WPS7154 Policy Research Working Paper 7154 Local Budget Transparency and Participation Evidence from the Kyrgyz Republic Damir Esenaliev Gregory Kisunko Governance Global Practice Group January 2015 Policy Research Working Paper 7154 Abstract The paper investigates determinants of civic participa- interest, awareness, and participation are positively asso- tion in local budget processes in rural areas in the Kyrgyz ciated with the age, education, employment, risk-taking Republic by using data from the Life in Kyrgyzstan survey, attitudes, trust, and social capital of respondents. The paper conducted in 2012. The analysis of the data suggests that documents that unawareness and lack of participation are although civic awareness and interest in local budget largely related to being female, of non-Kyrgyz ethnic origin, processes is relatively high, the participation rate in local inactive in the labor market, recent internal migrants, budgeting processes is low. The paper also shows that and residents of communities with poor infrastructure. This paper is a product of the Governance Global Practice Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at esenaliev@ sipri.org or gkisunko@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Local Budget Transparency and Participation: Evidence from the Kyrgyz Republic* Damir Esenaliev and Gregory Kisunko JEL classification codes: H31, H81. Keywords: budget process, Kyrgyzstan, Kyrgyz Republic, local budget, participation, transparency. Damir Esenaliev is a Sr. Researcher with the Stockholm International Peace Research Institute (SIPRI); Gregory Kisunko is a Sr. Public Sector Specialist with the World Bank. * The authors are grateful to colleagues at the World Bank: Mr. Alexander Kremer and Mr. Adrian Fozzard for their support of this work, to Ms. Sarosh Sattar and Ms. Asli Gurkan for their very useful comments and suggestions, and Mr. Nicholas A. Bian for his significant contribution to the final version of the paper. We also would like to thank the people of the Economic Transformation, Household Behaviour and Well-Being in Central Asia: The Case of Kyrgyzstan project for their cooperation and generosity. 1. INTRODUCTION Citizen participation is crucial in increasing transparency and accountability in the government budget-making process. Traditionally citizens are informed about budget decisions through statements from government officials or from media coverage. However, an active citizenry engagement with the budget-making process enhances government accountability and ensures the needs of various groups of society are considered (Ahmad & Weiser, 2006; Im, Lee, Cho, & Campbell, 2014; Irvin & Stansbury, 2004; Khagram, Fung, & de Renzio, 2013; Wampler, 2012). Increasing understanding of the importance of citizen engagement, especially at the local level, encourages governments and nongovernmental organizations to conduct budget-making process with greater transparency and accountability. The residents whose daily lives are positively affected by budget-related decisions are likely to remain engaged and active in community life (Khagram et al., 2013). This study contributes to the growing literature on participatory budgeting by looking at the demand side of participatory budgeting, quantifying the level of awareness and participation in the budget process by rural households in the Kyrgyz Republic and demonstrating the effect of characteristics of the active and non-active citizens. This paper used the results of the nationally representative data from the Life in Kyrgyzstan survey for 2012 to analyze the extent of citizen awareness and participation in the local budget processes, to identify the socioeconomic groups that participate most and least in the budget process, and to explore the determinants of awareness and participation. The study argues that civic awareness and interest in local budget processes is relatively high while the participation rate is relatively low. Interest, awareness and participation are positively associated with age, education, employment, risk-taking attitude, trust, and social capital. The findings suggest as well that lack of awareness and non-participation is attributable to females, non-Kyrgyz ethnic groups, inactive in labor market, internal migrants, and the residents of communities with poor infrastructure. This paper is a result of collaboration between the World Bank and the research consortium led by the German Institute for Economic Research (DIW Berlin) aimed at measuring attitudes, determinates and changes in the citizen participation in the local budget processes started under the World Bank project "Information Matters - Transparency and Accountability in the Kyrgyz Republic" (IMTAK). This project focused on rural areas and provided training to officials in village administrations (Ayil Okmotu), local elected leaders and representatives of civil society (active residents and members of 2 local NGOs). Consequently, this paper focuses on the Kyrgyz Republic’s rural population using the rural sample of the Life in Kyrgyzstan survey data. 1 The following section provides a brief overview of the fiscal transparency and the scope of the local budgets in the Kyrgyz Republic; sections three and four present methods and the data used; section five discusses the results; and section six concludes. 2. BUDGET TRANSPARENCY AND LOCAL BUDGET IN KYRGYZSTAN The Kyrgyz Republic ranks relatively poorly across governance and budget transparency indicators. The Worldwide Governance Indicators (WGI) – a research data set by the World Bank summarizing the views on the quality of governance – show little progress or a decline in rankings across its six indicators for the Kyrgyz Republic in the period from 1996 to 2013 (The World Bank, 2014).2 When viewing the WGI between 2006 and 2013, most measurements of governance deteriorated, especially in “Control of Corruption”, while only the “Voice and Accountability” indicator improved, though from a low base with 30 points out of 100. The Kyrgyz Republic ranked poorly as well on the Open Budget Index (OBI), which is an internationally accepted index on national budget transparency compiled biennially by the International Budget Partnership (IBP). 3 The Kyrgyz Republic’s score in OBI in 2012 was 20 (out of 100) – the result far below the average score of 43 in all 100 countries surveyed, but comparable with the average score of low-income economies (International Budget Partnership, 2012). To put this in the regional context, the Kazakh Republic ranked the highest with 48 points and Tajikistan the lowest with 17 points. The Kyrgyz Republic has made marginal progress since the inception of the OBI in 2006, with its score increasing from 8 points (out of 100) in 2008 to 20 points in 2012. 4 Placing key budget documents in the public domain can strengthen budget transparency, 5 with emphasis on 1 Compared to rural population, urban residents seem to be more interested, but are considerably less informed and less active in budget processes. Details of a comparison of responses between rural and urban population is provided in Appendix 1. 2 The six dimensions are Voice and Accountability, Political Stability and Absence of Violence, Government Effectiveness, Regulatory Quality, Rule of Law, and Control of Corruption. The WGI data are gathered from a number of survey institutes, think tanks, non-governmental organizations, international organizations, and private sector firms. Information about methodology, data sources, and country rankings can be found at http://info.worldbank.org/governance/wgi/index.aspx#home. 3 The IBP is a global research and advocacy program, which seeks to promote public access to budget information and the adoption of accountable budget systems. 