For Shared Growth Table 2: Central government �scal outcomes as percent of GDP, 1999/00-2006/07 1999/ 2000/ 2001/ 2002/ 2003/ 2004/ 2005/ 2006/ 00 01 02 03 04 05 06 07 Revenues 19.9 21.9 19.9 20.8 22.3 22.3 21.5 21.4 Expenditures 9.5 23.5 22.3 24.5 22.7 22.2 24.8 23.1 Recurrent 16.8 20.1 19.9 20.5 19.8 18.9 20.4 18.7 Development 2.6 3.4 2.4 4.0 2.9 3.3 4.4 4.4 Overall balance 0.1 -1.0 -2.6 -3.0 -0.3 0.1 -2.4 -1.6 Primary balance 3.3 2.2 0.3 0.3 2.2 2.4 0.3 0.7 Financing 0.0 0.5 2.6 3.0 0.0 -0.5 2.3 2.0 Domes�c 1.3 0.1 3.9 4.3 0.7 -0.5 1.8 3.0 External -1.5 0.8 -1.1 -1.0 -0.7 0.0 0.0 -0.2 Stock of domes�c debt, net 17.6 16.6 19.7 23.0 21.3 18.4 18.1 18.4 Source: Interna�onal Monetary Fund (IMF) As shown above, expenditures have been increasing second, the effect Kenya’s �scal policy has on the in line with GDP growth. In fact, expenditures have country’s GDP. The study uses quarterly observa- been growing at a higher rate than GDP growth, �ons from 1996 Q1 to 2008 Q3, the real GDP, and and the share reached almost 25 per cent in 2005- Consumer Price Index (CPI) series compiled by the 2006 from 19.5 per cent in 1999-2000. Kenya Na�onal Bureau of Sta�s�cs (KNBS) and the Central Bank of Kenya (CBK), and the �scal accounts Fiscal Policy and Macroeconomic Stability data from Quarterly Economic and Budgetary Re- Interna�onal experience and empirical research views compiled by Kenya’s Ministry of Finance. Real show that �scal policy tends to be expansionary exchange rate comes from the CBK. We use the CPI during booms and contrac�onary during reces- series rather than the GDP deflator to convert �scal sions, especially in developing countries. In other accounts into constant prices. This decision is driv- words, it tends to lean with the wind. Therefore, en by the fact that the KNBS started compu�ng the in developing countries, �scal policy is likely to ex- GDP deflator only in 2000 and, therefore, the data acerbate business cycle fluctua�ons, rather than is only available at annual frequency before 2007. smooth them. A pro-cyclical �scal policy, gener- ally speaking, is undesirable and can be eliminat- The methodology and structure of the paper is as ed in two ways: �rst, by crea�ng strong automa�c follows. First, cyclicality of Kenya’s GDP and �s- stabilizers (for example given a �xed tax rate, tax cal accounts are analyzed. We then compute and revenues usually fall in recessions and increase in analyze Kenya’s cyclically adjusted budget surplus. booms, while unemployment bene�ts rise in reces- Further, we characterize Kenya’s �scal policy and sions), and second, by adjus�ng the discre�onary observe that, in fact, it does not appear to be lean- components of the budget. The cyclicality of �scal ing with the wind, as one might have expected it policy and budget balance, in par�cular, is only an to. In the last sec�on, we construct and es�mate a issue if output responds strongly to changes in �s- small vector autoregressive (VAR) model of Kenya’s cal policy. Ilzetzki and Vegh (2008) show that it is economy, which allows us to evaluate the impact of indeed o�en the case for developing economies. Kenya’s �scal policy on GDP. The analysis suggests that a 1 per cent increase in Kenya’s cyclically ad- Focusing on the cyclicality of Kenya’s �scal policy, justed primary balance causes GDP to decline by this note considers two issues: �rst, whether Ken- less than 0.1 per cent. In other words, while �scal ya’s �scal policy indeed leans with the wind and, �ghtening causes GDP to contract, this effect is fair- 3 Fiscal Policies And Ins�tu�ons For Shared Growth ly weak, and factors other than �scal policy might linear �lter; (b) the Hodrick-Presco� (HP) �lter; (c) be more important determinants of GDP fluctua- the Beveridge-Nelson decomposi�on; and (d) the �ons. This �nding also implies that predic�ng the Peak-to-Peak trend. Annex 1: Figure A3 illustrates potency of the impact of a �scal s�mulus package the trends es�mated using these four techniques, will be difficult, although the component of �scal and Annex 1: Figure A4 presents devia�ons of Ken- policy studied in this paper suggests a small impact. ya’s real GDP from the trends. All the �gures men- Results suggest that the US Federal Funds rate and �oned below are to be found in Appendix Group A. Kenya’s monetary policy might be among the factors Annex 1: Figure A1, in this Appendix, presents the determining GDP fluctua�ons. However, this result real GDP series for the period of the study. must be interpreted with cau�on; more research is needed in the case of Kenya before a de�ni�ve con- As can be seen in Annex 1: Figure A4, the cyclical clusion can be arrived at. VARs of this nature deal components in three out of four cases are very simi- with �scal innova�ons (purely unpredictable �scal lar. The cyclical component resul�ng from the Bev- policy) and ignore two other important �scal policy eridge-Nelson decomposi�on is smaller in size and components: (i) systemic discre�onary policy; rou- exhibits somewhat different behavior. However, as �ne responses to changing economic condi�ons; Table 3 below illustrates, even this component is and (ii) automa�c policy dictated by laws and rules posi�vely correlated with the other three. Due to such as the tax code. these similari�es, only the HP method is used for further analysis. Business Cycle and Fiscal Aggregates GDP and Business Cycle in Kenya Trends and Cycles in Kenya’s Fiscal Accounts The analysis shows that there is a strong seasonality The analysis starts with the conversion of revenue and expenditure series into real terms, adjus�ng for pa�ern, with rela�vely high produc�on in the third seasonality in case a strong seasonality pa�ern is and fourth quarters, and rela�vely low produc�on in present, and de-trending using the HP, just as was the �rst and second quarters. To iden�fy this feature done for real GDP. Because the GDP deflator is not of the series and to make a seasonal adjustment, available at a quarterly frequency for the selected the SAS version of the US Census Bureau X-12-ARI- period, we use the CPI instead to convert the series MA procedure was used. The seasonally adjusted into constant 2001 prices. In addi�on to performing series is presented in Annex 1: Figure A2. Similarly this procedure for the total revenue and expendi- to the original series, it exhibits an upward trend. ture, we study each revenue and expenditure cat- To measure a business cycle, we use four different egory separately. Table 4 presents a summary of selected aggregate variables, while Table 5 provides methods of de-trending real GDP: (a) the Piecewise more detailed informa�on. Table 3: Correla�ons among various measures of the cyclical component of GDP, 1996-2008 Q3 Piecewise Linear HP Peak-to-peak BN Trend Trend Trend Decomposi�on Piecewise Linear Trend 1 HP trend 0.813 1 Peak-to-Peak trend 0.724 0.871 1 BN Decomposi�on 0.540 0.501 0.403 1 Source: World Bank staff es�mates 4 Fiscal Policies And Ins�tu�ons For Shared Growth Table 4: Summary budget �gures, 1996-2008 As a percentage of Budget category GDP Full sample Revenue Full sample Overall Surplus -2.1 -11.4 Primary Surplus 1.1 4.6 Revenue 19.5 100.0 Expenditure 21.5 111.4 Interest 3.2 15.9 Primary Expenditure 18.3 95.4 Source: World Bank staff es�mates Table 5: Components of public sector revenue and expenditure, 1996-2008 Budget category GDP Full sample Revenue Full sample Revenue 19.5 100.0 Import duty 2.2 11.0 Excise duty 3.0 15.6 PAYE 3.0 15.6 Other income tax 3.2 16.3 VAT local 2.4 12.7 VAT imports 2.2 11.2 Other revenue 3.5 17.6 Expenditure 21.5 111.4 Total recurrent expenditure 18.0 93.0 Domes�c and foreign interest 3.2 15.9 Pensions, etc 0.8 4.2 Wages and salaries 6.9 35.4 Other recurrent expenditure 7.1 37.3 Total development expenditure 3.5 18.4 Source: World Bank staff es�mates 5 Fiscal Policies And Ins�tu�ons For Shared Growth Revenue more so than GDP, as can be seen from Table 6. The Unlike the real GDP trend, the trend in total rev- other revenue category exhibits the highest vola�l- enue does not slope upwards, remaining instead ity (23.8 percent versus 1.5 percent for GDP), prob- rela�vely flat for the en�re study period (see Annex ably because its composi�on changes over �me. 1: Figure A5a). Such a pa�ern can be be�er under- stood once we examine the behavior of its compo- Expenditure nents. Other revenue and excise duty are the only Annex 1: Figure A7 presents a set of diagrams of the two categories that have rela�vely flat trend lines, expenditure components and their trends. We con- similar to that of total revenue. Pay As You Earn sider the total interest due, rather than the foreign (PAYE) and Value Added Tax (VAT) revenue, both on interest due, in sub-plot (d) since the foreign inter- imports and domes�c products, grow in signi�cance est due has a nega�ve entry and, therefore, we can- steadily through �me, replacing the revenue from not study its logarithm and compute the HP trend. import duty and other income tax revenue, which Both the domes�c and foreign component of the exhibit downward trends, such that the total does interest due fall over �me in real terms. Pensions not change substan�ally on average. and other recurrent expenditures are the only two categories that follow an increasing trend, while the The cyclical components of the revenue categories trend for wages and salaries is hump-shaped, with are plo�ed against the cyclical component of real a slow down star�ng in 2004 (see Annex 1: Figure GDP in Annex 1: Figure A6. A7 (f)). The trend for the total development ex- penditure, on the other hand, has a U-shape, with a Table 6 below provides the standard devia�ons for trough approximately in 2001 (see Annex 1: Figure the cyclical components of revenue and expendi- A7 (h)). This structural break can also be a�ributed ture, and their correla�on with the cyclical com- to the policy shi� associated with the change in ponent of GDP. The sample is further split into two government. The total expenditure exhibits a slight sub-samples: 1996-2002 and 2003-2008, where the posi�ve trend, with a slight accelera�on around second period corresponds to the period when the 2001-2002. The expenditure components have low- NARC government was in office. All the revenue er correla�on with the cyclical component of GDP components are posi�vely correlated with the cycli- than the revenue components. In the case of pen- cal component of GDP. For two components–excise sion-related expenditures, the correla�on coeffi- duty and PAYE–this rela�onship became stronger cient was actually -0.26 for the years of 1996-2002; over �me. The series is extremely vola�le, much that is, the pension payments during this period Table 6: Cyclical proper�es of public sector revenue and expenditure, 1996 Q1-2008 Q3 Correla�on with GDP Standard devia�on *Share of 1996-2008 1996-2002 2003-2008 Percent GDP (percent) Real GDP 1.00 1.00 1.00 1.5 1.5 Total expenditure 0.15 0.19 0.10 8.9 1.9 Total recurrent expenditure 0.15 0.17 0.13 8.2 1.5 Domes�c interest 0.19 0.11 0.39 24.4 0.6 Total interest due 0.18 0.12 0.38 23.1 0.7 Pensions etc. -0.09 -0.26 0.06 18.6 0.1 Wages and salaries 0.16 0.03 0.26 8.4 0.6 Other recurrent expenditure 0.02 0.08 -0.08 17.1 1.2 Total development expenditure 0.02 0.14 -0.12 50.2 1.8 Total revenue 0.25 0.30 0.23 8.1 1.6 Import duty 0.23 0.26 0.22 13.7 0.3 Excise duty 0.28 0.30 0.32 9.4 0.3 PAYE 0.17 0.10 0.32 7.2 0.2 Other income tax 0.03 0.10 -0.08 15.1 0.5 VAT local 0.31 0.34 0.29 10.2 0.2 VAT imports 0.10 0.09 0.09 8.9 0.2 Other revenue 0.11 0.20 0.02 23.8 0.8 Source: World Bank staff es�mates 6 Fiscal Policies And Ins�tu�ons For Shared Growth exhibited counter-cyclical behavior. The vola�lity of Therefore, we adjust for the cyclicality of import the expenditure cyclical components is also higher and excise duty, and local VAT. The adjustment is than that of the GDP component. Domes�c inter- performed according to the following formula: est payments, total interest payments, and total de- velopment expenditure have the highest vola�lity: 24.4, 23.1 and 50.2 per cent, respec�vely. Cyclically Adjusted Budget Surplus where is the total revenue, is primary ex- penditure, are the primary budget sur- Computa�onal Issues pluses before and a�er the adjustment, refers To compute the cyclically adjusted budget surplus, to the import duty revenue, is excise duty, and there is need to narrow down the list of variables we is the local VAT revenue; stand for want to adjust for. In other words, we must decide the es�mates of output elas�ci�es of , , and which components of the revenue/expenditure are Annex 1: Figure A9 presents overall and primary automa�c and which are discre�onary. To include a budget balance before and a�er the cycle adjust- component of revenue/expenditure in this ‘adjust- ment as a percentage of the HP trend of real GDP. ment list’, there needs to be a strong, a priori, reason While both adjusted and unadjusted overall bal- to believe it would behave pro-cyclically/counter- ance exhibit a very slight downward trend, there cyclically. In the case of Kenya, we have included all is a no�ceable decline in the primary balance over the revenue categories for adjustments. We believe �me. While it remained posi�ve throughout the late that no expenditure component can be expected to 1990s, it deteriorated in the �rst years of the 21st be moving closely with the business cycle, a priori; century, reaching -4.9 per cent in 2008 Q2. This be- that is, we classify all the expenditure components havior is primarily a result of an increase of the non- as discre�onary. Further, for each of the revenue interest government expenditure, combined with components, we es�mate its elas�city with respect no substan�al increase in the total revenues. Panel to the cyclical component of GDP (see Annex 1: Fig- (e) of Figure A9 depicts the cycle adjustment, also ure A8 for the relevant diagrams). For this purpose, as a percentage of the GDP trend. It never reaches we run a series of linear regressions described by 1 per cent in absolute value, though it is s�ll some- the equa�on below, where refers to the cyclical what signi�cant. For instance, the cycle adjustment component of the corresponding revenue category is nega�ve for most of the period of 2002 Q2–2005 and is the cyclical component of the real GDP. Q1, which implies that �scal policy was more expan- The regression es�mates are presented in Table 7. sionary than what the non-adjusted measure of the budget surplus indicates. The adjustment reaches the value of -0.51 per cent in 2002 Q3. Table 7: Es�mates of revenue elas�ci�es Characterizing Fiscal Policy Revenue Category Elas�city Kenya’s �scal policy does not appear to be pro-cy- Import Duty 2.13* (1.27) clical. As can be seen from Table 8, the unadjusted Excise Duty 1.79** (0.86) primary balance is posi�vely correlated with the PAYE 0.80 (0.67) cyclical component of GDP. This suggests that it is higher during expansions and lower during re- Other Income Tax 0.26 (1.44) cessions, which would suggest counter-cyclical- VAT Local 2.07** (0.92) ity instead. Some of this effect, however, is due to VAT Imports 0.57 (0.85) automa�c increase in revenues associated with ex- Other Revenue 1.78 (2.26) pansions. Once the primary budget balance is cycli- Source: World Bank staff es�mates cally adjusted, the correla�on coefficient decreas- The results show that the only two revenue catego- es. It is es�mated to be 0.08 for the full sample, ries that have elas�city coefficients signi�cantly dif- which suggests that the adjusted primary balance ferent from zero at 5 per cent are excise duty and is almost acyclical. Further, if we consider a sub- local VAT. Import duty has a P-value of 0.101; all the sample of 2003-2008, the correla�on coefficient other elas�ci�es are not signi�cant at 10 per cent. drops to -0.02, indica�ng that �scal policy became 7 Fiscal Policies And Ins�tu�ons For Shared Growth Table 8: Correla�on of the budget surplus measures with the cyclical component of GDP Full sample: 1996-2008 Shorter sample: 2003-2008 Cyclically Not Cyclically Not adjusted adjusted adjusted adjusted Overall balance 0.01 0.11 -0.05 0.09 Primary balance 0.08 0.16 -0.02 0.11 Source: World Bank staff es�mates somewhat more pro-cyclical under the new govern- ary �scal policy (�scal innova�ons that are purely ment. S�ll, the magnitude of the coefficient is too unpredictable) on Kenya’s economy, we construct small to suggest signi�cant pro-cyclicality. This con- and analyze a small VAR model (see Annex 2). The clusion is supported by the fact that, according to following variables are included in the model: Table 6 above, the expenditure components are not • The logarithm of the world price of oil (we strongly correlated with the cyclical component of use Europe Brent Spot Price FOB, expressed in GDP; besides, this correla�on is weaker than for the constant 2001 Ksh per barrel), pOt components of revenue. • The logarithm of real GDP in the US (in millions of 2000 chained dollars), yUt We also compute the discre�onary primary budget • The US Federal Funds rate (percent per year), balance and the �scal impulse. The discre�onary rUt balance is computed as follows: • The cyclically adjusted primary budget balance of Kenya (percent of HP trend GDP), δBt • The logarithm of Kenya’s real GDP (in millions where is the primary balance, t and x are the of constant 2001 Ksh), yt sample averages of revenue and primary expendi- • The logarithm of the real Kenya-US exchange ture as frac�ons of GDP, Y is real GDP, and Y* is the rate*t. HP trend of real GDP. The discre�onary balance is plo�ed in Annex 1: Figure A10 against the cyclically The price of oil is included as one of the variables adjusted primary balance. The two series are highly because Kenya’s trade balance and budget revenue correlated (with a correla�on coefficient of 0.998). closely depend on it, both through transporta�on The �scal impulse is computed as and costs and also directly, since Kenya trades petro- provides a measure of the stance in �scal policy. leum products interna�onally. The GDP of the US When the �scal impulse is nega�ve, �scal policy be- is included as a proxy for the world demand for comes more contrac�onary, and vice versa; a posi- Kenya’s exports (speci�cally, Kenya exports tea, cof- �ve �scal impulse indicates more expansionary �s- fee, apparel, household goods, etc to the US, UK, cal policy. The moving average of the �scal impulse and other countries). The US Federal Funds rate is is plo�ed against the moving average of the change used to account for the state of the world �nancial in the cyclically adjusted primary balance in panel market. Finally, real exchange rate is added to ac- (b) of Figure A10. Again, the two series are strongly count for other shocks to the Kenyan economy. The correlated; in fact, they are nearly iden�cal. Com- technical model is presented in Appendix Group A, paring them with the cyclical component of GDP, Annex 3. one can also conclude that Kenya’s �scal policy does not exhibit strong pro-cyclicality. The results of the es�ma�on are presented in two forms: plots of the impulse response func�ons (An- Impact of Fiscal Policy