4 Kyrgyzstan is likely to improve dramatically the OBI score in the next issue of the OBI index. The recently launched Open Budget Survey Tracker indicates that all but one budget document was made available to the public (http://www.obstracker.org/). 5 OBI notes eight key documents to make available to the public: Pre-Budget statement, Executive’s Budget Proposal, Enacted Budget, Citizens Budget, In-Year Reports, Mid-Year Review, Year-End 3 important, citizen-engaging documents such as the Citizens Budget (which is a simplified nontechnical presentation to enable broad public understanding of a government’s plans for raising revenues and spending public funds). The legislative efforts to improve transparency in the Kyrgyz government are often driven by assistance from development institutions, while the Government of the Kyrgyz Republic has only recently attempted to develop a long-term reform agenda in this area. One example is the adoption of the 2006 law “On access to information, maintained by state organizations and local self-governance bodies of the Kyrgyz Republic.” This law introduced a number of measures to make budget data and information readily available and ensured the general public access to government-held information. The law remains scarcely used largely not because of resistance or reluctance of public officials to implement it, but simply because of limited understanding by the residents of the local governance works, particularly the local budgeting rules, coverage and cycle. The country is divided into 31 cities and towns and 453 local self-government units (LSGs) – Ayil Aimaks (AA) which vary significantly in size, capacity and ability to raise funds for providing services to residents. Only a few LSGs are self-sufficient, i.e. able to derive enough revenue from local taxes, tax allocations from the national taxes, the rest of the LSG are receiving equalization grants from the central government that assist them in providing the public services in their respective territories. The LSGs are charged with delivery of a large number of public services to the local population, while managing only about a quarter (22.7% in 2012) of the consolidated budget, which incorporates the republican budget and budget of the LSGs. According to the Law on the Local Self-Governments these units are responsible for overall development of the territory, which includes economic development and attracting investment and grants; management of municipal property; management of the local budget; management of sewage systems, water treatment facilities and drinking water supply; maintenance of municipal roads; organization of street lighting; maintenance of cemeteries; landscaping and gardening in public areas; maintenance of parks, sport and recreation facilities; waste management; management of public transportation; protection of local cultural and historical sites; organization and maintenance of local libraries; establishment of rules on land use and enforcement of the regulations on urban planning and architecture; advertising issues; facilitation of public order; development folk art; creation of conditions for leisure; organization of activities for children and young people; development of physical culture and sports; facilitate prevention and mitigation of emergency situations. Reports, and Audit Report. Full description of the methodology and country reports can be accessed at http://internationalbudget.org. 4 While the LSGs are required to fulfill all these responsibilities and manage significant resources, not all LSGs have capability and staff to do these properly, while opportunities for professional development for LSG staff and elected officials are limited. The fiscal relations with the higher levels of government are also not necessarily clear, while transfers not always predictable. As a result, a significant proportion of services provided by the LSGs do not satisfy the population. According to the adopted Program for Development of the Local Self-Government in the Republic of Kyrgyzstan in 2013-2017 prepared by the State Agency for Local Self-Government and Interethnic Relations, this situation occurs due to the absence of a unified system of rules and procedures, insufficient reimbursement for certain services, lack of single set of requirements and responsibilities for local self-government units, etc. A lot of these problems stem from a poor understanding of how the local budgets are formulated, managed and executed and what are exact rights and responsibilities of LSGs by all local stakeholders – citizens, elected and appointed local officials and budget staff working at the local self-government units. 3. METHODS Life in Kyrgyzstan survey This paper builds upon previous surveys and analysis conducted in the Kyrgyz Republic. One primary example is the research project “Economic Transformation, Household Behavior and Well-Being in Central Asia: The Case of Kyrgyzstan” and its major product, the Life in Kyrgyzstan (LiK) panel survey (Brück et al., 2014). The project was conducted in 2010-2012 by the German Institute for Economic Research (DIW Berlin) and in cooperation with Humboldt-University of Berlin, Center for Social and Economic Research (CASE-Kyrgyzstan), and American University of Central Asia (AUCA). The project collected survey data, investigated well-being and behavior of individuals and households in the Kyrgyz Republic, and improved the research capacity within Central Asia. The initial sample of the survey - based on the 2009 Kyrgyz population census - included 3,000 households and 8,160 adult individuals in 120 communities. The households in the survey sample were obtained by stratifying two-stage random sampling with probabilities proportional to population size. The strata were formed by the two cities of Bishkek and Osh as well as the urban and rural areas of the seven oblasts, totaling 16 strata. The LiK sample is representative at the national, rural/urban and north/south levels, but not on the oblast level). The 2011 and 2012 waves of the LiK survey maintained a relatively low attrition rate of individuals and households. In the 2011 wave 2,863 households were interviewed; in 2012 this number declined slightly to 2,816 households interviewed. The LiK survey collected information at the individual, household, and community levels. Each of the questionnaires consisted of several topical sections. This paper exploits a rich set of individual, household and community characteristics found in the LiK 2012 survey, as well as lagged community 5 variables from the second wave of the LiK 2011 survey. Specifically, we use information from the 2012 module called “Local Budget” designed to collect information about awareness and participation of individuals in local budget processes (the questions in the module are illustrated in Table A.1 in the Appendix). On the individual level we use information about a person’s background, such as age, gender, ethnicity, education, labor market status, social capital, and subjective indicators, such as trust and political preferences. At the household level, we use data on household demographics, income, assets ownership, and location. We use the community level data pertaining to population size, locational and infrastructure characteristics. 