on Real Ac�vity: A Small nex 1: Figure A11) and a table with variance decom- Var Model of Kenya’s Economy posi�on (Table 9). Impulse response func�ons in this context shows the impact of an output shock Model Features on the cyclically adjusted primary balance. Panel (b) In order to assess the effect of random discre�on- of Annex 1: Figure A11 illustrates the response of 8 Fiscal Policies And Ins�tu�ons For Shared Growth the cyclically adjusted primary balance to a 1 per or unproduc�ve investment. Annex 1: Figure A12 cent posi�ve output shock. Due to one of the as- shows historical contribu�ons of the �scal shock sump�ons made, the adjusted �scal balance only component to output vola�lity, again sugges�ng responds contemporaneously to the external vari- low impact of �scal shocks on Kenya’s real GDP. ables and, therefore, the �rst element is zero. The �scal balance improves for two quarters and then Conclusions and Policy Implica�ons decreases. The ini�al posi�ve response only reach- This study has analyzed the cyclical proper�es of es 0.4 per cent of GDP, which suggests that Kenya’s Kenya’s �scal policy. First of all, Kenya’s �scal policy �scal policy is not pro-cyclical. Panel (a) of Annex does not lean with the wind. While some revenue 1: Figure 11(a)) suggests that �scal shocks have components adjust automa�cally over the business reasonably low persistence, with the effects of the cycle, increasing in booms and falling in recessions, original �scal shock negligible by quarter 5. At the the correla�on of the discre�onary budget compo- same �me, the output exhibits somewhat higher nent with the cyclical component of GDP is small. persistence, with the effects of the original shock As a result, the cyclically adjusted primary balance las�ng for 9 quarters, see Figure 11(d). appears to be acyclical for the period 1996-2007. Interpre�ng our �nding in the context of the broad- Finally, panel (c) illustrates that purely unpredict- er literature on this topic, the acyclicality of �scal able �scal shocks have very insigni�cant effects on policy suggests that Kenya’s �scal policy has been output. While �scal �ghtening causes real GDP to a growth-enhancing factor. As shown by a number contract, the response is smaller than 0.1 per cent of researchers, cyclicality is usually associated with in magnitude (in response to a 1percent change in vola�lity, which in turn lowers economic growth. the cyclically adjusted primary balance), which sug- The �nding that �scal policy has been acyclical in gests that the component of �scal policy we focus Kenya would imply that �scal policy did not amplify on in Kenya, the random discre�onary policy, has a extraneous shocks during the study period, and this limited effect on GDP. However, this �nding needs may have supported growth. However, it is worth- to be interpreted with cau�on. There are two other while emphasizing that �scal shocks by themselves �scal policy components: (i) systemic discre�onary are o�en cause of macroeconomic instabili�es, and policies, which are rou�ne responses to changing a major �scal shock would probably have led to economic situa�ons; and (ii) automa�c policies, macroeconomic instability in Kenya. governed by rules and laws, such as the tax code. A full assessment of the impact of overall �scal policy Second, the random discre�onary component of on GDP needs to take all three components into ac- �scal policy of Kenya does not appear to have a sub- count, and further research is needed in this area. stan�al effect on GDP, though this �nding requires a careful interpreta�on. While �scal �ghtening has Table 9 presents the variance decomposi�on of out- a contrac�onary effect, this effect is very small in put for the forecast error in the real GDP of Kenya. magnitude and rela�vely short-lived. However, the Again, �scal shocks do not appear to have an im- other components of �scal policy, namely the sys- portant effect on GDP. The importance of the out- temic discre�onary policy and automa�c policies, put shock decreases with the �me horizon, while oil need to be also considered to assess the impact of price shocks, shocks to the US GDP (and especially the overall impact of �scal policy on GDP in Kenya, shocks to the Federal Funds rate) become more im- which is a topic for further research. portant in the longer term. In fact, in the long term, the US monetary shocks explain over 70 per cent The results of the VAR analysis corroborate the �nd- of the variance in Kenya’s GDP. This seems to imply ing that �scal policy is acyclical in Kenya. In par�cu- that it is Kenya’s monetary policy, rather than �scal lar, the variance decomposi�on analysis suggests policy, that is an important factor in terms of out- that there is an important rela�onship between the put vola�lity. Further, the fact that Kenya’s random US Federal Funds rate and the cyclical component discre�onary �scal policy has had limited impact on of Kenya’s GDP. Therefore, Kenya’s monetary policy, the country’s real GDP seems to suggest that a con- and speci�cally its exchange rate policy, appears to siderable amount of Kenya’s government spend- be an important determinant of Kenya’s GDP fluc- ing either goes towards government consump�on tua�ons. Which factors may have contributed to 9 Fiscal Policies And Ins�tu�ons For Shared Growth these �ndings? One possibility is the �scal adjust- policy. ment Kenya experienced following the Goldenberg scandal. Expenditures between 2002 and 2007 re- Although somewhat surprising, the results of this mained fairly steady and ranged from 22-23 per study should be interpreted with cau�on. It is well cent, and these series do not exhibit much vola�lity. known that the quality of any data analysis is deter- Public borrowing was limited to about 1.8 per cent mined to a great extent, by the quality of underly- of GDP in that period, and the fact that there were ing data. In this respect, there are two main factors no major differences between policy announce- to consider: the general reliability of the data, and ments and �scal outcomes would suggest that �s- the somewhat short �me horizon. Without doubt, cal policy reduced uncertainty in general. This may it would be bene�cial to re-examine the issue once have contributed to this �nding, which suggests more data becomes available. that Kenya should avoid major changes to its �scal Table 9: Variance decomposi�on of output (percent of variance of the forecast error) Forecast Percent of variance due to shocks to Horizon Oil price US GDP Federal Fiscal GDP Real exchange (quarter) funds rate balance rate 1 3.3 0.0 14.1 0.0 82.6 0.0 2 3.3 1.1 24.2 0.3 69.0 2.0 3 3.0 4.3 39.5 0.2 47.7 5.2 4 2.5 4.5 53.4 0.3 34.1 5.1 6 1.8 3.6 69.7 0.3 21.3 3.4 8 3.0 2.8 74.8 0.2 16.5 2.7 12 5.2 3.0 75.3 0.2 14.0 2.3 16 6.0 4.5 73.5 0.2 13.6 2.2 8.9 5.8 70.2 0.2 12.9 2.1 Source: World Bank staff es�mates 11 2. A GLOBALLY COMPETITIVE AND PROSPEROUS KENYA: THE ROLE OF PUBLIC INVESTMENT Introduc�on The note starts with an overview of public invest- This note explores the effec�veness of public in- ment outcomes and the state of the physical capital vestment in Kenya and makes recommenda�ons stock in Kenya. The note also analyzes the composi- for improvements. It is the second in a series of �on and trends in development budget spending, budget policy notes exploring �scal policy, growth, and examines the budget process issues and the and poverty linkages. The note is based on a review extent of proposals for improvements in the short of the literature and publicly available �scal data. and medium term. It a�empts to examine the role of public invest- ment as a policy tool for: (a) achieving the aspira- Public Investment: Why It Ma�ers �ons set out in the government’s Vision 2030; (b) There are three core reasons why public investment achieving broad-based economic growth in Kenya; ma�ers. First, public investment provides goods and (c) improving service delivery for household and services to households, which directly improves welfare improvements. The note focuses on the their welfare. For example, public investments can ins�tu�onal issues that have adversely affected result in households gaining access to electricity or the performance of Kenya’s development budget, piped water and sanita�on, resul�ng in vastly im- such as the procurement procedures and the donor proved health status and quality of life. Second, disbursement delays, which have resulted in sig- public investment is posi�vely correlated with a ni�cant under-spending in recent years. In light of reduc�on in poverty–o�en through more indirect the 2009/10 �scal s�mulus package to bolster de- means. For example, investments in rural roads fa- velopment spending, expenditure under the devel- cilitate farmers ge�ng their products to new and opment budget is planned to grow from 6 per cent wider markets, or children to school. Third, public of GDP in 2007/08 to an average of 8 per cent of investment is also empirically posi�vely correlated GDP per annum in the next �ve years. This note also with economic growth through lowering the costs of provides some conclusions and recommenda�ons doing business and crowding in private investment. for improving the use of the development budget For example, cheaper and more reliable access to for sustainable development outcomes. While op- electricity, road and rail infrastructure lowers the era�ng and maintaining public investments is cri�- costs of doing business and creates new investment cal for the full bene�t of the ini�al investment to opportuni�es for future economic growth. Howev- be realized, a full analysis of the adequacy of the er, public investment can also crowd out private in- recurrent expenditure goes beyond the remit of vestment, par�cularly if it is �nanced domes�cally, the policy note, and provides an area for follow-up causing interest rates to rise and increasing the cost work that could be taken forward in the forthcom- to the private sector of undertaking its own invest- ing Public Expenditure Review (2010). ments. 12 A Globally Compe��ve And Prosperous Kenya Box 1: Public investment: Does it crowd in or crowd out private investment The role of the public sector in determining pri- health facili�es and educa�on services tend to vate investment and economic growth is covered complement private capital forma�on by eliminat- in several empirical studies. Da�ng back as far ing bo�lenecks. However, if such capital spend- as 1989, Aschauer found that public investment ing is �nanced by domes�c borrowing, this pushes crowds in private investment spending in both up the cost of capital and reduces private invest- developed and developing countries.² There has ments accordingly. Poorly designed public invest- been a resurgence of empirical research into the ment projects and/or those with low or marginal rela�onship as a result of endogenous growth rates of return would, therefore, be not only poor theory, which predicts that the output elas�city to value for public money but could poten�ally have private capital is considerably higher than predict- a doubly nega�ve effect through the impact on ed by the private capital share in na�onal income. domes�c interest rates. Crowding out can also oc- The posi�ve impact of public investment spend- cur with other inputs to investment projects. For ing found in both private investment and growth example, the costs in the construc�on industry equa�ons has been a�ributed to the presence of with a limited supply of contractors and profes- posi�ve spillover effect generated from the pro- sional services can increase drama�cally, or even vision of quasi-public goods, whose services are be totally absorbed by the public investment essen�al for the proper func�oning of a market program, raising the costs of such services to the economy. It is argued that public investment by private sector and making those investments less the state in roads, airports, ports, energy, water, viable. State of Infrastructure parts of the developing world. The differences are The stock of capital and access to related services par�cularly large in the case of paved roads, com- in Kenya lags behind the middle income countries munica�ons infrastructure and power genera�on to which it aspires, and is closer to the average in capacity. Furthermore, for these three key infra- Africa. Kenya’s indicators of public investment out- structure sectors, Africa has been expanding stocks comes lag considerably when compared to a sample much more slowly than other developing regions, of middle-income countries with respect to access and therefore the gap is widening.³ Even if the me- to sanita�on, telephone, internet and paved roads. dium-term targets for expected outcomes on many Kenya is closer to LICs and SSA averages (see Table of the infrastructure indicators are met, this would 10 below). Kenya should not feel comfortable with s�ll place Kenya in 2011 at or below today’s LIC av- being average; on just about every measure of in- erages. frastructure coverage, the Africa region lags other Table 10: Public investment outcome indicators, Kenya in comparison, 2008 Kenya Sub-Saharan Other Low Africa LIC Income Country average average Paved road density (kilometers per kilometer squared) 16 31 134 Total road density (kilometers per kilometer squared) 112 137 211 Main line density (lines per thousand popula�on) 7 10 78 Mobile density (lines per thousand popula�on) 30 55 76 Internet users (subscribers per thousand popula�on) 9.5 20 30 Genera�on capacity (megawa�s per million popula�on) 33 37 326 Electricity coverage (percent of popula�on covered) 18 16 41 Improved water (percent of popula�on covered) 57 60 72 Source: Yepes et al. (2008), World Development Indicators (2008) � See also Cardoso (1993), Ram (1986). � Foster (2008). 13 A Globally Compe��ve And Prosperous Kenya In the case of access to electricity, Kenya lags be- 3 to 5 percent of the total value of machinery and hind LIC averages by a considerable margin, and is equipment. The investment climate surveys show on a par with low income SSA averages. Although that Kenyan �rms suffer much higher power costs access in Kenya (18percent of households) is ahead (indirect costs and losses are about 7percent of of neighboring Uganda (9percent have access to sales) than �rms report in China, India and South electricity) and Tanzania (11percent have access to Africa. electricity), it lags behind the average for the low income countries at 41 per cent. Furthermore, the Transporta�on is the other top complaint of Ken- 2011 target to connect 33 per cent of households yan �rms as a result of high associated direct and to electricity is s�ll below the current average of 41 indirect costs. Unsurprisingly, �rms outside Nai- per cent with electricity in LICs in 2008 (see Table robi perceived this as a major problem more o�en 10). Africa’s largest infrastructure de�cit is in the than �rms located in the capital city. Firms in the power sector in terms of genera�on capacity, elec- manufacturing sector are par�cularly affected by tricity consump�on, and security of supply. Power the poor state of the transporta�on system. Inland consump�on at 124 kilowa� hours per capita per transport costs in Kenya are much higher than in year and falling is only a tenth of that found else- China and India. Indirect costs are also higher than where in the developing world. those of comparators, since Kenyan companies lose 2.6 per cent of their sales to spoilage and the� dur- The situa�on for roads, piped water and sanita�on ing transporta�on. Half of the transporta�on losses is closer to the averages for LIC and SSA. Kenya has a are due to transport delays, while the other half are rela�vely well extended road network, which is 112 due to the� during transporta�on. road miles per km2, close to the SSA average of 137 road miles per km2. However, only 14 per cent of The poor suffer more from lack of access to electric- the road network is paved. ity, water and sanita�on. About half of the popula- �on does not have access to water and sanita�on This infrastructure bo�leneck is a signi�cant con- services, and do not therefore meet the minimum straint to growth and compe��veness. In invest- standards of quality; the per cent of those meet- ment climate assessments, the private sector has ing the minimum standards is much lower in rural consistently iden��ed “poor and costly infrastruc- areas. In both rural and urban areas, the poorest ture as the binding constraint to business com- households have the lowest access to private piped pe��veness� in Kenya.� Around half of businesses water supply. On average, it takes 50 minutes in ru- surveyed in 2007 found infrastructure services ral areas and 20 minutes in urban areas to collect (transport and electricity) to be a major obstacle, water. There are also signi�cant regional varia�ons and this was higher than in the previous survey in in infrastructure access. In Western Province, for 2002. example, 6 per cent use piped water and only 4 per cent of the popula�on has access to electricity. This Electricity has become more of a problem for busi- is not only a rural phenomenon; the higher urban nesses over the last four years. As the rate of eco- averages hide a signi�cant rich: poor divide in terms nomic growth during 2003-2007 increased, so did of infrastructure access. In the urban slums, less electricity demand and therefore the reliability than 6 per cent have access to piped water, and less and cost of power supply worsen. Power outages than 3 per cent have access to a private latrine. have increased from an average of 16.4 hours per month to 33 hours per month between 2002 and There has, however, been improvement in access to 2007. According to the �rm surveys, 85 per cent re- infrastructure over the last decade (see Figure 9 and port experiencing power outages, and close to 80 10 ). The average distance to a motorable road, for per cent experience losses as a result. Two thirds of example, has fallen for both rich and poor house- �rms have their own generator to help resolve the holds. However, the pace of electri�ca�on has been issue. A generator is costly to run and the capital in- slow; an average 0.5 per cent of the popula�on vestment of a generator accounts for approximately gaining access each year since 1993. While elec- � World Bank Investment Climate Assessment 2002 and 2007. 