6 Empirical methods Descriptive analysis includes frequencies and means of the various indicators pertaining to the research questions of the study. We exploit both descriptive methods and regression models to investigate the relationship of dependent variables with various factors at the individual, household, and community levels. As responses to the questions on awareness and participation in local budget processes are given in several categories or in “Yes” and “No” responses, we respectively exploit ordered and binary logit regression models – the most suitable tools for such cases. The analysis also re-groups the multiple responses to binary responses (zero or one) to utilize binary logit models. 4. DATA Sample This paper analyzes rural households and individuals from the LiK survey. 7 In total, we exploit information from 5,145 adult members aged 18 and older, originating from 1,671 households (Table 1). The population, gender and ethnicity structure of respondents seem to be in concordance with the national level indicators. Individual, household and community characteristics We exploit a rich set of individual, household and community characteristics available from the LiK survey: (i) objective characteristics, such as age, gender, ethnicity, education, labor market status; and (ii) individual attitudes and preferences, such as trust, political preferences, and social links. At the household level, we mainly look at demographics and wealth indicators, such as aggregate income, land, and livestock. We also look at the reliability of public services, represented by electricity provision. The same methodology is followed in analyzing community data, where we look at 6 Because in the LiK survey for 2012 the community characteristics were not collected, we use the community level information from the LiK 2011 survey. 7 While we do not analyze involvement of population in budget processes in urban areas, Table A.2 gives a comparison of responses of rural and urban residents. 6 availability of key public services, such as medical service points, clean water sources, crop irrigation, and kindergartens. We describe these demographic and socio-economic characteristics at regional (North/South) and gender dimensions (Table 2). Table 1: Sample size, gender and ethnic composition Sample size of which, share of : Households Individuals aged 18+ Females Kyrgyz Total rural 1,671 5,145 0.52 0.72 North 723 2,075 0.51 0.67 Issyk-Kul 169 458 0.52 0.92 Naryn 99 290 0.48 0.89 Talas 99 233 0.49 0.98 Chui 356 1,094 0.52 0.45 South 948 3,070 0.53 0.75 Djalal-Abad 338 946 0.51 0.72 Batken 147 419 0.52 0.96 Osh 463 1,705 0.53 0.72 Source: LiK 2012 8 An overview of individual characteristics reveals some interesting (and statistically significant) differences on regional and gender lines. For example, the population in northern oblasts seems to have slightly more schooling compared to southern oblasts. Further, risk taking attitudes (e.g. establishing own business, etc.) are higher for individuals who are both male and who reside in northern oblasts. Labor market participation is expectedly higher for males than for females with almost 80 percent of males active in labor markets compared to 49 percent of females. This suggests the inactivity of rural women in labor markets could be an important factor of participation in communal life. Social capital, proxied here by reliance on social networks and duration of residency, is a consideration in the analysis as an indicator of better opportunities and a proactive attitude of individuals including interest and participation in community issues. As Table 2 demonstrates, the extent people rely on each other and duration of residency, which were used to proxy social capital in this paper, seems to be generally stronger among males and among residents in the south. Trust in public institutions is a crucial factor for strong citizen participation in local budget processes therefore suggesting that lower level of trust leads to less citizen interest and participation. The survey data reveals an interesting fact: residents in southern oblasts express greater trust in people in their 8 For details, see http://www.diw.de/de/diw_01.c.100313.en/forschung_beratung/projekte/projekte.html?id=diw_01.c.345525.en 7 communities and to local and central government. Surprisingly, female respondents exhibit a higher level of trust in public institutions compared to male respondents. Political preferences and voicing political opinions may play a similarly important role as trust does in civic participation. In this regard, we consider respondents’ political system preferences and participation in the last presidential and local elections. Evidently, the current party-based political system in the Kyrgyz Republic has low popularity, especially in the south: only 15 percent of rural respondents favor it. In contrast, about a fifth of respondents like the current political system in the Russian Federation, a home for a majority of Kyrgyz migrant workers. Participation (or intention to participate) in both types of elections was high at about 87 and 78 percent of all respondents in the presidential and local elections, respectively. Rural household composition differs markedly between the Kyrgyz Republic’s northern and southern oblasts. Households in the south oblasts are larger, despite the fact that labor migration is prevalent. About 25 percent of rural south households have a migrant worker whereas in the north this proportion is only 5 percent of households. The difference in migration patterns between the two regions largely reflects the availability of land, which is the main economic resource for rural population in the Kyrgyz Republic. We hypothesize that economically advantaged households are more interested and more willing to participate in local budgets to address communal and private issues. Alternatively, we hypothesize that poorer households will have less interest and fewer opportunities in participation. To examine these hypotheses we employ a number of economic indicators, including household income and assets. Household economic indicators presented in Table 2 indicate there are notable regional differences in incomes and assets; however there is practically no difference in gender. For example, income per capita in northern households is slightly higher than in southern households. In addition, sources of income in the regions vary, with remittances constituting about one fifth of household income in southern oblasts against 1 percent in northern oblasts. In terms of assets, households in northern areas are in advantageous position with larger plots of land and more livestock. An asset index, calculated by principal component analysis on basis of ownership of 13 household items, confirms the relative advantage of rural households in the northern regions. 