14 A Globally Compe��ve And Prosperous Kenya tri�ca�on across the country increased by 50 per Before looking at the sectoral composi�on of the cent between 1993 and 2003, it has since slowed development budget in more detail in the next sec- down and access remains low especially in rural ar- �on, we look at the trend and comparison of Kenya’s eas. Na�onally, access to safe water stagnated un�l indicators of human capital stocks. We �nd that in- 2003, and actually fell in Western Province during dicators of the health status, level of educa�on and this �me. employability of Kenyans to be around the SSA and LIC averages (see Table 11 below). However, once Stock of Human Capital again, we �nd that Kenya lies behind several African On several indicators of human capital stock, Kenya countries (such as Ghana) and the Asian middle-in- appears to be no more than average for the SSA re- come countries that we have used for comparison. gion and LICs. A broader concept of public invest- The full table of indicators is summarized in Appen- ment should include investment in human capital. dix Group B, Annex 2 Table B2. Figure 9: Provincial distribu�on of service gains, Figure 10: Reduced distances to various 2003-2005/06 infrastructure, 1997-2007 Source: World Bank (2008b) Source: Nyoro etal 2008 Table 11: Indicators of human capital, Kenya and SSA and LIC averages Sub-Saharan LIC Kenya Africa average average Employment to popula�on ra�o, 15+, total (percent) 73 64 65 Employment to popula�on ra�o, ages 15-24, total (percent) 59 49 51 Primary educa�on, dura�on (years) 6 6 6 Immuniza�on, DPT (percent of children ages 12-23 months) 81 73 77 Immuniza�on, measles (percent of children ages 12-23 months) 80 73 76 Life expectancy at birth, female (years) 55 52 59 Life expectancy at birth, male (years) 53 50 56 Life expectancy at birth, total (years) 54 51 57 Mortality rate, infant (per 1000 live births) 80 89 80 Newborns protected against tetanus (percent) 73 75 78 Source: World Development Indicators (2008) 15 A Globally Compe��ve And Prosperous Kenya Overview of the Development Budget of the development budget shows that acquisi�on of non-�nancial assets and capital grants are the A Note on Classi�ca�on largest components (see Table 12). Compensa�on Public investment is the public sector spending of employees are less than one per cent on aver- that typically adds to the capital stock; it is approxi- age, and goods and services consump�on averages mated by the development budget in Kenya. Pub- 10 per cent of the budget. In 2008/09, spending on lic investment is “capital� spending or “acquisi�on the acquisi�on of non-�nancial assets was largely of non-�nancial assets� in the budget according to from the development budget (89percent). Howev- Government Finance Sta�s�cs (GFS 1986 and 2001) er, the development budget includes other types of Manual classi�ca�on. No such classi�ca�on ex- spending, and only 63 per cent of the development ists in the Kenya �nance sta�s�cs. Instead, public budget was on acquisi�on of non-�nancial assets in investment spending is approximated by spending 2008/09, and this was as low as 41 per cent of the under the “development� budget. In any par�cular development budget the year before.� year, the non- “capital� part of the development budget can be sizeable. For example, the 2007/08 There is poten�ally a considerable amount of pub- development budget es�mates include provision lic investment spending that takes place off-budget. for Ksh 41.5 billion, or 20 per cent of the develop- Briceno-Garmendia and Foster (2007) take a more ment budget, as the government’s equity par�cipa- comprehensive look at infrastructure expenditures �on in the na�onal telecom company prior to pri- by looking at the infrastructure spending of several va�za�on. non-budget ins�tu�ons.� They �nd that only around 30 per cent of infrastructure spending in Kenya is on budget. This varies according to sector, from less There is some misclassi�ca�on, but possibly less than 10 per cent of energy expenditures to almost than other countries in East Africa. The development 70 per cent of transport expenditures. Given that budget comprises all donor-�nanced spending, re- about half of infrastructure spending is public in- gardless of whether these are recurrent costs or vestment, and many of the non-budget ins�tu�ons capital investments, plus “capital� projects �nanced are state-owned enterprises, this could imply that by the government’s own sources, Cons�tuency considerable off-budget public investment is taking Development Funds, and transfers to other parts place. of the public sector. The donor-�nanced projects can also contain a considerable amount of funding It is not surprising that a substan�al share of pub- lic investment is off budget and nor is it necessarily for recurrent expenditures, such as for goods and sub-op�mal. Public investment in infrastructure is services, such as consultants and pharmaceu�cals o�en carried out through the relevant state-owned in health projects. However, an exercise to compile enterprises, rather than the central government. public investment spending in infrastructure found This is par�cularly the case when such investments that less than 5 per cent of current expenditures can be recouped through the sale of services to the were misclassi�ed as development, and only ap- private sector, but may also require transfers from proximately 1 per cent of capital spending was mis- the budget and implicit and explicit loan guarantees classi�ed as recurrent. The sum of misclassi�ca�on when the capital for those investments is raised. problems in the infrastructure sector was less than Most of the pressing demands for infrastructure in Kenya—such as electricity—are provided by par- other countries in the region.� astatals. Therefore, analysis of the infrastructure that public enterprises �nance and the scope for The development budget is predominantly, but not full cost recovery, including the servicing of their only, capital spending. The economic composi�on capital, should be undertaken to more accurately � Briceno-Garmendia and Foster (2007). � Government of Kenya (2009). It will be useful in follow-up work to explore further the use of capital grants and transfers included in the development budget. 16 A Globally Compe��ve And Prosperous Kenya Table 12: Economic composi�on of development budget, 2004/05-2008/09 as percent of total 2004/05 2005/06 2006/07 2007/08 2008/09 Compensa�on of employees 0.49 0.52 1.14 0.64 0.83 Use of goods and services 8.82 14.72 8.47 7.38 10.61 Subsidies and grants 16.23 5.86 6.23 6.04 5.76 Capital grants 15.92 27.60 14.39 17.62 13.10 Acquisi�on of non-�nancial assets (net) 58.00 44.33 47.29 40.83 62.97 Others 0.54 6.97 22.48 27.49 6.72 Source: Government of Kenya (2009) assess the full poten�al for Kenya to raise the funds the �scal outcomes since the beginning of the dec- required for infrastructure investments. This analy- ade and the medium term plan. sis was not undertaken here due to the difficulty of accessing the relevant data.� In real terms, the development budget has increased almost four-fold since the beginning of the decade. Previous World Bank research on infrastructure in Figure 10 shows that total development spending East Africa indicated that u�li�es in Kenya incurred has increased from Ksh 25 billion to Ksh 97 billion substan�al quasi-�scal de�cits, resul�ng in insuffi- between 2000/01 and 2008/09 in constant prices. cient investment and maintenance, which results in Between 2006/07 and 2007/08, there was a real a deteriora�ng capital stock. Further analysis of this increase of 50 per cent in public investment, and should be taken forward in the Public Expenditure there is an es�mated further 28 per cent increase Review 2010. Addi�onal work is also being done in 2008/09. separately on the role of the private sector in public investment. While Kenya has reversed the decline in public in- vestment that occurred in the 1990s, it has yet to Increasing Public Investment Kenya reach the historically high rates of investment seen The Kenyan development budget has risen from 3.0 in middle-income country comparators. Using the per cent of GDP in 2003/04 to 6.6 per cent of GDP IMF’s GFS01 �scal dataset, we compared Kenya’s in 2007/08, and an es�mated 9.3 per cent of GDP spending on non-�nancial (mostly �xed) assets in 2008/09. It is forecast to stay at approximately over �me with several comparators (see Figure 11 8.0 per cent of GDP over the medium-term. This in- ). We �nd that as a percentage of GDP, there was a crease in development spending has been brought considerable decline in public investment spending about by �scal space created by an increase in rev- over the 1990s, a trend that was reversed in the fol- enue from 21 percent of GDP in 2003/04 to 24.4 per lowing decade. However, at less than 4 per cent of cent of GDP in 2008/09, and restraint on the growth GDP in 2007/08, this is s�ll below the investment of recurrent expenditure (maintained at an average spending as a percentage of GDP of Indonesia and of 20 percent of GDP over the period). The forecast Malaysia in the 1990s, and South Africa’s average shows that a combina�on of increased domes�c since 2000. As richer and larger economies (in terms and external borrowing will be required to accom- of GDP), this also means that in per person terms, modate a higher level of development budget in this is considerably lower. propor�on to GDP. Table 13 shows the evolu�on of � These include the power u�li�es and several private and state-owned enterprises providing the rail, airport, and telecom services. � Comprehensive details for the 161 state corpora�ons on their expenditure and non-tax revenues are not included in the �scal re- ports, either in consolida�on with other central government expenditure, or shown in a separate annex of the document, or shown in a separate document presented to the legislature and published at the same �me as the �scal reports. 17 A Globally Compe��ve And Prosperous Kenya Table 13: Fiscal space for increased development spending 2002/03-2008/09 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 Revenue 19.4 21.0 21.6 20.5 21.6 22.0 24.4 21.6 21.7 21.8 Expenditure 24.3 23.3 22.6 25.2 24.3 27.3 32.1 26.6 26.8 27.0 Recurrent 20.3 20.2 19.0 20.2 17.8 20.6 22.5 18.4 18.6 18.6 Development 3.9 3.0 3.3 4.4 4.6 6.6 9.3 7.9 8.0 8.2 Net lending 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 Balance -4.9 -2.3 -1.0 -4.7 -2.7 -5.2 -7.6 -5.0 -5.1 -5.7 Financing 3.2 0.0 -0.5 2.4 2.1 -0.4 -2.9 3.5 3.4 3.4 Of which Domes�c -1.1 -0.7 -0.1 0.1 -0.1 0.3 1.2 1.9 1.5 1.5 External 4.3 0.7 -0.5 2.3 2.2 -0.7 1.7 1.6 1.9 1.9 Source: Government of Kenya, Budget Strategy Paper 2009; Quarterly Economic Budget Review Third Quarter 2008/09; IMF Staff Report for the 2008 Ar�cle IV Notes: Data from 2002/03 to 2007/08 are actuals, data for 2008/09 are es�mates, and data for 2009/10 to 2011/12 are projec�ons Figure 11: Rise in development spending, 1999/00-2011/12 Manpower and ICT Health Jus�ce, Law & Physical Infrastructure - 2004/05 2005/06 2006/07 2007/08 2008/09 Source: IMF and MoF Notes: Constant 2001/2 prices 18 A Globally Compe��ve And Prosperous Kenya While Kenya’s total spending on infrastructure is Priori�es are Shi�ing Towards Physical high compared to East Africa (10 percent of GDP), Infrastructure only 20 per cent is for public investment. Given The composi�on of development spending shows the limita�ons of GFS-classi�ed data with respect a shi� towards physical infrastructure in recent to iden�fying infrastructure spending, a new �s- years. The sectoral composi�on of Kenya’s develop- cal dataset for Africa was compiled for an earlier ment spending as reported in Government of Kenya World Bank report.� Their �ndings show that there (2009) shows that the shares have shi�ed (see Fig- is rela�vely low public investment in infrastructure ure 12 and Table 15) to reflect an increase in physi- in Kenya (only 2percent of GDP). While spending on cal infrastructure, largely at the expense of shares infrastructure in propor�on to GDP is around 10 per to health, educa�on and governance sectors. cent in Kenya—the highest in the region and slightly higher than middle income countries in other re- Taking a longer �me perspec�ve, and comparing gions—the low share that goes to public investment the averages for two periods, pre- and post- 2003, means that Kenya is much lower than Uganda and both physical infrastructure and public administra- Rwanda, where over half of infrastructure spending �on made the biggest increases in the share of the is on public investment. development budget. As shown in Figure 13, physi- cal infrastructure averaged less than 30 per cent of Low public investment compared to the region is development spending in the 1999-2003 period, con�rmed by the government’s Public Expenditure and the share increased to 39 per cent of spend- Review (2009). The Review reports that Kenya has ing for the period 2004-07. Educa�on also gained been behind other countries in the East African from 6.1 per cent to 7.8 per cent of spending. The Community on capital investment. Kenya’s develop- biggest increase, however, was in public administra- ment budget has been rela�vely low compared to �on, which doubled its share of spending from 16.5 Tanzania and Uganda. In 2004/05 and 2005/06, Tan- per cent to one third of development spending. This zania and Uganda were inves�ng more than double reflects, in part, the increase in development budg- the level in Kenya in propor�on to GDP. Table 14 et funds going through the Ministry of Finance, in- shows how Kenya compares, and the progress that cluding for equity investments as men�oned in the has been made since then as development spend- notes on classi�ca�on sec�on. The sector shares for ing increases. agriculture and rural development and public safe- Figure 12: Acquisi�on of non-�nancial assets, Kenya and comparators, 1990-2007 Source: GFSO1 � Briceno-Garmendia and Foster (2007). 19 A Globally Compe��ve And Prosperous Kenya Table 14: Kenya and comparators, development spending as percent of GDP Tanzania Uganda Kenya 2004/05 7.2 8.8 3.0 2005/06 9.1 8.0 4.1 2006/07 7.3 1.3 6.2 Source: Government of Kenya (2009) Figure 13: Share of development spending by sector, per cent of total 2004/05-2008/09 Source: Economic Survey 2009; East African Facts ans Figures, 2006 Table 15: Share of development budget by sector, 2004/05 to 2008/09 (Ksh millions) 2004/05 2005/06 2006/07 2007/08 2008/09 Produc�ve Sector 19,184.99 20,881.59 27,446.74 3,321.38 42,951.18 Public Administra�on 26,868.65 36,976.33 5,082.04 87,692.61 6,885.95 Physical Infrastructure 34,737.27 45,675.37 67,232.75 96,610.03 155,027.35 Governance, Jus�ce, Law and Order 47,934.99 57,083.54 49,259.73 64,370.81 75,020.24 Health 19,158.35 23,006.98 27,530.04 29,307.68 34,636.53 Educa�on 81,041.72 92,601.12 103,856.74 121,315.23 138,780.31 Na�onal Security 25,142.07 30,709.25 31,050.23 44,890.84 48,911.46 Informa�on & Communica�on Technology 1,307.65 2,008.09 2,761.74 4,531.22 8,995.72 Manpower and Special Programs 2,575.67 2,928.59 14,756.10 17,599.91 37,154.79 Total 257,951.36 311,870.86 378,976.11 469,639.71 598,363.53 Source: Government of Kenya Public Expenditure Review 2009 ¹� Although the infrastructure de�cit is well documented, it would be interes�ng for future work to look at the composi�on of the infrastructure investment, since the Foster (2008) found that Kenya (like others in the region) showed signs of under-investment in the power sector, and poten�al over investment in the water sector. 20 A Globally Compe��ve And Prosperous Kenya ty and law and order declined signi�cantly. While ment budget, falling from 73 per cent of expendi- increasing in real Kenya shilling terms, the share ture to 41 per cent of the development budget in of net development spending in the health sector 2008/09. fell to 5 per cent, on average. In the last two years, the budget shows a slight shi� back to priori�zing In terms of annual commitments to the Govern- the social sectors—or “human resource develop- ment of Kenya, the largest development partner is ment�—as the sector has been de�ned in the Me- currently the World Bank (officially the Interna�on- dium Term Expenditure Framework (MTEF). al Development Agency-IDA). In real terms, there has been a decline in resources Figure 14 shows the loan/grant mix of donor as- to the social sectors. Reclassifying the agency or in- sistance as captured by External Resources Depart- s�tu�onal breakdown into standard sectors and ad- ment (ERD) data on donor commitments. The top jus�ng to constant prices shows that not all sectors two donors: IDA and the African Development Bank have seen a real increase in development budget made up 43 per cent of total donor commitments spending. The increase in spending on physical in- in 2008/09, followed by France, China and the EC, frastructure has been maintained, and so too the who each have around 6 per cent of total annual rise in public administra�on development spend- commitments. ing. Health and educa�on, however, have been de- clining in real terms in the most recent years. Development partner �nance is predominantly for the infrastructure ministries. Figure15 shows the Role of Development Partners main government counterparts listed in the 2008/ The overall role of development partners in �nanc- 09 budget as the main recipient for donor funds. ing the development budget is decreasing. Annual As we might expect, the ministries responsible for development partner grant expenditures rose from physical infrastructure (Ministry of Roads, Ministry Ksh 4.4 billion to Ksh 33.8 billion between 2002 and of Water and Irriga�on and Ministry of Energy) are 2008, and loan disbursements from Ksh 14 billion to the most signi�cant bene�ciaries and implemen�ng Ksh 47 billion (see Table 17). However, the donors agencies for development partner-funded projects are no longer �nancing the majority of the develop- and programs. Figure 14 : Ministerial Spending 2000 Prices: Spending shi� to physical infrastructure Source: MoF Key: Agric.and R. Dev- Agriculture and Rural Development; P. Infrast. - Physical Infrastructure; PBSLO- Public Safety Law and Order; P. Admin.-Public administra�on. 21 A Globally Compe��ve And Prosperous Kenya Table 16: Sectoral composi�on of development expenditures (Ksh millions) 2004/05 2005/06 2006/07 2007/08 Public Administra�on 18,496 15,191 30,280 57,555 Physical Infrastructure 15,853 22,543 35,512 33,517 Agriculture and Rural Development 1,652 3,629 5,409 7,021 Educa�on 3,554 5,846 5,492 4,536 General Economic Services 1,296 962 4,794 4,098 Public Safety Law and Order 5,017 4,399 5,659 3,704 Health 595 2,939 2,163 2,748 Total Development 46,463 55,508 89,310 113,179 Source: Ministry of Finance Notes: Constant 2004 prices Table 17: Donor-�nanced development expenditures, 2002/03-2008/09, Ksh billions 1999/00 2000/01 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 Grants 4.4 11.6 6.8 14.9 15.8 14.9 20.1 15.5 25.4 33.8 Projects 4.4 5.7 5.4 14.5 11.4 14.9 15.7 15.5 21.7 33.8 Programs - 6.0 1.5 0.5 4.4 - 4.4 - 3.8 0.0 Loans 14.0 13.7 10.0 7.5 11.6 7.2 9.0 10.6 22.5 47.4 Projects 13.9 9.7 10.0 7.5 5.6 7.2 7.4 10.6 21.2 47.4 Programs - 4.0 0.0 0.0 6.0 0.0 1.6 0 1.3 0.0 Total 18.4 25.3 16.8 22.4 27.4 22.1 29.1 26.1 48.0 81.2 Source: Ministry of Finance, Kenya Figure 15: Total donor commitments 2008/09 Source: World Bank staff es�mates 22 A Globally Compe��ve And Prosperous Kenya Figure 16: Government departments with donor-�nanced projects, 2008/09 budget Key: MOR-Ministry of Roads; MOWI-Ministry of Water and Irriga�on; MOE-Ministry of Energy; MOPBHS-Ministry of Public Health and Sanita�on; MOA-Ministry of Agriculture; MOEDU-Ministry of Educa�on; MOF-Ministry of Finance; CAB-Cabinet Office; MOSSP-Ministry of Special Programs; MOPND-Ministry of Planning and Na�onal Development and Vision2030; MOT-Ministry of Transport. Development Budget Performance �ons) that lie beyond the control of their managers. A well managed public spending program would Absorp�on Capacity Problems allow budget managers to switch resources from The development budget has suffered from poor stalled projects to those where implementa�on execu�on rates in the past, raising concerns that can be accelerated, thereby par�ally offse�ng the the budgeted increase in spending would not trans- impact on the overall capital spending program of late into outcomes due to absorp�on constraints. such delays. In countries where a substan�al share Public investment is o�en for projects that are im- of public investment comprises externally �nanced plemented over a number of years and, therefore, projects, levels of under-spending have tended to require a medium-term approach for planning and be signi�cantly higher due to both the complex- public �nance alloca�ons. This makes these invest- ity of development partner procedures and lack ments difficult to accommodate in a single budget of flexibility to switch resources between projects cycle, and raises the risks that mul�-year invest- when delays occur. Over-op�mis�c and unrealis�c ment projects are not sustained over �me, or may budge�ng are also frequent problems. not receive sufficient annual alloca�ons to keep the project on schedule. The annual cycle does not lend The budget process could be improved to ensure itself to planning and procuring for public invest- that high return projects are priori�zed and sustain- ment projects, which consequently results in a low able. The development budget is developed at the execu�on of the budget and eventually leads to se- same �me as the recurrent budget, according to the vere delays in implementa�on and stalled projects. standard budget cycle from 1 July to 30 June. The Even in highly developed budget systems, there is development budget is prepared separately with a tendency to under-spend on the capital program no linkages to the recurrent budget. The budget cy- compared to recurrent or opera�onal budgets. cle is captured in Figure 16 and the key players and Projects may experience delays for a variety of rea- stages in the budget cycle are explained in Appen- sons (such as weather condi�ons or poli�cal disrup- dix Group B, Annex 1. 23 A Globally Compe��ve And Prosperous Kenya Box 2: Public investment management: A note on good prac�ce Drawing on World Bank country case studies on public investment management (PIM) and the devel- opment of a new framework for analysis, we put forward the following stylized rules for good public investment management prac�ce: • Clear centrally-issued guidelines: Guidelines for • Role of Public-Private Partnerships (PPPs): the development of public investment projects, There is general acceptance that PPPs can, for which are then rigorously applied, are a cri�cal a number of reasons, accelerate the delivery of aspect of a well func�oning PIM system. highly valued social services and infrastructure. The availability of PPP-related �nancing and • Provide training: There is need to develop use may need to be considered in the context pragma�c systems based on how the rest of of the availability of relevant exper�se to assess the budget management works, and to ensure PPP arrangements, the poten�al for op�mal adequate training for prac��oners in their use, risk transfer, and value for money considera- and the poli�cal will to enforce the system. �ons. The Australian case usefully highlighted a number of technical issues associated with • Evidence based research: Availability of good managing a PPP por�olio. data is a cri�cal component in the evalua�on of the efficiency of speci�c public investment • To conclude there is general agreement that ef- projects, and for reviewing the performance of fec�ve PIM needs to be supported by the as- PIM systems as a whole. Some countries, most sembly of good data, access to well-trained of- notably Chile and Korea, are able to assemble �cials, and the need for support of government and publish a great deal of investment-related at the highest level. data. Source: PIM Conference Seoul, Korea, 21 November 2008 24 A Globally Compe��ve And Prosperous Kenya Figure 16: The budget process 1. Ministry of Finance 2. Line ministries prepare ini�ates budget cycle budget submissions SWGs 3. SWGs decide sector 6. Parliament discusses priori�es and provide and approves budget recommenda�ons to the EABC. The EABC reviews overall budget ceilings and dra�s BSP 5. Ministry of Finance 4. EBSC prepares BSP and �nalizes budget es�mates oversees sectoral budgets prior to submission to Parliament 25 A Globally Compe��ve And Prosperous Kenya The main de�ciencies in the budget-approval cycle tries in addi�on to direct nego�a�on also have a that relate to the performance of the development publicity campaign to increase the importance of budget are: their programs. • Development projects are not priori�zed accord- • The �nal decision on alloca�ons rests with the ing to economic appraisal, ment budget, projects Ministry of Finance, and this has also led to “cuts� that require more than one year to be implement- in the development budget for projects that the ed are not required to develop medium-term �- line ministry would not have themselves cut. nancing implica�ons (neither capital alone, nor capital and recurrent) consistent with the MTEF On average, one third of the development budget and outer year ceilings. is not spent each year. As Table 18 shows, the ex- ecu�on rate of the development budget has been • Development projects are not clearly iden��able improving from around 40 per cent of the budget in in the development budget es�mates or govern- 2001/02 to over 70 per cent by 2006/07. However, ment accounts, hampering monitoring of devel- budget execu�on rates vary signi�cantly by sector opment project progress. and by year. The execu�on rate in the health sec- tor is especially low, and only 22.4 per cent of the • While the Sector Working Groups (SWGs) no�on- budgeted development expenditures was spent in ally apply a sectoral priori�za�on �lter for devel- 2006/07. Despite the a�empts to improve execu- opment projects within a sector, the direct nego- �on rates, there remain signi�cant problems of �a�ons that take place later in the cycle serve to under-performance in the development budget. negate the relevance of this step. Projects take too long to �nish and ci�zens fail to bene�t from the spending on �me. • The budget is highly contestable, and line minis- Table 18: Development budget execu�on rates, 1999/00-2006/07 1999/ 2000/ 2001/ 2002/ 2004/ 2005/ 2006/ 2007/ 2000 2001 2002 2003 2005 2006 2007 2008 Agriculture and Rural Development 21.4 59.9 78.0 100.3 25.7 60.9 69.0 56.3 Public Safety, Law and Order 47.8 87.2 48.6 48.9 60.5 52.7 55.7 39.0 Health 36.9 23.9 57.8 22.3 13.3 32.7 22.4 26.0 Educa�on 49.4 40.2 52.3 51.4 60.4 66.2 59.3 51.5 Physical Infrastructure 50.6 68.2 28.7 24.6 70.0 72.0 73.3 45.1 General Economic Services 59.3 28.9 62.7 70.6 65.3 45.2 80.5 46.8 Public Administra�on 80.2 87.4 11.4 42.1 82.9 85.9 91.3 97.2 Total Development Budget 45.9 65.5 40.4 42.8 64.6 66.7 71.8 61.6 Total All Budget 87.5 89.5 86.7 89.5 86.7 81.3 88.6 n/a Source: Ministry of Finance (Kenya), as reported in the Concept Note 2007/08 Bank staff es�mates Notes: 1- n/a indicates data unavailable., 2- Figures for 2003/04 Missing 26 A Globally Compe��ve And Prosperous Kenya There is signi�cant devia�on between budgets and Donor disbursements differ drama�cally from what spending (even excluding donor �nancing) at the lev- is budgeted, pulling down development budget ex- el of the spending agency, but this is improving. The ecu�on rates. This is examined in more detail in the 2008 Public Expenditure and Financial Assessment next sec�on, which deals speci�cally with execu- (PEFA) provides informa�on on the performance of �on problems. Although there are problems with the development budget �nanced by development disbursements, there are also serious problems in partners. Across the board, there was an average the repor�ng, which makes assessing the extent of devia�on of 13.2 per cent between development execu�on very challenging. In the Public Expendi- budget and spending in 2006/07. This is down from ture and Finance Assessment (PEFA) 2008, the indi- 36 per cent in 2005/06. There is considerable vari- cators for �nancial informa�on provided by donors a�on, but there are also improvements at the level scored a D+ and concluded that there is insufficient of agencies, par�cularly in infrastructure ministries repor�ng by donors for budget es�mates and ac- such as energy, roads, transport, water and irriga- coun�ng. At least half of the donors provide budget �on. es�mates using their own classi�ca�on, which are not consistent with the government’s budget clas- With some priori�zed effort, the total devia�on si�ca�on. Donors provide quarterly reports within between government development budgets and two months of end-of quarter disbursements for at spending could be further improved. Only six minis- least 50 per cent of the externally �nanced projects tries account for three quarters of the development in the budget, but less than 50 per cent of aid funds budget devia�on. Not all government ministries to government are managed through na�onal pro- have an equal share of the development budget, cedures. and therefore some devia�ons ma�er more than others for the aggregate budget performance. As Part of the problem is over-budge�ng for expected Figure 17 shows, as a share of the total extent of donor disbursements. In fact, there are several re- development budget devia�on, there are six min- por�ng differences that confuse the calcula�on of istries that warrant special a�en�on: Ministry of donor-�nanced project execu�on rates. According Finance, Ministry of Roads, Office of the President, to the Survey on the Paris Declara�on in 2008, only Ministry of Defense, Ministry of Special Programs about 60 per cent of what had been recorded in the and Ministry of Energy, which together comprise al- budget as development partner expenditures was most 75 per cent of the annual development budget due to be disbursed according to the development devia�on. Figure 18 shows the top under-spenders partner’s own records. Of this, as Table 19 shows, in 2004-2007. there was not only a signi�cant varia�on between budgets and disbursements, but there was also a Progress has been made in comple�ng stalled large degree of varia�on between what enters the projects, which might explain the improvements in government’s budget records and what donors be- execu�on rates. A�er the adop�on of the MTEF, the lieve they will disburse, and also between what the government’s policy was to not only increase devel- government has accounted for as having received, opment spending, but also to increase the comple- and what the donor accounts for as disbursements. �on rates in programs and projects. A special allo- The extent of the problem by donor is summarized ca�on of Ksh 2 billion was speci�cally allocated for in Table 19. stalled projects each year from 2003/04, and was es�mated to take a total of Ksh 8 billion. A total of Alignment with Na�onal Goals Ksh 6 billion has been budgeted over the period While Kenya has a well established and ins�tu�on- 2004/05-2007/08, but only about 63 per cent of alized process of drawing up plans to implement funds were absorbed. Although execu�on rates for a na�onal agenda, the process for aligning spend- these projects was below 50 per cent the �rst year, ing with these plans is not so well established. Vi- it increased to 100 per cent in 2007/08, and a total sion 2030, the government’s new long-term de- of 130 projects had been completed by the end of velopment plan, aims to transform Kenya into a that year. This might explain the improved execu- middle-income, industrialized country by 2030. To �on rates observed in Table 18.¹¹ ¹¹ Stalled projects are mainly construc�on of office blocks, staff houses, health facili�es and police sta�ons. 27 A Globally Compe��ve And Prosperous Kenya achieve this vision, a target of 10 per cent annual level to align spending to priori�es. GDP growth will need to be sustained. Under the economic pillar of Vision 2030, six key focus sectors An es�mated Ksh 500 billion is needed for invest- are iden��ed to drive these annual growth targets: ments over the next �ve years. The medium-term (a) tourism, (b) agriculture, (c) manufacturing, (d) plan (MTP) for 2008-2012 captures the policies wholesale and retail trade, (e) business process out- and “flagship projects� by which the goals of Vision sourcing, and (f) �nancial services. In the social pil- 2030 are to be achieved. In order to implement the lars, the focus is on: (a) educa�on and training (b) flagship projects, the government es�mates that health (c) water and sanita�on, (d) environment, Ksh 500 billion will need to be invested, of which (e) housing and urbaniza�on and, (f) gender, youth the government will invest 50 per cent over the and vulnerable groups. The Vision is to be imple- next �ve years. The government expects the other mented through a series of �ve-year medium term 50 per cent of the �nance to come from local and plans. There are various checks during the annual foreign private investors, mainly through Public- budget cycle that budgets are aligned with sector Private Partnership (PPP) arrangements. The MTP and na�onal plans, but this is not well-ins�tu�onal- iden��es priori�es and ac�ons that will require ized, and the direct nego�a�ng between Ministry of public �nancial commitments and these are sum- Finance and line ministries during the process could marized in Box 4. serve to undermine the a�empts made at sector Figure 17: Budget devia�on by ministry Source: World Bank staff es�mates Figure 18: Top 5 under-spenders Source: World Bank staff es�mates 28 A Globally Compe��ve And Prosperous Kenya Box 3: Vision 2030 and the New Coali�on Government Vision 2030 was �rst dra�ed prior to the 2007 star�ng the Vision 2030 period with growth rates General Elec�on. Following the disputed out- of 7 per cent or more (as in 2007), the growth come of the General Elec�ons in December 2007, rate in 2008 was only 1.3 per cent. the poli�cal and socio-economic environment changed signi�cantly. Although Vision 2030 was Furthermore, the 2008 downturn in the global revised in early 2008 and has been adopted by economic environment poses a risk to the strat- the new coali�on government, the revisions ar- egy as the tourism industry globally has taken guably did not take sufficient considera�on of the a down turn, and poten�ally through lower FDI impact and consequences of the post-elec�on and/or remi�ances. So far, neither of these im- violence. pacts have been signi�cant, but the risks remain. Both tourism and agriculture have suffered and Vision 2030 has not, therefore, got off to the will require signi�cant efforts to recover back to promising start envisaged at the �me of dra�- 2007 status. Physical damage to public and private ing. Furthermore, the poli�cal situa�on remains property, and the� of public property (the energy fragile, with a number of issues (e.g. “item four sector experienced losses due to the the� and agenda� of the na�onal peace accord) requiring destruc�on of transformers and other expensive the a�en�on of poli�cians and public servants equipment) are all near-term priori�es that are over the next �ve years before the next elec�on likely to have budgetary implica�ons. Instead of is due in 2012. Table 19: Budgeted and disbursement varia�on, 2006/07 US$ millions Donors scheduled Disbursements Total Of which: for Government disbursements to included in disbursed government accounted as the government government aid sector received sector budget World Bank 182 184 66 66 128 EC 52 101 56 52 47 DfID 65 45 84 37 32 GFATM 67 65 52 52 28 AfDB 90 90 17 16 16 Germany 49 40 29 27 14 Sweden 26 25 35 27 14 France 30 29 25 24 10 Denmark 26 19 38 19 9 Japan 9 9 68 35 8 Others 123 86 270 92 28 Total 719 693 740 447 334 Source: 2008 Survey on Monitoring the Paris Declara�on 29 A Globally Compe��ve And Prosperous Kenya Box 4: Vision 2030, the medium term plan and spending needs FOUNDATIONS Na�onal Terrestrial Fibre Op�c Network Project. Physical Infrastructure: Increase the stock of phys- ical infrastructure, so that there are 64,500km of Financial Services: No “big �cket� expenditure well maintained roads. A total of Ksh 186 billion items iden��ed. is an�cipated for road construc�on and upgrade during 2008-2012. The PPP policy is expected to SOCIAL PILLAR expedite private sector par�cipa�on; e.g. there Educa�on: Construct and fully equip 560 second- will be concessions for main toll roads to be built ary schools, build at least one boarding primary by the private sector. There are plans for a new school in each cons�tuency of the ASAL districts, transport corridor to be built, linking Lamu, Ethio- recruit an addi�onal 28,000 school teachers. pia, Southern Sudan and Somalia, to a second port to be constructed at Lamu through Build, Own, Health: Introduce community-level health units, Operate and Transfer (BoT) arrangements at a supported by a well trained cadre of Community– cost of US$ 15-20 billion. A free port in Mombasa Owned Resource Persons and Community Health will also be developed. The other major transport Extension Workers. infrastructure projects will be the development of a Rapid Bus and light rail system in the Nairobi Water: Rehabilitate and protect forests in �ve wa- Metropolitan area, which is expected to serve as ter towers, implement water storage and harvest- a prototype for the other main urban areas in the ing program, construct two large mul�-purpose country. dams with capacity to store 2.4 billion cubic me- ters, and develop the sanita�on and urban sewer- Energy: Expand the network to connect a million age programs. households at a cost of Ksh 84 billion in �ve years; connect Kenya to the Southern Africa power pool Gender, Vulnerable Groups and Youth: Increase through Tanzania at a cost of US$ 110 million in the Women Enterprise Fund and the Youth Enter- two years; and undertake other projects such as prise Fund; facilitate the training of young people geothermal, solar and wind power. Provide solar in technical, voca�onal and entrepreneurial skills, generators to 74 public ins�tu�ons at Ksh 180 mil- expand polytechnics; and employ youth in labor lion. intensive projects (e.g. road building and tree plan�ng). ECONOMIC PILLAR Tourism: Facilitate the crea�on of 3 resort ci�es. Housing: Provide incen�ves to the private sector to build 200,000 housing units annually by 2012, Agriculture: Irrigate an addi�onal 1.2 million hec- and to individuals through a secondary mortgage tares of land for crop produc�on, of which 600,000 �nance corpora�on. Encourage local authori�es in Arid and Semi-Arid Lands (ASALs), and establish to provide serviced land and construct low cost �ve disease-free zones. housing units through PPP arrangements. Manufacturing: Set up a Special Economic Cluster POLITICAL PILLAR ini�ally in Mombasa, and a second in Kisumu for Governance, Peace Building, and Conflict