8 Table 2: Individual, household and community characteristics Indicator Average North South Female Male (in ratio, if not indicated otherwise) Personal characteristics Age, years 40.4 40.7 40.1 40.7 40.0 Kyrgyz 0.72 0.67 0.75 *** 0.71 0.73 Uzbek 0.12 0.01 0.20 *** 0.12 0.12 Of other ethnicity 0.15 0.32 0.05 *** 0.16 0.15 Years of schooling 10.3 10.4 10.3 ** 10.3 10.4 Risk attitude, 0->10 scale 4.7 5.3 4.3 *** 4.2 5.3 *** Labor market status Employed 0.61 0.60 0.61 0.46 0.77 *** Inactive 0.37 0.37 0.36 0.51 0.21 *** Unemployed 0.03 0.03 0.02 0.03 0.02 Social capital Lived in the community, years 38 36 40 *** 38 39 * Member of a social group 0.12 0.13 0.10 *** 0.11 0.12 Can get financial help from others 0.81 0.76 0.85 *** 0.79 0.84 *** Trust, in a scale from 1->4 Trust in people in community 3.15 3.00 3.25 *** 3.14 3.15 Trust in local government 2.70 2.51 2.84 *** 2.74 2.66 *** Trust in central government 2.61 2.56 2.65 *** 2.66 2.56 *** Trust in police 2.33 2.30 2.35 * 2.38 2.28 *** Political attitudes Supports current political system 0.15 0.21 0.11 *** 0.14 0.15 Supports Russian political system 0.20 0.23 0.18 *** 0.17 0.23 *** Has no political system preference 0.23 0.18 0.27 *** 0.27 0.20 *** Voted in last Presidential election 0.87 0.86 0.87 0.86 0.88 ** Is going to vote in local elections 0.78 0.76 0.79 ** 0.76 0.80 *** Household (HH) demographics Female head 0.19 0.22 0.17 *** 0.24 0.14 *** HH size, people 5.9 5.5 6.3 *** 5.9 5.9 Number of children 2.1 1.9 2.2 *** 2.1 2.1 Number of migrants 0.3 0.1 0.4 *** 0.3 0.2 * Household economics HH income, '000 Soms/month 18.4 18.2 18.5 18.1 18.7 Income per person, '000 Soms/month 3.5 3.7 3.3 *** 3.4 3.5 Remittances, '000 Soms/month 2.1 0.2 3.4 *** 2.2 2.1 Land size, ha 1.2 1.9 0.8 *** 1.2 1.3 Livestock units, sheep equivalent 22.7 25.2 21.0 *** 22.0 23.4 Number of cars 0.45 0.46 0.44 0.43 0.47 ** Asset index, HH level mean=100 108.1 133.0 91.3 *** 106.2 110.2 Housing value, mln.Soms 0.90 0.86 0.92 *** 0.90 0.89 Community characteristics1 Population, '000 people 6.5 4.3 8.0 *** 6.5 6.5 Kindergarden is available 0.48 0.46 0.49 * 0.48 0.48 Medical point is available 0.51 0.49 0.53 ** 0.51 0.52 Frequent energy interruptions 0.12 0.09 0.14 *** 0.12 0.11 Clean water availability, % 80 87 75 *** 81 80 Irrigation water available, % 76 80 74 *** 76 76 Number of attending people 153 174 140 *** 155 151 Sources: Life in Kyrgyzstan (LiK) Survey, 2011 and 2012 1/ Community data are from LiK 2011 wave The mean differences for North/South and female/male dimensions are tested using t-test. Significant differences are indicated by * p<0.1, ** p<0.05, *** p<0.01. 9 Community characteristics also differ on a regional dimension. On average, the communities in southern areas are almost twice as large as those in the north. Also communities in the south are more disadvantaged in terms of access to clean water and irrigation. Communities in both regions seem to have marginal difference in availability of public services such as kindergartens and medical points. However, the average number of participants who attended community meetings in northern communities was considerably higher than the number of participants in southern communities. This may have implications for both awareness and participation in local budget processes. Awareness about and participation in local budget processes The responses to questions in the local budget module revealed a moderate level of awareness and interest in local budget processes, but an extremely low level of participation. Generally every second respondent expressed an interest in learning how the local budget funds are spent, with only half of respondents indicated little interest (Table 3). However, only every third of respondents thinks being informed about local budget issues is important, with a majority of them having limited information; 31 percent of respondents know that local governments are supposed to provide information about their respective budgets; and while 42 percent of responses are positive about accessibility of budget information, most of these respondents felt that information is accessible with difficulties. Low degree of civic participation in local budget processes seems to stem from the lack of confidence in the effectiveness of citizen participation. Only about 12 percent of respondents believe that “an ordinary citizen can influence budget processes”. Positive responses on whether “anyone from your household tried to influence budget process” constitute only 3.5 percent. Likewise, only about 5 percent of respondents could report the date of the last budget hearings and only 5 percent of them personally participated in the budget hearings. 9 9 Additionally, we compare the responses to local budget related questions between rural and urban residents. We document that local urban population has a higher level of interest, but lower degree of awareness. Whereas the perception of urban residents about impact of civic participation and access to budget information is comparable to the responses of rural residents, the data demonstrate a very low share of urban residents participating in the local budget events, such as budget hearings. 10 Table 3: Awareness about and participation in local budget processes Percentage of responses Indicator No Yes Yes sub-categories Very Fair Little Informed about the local budget 65.5 34.5 1.2 10.1 23.2 Interested in how the local budget is spent 48.6 51.4 4.8 22.5 24.0 Knows that local government should provide info 69.0 31.0 about local budget Information on local budget is accessible 57.5 42.5 1.3 8.8 32.4 An ordinary citizen can influence budget process 88.3 11.7 Someone from my household tried to influence 96.5 3.5 local budget process in the last 12 months Knows the date of the last budget hearing in the 94.7 5.3 local government Participated in budget hearing of Aiyl Aimak 96.0 4.0 during the last 12 months Source: LiK 2012 5. RESULTS In our descriptive and regression analysis we separate the awareness and interest questions from the participation questions. First, we consider the awareness and interest questions. For clarity and succinctness, we look at descriptive statistics using binary responses. Interest in and knowledge about local budget issues Personal characteristics and labor market status play a significant role in knowledge about and interest in local budget processes (Table 4 below; see also table A.1 in the Appendix for exact wording of the survey questions referring to interest and knowledge). The individuals who are knowledgeable and interested tend to be 2-5 years older than those who responded ‘No’. Females tend to know and be interested less compared with males. With the context of rural areas, there seems to be disproportionally fewer individuals from non-Kyrgyz ethnic groups that have knowledge and interest. Level of education seems to be an important factor in both knowledge and interest: the respondents who self-report interest and knowledge tend to have about one more year of schooling compared with individuals who responded negatively. It is also worth noticing, that there is a difference in subjective risk-taking attitudes: positive responses are given by more risk-taking individuals. 11 Table 4: Awareness about and interest in local budget processes Indicator Informed about Interested in how Knows that local (in ratio, if not indicated otherwise) the local local budget government should budget expenditures are inform about its done budget No Yes No Yes No Yes Personal characteristics Age, years 39.5 42.0 *** 38.8 41.8 *** 38.7 44.0 *** Female 0.56 0.45 *** 0.57 0.47 *** 0.55 0.46 *** Kyrgyz 0.67 0.82 *** 0.69 0.75 *** 0.67 0.83 *** Years of schooling 10.0 10.9 *** 9.9 10.8 *** 10.0 11.0 *** Risk attitude, 0->10 scale 4.4 5.3 *** 4.2 5.1 *** 4.3 5.5 *** Labor market status Employed 0.57 0.69 *** 0.55 0.66 *** 0.57 0.69 *** Inactive 0.40 0.30 *** 0.41 0.32 *** 0.40 0.30 *** Unemployed 0.03 0.01 *** 0.03 0.02 *** 0.03 0.01 *** Social capital Lived in the community, years 37 41 *** 37 40 *** 37 42 *** Member of a social group 0.09 0.17 *** 0.09 0.14 *** 0.09 0.17 *** Can get financial help from others 0.76 0.91 *** 0.73 0.89 *** 0.78 0.90 *** Trust, in a scale from 1->4 Trust in people in community 3.1 3.3 *** 3.1 3.2 *** 3.1 3.3 *** Trust in local government 2.7 2.8 *** 2.7 2.7 2.7 2.8 *** Trust in central government 2.6 2.7 *** 2.6 2.7 *** 2.5 2.8 *** Trust in police 2.3 2.4 *** 2.3 2.4 *** 2.3 2.5 *** Political attitudes Supports current political system 0.15 0.15 0.13 0.16 *** 0.13 0.19 *** Supports Russian political system 0.17 0.25 *** 0.16 0.24 *** 0.19 0.22 ** Has no political system preference 0.29 0.14 *** 0.33 0.15 *** 0.28 0.14 *** Voted in last