Man- manufacturing establishments, and set up at least agement: Pilot a na�onal CCT/Camera Surveil- �ve Small and Medium Enterprises (SMEs) Indus- lance project in Nairobi, Mombasa, Kisumu, and trial Parks and Specialized Economic Zones. Nakuru; build an addi�onal 20,000 housing units for Police Staff Housing; and establish a Na�onal Wholesale and Retail Trade: Build a free trade Security Data Center and roll out the Na�onal port at Mombasa, construct wholesale, retail and Community Policing Ini�a�ve. An Independent hawkers markets in selected urban areas. Truth, Jus�ce and Reconcilia�on Commission (TJRC) and a permanent Ethnic and Race Rela�ons ICT and BPOs: Establish a Business Processing Commission of Kenya will commence in 2008/09. Outsource Park and digital villages; implement the 30 A Globally Compe��ve And Prosperous Kenya There is a disconnect between the na�onal plan and pia in the Ministry of Roads sec�on of the printed the sector plans, causing confusion in spending pri- budget es�mates. ori�za�on. The sectors and line ministries are sup- posed to adjust their plans, projects and programs Despite the infrastructure focus of the na�onal goals, to �t with Vision 2030, since it was not a bo�om-up the recent development budget does not show an development and, therefore, there is a difference increase in the share of resources to the sector. It between sectoral plans and priori�es and the na- is possible to determine the shares of the develop- �onal plans. Given the approach taken, it would ment budget according to the MTEF Sector Classi�- make sense for the clearly iden��able priori�es with ca�on. There is li�le change in overall sector com- public �nance implica�ons to be brought to bear posi�on of spending in the last two budget cycles on the budget process. Instead, the medium-term (see Figure 19 below). The need for signi�cant new plan, the sectoral plans and the development budg- public investment in physical infrastructure, while ets are not always consistent. For example, while increasing in real terms, is not evident in a sizeable the Vision 2030 and 2008-2012 medium-term plan increase in the share of total spending, which has iden��ed �ve key water catchment areas as priority actually fallen in the 2009/10 budget. This might areas, the dra� Environment Sector Medium Term reflect the fact that the budget process involves a Strategy for 2008-12 and the budget itself give the fair amount of back and forth between the center �ve areas very low priority for public �nance. Since and the line ministries, which makes the se�ng of the printed es�mates do not show project level in- strategic direc�ons and s�cking to them much more forma�on, the classi�ca�on system is not useful for difficult. It also reflects, to some degree, the extent iden�fying which expenditures are aligned to which to which the environment has changed since Vision of these na�onal (or sectoral) objec�ves. There is 2030 was dra�ed, and the need to focus on equity no men�on, for example, of Vision 2030 aspira�ons issues and on the challenges in the social sectors. for a new transport corridor linking Lamu to Ethio- Figure 19: Sectoral budgets 2008/09 and 2009/10 Source: Government of Kenya, Printed Es�mates Key: P. Admin.-Public administra�on; HRD- Human Resources Development; EWS- Environment, Water and Sanita�on; RIT- Research, Innova�on and Technology; GJLOS-Governance, Jus�ce, Law and Order; TTI-Trade, Tourism and Industry. 31 A Globally Compe��ve And Prosperous Kenya What is also missing from the budget process is the of the disbursement difficul�es in IDA-�nanced link to the �nancing gap required from the private projects. Not all of these were issues for all of the sector through PPP arrangements. It is not clear projects but in total illuminate a very challenging if line ministries are required to priori�ze the PPP environment in improving the effec�ve implemen- projects or to explore this as a funding op�on as part ta�on of exis�ng projects (See Box 5). of the budget process. This is the subject of a paral- lel report being produced by the World Bank, and Project execu�on takes a number of detailed steps, an area for future work since there is much more which require �me. However, several speci�c prob- scope for proac�vity that could unlock the unreal- lems have led to incredibly long delays in disburse- ized poten�al. The government, for example, could ments on both the government and on the devel- ensure that development expenditure is aligned opment partner sides. The government’s internal with the private investments required to produce audit department and the World Bank point to the the flagship projects and other Vision 2030 aspira- following overriding weaknesses: �ons, and that the scope for leveraging private �- nance is ac�vely considered during the budget al- • Long procedures and slow processing of doc- loca�on process. umenta�on. Project Management • Accoun�ng challenges as a result of: (i) in- Poor project management is at the heart of the complete �nancial management systems; poor development budget execu�on rates. World (ii) inadequate accoun�ng for pooled donor Bank (2007b) found that problems of large delayed funds; and (iii) unsa�sfactory value for mon- payments, pending bills and arrears weakened lo- ey, fraud and poten�al corrup�on concerns in cal (technical) absorp�ve capacity. Local contrac- projects. tors, par�cularly in roads, suffer from poor cash flow and were unable to successfully complete • Lack of project staff and high staff turnover on projects on �me, resul�ng in reduced quality and both sides. scope, as well as �me and cost overruns. Overall, projects �nanced 100 per cent by the government • Complex project design, which hampers im- are roughly twice as likely (34 percent of projects) plementa�on. not to be completed on �me as compared to the externally funded projects (16 percent of projects). • Doub�ul project sustainability. Within infrastructure sectors, water resource man- agement projects seem to face more difficulty for The vast majority of contracts in the roads por�olio �mely comple�on (for both government and exter- were awarded without the requisite depth of plan- nally funded projects). The average delay in water ning, feasibility study, site inves�ga�ons, proper de- resource project implementa�on is three years.¹² sign details, and engineering documenta�on (see It is, therefore, not surprising that there are such World Bank 2007c for a review of the project man- capacity constraints in light of the substan�al in- agement issues in the roads sub-sector). The prob- crease in spending under the development budget lems iden��ed in this report include: rushed ini�al observed in earlier sec�ons and shown in Figure 14 designs; delays in star�ng project implementa�on in the execu�ve summary. It appears that the weak because of poor ini�al prepara�on prior to budget absorp�ve capacity will neutralize the poten�al im- entry, leading to �me and cost overrun; insufficient pact of increased spending. �eld inves�ga�ons and oversight mechanisms re- sul�ng in inappropriate engineering interven�ons; Delays in project implementa�on, donor disburse- over-designing; and waste of resources in some cas- ments and u�liza�on of resources are common and es. The report also notes irregulari�es such as the arise from a large number of complex problems. issuing of varia�on orders to ini�ate new projects. Thirty eight weaknesses were iden��ed by a joint Typically, these varia�ons would be added onto on- Government of Kenya-World Bank review (2009) going contracts, regardless of government policies, the project’s own plans and appraisal, and the ca ¹² World Bank (2007b). 32 A Globally Compe��ve And Prosperous Kenya Box 5: Disbursement Problems: The Case of the World Bank The problems causing disbursement delays are complex, with some ins�tu�on-speci�c, while others are project-speci�c. 38 dis�nct problems, iden��ed by a joint Government of Kenya-World Bank re- view (2007c) of IDA �nanced projects are summarized in six major groups below. Problem 1: Limited understanding of the pro- goods and services are procured. cedures. The disbursements procedure for each project is de�ned and documented in a “Disburse- Problem 4: Inability to suitably account for funds ment Le�er�, which forms part of the legal loan due to a weak repor�ng capacity. Since disburse- agreement between the government and the ment approvals are stringently dependent on cor- Bank. This le�er is normally a key part of project rect repor�ng, some delays in disbursement are nego�a�ons. However, those nego�a�ng do not driven by weak repor�ng capaci�es. Cases of dis- always make it available to those implemen�ng bursement requests that have been rejected and the project. As a result, implementers of projects amended over and over again before approval is are not always aware of what needs to be done �nally possible, are frequent. before a disbursement can be “signed-off�. Problem 5: Compounded delays in transferring Problem 2: Not having the right disbursement funds to the project. Projects do not receive an- methods for the project. Several disbursement �cipated funds on �me if a delay occurs in any methods (e.g. direct payments, reimbursements, one, or more, of the following places during the transfers to designated accounts, advances, etc) protracted ‘request-transfer’ process: (i) if Task are available for projects, and they have differ- Managers or the loans department holdup ap- ing purposes, implica�ons, requirements and provals at the World Bank; (ii) ) if the Central Bank complica�ons. Careful selec�on during project (once funds are disbursed by the Bank) does not design supports future success in implementa- release funds from the interna�onal reserve to �on and disbursements. In par�cular, not match- the Treasury, and for the PIUs, fast enough; (iii) if ing a project’s speci�c environment, plans, and the project operates an off-shore account, which capacity to the most appropriate or suitable dis- takes much more �me to bring the funds into Ken- bursement methods holds back project poten�al. ya; (iv) if Treasury does not release funds from its Simpler methods are o�en not used; the more cash reserves to line ministries fast enough; and complicated procedures are more ac�vely pur- (iii) if line ministries decide to allocate in piece- sued. In contrast, the Kenya Educa�on Sector de- meal amounts to project implementa�on units. signed a one-tranche disbursement for the �rst Kenyan parastatals, surprisingly, do not suffer as year of the project (advance payment followed by much from these delays. This is largely because spending accoun�ng), which has supported quick they are much more independent of central gov- disbursements. ernment, and they hold special accounts that are directly managed by the Project Implementa�on Problem 3: World Bank’s and the government’s Units (PIUs). accoun�ng requirements are not always com- pa�ble. Frequently, Bank teams wish to see ac- Problem 6: A Bank structure that’s unhelpful for counts that show how the �nances are aligned task teams. The Bank’s internal organiza�onal with a project’s speci�c objec�ves and with the structure does not allow for daily interac�ons be- classi�ca�on (e.g. goods, works and services) of tween the �nancial management specialists, the the procurements made. However, this approach procurement specialists and the loan officers (re- is not consistent with government’s classi�ca�on sponsible for disbursements). In part, this is con- system. Moreover, it is difficult to dis�nguish the sciously designed to ensure a clear division of ac- Bank’s funds from that of other donors in basket countabili�es between the approving, spending arrangements, or even from government’s own and release of funds roles. However, in prac�ce, funds. Bank-supported disaggregated policy ob- this generates considerable confusion and delay jec�ves are o�en not the “result� alongside which for the task teams. 33 A Globally Compe��ve And Prosperous Kenya por�olio that may not have been well appraised, or pacity of the contractor. The roads sector review also notes that there are no independent gate-keepers aligned with ministerial priori�es show problema�c to screen project proposals, and decide whether a implementa�on and spending, largely due to: (i) in- project meets the necessary level of prepara�on or sufficient a�en�on given the project’s low policy/ economic jus��ca�on. Furthermore, it is not easy priority status; or (ii) the difficul�es in implement- to make judgments on the projects iden��ed or ing a poorly appraised and designed project. Other selected when an overall road sector development projects that are equally problema�c are those that policy and plan is not available. spread their budgets too thinly across geographical loca�ons to accommodate the social-poli�cal con- Implementa�on of public investment projects cerns of poten�al bene�ciaries. are o�en delayed by the lengthy and late started procurement processes. Since a project is not ap- Improving the Role of the Development Budget proved un�l it is approved by Parliament, no sub- A piece meal or quick-�x approach will not resolve stan�al project prepara�on, including procurement the development issues iden��ed above. Instead, processes, can be effected in advance. Ministries, a comprehensive strategy is needed and one that for example, cannot �nance preparatory feasibility addresses, among other considera�ons, the follow- studies un�l the budget is approved. This means ing speci�c recommenda�ons and conclusions from that the procurement processes, which typically this report: take several months from ini�a�on to conclusion, (a) Ensuring consistency between sectoral and are launched during the project’s core implemen- na�onal priori�es. There is s�ll a need to re- ta�on period, thus limi�ng the project’s spending view the consistency between the medium ability from year one. Spending is also compro- term plan and the sectoral plans. In order to mised when annual budget ceilings do not allow a get spending in line with priori�es, it is �rst project to undertake all its required procurement important that the priori�es are clear, and processes. And even if supplementary funds are currently they are not. requested (typically submi�ed half-way through a �scal year), the supplementary budget is frequently (b) Strengthening spending alignment with na- approved at the end of the year (usually during the �onal priori�es. The various departments last two months). This, of course, does not give the within Treasury and in the Ministry of Plan- project �me to ini�ate, complete and reflect any ning and Na�onal Development departments procurement spending before year-end. Finally, need to work together to ensure that budg- where development partners are involved, addi- et alloca�ons are consistent with the stated �onal procedures and actors further complicate na�onal priori�es. For example, the Vision projects’ procurement implementa�on and spend- 2030 flagship projects that are brought into ing capacity. the �ve-year medium term plan should be screened and priori�zed for new discre�on- Inadequate capacity in project management and ary development budget alloca�ons. lack of technical staff are also key causes for poor (c) Improving the role of public �nance as lever- implementa�on of the development budget.¹³ The age. The extent to which new projects receive crea�on of new administra�ve districts has spread public �nancial backing needs to be looked at the limited resources available even thinner, par- in the context of leveraging private �nancing. �cularly in provincial administra�on and services The PPP policy needs to be �nalized, and the in educa�on, health and agriculture. There is a guidance for taking this forward brought into high turnover of staff and/or understaffing in many the annual budget cycle. project units, and the few technical and project management staff available are stretched over all (d) Strengthening spending alignment with sec- projects. toral priori�es. Within sectors, the SWGs are required to make the necessary trade-offs and Inadequate project appraisals or designs lead to provide advice on sectoral policies. Treasury poor implementa�on and spending. Projects in the should discon�nue the prac�ce of reoreopen- ¹³ Kirira (2009) 34 A Globally Compe��ve And Prosperous Kenya ing up budget alloca�on discussions with in- for all proposed development projects. This, dividual line ministries once the SWGs have and other cri�cal requirements, could be pre- completed their recommenda�ons. Instead, sented in clear and mandatory guidelines that any addi�onal resources should be given to are issued alongside the budget circular. Ex- the SWGs to decide on priority realloca�ons. amples of other appraisal requirements that would be listed in the guidelines include: in- (e) Proac�vely dealing with stalled projects. The forma�on on es�mated project costs (both progress with implemen�ng stalled projects �xed and recurrent over a reasonable �me- should be maintained, and the remaining frame), the likely bene�ts of the project, the ones reviewed with a view to hal�ng projects alterna�ve interven�ons considered, the envi- that are no longer a priority. ronmental and social impacts (if any), and the (f) Improving implementa�on of sectoral strate- staffing and implementa�on arrangements. gies. The SWGs should be required to report Such guidelines, including templates for some to the Economic and Budget Steering Com- of the informa�on sought, could be adapted mi�ee (EBSS) on the extent to which their from countries that have already adopted the budget submissions meet the targets laid out approach. in their own sector plans, and the na�onal medium-term plan. (i) Improving the pace of implementa�on. Project budgets, and especially those with mul�-year (g) Introducing more effec�ve project screening implementa�on and expenditures, should be and appraisal. A standard format and screen- protected and approved. There should be an ing process for the proposed new projects is assump�on in the MTEF, going forward, that needed in the budget circular. This should be exis�ng commitments will be honored and designed in a way that ensures that all pro- this should come off the top before ceilings posed new projects are fully assessed for their are issued for the development budget. This coherence with sectoral plans, the MPER rec- would help get rid of projects that have stalled ommenda�ons and the iden��ed priori�es or are taking too long to deliver because of in- across the sector (rather than across minis- sufficient budget alloca�ons. tries) before inclusion in the budget. (j) Accelera�ng implementa�on and execu�on. (h) Improving and standardizing project apprais- Prepara�on of technical drawings, tender al and design. A robust economic appraisal, documents and feasibility studies should all and one that draws relevant lessons and ap- start long before budget approval so that im- plica�ons from other parts of the country and plementa�on can begin at the start of the �s- the world should be a minimum requirement cal year, and at the start of the project. 