Presidential election 0.84 0.92 *** 0.83 0.90 *** 0.84 0.92 *** Is going to vote in local elections 0.75 0.84 *** 0.74 0.82 *** 0.75 0.85 *** Household (HH) characteristics HH size, people 6.0 5.9 6.1 5.8 *** 6.1 5.5 *** Income per person, '000 Soms/month 3.4 3.7 *** 3.3 3.7 *** 3.4 3.7 *** Land size, ha 1.1 1.4 *** 1.1 1.4 *** 1.2 1.4 *** Livestock units, sheep equivalent 21.0 25.8 *** 21.7 23.6 * 21.1 26.3 *** Number of cars 0.42 0.49 *** 0.41 0.49 *** 0.43 0.48 *** Asset index, HH level mean=100 98.0 127.2 *** 90.6 124.7 *** 98.8 128.8 *** Housing value, mln.Soms 0.93 0.83 *** 0.93 0.86 *** 0.91 0.86 ** 1 Community characteristics Population, '000 people 6.3 6.9 *** 6.6 6.4 6.7 6.0 *** Kindergarden is available 0.45 0.52 *** 0.46 0.49 ** 0.47 0.50 * Medical point is available 0.47 0.59 *** 0.49 0.54 *** 0.49 0.56 *** Frequent energy interruptions 0.14 0.08 *** 0.16 0.08 *** 0.14 0.06 *** Clean water availability, % 80 81 77 83 *** 82 77 *** Irrigation water available, % 78 72 *** 81 72 *** 74 79 *** Number of attending people 145 168 *** 147 159 ** 149 163 ** Sources: Life in Kyrgyzstan (LiK) Survey, 2011 and 2012 Note: 1/ Community data are from 2011 wave. Significant differences are indicated by * p<0.1, ** p<0.05, *** p<0.01. 12 Labor market status seems to have a considerable impact on both knowledge and interest. Employed individuals are more likely to have knowledge and interest; those who are not employed (unemployed and inactive) are more likely to be uninformed and not interested. We find that social capital, level of trust and political attitudes play important roles in awareness and interest in local budget processes. The individuals who have larger social networks and belong to formal or informal social groups are more likely to respond positively. Community characteristics suggest more informed and interested individuals originate from smaller communities with better access to public services, such as healthcare and childcare as well as with a more reliable energy supply. Communities that have a designated meeting place proved to have residents that are more knowledgeable and interest in local budgets and the number of residents interested in these issues are also, on average, higher in communities where meeting places exist. The regression analysis of knowledge about and interest in the local budget process The regression analysis using a binary logit model largely supports the descriptive analysis. Most prominent factors that positively affect interest and awareness are largely related to a more proactive social and economic role of an individual. The factors that negatively affect interest and awareness seem to be determined by gender and ethnicity, as well as by external factors, such as residence in certain regions and availability of public services. As illustrated in Table 5, those individuals who have no interest or awareness about local budget processes tend to be female originating from non-Kyrgyz ethnic groups, be inactive or unemployed in labor markets, be a recent internal migrant, have no clear political preferences, and originate from multi-ethnic oblasts. The fact that the duration of residency plays some role in responses may indicate that recent internal migrants – quite rare phenomenon in rural Kyrgyz Republic – may be isolated from local public life. A higher individual trust to neighbors or public institutions seems to correlate with a higher degree of interest and awareness. In general, the level of trust is higher to local administrations that to central government, and is lowest for the police. Political proactivity plays some role in interest and awareness about local budget processes. For example, those who voted in the last presidential or local elections are more likely to be aware and have an interest in local budgets. In general, positive responses are higher for those who are supportive of the current party-based Kyrgyz political system – despite the fact that only about 15 percent of respondents are sympathetic to it. 13 Table 5: Marginal effects of the logit model of awareness and interest Indicator Interested in how Informed about Knows that local budget the local information about expenditures are budget issues local budget should done be available Personal characteristics Age 0.002 *** 0.002 *** 0.004 *** Female -0.050 *** -0.057 *** -0.040 *** Uzbek -0.116 *** -0.148 *** -0.106 *** Russian -0.008 -0.010 -0.042 Other ethnicity -0.010 -0.144 *** -0.127 *** Years of schooling 0.020 *** 0.020 *** 0.024 *** Risk attitude 0.011 *** 0.010 *** 0.012 *** Labor market status Inactive -0.041 ** -0.047 *** -0.063 *** Unemployed -0.090 * -0.095 ** -0.141 *** Social capital Internal migrant -0.070 * -0.042 0.090 ** Member of a social group 0.032 0.035 0.035 * Can get financial help from others 0.190 *** 0.171 *** 0.084 *** Trust in local government -0.013 0.012 0.015 * Political attitudes Supports current political system -0.078 *** -0.092 *** 0.030 No preference for political system -0.152 *** -0.134 *** -0.087 *** Is going to vote in local elections 0.050 *** 0.048 ** 0.040 ** Household characteristics Female HH head 0.009 0.022 0.059 *** Children 0.018 *** 0.019 *** 0.002 Income per person 0.002 0.003 -0.004 Land size -0.001 0.012 ** 0.000 Livestock 0.000 * 0.000 * 0.000 Asset index 0.000 *** 0.000 *** 0.000 *** Community characteristics1 Population 0.003 * 0.003 * -0.002 Kindergarden is available 0.026 0.018 0.030 Medical service is available -0.021 0.050 ** 0.025 Frequent energy interruptions -0.163 *** -0.122 *** -0.125 *** Oblasts Issyk-Kul 0.147 *** 0.027 0.026 Naryn -0.092 ** -0.026 -0.128 *** Batken -0.049 0.011 0.076 ** Osh -0.094 *** -0.049 * -0.153 *** Talas 0.294 *** 0.019 0.030 Chui 0.082 ** -0.131 *** -0.091 *** Pseudo R2 0.152 0.149 0.172 No of observations 5,042 5,042 5,042 Source: Life in Kyrgyzstan (LiK) Survey, 2012 Note: 1/ Community data are from 2011 wave Standard errors are corrected for cluster effects at the household level. Significant estimates are indicated by * p<0.1, ** p<0.05, *** p<0.01. ‘Kyrgyz’, ‘Employed’, and ‘Jalalabad oblast’ are the reference groups. Household characteristics indicate that awareness and interest may be driven very much by household economic status. Proactive individuals seem to originate from households that have fewer members, 14 enjoy higher income, possess larger plots of land, and own more livestock and other forms of assets. These characteristics suggest members of relatively poor households could be less active and uninterested in local formal institutions and processes, and thus, could be more isolated and under- voiced compared to better off households. There is a positive association of personal characteristics, such as age, education, and risk-taking attitude with awareness and interest in local budgets. However, females or individuals from non- Kyrgyz ethnic groups have lower level of knowledge and interest in local budgets. Low proactivity of women can be explained by traditional gender roles with dominant economic role of males in households, whereas women perform household chores and provide care for children and elderly. Risk-taking individuals tend to be interested and more knowledgeable about local budgets, though the effect is not particularly large. Alternatively, a negative association of interest and awareness about local budget processes with ethnic non-Kyrgyz, unemployed, and internal migrant status may indicate about some form of discrimination or isolation of these groups from the community life. A stronger social capital and higher level of trust are positively associated with familiarity and interest to local budget processes. The coefficients of