35 3. PRO-POOR SPENDING IN KENYA: A REVIEW OF THE TARGETED SUBSIDY PROGRAMS Background food prices, low economic growth, and unemploy- Recovery from post-elec�on violence experienced ment, the number of poor and vulnerable Kenyans in the �rst quarter of 2008 was muted by new exog- will increase un�l the economy is on a path to full enous shocks later in the year, including the drought recovery. In 2008, for instance, real GDP growth of and global �nancial crisis. Economic growth slowed 1.7 per cent and popula�on growth of 2.7 per cent to 1.7 per cent in 2008, down from 7 per cent in meant a decline in per capita incomes of around 1 2007, and is projected to be only 3 per cent in 2009. per cent.¹� Infla�on rose ini�ally due to the violence-driven dis- rup�ons to supply chains and rising global energy In 2006, poverty es�mates based on the Kenya Inte- prices, and more recently due to rising interna�onal grated Household Budget Survey (KIHBS, 2006/05) food and fer�lizer prices. Overall month on month indicated that 16.3 million Kenyans were food-poor infla�on was 16 per cent in 2008 and 9.4 per cent and could not meet the cost of a basic food bun- in 2009.¹� dle. The World Bank’s Kenya Poverty and Inequality Assessment (KPIA) showed that although poverty Kenya remains vulnerable to further domes�c and declined between 2000 and 2006, the poverty inci- external shocks, as it does not have assured sources dence is s�ll high at 47 per cent. Furthermore, rural of external �nance from either official or private poverty is much higher at 49.7 per cent compared to sources. Food supply shortages will maintain in- 34.4 per cent in urban areas. The na�onal numbers fla�onary pressure in 2009, at least un�l the �rst mask important regional differences; in the Coast harvest, which takes place in the last quarter of the and North Eastern provinces, poverty incidence is year. Increased imports of maize will deplete for- es�mated at 70 per cent and 74 per cent, respec- eign exchange reserves at a �me when export earn- �vely, compared to 22 per cent in Nairobi and 31 ings, remi�ances and capital inflows are all damp- per cent in Central Province.¹� However, although ened by the nega�ve impact of the global �nancial North Eastern Province has the highest level of pov- crisis. As a result of the combined effect of high erty, the province only contributes 4.9 per cent to ¹� The popula�on growth rate in 2008 is es�mated at 2.7 per cent (Economic Survey, 2009). ¹� Current es�mates by the Kenya Food Security Group (KFSG) indicate that about 10 million people in Kenya are food insecure and vulner- able to domes�c and external shocks. The es�mates indicate that out of the 10 million, 4.1 million are in urban areas. These es�mates are compiled based on the Early Warning System (EWS), which is based on rainfall pa�erns, and updated every six months. ¹� De�ned as consump�on levels below the Minimum Dietary Energy requirements (MDER), i.e. the amount that is considered adequate to meet individual daily energy needs for light ac�vity and good health for age and sex. ¹� Food Insecurity Assessment in Kenya: Based on Kenya Integrated Household Budget Survey 2005/06, KNBS 2008. 36 Pro-poor Spending In Kenya the total number of poor since it has a low popu- line at 15 per cent for urban areas and 22 per cent la�on. The provinces with the highest numbers of for rural areas.� The lowest incidence of cri�cal food poor people are Ri� Valley (25 percent of the poor),poverty was in Nairobi at 2 per cent, and the highest and Eastern (17 percent), Nyanza and Western (14 in Western Province at 35 per cent (see Figure 21). percent each)—see Figure 20. However, none of the es�mates cited have been up- dated to take into account the external and domes- Furthermore, data from KIHBS show that 51 per cent �c shocks that have taken place since 2005/06. of the popula�on is undernourished,¹� and that the prevalence is much higher in the rural areas at 57 In an a�empt to cushion the vulnerable against ris- per cent compared to 39 per cent in urban areas.¹� ing food prices, the government introduced a gen- The report on Food Insecurity Assessment (2005/ eralized maize subsidy scheme in November 2008. 06) in Kenya es�mated the number living below the The scheme had two components; the �rst was a cri�cal food poverty line at 15 per cent for urban policy to sell maize to millers through the Na�onal areas and 22 per cent for rural areas. The lowest Cereals and Produce Board (NCPB) at below-market incidence of cri�cal food poverty was in Nairobi at prices. The millers were then expected to pass-on 2 per cent, and the highest in Western Province at the subsidy and sell the flour at a price below pre- 35 per cent. However, none of the es�mates cited vailing retail prices to everyone regardless of income have been updated to take into account the exter- status. However, rather than sell directly to the mill- nal and domes�c shocks that have taken place since ers, the NCPB sold the maize to brokers, who then 2005/06.Assessment (2005/06) in Kenya es�mated sold to the millers. Ini�al reports also indicated that the number living below the cri�cal food poverty there was leakage across borders.²� Figure 20: Regional poverty levels (KIHBS 2005/06) Source: Kenya Poverty and Inequality Assessment (KPIA), World Bank (2009) Notes: Size of bubble shows contribu�on to total poverty ¹� The propor�on of the popula�on whose income is lower than the cost of a macronutrient balanced food basket equivalent to minimum dietary energy requirement. ²� At the �me of dra�ing this policy note, a forensic audit was in progress. 37 Pro-poor Spending In Kenya Figure 21: Prevalence of cri�cal food poverty (KIHBS 2006/06) Source: Food Insecurity Assessment in Kenya: Based on Kenya Integrated Household Survey 2005/06, KNBS, April 2008 As a result of design flaws, the generalized food men�ng and funding agencies. The programs re- subsidy scheme implemented by the government viewed are listed in Table 20,²¹ with a more detailed in 2009 was too expensive. As a generalized sub- descrip�on provided in Table 21. The data provided sidy scheme, the programme was designed to be by agencies includes total popula�on coverage dis- propor�onal, conferring flat-bene�ts to the rich aggregated at district level. Some of the programs and the poor, although the poor would have likely target households, while others target individuals. gained more as a propor�on of their income or ex- To facilitate comparison and analysis, household penditures. A�er a cri�cal analysis, the scheme was coverage was converted to individual coverage, as- withdrawn by the government. However, the gov- suming an average household size of seven in the ernment also made a commitment to develop an ASAL areas where most of the programs are located. alterna�ve that would be more efficient and be�er District level data are aggregated to the provincial targeted. level for presenta�on purposes.²² For World Food Program (WFP) implemented programs, the budg- This review of the exis�ng targeted programs is the ets are based on tonnage delivered and calculated �rst step towards the process of designing a new based on es�mated overall costs. Therefore, only program. It iden��es the gaps in coverage by cat- average total cost per ac�vity was provided rather egories and geographical representa�on, which all than at district level. To get a proxy for spending at need to be taken into considera�on. It covers 14 district level, we divided the total spending by the major programs providing in-kind or income sup- popula�on covered in each district. port through cash or workfare interven�ons. It also illuminates the popula�ons and geographic areas Kenya has a high number of poor and vulnerable reached by the exis�ng programs (�scal year 2008/ individuals and households, but social safety nets 09). Finally, the review offers some concluding in- to cushion the poor are limited and fragmented. sights and lessons from global experiences. Although this review is not exhaus�ve, as it covers only 12 ongoing programs, it shows that those cov- Overview of Exis�ng Programs ered by targeted programs amount to a popula�on This note reviews 12 targeted cash and in-kind sub- of about 6.6 million people (Table 20). A summary sidy programs based on data provided by imple- of the programs is provided in Table 21. ²¹ The Table has 14 programs but data on special programs and hospital waiver schemes is not yet available. ²² The district level data is available from the authors on request. 38 Pro-poor Spending In Kenya Table 20: Targeted schemes in Kenya: Popula�on coverage (numbers’000) Program Districts 2008/09 Individuals Scale up ²� Households ‘000 Individuals ‘000 OVC Cash Transfer 1. Program 47 44,668 134 376 Hunger Safety Net 2. Program 4 12,000 60 360 Food Distribu�on 3. Emergency Opera�ons 26 2,581 2,581 Regular and Expanded 4. School Feeding 19 1,076 1,076 Homegrown School 5. Feeding 28 743 743 Most Vulnerable Children 6. Supplementary Feeding 200 200 and Mother and Child 7. Health Program 20 340 340 HIV/Aids Nutri�on 8. Feeding 4 77 77 9. Voucher Schemes 3 - 10. Njaa Marufuku Kenya 31 31 Na�onal Accelerated Agricultural Inputs Access Programme 11. (NAAIAP) 53 626 626 12. Special Programs - - 13. Kazi kwa Vijana 35 80 14. Support to the Elderly 3 1.8 Total 56,668 6,065 6,605 Source: World Bank staff es�mates ²³ The food ra�ons vary by type of district (arid districts receive more than semi-arid districts). However, for simplicity, we assumed uni- form ra�ons, but the overall total remains as provided by WFP. Therefore, the es�mates slightly underes�mate the share of bene�ts received in arid districts. ²� This is a rough es�mate, assuming the average household size is seven. From this review, it is not possible to know if some households are covered by different programs, which would imply there could be an element of double coun�ng. ²� Scale up details available for OVC and Hunger Safety Net. For the other programs, we retain the �scal year 2008/09 numbers. 39 Program Lead Agency Objec�ves Targe�ng criteria Sources of Implemen�ng Delivery Coverage Funds agency 1. Orphaned and Ministry of Gender Increase school enrollment Extremely poor households with Government; Ministry Cash transfer Ksh 47 districts; 44,000 Vulnerable and Children and OVC UNICEF; DfID 1,500 per month households Children (OVC) Social Development Reduce child mortality and World Bank morbidity Promote House Hold nutri�on 2. School Feeding Ministry of Educa�on Increase enrollment DFID Government; DfID; WFP Food 19 districts; 1 million Program World Bank children Ensure children stay in school 3. Most Vulnerable Ministry of Educa�on Retain vulnerable children in school High prevalence of HIV/AIDS DfID; World Bank; Ministry Cash Ksh 100,000 4 schools in each Children UNICEF per school per year district Return MVC dropouts back to High numbers of MVC school Poor health and nutri�on Enroll MVC indicators High poverty incidence 4. General Food Ministry of Northern Targe�ng chronically food Geographic vulnerability WFP Food 26 districts; 2.5 Distribu�on Kenya and Other Arid insecure households in ASAL million people (Emergency) Lands areas Community based 5. Hunger Safety Net Ministry of Northern Provide regular, predictable and Community-based social DfID Oxfam Cash transfer of Ksh 4 districts Kenya and Other Arid guaranteed amount of cash to pensions dependency ra�o HELPAGE 1,075 per month 60 000 households Lands chronically food insecure house- when fully scaled up holds 6. Food for Assets WFP Food ra�ons Same as general food 7. Njaa Marufuku Ministry of Agriculture Provide regular, predictable and Community-based social DfID Cash transfer of Ksh 4 districts guaranteed amount of cash to pensions dependency ra�o Oxfam 1,075 per month 60 000 households chronically food insecure house- HELPAGE when fully scaled up holds 8. Na�onal Accelerated Ministry of Agriculture Improve household/na�onal food Small scale resource poor Government of Ministry Vouchers for inputs 53 districts; 89,000 Agricultural Inputs security and re-investment farmers growing maize Kenya; World Bank worth 8,300 for households Access Programme plan�ng season (NAAIAP) 9. Supplementary WFP Food ra�ons 20 districts; 340,000 feeding and mother people and child health program Table 21: Selected targeted programs in Kenya 10. HIV/AIDS nutri�on WFP Food ra�ons 4 districts; feeding 77,000 people 11. Voucher schemes Ministry of Health Improve maternal and neonatal Community-based social Government of Vouchers health pensions dependency ra�o Kenya; UNICEF Reduce mortality and morbidity 12. Kazi kwa Vijana Office of the Prime To provide immediate relief to Unemployed youth 18-35 years Government of Various Daily wage: Ksh 223 35,000 youth Minister young people from the current Kenya ministries for urban, Ksh 205 crisis through employment in municipali�es, public works. Ksh 125 for other 13. Support to the Ministry of Gender and Provide income support to the Government of Ministry Cash transfer 3 districts,; 300 elderly Children, and Social elderly Kenya Ksh 1,000 per people Development month ²� The coverage will increase to 120,000 households by 2012. ²� Includes homegrown school feeding program expected to start in May 2009. ²� Bank funding is a one-off disbursed in �rst half of 2009. 40 Pro-poor Spending In Kenya ²� Details on the program are s�ll being worked out. The informa�on provided here is tenta�ve. Pro-poor Spending In Kenya The exis�ng programs can be grouped into three ing administra�ve costs) will increase to about broad groups based on their method of delivery; Ksh 2.3 billion by 2013. i.e. in kind transfers, cash transfers, and voucher schemes. The following sec�on describes the vari- • HSNP: Launched in the �rst quarter of 2009, this ous programs that exist within these three groups. is a ten-year uncondi�onal cash transfer program providing Ksh 1,075 per month (transferred every a) In-kind Transfers. Emergency Relief Opera�ons two months) to extremely poor and vulnerable (EMOP) are rou�nely undertaken as part of a hu- people in arid and semi-arid districts in Wajir, manitarian response during food shortages, and Mandera, Turkana and Marsabit. The program is cover the chronically food-insecure popula�on in on a pilot phase, with a coverage of about 12,000 the ASALs. The review shows that the most signi�- households, but it is projected to cover 60,000 cant programs in this category are the emergency households, or about 300,000 people by Decem- relief opera�ons for the arid regions and the school ber 2009. The program will be scaled up gradu- feeding program, which jointly cover about 4.4 mil- ally to cover 300,000 households, and 1.5 million lion people (including the homegrown school feed- people by 2017. The program is funded by DfID ing program, which starts in May 2009). The emer- with a total grant of £80 million over 10 years. gency opera�ons (EMOP) have been implemented since 1989. Table 21 provides a quick overview of • Homegrown School Feeding Program: In a de- the objec�ves and the targe�ng criteria of the on- parture from food ra�ons, this program transfers going programs. The EMOP programs are concen- cash to qualifying schools in areas where markets trated in the rural areas, par�cularly in the ASALs, func�on; i.e. where food is available in the mar- with limited interven�ons targe�ng the vulnerable ket. The schools are expected to use the funds to in the urban areas. The programs are implemented purchase food, which will be cooked and served by WFP, on behalf of the government and develop- in school. It is expected that this new approach ment partners. will reduce the administra�ve costs of the school feeding program and provide incen�ves for in- b) Cash Transfer Programs. The review covers six creased food produc�on. The program, covering cash transfer programs: the Orphans and Vulnera- 28 districts and about 700,000 children, is funded ble Children (OVC), the Hunger Safety Net Program by the Government of Kenya, and in �scal year (HSNP), support to the elderly, homegrown school 2008/09 bene�ted from a one-off JICA grant. feeding program, Most Vulnerable Children (MVC), and the Njaa Marufuku program. The �rst three • Support to the Elderly: This program is imple- programs men�oned target households, and the lat- mented by the Ministry of Gender and Children ter three target schools. Four of these schemes are and Social Development. It transfers Ksh 1,000 described in more detail below. monthly to 300 households in three districts. • OVC: This is a cash transfer program conceptual- c) Voucher Schemes. The Na�onal Accelerated ized in 2002 and piloted from 2004, with 7 districts Agricultural Inputs Access Program (NAAIAP), un- and 17,000 households covered. An expanded pi- der the Ministry of Agriculture, provides vouchers lot started in 2007 covering an addi�onal 30,000 worth Ksh 7,300 to small-scale resource poor farm- households. Funding is received from Govern- ers growing maize to boost food produc�on. The ment of Kenya, UK’s Department for Interna�on- Government of Kenya funds the program, and the al Development (DfID), and the United Na�ons World Bank provided a one-off grant of US$ 5 mil- Children’s Fund (UNICEF). Further scale up is in lion in 2008/09. the pipeline from World Bank funding, which will increase coverage to 125,000 households by the Figure 22 shows the programs with poten�al for year 2013. At the current cash transfer of Ksh scale up. 1,500 per household, the annual budget (exclud- 41 Pro-poor Spending In Kenya Table 22: Programs with poten�al for scale up EMOP Has the widest popula�on, and signi�cant geographic The program has high administra�ve coverage spanning 26 districts costs Has been opera�onal for several years Requires an urban model The implemen�ng agent, WFP, has wide experience in targe�ng food aid Has signi�cant support from both Government of Kenya and donors The Regular and Program ranks second in popula�on coverage The program has high administra�ve Expanded School costs Feeding Program The implemen�ng agent, WFP, has wide experience in targe�ng food aid Limited coverage in the slums where most of the schools are informal Retains children in school The Homegrown The program is likely to be more efficient than the The program has just started, and School Feeding regular school feeding program requires to be reviewed before it can Program be scaled up Has a signi�cant geographic coverage in 28 districts OVC Cash Transfer The program has the widest geographic coverage The program targets only households in 28 districts with OVCs Has developed a good system for targe�ng Has been in opera�on since 2002 and successfully piloted Has signi�cant support from donors and uses government systems Retains children in school Hunger Safety Net Has developed a comprehensive mechanism for The program is rela�vely new, s�ll in targe�ng the pilot phase Has an excellent Management Informa�on System It has limited geographic coverage in only 4 ASAL districts Kazi kwa Vijana The program is currently covering 35,000 youth The program is rela�vely new, s�ll in the pilot phase Enjoys high poli�cal support Bene�ts only the youth Its projects create/maintain public assets The program is expensive ³� Source: World Bank staff es�mates ³� The expenditure by June 2009 shows that 33 per cent of the total cost of the program was spent on wages for the youth. 42 Pro-poor Spending In Kenya Expenditure Es�mates 2008/09 The total spending on the targeted programs re- The total spending from the programs reviewed viewed is about 0.9 per cent of GDP, with the Gov- here for �scal year 2008/09 was about Ksh 22 bil- ernment of Kenya contribu�ng about Ksh 5 billion lion (see Table 23), and the same amount is es�- of this, or 23 per cent. The Government’s contribu- mated for 2008/09. This total captures the 14 pro- �on translates to about 0.2 per cent of GDP or 0.9 grams and covers both Government of Kenya and per cent of total government revenue, and about donor contribu�ons, and includes the recently 0.7 per cent of total expenditure. launched Kazi kwa Vijana program with about Ksh 3.4 billion. Total spending includes both the wage Global es�mates indicate that spending on social and non-wage expenditures, for example purchase safety nets varies substan�ally by country, but aver- of seedlings under the component implemented by ages about 1.9 per cent of GDP. The programs re- the Ministry of Youth Affairs and Sports. viewed here show that Kenya is currently spending about 1.0 per cent of GDP (including external fund- Figure 22 shows the budget and popula�on shares ing). Low income countries prone to poor harvests, covered by the programs under review (excluding such as Malawi and Ethiopia, spend 4.5 per cent of Kazi kwa Vijana). The most signi�cant programs are GDP when external �nancing is included (see Table the emergency relief opera�ons, the school feed- 25). However, when external aid is excluded, ex- ing program, and the NAAIAP. When the regular penditure on social safety nets in the two countries and homegrown school feeding programs are com- is es�mated at about 0.5 per cent of GDP. Although bined, they account for about 27 per cent of the to- Kenya has other poverty reducing programs, such tal popula�on covered. as the Cons�tuency Development Fund (2.5 per- Table 23: Expenditure levels by program, 2008/09 (Ksh million) Program Government Donor Total of Kenya OVC Cash Transfer Program 95 803 898 Hunger Safety Net program - 155 155 Food Distribu�on: Emergency Opera�ons 258 10,025 10,283 Regular and expanded school feeding 206 2,387 2,593 Home grown school feeding (es�mates) 400 180 580 Most Vulnerable Children (MVC) - 1 1 Supplementary feeding and mother and child health program - 1,833 1,833 HIV/AIDS nutri�on feeding - 640 640 Health voucher schemes - Njaa marufuku Kenya 128 30 158 Na�onal Accelerated Agricultural Inputs Access Program (NAAIAP) 300 400 700 Special programs - Kazi kwa Vijana 3,400 - 3,400 Support to the elderly 4 - 4 Total 4,790 16,453 21,243 Source: World Bank staff es�mates 43 Pro-poor Spending In Kenya Figure 22: Targeted programs popula�on and shares of total expenditures 2008/09 (percent) Notes: 1. The budget in the WFP programs includes administra�ve costs, which are not included in the other programs. 2. The Kazi kwa Vijana program is not presented here because different ministries present the data in different ways. Further work is required to make the data comparable. Table 24: Fiscal opera�ons (Ksh billions) 2008/2007 GDP 2,299 Total revenue 513 Total expenditure 673 Targeted Programs 21 Government of Kenya 4.8 Donor 16 Targeted Programs: Ra�os Percent of GDP (Total) 0.9 GoK contribu�on Percent of GDP 0.2 Percent of total Expenditure 0.7 Percent of Revenue 0.9 Source Budget Outlook Paper 2009 Notes: 1. Shares decline in the medium term because GDP and total expenditure growth rates are higher than spending on targeted programs. 44 Pro-poor Spending In Kenya cent of revenue) and other core poverty programs, ing by iden��ed vulnerable groups and in the sec- government spending on direct social safety nets ond sec�on we map regional spending by regional is rela�vely low, but not signi�cantly different from poverty pro�les. the spending by other governments in the region. Categorical coverage³¹ Global experiences and research show that a well- The analysis shows that ASAL communi�es account designed cash transfer program with sufficient for about 70 per cent of total expenditure of food coverage can: (i) reduce chronic poverty while pro- and income-support targeted schemes (see Figure grams to alleviate the causes are being put in place; 23). OVCs also receive signi�cant a�en�on, covered (ii) reduce transitory poverty for households faced by �ve of the programs reviewed. They account for by unmanageable shock(s); and (iii) overall, prevent 14 per cent of the vulnerable popula�on but, from long term consequences of poverty, such as hunger this analysis, receive 26 per cent of the funding. and malnutri�on, which are especially damaging to However, as indicated in Table 21 above, the OVC children. and MVC programs have mul�ple objec�ves, and the key is to enroll and retain children in school. Several instruments are available in the delivery of food subsidy schemes: in-kind food transfers, vouch- The pa�ern of spending is mainly informed by the ers and food stamps, workfare and cash transfer vulnerability assessment from the Kenya Food Se- programs. Although cash transfers have been iden- curity Group based on the Early Warning System, ��ed as the most efficient, alterna�ve instruments which does not comprehensively cover the urban are jus��ed where markets do not work, or where areas. the assistance is only available in-kind, and where targe�ng is difficult. Administra�ve costs are usual- Regional coverage ly higher for in-kind transfers: the programs require In this sec�on, we map spending versus needs transport, storage, and other administra�ve costs, by regions. We use the regional poverty pa�erns and the commodi�es are worth less than the over- based on KIHBS 2005/06.³² We map the data to- all cost to government. Workfare programs provide gether with spending data summarized in Sec�on 2 a good safety response during crisis and, although in Figures 24 and 25. Figure 17 shows the number they create assets, they can be rela�vely expensive of poor people by province based on KIHBS data, and their efficiency can only be determined by the and the expenditure per capita by province. Figure quality of assets created. 25 shows the poverty incidence and total spending by province. The two �gures show that the level of Spending versus needs spending is guided by the poverty incidence, and In this sec�on, the review matches needs in two di- not by the total number of poor people. mensions; in the �rst sec�ons, we aggregate spend- Table 25: Regional spending social safety nets Social Assistance Expenditures Total percent of GDP Year Senegal 0.2 2004 Madagascar 0.9 2002 Kenya 0.9 2008/09 South Africa 3.2 2002/03 Malawi 4.4 1999-2000 Ethiopia 4.5 2001/02 Mauri�us 5.3 2001/02 Source: World Bank staff es�mates ³¹ Further work is required to iden�fy categories of vulnerable groups without double coun�ng. This review is tenta�ve un�l the groups are clearly iden��ed. 45 Pro-poor Spending In Kenya Figure 23: Spending on different categories of vulnerable groups Source: World Bank staff es�mates The overview in Sec�on 2 indicated that Nairobi programs in Kili� and Malindi districts, while the av- has the lowest number of poor people, followed by erage number of programs for ASAL areas is about North Eastern and Central provinces (see Figure 21). 4 per district. Thus, although the poverty headcount is very high in North Eastern Province, the number of poor peo- Comparing the results with analysis based on house- ple is rela�vely low, since this is a low popula�on hold data (KIHBS 2005/06), we �nd that 55 per cent density province. However, the region has limited of the households in North Eastern Province, 16 per economic opportuni�es and is highly vulnerable to cent in Coast, 19 per cent in Eastern, and 2.4 per weather pa�erns, and this explains the coverage by cent in Ri� Valley received transfers from the gov- the EWS assessment. ernment (see Table 26). The household level data closely mimics the spending pa�ern on transfers The analysis shows that most of the exis�ng pro- emerging from this review. Households in Nairobi, grams are mainly concentrated in four provinces: Western and Nyanza provinces are in the lower end Coast, North Eastern, Ri� Valley and Eastern. How- of the scale, where 1.6 per cent of the households ever, from Figure 24 and 25, it is evident that there in Nairobi received transfers, 2.7 per cent in West- is clear mismatch between the number of poor ern and 5.2 per cent in Nyanza. people and the level of funding in Nyanza and West- ern provinces. The spending on these programs is The summary in Table 26 also shows that the av- skewed in favor of North Eastern and Coast prov- erage transfers in Nairobi and Nyanza are rela�vely inces, where Ksh 3,000 and Ksh 1,300 per capita high, but the coverage is limited: 1.6 per cent and 5.2 (respec�vely) is spent annually on the programs per cent, respec�vely. A compara�ve analysis of the reviewed here. On the low end of the scale is Cen- receipts by poor and non-poor households shows tral and Nyanza provinces, where annual per capita that it is only in Nyanza and North Eastern provinces spending is about Ksh 42 and 51, respec�vely (see where the average transfer to the poor households Figure 24). is much higher than the average transfer to the non- poor households (see Appendix Group C, Annex for There is a mismatch between regional spending transfers by quar�le and province). per capita and the levels of poverty and vulnerabil- ity. Spending per capita is highest in North Eastern Rural versus Urban Coverage and Coast provinces. Districts in the Coast have the The data provided is not disaggregated by rural- highest number of programs, for example with 7 urban regions, so we use spending in Nairobi and �� The propor�on of the popula�on whose income is lower than the cost of a macronutrient balanced food basket equivalent to minimum dietary energy requirement.m. 46 Pro-poor Spending In Kenya Mombasa as proxy for spending in urban areas. Nai- urban coverage. The limited coverage can be a�rib- robi is mainly covered by WFP programs, with a to- uted to lack of a comprehensive na�onal policy and, tal coverage of about 170,000 people. Mombasa is as a result, coverage is mainly ad hoc and informed covered through the school feeding program, with a by the early warning system that covers only the total coverage of about 50,000 people. Total spend- arid and semi-arid lands. There is also limited un- ing in both regions is about Ksh 800 million, or 5 per derstanding of the appropriate mechanisms for cent of total spending.�� reaching the poor in urban areas, and this could ex- plain the limited response by both the government The review demonstrates that there is a strong ru- and donors. ral focus in the provision of safety nets and limited Figure 24: Regional poverty (numbers ‘000) and Figure 25: Poverty incidence and total targeted expenditure per capita spending by province Per ca ita (Ksh) - Notes: Regional coverage excludes Kazi kwa Vijana; the data requires Source: World Bank staff es�mates clearer mapping Table 26: Average government transfer in Kenya by province and poverty status Average for Percent of household with Average for household with household receipts only receipts only receiving Poor Non-poor Nairobi 2,956 1,000 3,105 1.6 Central 955 613 984 16.9 Coast 2,316 688 2,649 15.7 Eastern 1,616 1,946 1,576 19.2 North Eastern 3,729 4,045 3,563 54.6 Nyanza 2,930 4,055 2,654 5.2 Ri� Valley 1,057 837 1,215 2.4 Western 182 1,067 1,974 2.7 Source: Computa�on from KIHBS 2005/06 �� Excluding Kazi kwa Vijana. 47 Summary and Conclusion From this review, it is evident that Kenya has a sig- skewed in favor of crisis-prone geographic areas ni�cant number of programs funded and managed with limited a�en�on to other vulnerable groups, by different agencies, but coordina�on is limited. par�cularly the urban poor. In Kenya, the scope and While donor interven�ons close the gap where coverage of a new program is likely to be constrained there is limited government ac�on, the current ap- by limited �scal space. On the other hand, inade- proach increases administra�ve overheads and lim- quate response to cushion the vulnerable groups, its the poten�al for efficiency gains through econo- par�cularly in strengthening their household coping mies of scale. The government led ini�a�ve in the mechanism during crisis, can delay future prospects design of a new-targeted program provides an op- for poverty reduc�on, for example when children portunity for reforms, and be�er coordina�on and nutri�on requirements are not met, or children harmoniza�on among development partners in line drop out of school to look for work. With the high with the principles of the Kenya Joint Assistance number of poor in Kenya, it is unlikely that a subsidy Strategy (KJAS). scheme can comprehensively cover all the poor. In this regard, short-term response to crisis, including The review also reveals that there are several target- recurrent droughts, needs to be complemented by ing approaches currently in use in Kenya. Although long-term development strategies for poverty re- the scale and coverage by programs differ, there is duc�on. The global �nancial crisis has opened up sufficient implementa�on experience in the coun- policy dialogue and provides a good opportunity for try to inform the design of a new program or the a more coordinated approach between government scale up of exis�ng one(s). The experience and les- and donors in the design of social safety nets. sons from the OVC and the Hunger Safety Net cash transfer programs provide a useful star�ng point The generalized food subsidy scheme implemented for the design of a cash transfer scheme. The WFP by the government in 2008/09 was too expensive emergency relief opera�ons and school feeding pro- for the exchequer and had to be suspended to be grams also offer useful star�ng points for the design replaced by a more comprehensive, efficient and of in-kind transfers, if this is selected as the appro- well-targeted scheme. Targe�ng concentrates re- priate instrument. Although s�ll in its infancy, the sources on a speci�ed and a poorer set of bene�- Kazi kwa Vijana program and WFP’s food for assets ciaries to achieve the desired impact at the lowest program provide useful founda�on and poten�al cost. Global experience shows that the design of a for scale up for a workfare scheme. Overall, given comprehensive and effec�ve targe�ng system takes the diversity of iden��ed vulnerable groups, it may at least nine months, some�mes even longer to put be necessary to develop a scheme that combines in place. However, when such a system is in place, it different instruments, for example cash transfer is available for use by many programs, for instance and food for work. The reform of Ethiopia’s emer- healthcare, schooling, housing, cash transfers and gency food relief program to the Produc�ve Safety even in-kind transfers. In this regard, efficiency in- Net program (PSNP) is a good example.³� creases when the overhead costs for se�ng up the system are shared among many programs.�� In Ken- The review shows that the exis�ng targeted pro- ya, where a large propor�on of the popula�on lives grams have limited geographic coverage, and main- below the poverty line, determining a cutoff point ly respond to drought and headline crisis. As a re- can be complex, and mul�ple targe�ng approaches sult, the approach is fragmented and the coverage maybe used. ³� Ethiopia’s emergency food relief program was transformed to a development oriented safety net, providing more than food support through two components: labor-intensive public works that address the underlying causes of food insecurity, and through grants to households who cannot undertake public works, such as orphans, pregnant and lactating mothers, the elderly, and people living with HIV/AIDS. ³� Grosh et al. (2008). 48 References 2008 Survey on Monitoring the Paris Declara�on Briceno-Garmendia and Foster (2007), Fiscal Space for Infrastructure in Africa Burnside, Craig and Yuliya Meshcheryakova (2005), ‘Mexico: A Case Study of Procyclical Fiscal Policy’, in Fiscal Sustainability in Theory and Prac�ce, Craig Burnside (ed), Washington DC: World Bank Cardoso, Eliana. (1993) Private Investment in La�n America, Economic Development and Cultural Change, 41, July 1993, pp. 833–48 Foster, Vivian, (2008) “Overhauling the Engine of Growth: Infrastructure in Africa,� Africa Infrastructure Country Diagnos�c Study�. World Bank, Washington DC Gavin, Michael and Roberto Pero� (1997), ‘Fiscal Policy in La�n America’, NBER Macroeconomics Annual Government of Ireland (2005), Guidelines for the Appraisal and Management of Capital and Expenditure Proposals in the Public Sector Government of Kenya (2009), Public Expenditure Review, Nairobi: Government of Kenya Government of Korea (2007), Guidelines for Total Project Cost Management System, Korea, 2007 Government of UK Treasury Green Book, Appraisal and Evalua�on in Central Government Grosh Margaret, del Ninno Carlo, Tesliuc Emil, and Ouerghi Azedine, (2008), For Protec�on and Promo�on: The Design and Implementa�on of Effec�ve Safety Nets. World Bank Washington Dc Ilzetzki, Ethan and Carlos A. Vegh (2008), ‘Procyclical Fiscal Policy in Developing Countries: Truth or Fic�on?’, NBER Work- ing Paper 14191 Kirira, Njeru (2009), “Development Budget in Kenya�. Background Paper to this Report Nyoro, James, Tavneet Suri and Be�y Kibaara (2008). “Income, Poverty and Income Dynamics in Kenya.