variables that measure social networks, such as reliance of a person on other individuals in case of financial emergency and membership in social groups, tend to be largely positive and statistically significant. Similarly, more trust to local population is associated with more knowledge and interest in budget processes. Consequently, lack of political preferences is strongly and significantly associated with low awareness and interest in local budgets. Household demographic factors seem to play some role in awareness and interest in local budgets. More interest and awareness is observed among members of households that have more children and are headed by a female. There is no correlation between wealthier households being more aware and having more interest in local budget issues: superior incomes and assets holdings such as livestock do not seem to play significant roles. However, there is some positive, though very small, effect from land ownership - the main asset of rural households. The asset index, that represents ownership of various types of household assets – such as furniture, appliances, and access to key services - seems positively and robustly associated with all the variables of budget participation, albeit with a very small coefficient. Communities that are better off in terms of infrastructure and public services tend to have more interested and aware population. For example, this is the case for communities that have child care and medical service facilities. Those households that experience frequent outages in energy supply tend to have members with low interest and awareness. 15 Our analysis hints that population of multi-ethnic and more densely inhabited regions such as Osh and Chui oblasts tends to less informed in local budgets. Population of smaller oblasts, such as Talas, tends to be informed and interested more, however this cannot be generalized to other smaller regions such as Naryn and Batken. Participation in local budget processes Interest and awareness of individuals about local budgets does not necessarily mean that they are also active in public events related to budget process, such as budget hearings. Participation in such events implies a reasonably good understanding about the budget cycle and budget documents/reports, with participants ready to devote their time to attend such events. In this section, we look at whether an individual attempted to influence local budget processes or participated in budget hearings in the last 12 months and if a respondent was aware of the date of the recently conducted budget hearings. The proportion of respondents who participated in local budget events is considerably lower than the proportion of respondents who are informed and interested in budget issues, as it was illustrated in Table 3 earlier. For example, only four percent of respondents took part in local budget hearings, while 51 percent of respondents indicated their interest in local budgets. Despite this large difference in interest and awareness in local budget issues and participation in the related events, we find that individual characteristics, attitudes and preferences play a more prominent role in participation compared to household and community characteristics (Table 6). We document that the participants in the budget processes are individuals who are relatively older, have more schooling and belong to the Kyrgyz ethnic group. For instance, the participants in local budget hearings are an average of seven years older than non-participants and have in average 1.4 more years of formal schooling. We find that labor market status, social capital and political proactivity of participants is also consistently different between participants and non-participants. Additionally, we note that the differences in demographic, social and economic indicators between participants and non-participants tend to be relatively large compared to the differences observed in the analysis of awareness and interest. Analysis of household and community characteristics suggest that economic and infrastructure factors play a stronger role than household demographic factors. The participants in the budget processes tend to originate from wealthier households if to judge from productive asset holdings such as land and livestock. Their communities also experience fewer energy interruptions and enjoy wider access to public services, such as childcare, clean water and irrigation. 16 Table 6: Participation in local budget events Indicator Someone from my Knows the date of Participated in (in ratio, if not indicated otherwise) household tried the last budget budget hearing of to influence local hearing in the local Aiyl Aimak during budget process in government the last 12 months No Yes No Yes No Yes Personal characteristics Age, years 40.2 43.3 ** 40.1 44.8 *** 40.1 47.0 *** Female 0.52 0.39 *** 0.53 0.34 *** 0.53 0.30 *** Kyrgyz 0.72 0.87 *** 0.71 0.91 *** 0.71 0.89 *** Years of schooling 10.3 11.2 *** 10.2 11.7 *** 10.3 11.7 *** Risk attitude, 0->10 scale 4.6 6.6 *** 4.6 5.9 *** 4.6 6.2 *** Labor market status Employed 0.60 0.70 *** 0.60 0.80 *** 0.60 0.80 *** Inactive 0.37 0.28 ** 0.38 0.19 *** 0.37 0.19 *** Unemployed 0.03 0.02 0.03 0.01 ** 0.03 0.01 Social capital Lived in the community, years 38 41 ** 38 44 *** 38 45 *** Member of a social group 0.11 0.27 *** 0.10 0.39 *** 0.11 0.35 *** Can get financial help from others 0.81 0.88 ** 0.81 0.93 *** 0.81 0.91 *** Trust, in a scale from 1->4 Trust in people in community 3.2 3.0 *** 3.1 3.3 *** 3.1 3.2 Trust in local government 2.7 2.8 * 2.7 2.9 *** 2.7 3.0 *** Trust in central government 2.6 2.7 2.6 2.8 *** 2.6 2.9 *** Trust in police 2.3 2.3 2.3 2.4 2.3 2.4 Political attitudes Supports current political system 0.15 0.14 0.14 0.21 *** 0.15 0.18 Supports Russian political system 0.20 0.29 *** 0.20 0.24 * 0.20 0.25 * Has no political system preference 0.34 0.27 ** 0.24 0.11 *** 0.34 0.32 Voted in last Presidential election 0.86 0.94 *** 0.86 0.95 *** 0.86 0.96 *** Is going to vote in local elections 0.77 0.93 *** 0.77 0.88 *** 0.77 0.91 *** Household (HH) characteristics Female head 0.19 0.12 ** 0.19 0.17 0.19 0.19 HH size, people 5.9 5.6 * 5.9 5.8 6.0 5.4 *** Income per person, '000 Soms/month 3.5 3.6 3.4 4.6 *** 2.0 3.8 Land size, ha 1.22 1.68 *** 1.19 1.97 *** 1.21 1.87 *** Livestock units, sheep equivalent 22.0 40.9 *** 21.8 37.3 *** 22.1 37.7 *** Number of cars 0.45 0.49 0.44 0.53 *** 0.45 0.49 Asset index, HH level mean=100 107.0 138.6 *** 105.4 152.2 *** 106.6 142.7 *** Housing value, mln.Soms 0.90 0.76 *** 0.90 0.81 ** 0.90 0.80 ** Community characteristics1 Population, '000 people 6.5 5.4 ** 6.5 6.6 6.5 5.7 * Kindergarden is available 0.48 0.52 0.47 0.59 *** 0.47 0.59 *** Medical point is available 0.51 0.51 0.51 0.52 0.51 0.49 Frequent energy interruptions 0.12 0.08 0.12 0.05 *** 0.12 0.05 *** Clean water availability, % 80 75 ** 80 86 *** 80 80 Irrigation water available, % 76 85 *** 76 79 * 76 84 *** Number of attending people 153 162 154 151 155 126 * Sources: Life in Kyrgyzstan (LiK) Survey, 2011 and 2012 Note: 1/ Community data are from 2011 wave. Significant differences are indicated by * p<0.1, ** p<0.05, *** p<0.01. 17 This comparative analysis provides an illustration of the important factors of participation in local budget processes. The following sub-section describes results of the regression analysis in more details. Regression analysis of participation The regression model of participation in local budget processes based on a binary logit model reveals that largely the same factors at play. We find that participation is positively associated with older age, better education, higher risk-taking attitudes, membership in social groups, and higher trust in local government (Table 7). The negative determinants of participation are associated with being a female, originate from a non-Kyrgyz ethnic group, being inactive in labor market, and have unclear political preferences. The regression analysis confirms that education, employment, and social status are key determinants of individual and household standing in the community. While the descriptive analysis presented earlier reveals that superior economic status of respondents is robustly associated with proactive behavior of individuals, we do not find compelling support from the regression analysis for the hypothesis that the economic position of households has a positive relationship with participation in local budget processes. While land size seems to have marginally positive association with participation, there is no effect from the income or asset variables. The regional dummies reveal that only Osh oblast has negative and strongly significant coefficients (compared to Jalalabad oblast – the reference region). We do not have a compelling explanation for this; however, one possible reason for these differences is the post-conflict environment of the region. Osh oblast was a major area affected by the interethnic violent conflict between ethnic Kyrgyz and Uzbeks in June 2010. Despite the reconstruction and peacemaking efforts made after the conflict, the local population may still suffer from lack of trust in each other, including trust in local administrations. 18 Table 7: Participation in local budget events Indicator Someone from my Knows the date of Participated in household tried to the last budget budget hearing of influence local hearing in the local Aiyl Aimak during budget process in government the last 12 months the last 12 months Personal characteristics Age 0.000 ** 0.001 *** 0.001 *** Female -0.010 ** -0.031 *** -0.027 *** Uzbek -0.002 -0.032 -0.003 Russian -0.035 0.000 -0.025 Other ethnicity -0.038 ** -0.057 ** -0.060 *** Years of schooling 0.005 *** 0.008 *** 0.007 *** Risk attitude 0.006 *** 0.003 ** 0.003 ** Labor market status Inactive -0.007 -0.032 *** -0.028 *** Unemployed 0.004 -0.027 -0.012 Social capital Internal migrant 0.010 -0.007 0.004 Member of a social group 0.024 *** 0.056 *** 0.023 *** Can get financial help from others 0.016 0.028 0.010 Trust in local community -0.010 ** 0.007 -0.001 Trust in local government 0.007 * 0.007 0.011 *** Political attitudes Supports current political system -0.003 0.000 -0.003 No preference for political system 0.012 -0.019 -0.032 *** Is going to vote in local elections 0.040 *** 0.008 0.017 * Household characteristics Female HH head -0.010 0.016 0.016 ** Children 0.000 0.005 * 0.000 Income per person -0.001 0.002 ** 0.000 Land size 0.001 0.004 ** 0.002 * Livestock 0.000 0.000 0.000 Asset index 0.000 0.000 * 0.000 1 Community characteristics Population 0.000 0.002 ** 0.000 Kindergarden is available -0.009 0.022 ** 0.013 * Medical service is available 0.002 -0.029 ** -0.013 Frequent energy interruptions 0.004 -0.014 -0.010 Oblasts Issyk-Kul -0.033 ** -0.026 -0.033 ** Naryn 0.015 0.001 0.021 * Batken -0.023 * 0.005 0.011 Osh -0.039 *** -0.048 *** -0.036 *** Talas -0.037 * 0.027 0.010 Chui -0.009 -0.030 -0.016 Pseudo R2 0.163 0.214 0.227 No of observations 5,042 4,859 5,042 Source: Life in Kyrgyzstan (LiK) Survey, 2012 Note: 1/ Community data are from 2011 wave Standard errors are corrected for cluster effects at the household level. Significant estimates are indicated by * p<0.1, ** p<0.05, *** p<0.01. ‘Kyrgyz’, ‘Employed’, and ‘Jalalabad oblast’ are the reference groups. 19 6. DISCUSSION The descriptive and regression analyses reveal that higher social status or a position of an individual – such as age, education, and employment – plays a positive role in interest, awareness, and participation in these processes. Profiles of those individuals who are not interested in and not informed about local budgets are also rather telling. These are females, residents from non-Kyrgyz ethnic groups, especially Uzbeks; recent internal migrants; and inactive/unemployed. This is in line with the persisting gender inequality, poor communication to the country’s minority groups, and uprooted status of internal migrants in a country where belonging to a community is very important. It is within this context that targeted information campaigns for women, youth, internal migrants, or ethnic groups in their native language could mitigate this perceived isolation and discrimination. The analysis further reveals that a portion of households has no single adult members interested, knowledgeable or active in local budget processes. We find that about 40 percent of households are completely unaware and not interested in what is happening to the money that is supposed to be raised from and spent on their communities: no single adult member responded affirmatively to the local budget related questions. These households are probably extreme cases compared to households where at least one member is informed. The households with at least one informed and knowledgeable member may also serve as focal points to other household members or to households in nearby neighborhoods. Furthermore, the analysis suggests an indirect correlation between local budget knowledge, interest and participation with a socioeconomic status or poverty dimension. The individuals who express a higher level of knowledge, interest and participation in local budgets are distinctly better off economically compared to less active individuals. However, the regression analysis points to activeness being not directly related to economic status, but indirectly through other individual factors, such as education, employment, and social connections. Certainly, from a poverty reduction perspective, engaging the poor and disfranchised in the budget process, arming them with budget literacy and access to information is the first step to allowing them an opportunity to voice their collective needs at the local level. This analytical element is important as policy makers and development institutions are interested in organizing participation of all population groups, and especially those who are traditionally are left out of decision making, in the local governance and local budget discussions and decisions. While we conclude interest and awareness about local budgets is relatively high in the Kyrgyz Republic, the question is how to make citizens actively involved in local budget formulation and how the execution of empowered participation is more likely to be achieved at the local level (Khagram et 20 al., 2013). As it is an iterative process for local populations and local administrations, the importance of continuous awareness raising efforts for both parties seems warranted (Berner, 2001). Successful experience in affecting the local budget decisions and implementation may improve the trust in and credibility of local administrations. In turn, an interested and actively engaged citizenry provide incentives for local administrations to work transparently and gain efficiency through open exchange with the service. Evidently, information sources for raising awareness are crucial in reaching various demographic and social groups. Television as a channel to deliver information about local budgets seems, in the Kyrgyz context, to be most promising as about 94 percent of respondents in the LiK survey indicated television as a main source of news and information. Newspapers are only used as a source of information by a quarter of respondents, and are thus less effective in targeting the population. About 60 percent of people also use informal sources of information by talking to neighbors, friends and colleagues. This quantitative study brings new evidence and highlights important policy related aspects of citizen interest and participation in local budgets in rural Kyrgyz Republic by using a novel and multi-topic household and individual survey. However, the study does not go beyond descriptive statistics and determinants of citizen proactivity. It also does not look into mechanisms and intensity of civic engagement and participation, and cannot speculate that a higher participation leads to better local development outcomes. The Kyrgyz Republic has been an unusual country in the Central Asia region that went far ahead with governance and fiscal decentralization and encouraged large participatory- based rural projects supported by development institutions (such as Village Investment Project, supported by the World Bank). These developments and advancement of research infrastructure, such as the Life in Kyrgyzstan panel survey, may shed in the future more light on how a bottom-up governance system responsive to citizens may work more efficiently. 