� Tegemeo Work- ing paper Ram, Ra�. (1986) “Government Size and Economic Growth,� American Economic Review, 76, March 1986, pp. 191–203 Talvi, Ernesto and Carlos A. Vegn (2005), ‘Tax Base Variability and Procyclical Fiscal Policy in Developing Countries’, Journal of Development Economics 78: 156-190 World Bank (2002), Investment Climate Assessment. Washington DC ______(2007a), Investment Climate Assessment. Washington DC ______(2007b) Kenya: A Pilot Study on Infrastructure Public Expenditure Efficiency. Washington DC ______(2007c) Kenya, Maintaining and Rehabilita�ng the Road Network for Growth. ______(2008a) Kenya, Accelera�ng and Sustaining Inclusive Growth. ______(2008b) Kenya, Poverty and Inequality Assessment. Report No 44190-KE, Washington DC ______(2008c) World Development Indicators (2008), World Bank, Washington DC 49 APPENDIX Appendix Appendix Group A: Fiscal Policies and Ins�tu�ons for Shared Growth Annex 1: Figures Figure A1: Real GDP, not seasonally adjusted 1 Source: World Bank staff es�mates Figure A2: Seasonally adjusted real GDP - Source: World Bank staff es�mates 51 Appendix Figure A3: Trends in Real GDP, 1996-2008 Source: World Bank staff es�mates 52 Appendix Figure A4: Devia�ons from trend in real GDP, 1996-2008 Source: World Bank staff es�mates 53 Appendix Figure A5: Public sector revenue in Kenya, 1996-2008, billions 2001 Ksh Source: World Bank staff es�mates 54 Appendix Figure A6: Cyclical components of Kenya’s public sector revenue, 1996-2008 Source: World Bank staff es�mates 55 Appendix Figure A7: Public sector expenditure in Kenya, 1996-2008 Source: World Bank staff es�mates 56 Appendix Figure A8: Cyclical components of public sector expenditure, 1996-2008 Source: World Bank staff es�mates 57 Appendix Figure A9: Measures of cyclically adjusted budget balance in Kenya, 1996-2008 Source: World Bank staff es�mates 58 Appendix Figure A10: The discre�onary balance and the �scal impulse, 1996-2008 Source: World Bank staff es�mates 59 Appendix Figure A11: Impulse response func�ons Source: World Bank staff es�mates 60 Appendix Figure A12: Devia�ons of output from trend, and the es�mated component due to �scal shocks, 1996-2008 - Source: World Bank staff es�mates 61 Appendix Annex 2: Filter Construc�on Annex 2 discusses the �lter construc�on in more detail. Table A1: Es�mates of a Piecewise Linear Trend in the logarithm of seasonally adjusted real GDP, 1996-2008 Q3 Coefficient Es�mate Standard Error t-sta�s�c Constant 12.33 0.012 1032 Post-2002 Q2 dummy -0.261 0.060 -4.37 Trend 0.0051 0.0010 5.06 Trend * post-2002 Q2 dummy 0.0086 0.0019 4.46 Source: World Bank staff es�mates (A) Piecewise Linear Filter The model was es�mated using the standard SAS The piecewise trend and the devia�ons of Kenya’s procedure, and the speci�ca�on ARIMA(1,1,0) was real GDP from this trend are presented in panels (a) chosen as it minimizes the Bayesian Informa�on of Annex Figures 3 and 4, respec�vely. To construct Criterion (MINIC op�on in SAS). Therefore, the fol- the trend, the following regression equa�on was lowing model was es�mated: es�mated: where , Yt is real GDP, for and The es�mates for a� and for our sample are -0.2375 for , and represents 2002 Q2. The and 0.0082, respec�vely, which shows low persist- date t* was chosen as the break date since it maxi- ence in the GDP growth rate of Kenya. mizes the t-sta�s�c for the es�mate of the coeffi- cient . Annex Table 1 presents the es�mated coef- The trend for the logarithm of real GDP in this case �cients and the corresponding sta�s�cs. No other can be presented as: signi�cant breaks in the trend were found. (B) The Hodrick-Presco� Filter The Hodrick-Presco� �lter is constructed by choos- As can been seen in panel (c) of Annex Figure 3, the ing a seeries to minimize trend obtained in such a way follows the original series very closely. This is yet another illustra�on of low persistence in the GDP growth rate. Annex Figure 4 panel (c) depicts the devia�on of real GDP We use the value of 1,600 for , which is standard from the trend, that is . in the literature for quarterly frequency data. Panels (b) of Annex Figures 3 and 4 illustrate the resul�ng (D) Peak-to-peak Trend trend and cyclical component of real GDP. We used an ad hoc procedure to select the following peaks in Kenya’s real GDP: 1996 Q3, 1999 Q4, 2004 Q2, (c) Beveridge-Nelson Decomposi�on and 2007 Q3. The resul�ng trend and the cyclical com- The �rst step of this method is to �t an ARIMA(p,1,q) ponent of GDP are plo�ed in panel (d) of Annex Figures model for the real GDP series, as follows: 3 and 4. Due to the de�ni�on of the trend, the cyclical component in this case will always be non-posi�ve. 62 Appendix Annex 3: VAR Model of Kenya’s Economy Denote the vector of variables exogenous to Kenya’s economy (out of Kenya’s control) by xt, and those and , where endogenous to Kenya by wt: Ourfurtherassump�onsareasfollows: , , and Further, denote by zt all the six variables: For more detail and jus��ca�on of these assump- �ons, see [1]. For our es�ma�ons, we chose k = 2; we also include Then, we consider the following “structural� VAR a linear �me trend in the model. model: The output of the es�ma�on is presented in two forms: plots of the impulse response func�ons (An- nex Figure 11) and a table with variance decom- posi�on (Table 9). From the reduced form repre- where , is senta�on of VAR above, we can obtain the moving a polynomial of order k in the lag operator, and is a average representa�on of zt: vector of mutually orthogonal serially uncorrelated shocks, such that their variance-covariance matrix D is diagonal. The structural model above can be transformed into a reduced form model by pre-mul- �plying both sides of the equa�on by : The term “impulse response� refers to the ele- ments of matrices . For instance, the elements correspond to the impulse responses of the cyclically adjusted primary balance to an output Where and shock. The subscript i refers to the number of peri- is the new vector of error terms, which ods that passed a�er the ini�al shock. Panel (b) of now can be poten�ally correlated with each other. Annex Figure 11 illustrates the response of the cycli- The variance-covariance matrix of ut is . To cally adjusted primary balance to a 1 per cent posi- be able to back out an es�mate of matrix B, we must �ve output shock. Due to one of our assump�ons, impose addi�onal restric�ons on its structure. Once the adjusted �scal balance only responds contem- that is done and once we have chosen the order of poraneously to the external variables xt; therefore, the VAR, k, the model parameters can be es�mated the �rst element of the impulse response is zero. by running OLS regressions, equa�on by equa�on. A�er that, the adjusted �scal balance improves for Our restric�ons on matrix B follow those in [1]. Spe- two quarters before slightly decreasing, and then ci�cally, we normalize all of its diagonal elements the effect of the output shock dissipates by the end to be unity. We also assume that the world oil price of year three. The ini�al posi�ve response reaches is determined exogenously and none of the other 0.4 per cent of GDP. This supports the conclusion �ve variables has a contemporaneous effect on it. that Kenya’s �scal policy does not appear to be pro- In other words: cyclical. In addi�on to that, one could compare the 63 Appendix response of output and cyclically adjusted �scal and the forecast error, therefore, is balance to different shocks. If both responses have the same sign, it means that �scal policy is leaning against the wind. This is indeed the case for Kenya; Therefore, the variance of each of the six elements most of the shocks causing expansions also cause of the forecast error can be presented as a sum of an improvement in the adjusted primary balance. six components, each represen�ng the contribu�on of one of the six elements of εt+p,…, εt+1. Table 9 Panel (a) suggests that �scal shocks have reasonably contains the results of variance decomposi�on for low persistence, with the effects of the original �s- the forecast error in real GDP of Kenya. Again, �scal cal shock negligible by quarter 5. At the same �me, shocks do not appear to have an important effect on the output exhibits somewhat higher persistence, GDP. The importance of the output shock decreases with the effects of the original shock las�ng for 9 with the �me horizon, while oil price shocks and quarters (panel (d)). shocks to the US GDP, and especially shocks to the Federal Funds rate, become more important in the Finally, panel (c) illustrates that �scal shocks have longer term. In fact, in the long term, the US mone- very insigni�cant effects on output. While �scal tary shocks explain over 70 per cent of the variance �ghtening causes real GDP to contract, the re- in Kenya’s GDP. This seems to imply that it is Ken- sponse is smaller than 0.1 per cent in magnitude (in ya’s monetary policy, rather than �scal policy, which response to a 1percent change in the cyclically ad- is an important factor in terms of output vola�lity. justed primary balance), which suggests that �scal Further, the fact that Kenya’s �scal policy has a very policy in Kenya has very limited effect on GDP. limited impact on the country’s real GDP seems to suggest that most of Kenya’s government spending Table 9 presents the variance decomposi�on of out- either goes towards government consump�on or put. The variance decomposi�on shows the contri- unproduc�ve investment. bu�on of each component of εt to the forecast er- ror for zt. In par�cular: Annex Figure 12 shows historical contribu�ons of the �scal shock component to output vola�lity, a p-step ahead forecast of zt is again sugges�ng low impact of �scal shocks on Ken- ya’s real GDP. 64 Appendix Appendix Group B: A Globally Compe��ve and Prosperous Kenya-The Role of Public Investment Annex 1: The Budget Cycle and Key Players 1. The Budget Outlook Paper (BOP) is produced by budget, but this prac�ce was discon�nued when Treasury in December/January each year. The the MTEF was introduced in the budget process. BOP provides sectoral indica�ve ceilings. The As a result, the prac�ce of reviewing and compil- ceilings are expected to provide guidance for the ing the list of public investment projects across all Sector Working Groups and line ministries as they sectors was discon�nued. prepare their budgets. The Treasury’s Medium Term Expenditure Framework (MTEF) Secretariat Line Ministry works on the medium-term framework. Though 5. Line ministries are required to produce a Ministe- the MTEF provides a medium-term perspec�ve, rial Public Expenditure Review (MPER) at the start in effect the outer year expenditure ceilings do of the budget process (usually during September- not have any bearing on the current year’s budg- November, once the previous �scal year accounts et process. are completed and data on budgetary spending is con�rmed). The line ministries are required to 2. The Budget Supply Department (BSD) in the Min- prepare their itemized budgets on the basis of istry of Finance coordinates the budget prepa- the development budget resources available to ra�on. The BSD issues a budget circular for all them. Projects to be �nanced under the develop- departments to begin prepara�on, and issues ment budget are iden��ed by the line ministries ceilings that are broken down into gross expen- as part of the budget process, taking into consid- ditures that are either grant, loan or domes�cally era�on the analysis in the MPERs. �nanced. 6. Proper project iden��ca�on and planning, such 3. Development projects are not clearly iden��able as an economic appraisal, is not done at this in the budget es�mates or government accounts, stage. The budget ceiling constrains the ini�a�on hampering monitoring of development project of new projects, but there is no structured proc- progress. ess of priori�za�on in line with strategic priori- �es. In fact, officials in a ministry may not even Prior to the adop�on of the GFS classi�ca�on, know how a development project ends up in the the budget used to clearly iden�fy which devel- budget. There is no detailed cos�ng of individual opment projects were to be �nanced. However, projects required for either the current �scal year since the introduc�on of the GFS format, the or outer years. Cos�ng of projects, if it is done at format of the development budget more closely all, is done at the �me the project goes to tender reflects that of the recurrent budget. Currently, in order to establish a benchmark during the ap- the development budget es�mates do not clearly praisal. show the name of the project for which each line item development expenditure is intended (only 7. Detailed project appraisal is not required at any the purpose of the expenditure, e.g. new con- stage prior to the formal approval of the develop- struc�on, domes�c travel etc). The development ment budget. All development partner-�nanced budget printed es�mates are publicly available projects are also included in the development and include all the line item expenditures that budget and, in the course of their prepara�on, comprise two volumes (and an astonishing 800+ appraisal may be undertaken by the donor or �n- pages), but the classi�ca�on system means it is ancier. Without systema�c appraisal, however, difficult to track progress with project implemen- the current process does not allow for the pri- ta�on based on the printed es�mates or other ori�za�on of development projects on the basis Ministry of Finance �nancial reports on actual of cost effec�veness or cost bene�t analysis. An expenditures. a�empt to quan�fy or address bene�ts and al- terna�ves are also not required prior to inclusion MTEF Secretariat in the budget. As such, there is no scope to pri- 4. Prior to adop�ng the MTEF approach, the Min- ori�ze projects on the basis of standard appraisal istry of Planning was responsible for developing techniques. a Public Investment Programme as part of the 65 Appendix Sector Working Groups Na�onal Development and Vision 2030. It is 8. Sectors cover a group of line ministries dealing chaired by the Economic Secretary and the BSD with similar or inter-related issues. The Sector provides the secretariat. The role of EBSC is to Working Groups (SWGs) are chaired by one of the coordinate and monitor the overall budget es�- Permanent Secretaries from the sectors, and in- mates against macroeconomic considera�ons. clude officers from the other line ministries, BSD The EBSC reviews the expenditure proposals and the Ministry of Planning, and can also include coming from the SWGs, holds sector dialogues representa�ves from development partners, sec- with the SWGs, and makes recommenda�ons tor NGOs, and the private sector. This broader at- to Cabinet through the dra� Budget Strategy tendance helps to improve communica�on and Paper (BSP) on the �nal alloca�on of resources build partnership with other stakeholders during to ministries and programs within ministries. the budget process. 11. The EBSC is responsible for reviewing the de- 9. The SWGs play a key role in the development of tailed es�mates before they are circulated the budget. They provide a cri�cal assessment of through the Budget Strategy Paper. The EBSC the Ministerial Public Expenditure Review submis- also plays a role during budget execu�on as it sions, par�cularly with regard to na�onal goals. reviews and monitors implementa�on of minis- They are responsible for reviewing the alignment tries’ spending plans and expenditure data sub- of sector objec�ves and strategies with na�onal mi�ed by line ministries. goals; coordina�ng sector ac�vi�es leading to the Budget Strategy Paper and the Annual Budget development of the Ministerial Public Expendi- ture reviews and �nalizing the SWG reports. The 12. The budget ceilings set early in the budget proc- SWG, in effect, iden��es the sector priori�es and ess are rarely adhered to. As in most countries, ranks them according to their relevance to na- the budget process becomes compe��ve and �onal goals; iden��es sector programs that are various techniques are used to a�empt to en- a priority, and those that are “core poverty pro- large ones share, such as minister’s publicly grams�; and ensures that these are funded. By campaigning and/or complaining about their working at the sector level, the SWG pays a�en- allotment. The Treasury has also opened up a �on to linkages that cross line ministry boundaries window of renego�a�ng the ceilings with min- and ensures that ministerial spending plans take istries directly, even a�er the SWGs have met account of such linkages. The SWG is supposed to and agreed on the alloca�ons. This clearly un- take a comprehensive look at �nancing of sector dermines the work of the SWGs. As a result, priori�es, including the budget, extra-budgetary there is a tendency for the �nal budget to be funds, external resources and appropria�ons-in- some 10 to 15 per cent higher than the BSP. aid. While line ministries are not intended to un- 13. This could be caused by several likely factors. dertake a cos�ng exercise, the SWGs analyze cost The BSP is not completed with full informa�on, implica�ons of various proposals mindful of the or the Treasury prefers to be the �nal arbiter ceilings that have been set in the sector. of resources and probably sets asides resources for it, thus reinforcing the incen�ve for minis- Economic and Budget Steering Commi�ee tries to contest their budget ceilings. The sum 10. The Economic and Budget Steering Commi�ee is to undermine the system of se�ng ceilings, (EBSC) comprises key departments of the Min- and to undermine the process for intra- and in- istry of Finance, and Ministry of Planning and ter-sectoral resource alloca�ons. 66 Appendix Annex 2. Sta�s�cal Annex Table B1: Infrastructure indicators, Kenya and country comparators South Malaysia Indonesia Tanzania Uganda Ghana Kenya Africa Improved sanita�on facili�es 33 33 10 42 59 94 5 (% of popula�on with access) Improved water source 55 64 80 57 93 99 80 (% of popula�on with access) Improved water source, rural 46 60 71 49 82 96 71 (% of rural popula�on) Improved water source, urban 81 90 90 85 100 100 89 (% of urban popula�on) Telephone lines (per 100 people) 0.4 0.5 1.6 0.7 9.7 16.4 7.9 Fixed broadband subscribers 0.0 0.1 0.0 0.8 3.8 0.1 (per 100 people) Internet users (per 100 people) 1.0 5.8 3.8 8.0 8.3 55.7 5.8 Mobile and �xed-line telephone subscribers 21.0 44.2 34.0 30.9 98.1 104.3 44.2 (per 100 people) Roads, paved (% of total roads) 9 23 15 14 17 96 55 Source: WDI 2008, Latest Year Available between 2004-2007 67 Appendix Annex 2. Sta�s�cal Annex Table B2: Human capital indicators, Kenya and country comparators Ghana Indone- Kenya Korea Malaysia South Tanzania Uganda sia Rep. Africa Employment to popula�on ra�o, 15+, total (%) 65 62 73 59 61 41 78 83 Employment to popula�on ra�o, ages 15-24, total (%) 40 41 59 29 44 14 70 76 Primary educa�on, dura�on (years) 6 6 6 6 6 7 7 7 Primary comple�on rate, total (% of relevant age group) 71 99 93 101 98 92 85 54 Life expectancy at birth, female (years) 60 73 55 82 77 52 54 52 Life expectancy at birth, male (years) 60 69 53 76 72 49 51 51 Life expectancy at birth, total (years) 60 71 54 79 74 50 52 51 Literacy rate, adult total (% of people ages 15 and above) 65 92 n/a n/a 92 88 72 74 School enrollment, secondary, female (% net) 43 61 42 93 n/a n/a n/a 15 School enrollment, secondary, male (% net) 47 60 43 99 n/a n/a n/a 17 Immuniza�on, DPT (% of children ages 12-23 months) 94 75 81 91 96 97 83 64 Immuniza�on, measles (% of children ages 12-23 months) 95 80 80 92 90 83 90 68 Mortality rate, infant (per 1,000 live births) 73 25 80 4 10 46 73 82 Pregnant women receiving prenatal care (%) 92 93 n/a n/a 79 0 78 94 Incidence of tuberculosis (per 100,000 people) 203 234 353 90 103 948 297 330 Newborns protected against tetanus (%) 88 87 74 n/a 89 72 88 85 Physicians (per 1,000 people) 0.15 n/a 0.14 n/a n/a 0.77 n/a 0.08 Tuberculosis cases detected under DOTS (%) 36 68 72 14 80 78 51 51 Source: WDI 2008, Latest Year Available between 2004-2007 68 Appendix Appendix Group C: Pro-poor Spending in Kenya: A Review of Exis�ng Targeted Subsidy Programs Annex 1: Average government transfers to households by expenditure quar�les (adult equivalent) Percent receiving Quar�le 1 Quar�le 2 Quar�le 3 Quar�le 4 transfers Nairobi 1,000 1,518 1,000 4,930 1.6 Central 788 770 795 1,376 16.9 Coast 2,920 803 2,467 3,455 15.7 Eastern 1,744 2,350 1,049 1,451 19.2 North Eastern 3,970 4,037 2,324 4,707 54.6 Nyanza 3,469 3,957 1,284 2,373 5.2 Ri� Valley 1,127 960 1,106 726 2.4 Western 2,143 1,034 1,402 3,308 2.7 Na�onal 2,396 2,069 1,359 2,088 Source: World Bank staff es�mates 69