21 7. REFERENCES Ahmad, R., & Weiser, E. T. (2006). Fostering Public Participation in Budget-making: Case Studies from Indonesia, the Marshall Islands, and Pakistan. Berner, M. (2001). Citizen Participation in Local Government Budgeting. Popular Government, Spring. Brück, T., Esenaliev, D., Kroeger, A., Kudebayeva, A., Mirkasimov, B., & Steiner, S. (2014). Household survey data for research on well-being and behavior in Central Asia. Journal of Comparative Economics, 42(3), 819–835. doi:10.1016/j.jce.2013.02.003 Im, T., Lee, H., Cho, W., & Campbell, J. W. (2014). Citizen Preference and Resource Allocation: The Case for Participatory Budgeting in Seoul. Local Government Studies, 40(1), 102–120. doi:10.1080/03003930.2013.812963 International Budget Partnership. (2012). Open Budget Survey 2012. Irvin, R. A., & Stansbury, J. (2004). Citizen Participation in Decision Making: Is It Worth the Effort? Public Administration Review, 64(1), 55–65. doi:10.1111/j.1540-6210.2004.00346.x Khagram, S., Fung, A., & de Renzio, P. (2013). Open Budgets: The Political Economy of Transparency, Participation, and Accountability. Brookings Institution Press. The World Bank. (2014). Worldwide Governance Indicators: Country Data Report for Kyrgyz Republic, 1996-2013. Washington D.C. Wampler, B. (2012). Participatory Budgeting: Core Principles and Key Impacts. Journal of Public Deliberation, 8(2), 1–13. 22 8. APPENDIX 1: Table A.1: Local Budget module of the Life in Kyrgyzstan survey for 2012 8.D. LOCAL BUDGET Now I would like to ask your opinion and involvement in the process of budget decision of local government authorities Very well 1 How do you think, how informed are you about the Well 2 I828 budget of your local government body (Aiyl Aimak Badly 3 or municipal government)? Absolutely not informed 4 Please estimate, how much it is interesting to you, Very interesting 1 how your local government body (Aiyl Aimak or municipal Interesting 2 I829 Little interesting 3 government) spends your money from taxes and payments? Absolutely not interesting 4 Do you know that your local government body should Yes 1 provide local population with I830 No 2 information about local budget? Easily accessible 1 Estimate, please, how accessible is information about Accessible 2 the I831 Accessible with difficulties 3 budget of Aiyl Aimak for population? Inaccessible 4 How do you think, can an ordinary citizen influence Yes 1 budget process of his/her local I832 No 2 government body (Aiyl Aimak or municipal Did you or members of your household try to influence the budget process of your local Yes 1 I833 government body (Aiyl Aimak or municipal government) during the last 12 months No 2 (since November 2011)? When was the last budget hearing in your local government body (Aiyl Aimak or I834 ._ (month. year) municipal government)? Did you participate in budget hearing of your Aiyl Aimak Yes 1 during the last 12 months I835 (since November 2011)? No 2 Influences strongly 1 In your opinion, how much does the participation of citizens in the process of local budget production Influences few 2 and I836 implementation influence on quality of governmental Doesn't influence 3 services, provided on local level? Do not know 99 23 Table A.2: Overview of key variables Variable Definition Personal characteristics Age Age of an individual in full years Female 1=household head is female, 0=male Kyrgyz, Uzbek, etc 1=household head is Kyrgyz (Uzbek,..), 0=otherwise Years of schooling Reported years of schooling Internal migrant Number of years a person lived in community Risk attitude Self-assessed risk-taking attitude in a scale from 1 to 10 Trust questions Self-assessed trust level in a scale from 1 to 4 Household characteristics Household size Number of members listed by a household; external migrants and students are excluded Migrants abroad Number of adults in labor migration Income per capita Household income from all sources/household size, in thousand Soms Land Size of owned land in hectares Cars Number of cars/vans a household owns Livestock Livestock units in sheep equivalent; 1 cow = 5 sheep Asset index An index calculated by means of the first principal component analysis based on 13 household assets and access to services. The mean index at household level is 100. Location and community South oblast 1=if Batken, Jalalabat, or Osh oblasts, 0=Chui, Issyk-Kul, Naryn or Talas oblasts Population Population in the community, in thousands Frequent energy interruptions Energy interrupted several times per week Sources: LiK 2012 and 2011 Table A.3: Responses to local budget questions by rural and urban individuals LiK 2012 Indicator LiK 2012 rural urban Informed about the local budget 34.5 27.3 *** Interested in how the local budget is spent 51.4 60.7 *** Knows that local government should provide info about 31.0 15.4 *** local budget Information on local budget is accessible 42.5 43.4 An ordinary citizen can influence budget process 11.7 13.2 ** Me/my household members tried to influence local budget 3.5 2.5 ** process in the last 12 months Knows the date of the last budget hearing in the local 5.8 3.2 *** government Participated in budget hearing of Aiyl Aimak during the 4.0 1.6 *** last 12 months 5,145 3,022 *** No. of observations Source: LiK 2012. The mean differences in responses of rural and urban respondents are tested using t-test. Significant differences are indicated by * p<0.1, ** p<0.05, *** p<0.01. 24 Table A.4: Responses to local budget questions by oblasts Percentage of positive responses Issyk- Indicator Jalalabad Naryn Batken Osh Talas Chui kul Informed about the local budget 53.9 40.2 41.7 44.9 29.9 48.9 19.9 Interested in how the local budget is 75.5 52.4 47.2 50.1 37.8 82.4 56.4 spent Knows that local government should 53.7 36.7 31.7 50.1 18.2 54.9 24.1 provide info about local budget Information on local budget is 56.8 49.4 43.8 52.5 35.4 68.2 32.0 accessible An ordinary citizen can influence 19.7 16.1 25.2 14.8 4.0 14.2 11.1 budget process Me/my household members tried to influence local budget process in the 3.5 5.0 14.1 4.3 1.4 3.0 2.6 last 12 months Knows the date of the last budget 5.9 6.6 12.4 10.3 3.5 15.9 2.9 hearing in the local government Participated in budget hearing of Aiyl Aimak during the last 12 2.8 4.4 13.1 8.4 1.9 8.6 2.3 months 458 946 290 419 1,705 233 1,094 No. of observations Source: LiK 2012 The mean differences in responses of North and South respondents are tested using t-test. Significant differences are indicated by * p<0.1, ** p<0.05, *** p<0.01. 25