WPS5181 Policy Research Working Paper 5181 Citizen-centric Governance Indicators Measuring and Monitoring Governance by Listening to the People and Not the Interest Groups Maksym Ivanyna Anwar Shah The World Bank World Bank Institute Governance Division January 2010 Policy Research Working Paper 5181 Abstract Governance indicators are now widely used as tools for governance and implementing a uniform and consistent conducting development dialogue, allocating external framework for measuring governance quality across assistance, and influencing foreign direct investment. This countries and over time based on citizens' evaluations. paper argues that available governance indicators are not Using data from the World Values Survey (and other suitable for these purposes as they do not conceptualize sources) we implement this framework into practice governance and fail to capture how citizens perceive the and build citizen-centric governance indicators for 120 governance environment and outcomes in their countries. countries over the period 1994 to 2005. The paper attempts to fill this void by conceptualizing This paper--a product of the Governance Division, World Bank Institute--is part of a larger effort in the department to develop analytical methodologies for governance assessments to guide public sector reform efforts in developing countries. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at shah.anwar@gmail.com. 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 Citizen-centric governance indicators: Measuring and monitoring governance by listening to the people and not the interest groups Maksym Ivanyna Bavarian Graduate Program in Economics (Regensburg, Germany) and World Bank Anwar Shah, World Bank Keywords: citizen-centric governance indicators, rule of law, fair governance, service delivery, quality of life, trust in government, government oversight, public opinion JEL-classification: H10, H11, H83, I31, O10 Citizen-centric governance indicators: Measuring and monitoring governance by listening to the people and not the interest groups Maksym Ivanyna Bavarian Graduate Program in Economics (Regensburg, Germany) and World Bank Anwar Shah World Bank 1 Introduction Since the publication of pioneering work on measuring governance quality by Huther and Shah (1998), there has been a proliferation of composite worldwide governance indicators purporting to measure various aspects of governance quality (see Arndt, 2008, for the history and politics of governance ratings). The growth of these indi- cators has been spurred by generous support by the development assistance commu- nity, especially multilateral development finance agencies, and the infinite appetite of media and the academic community for governance assessments and country rank- ings. Governance indicators are now being used as tools for conducting development dialogue, allocating external assistance and influencing foreign direct investment. Each new indicator series is now released with great fanfare from major industrial country capitals and the popular press uses these indicators to name and shame individual countries for any adverse change in rank order over time or across coun- tries. The development assistance community is increasingly using these indicators in making critical judgments on development assistance. The World Bank's Interna- tional Development Association (IDA) allocation - a window of subsidized lending to the developing world and the United States Agency for International Develop- ment's Millennium Challenge Account use various governance indicators as criteria for allocating external assistance. At the same time, some of the recent findings of these indicators have also led to much controversy and acrimony and thereby con- The views expressed in this paper are those of the authors alone and should not be attributed to World Bank and its Executive Directors. Authors are grateful to the participants of Advanced Aca- demic Update "Governance Indicators and Assessments - Impact and Future Trends" (Maastricht Graduate School of Governance, Netherlands, January 21-22, 2009), Bavarian Graduate Program in Economics Research Workshop (Bayreuth, Germany, January 22-23, 2009) for comments on earlier versions of this paper. Comments are welcome and should be addressed to: ashah@worldbank.org. 1 tributing to complicating the dialogue on development effectiveness.1 In view of the influential nature of these indicators and potential to do harm if judgments embod- ied in these indicators are biased and erroneous, it is imperative that they capture critical dimensions of the quality of governance and all countries are evaluated using uniform and reasonably objective assessment criteria. Do the existing indicators meet this test? While the literature on this subject is woefully inadequate and thin, four widely used indicators - namely the World Bank's Worldwide Governance Indicators (WGIs), Overseas Development Institute's World Governance Assessments (WGAs), Mo Ibrahim Foundation's Indexes of African Governnace (IIAGs) and the United Nations Economic Commission for Africa's African Governance Report Indicators (AGRIs) - all lack a conceptual framework on governance, lack a citizen-based evaluations and have time and country assess- ment inconsistencies, making their rankings suspect. A number of recent papers have been especially critical of WGIs for lacking "concept" (implying lack of clar- ity in conceptualization) and "construct" (implying lack of clarity in measurement) validity, sample bias (mostly interest group views), lack of transparency and time inconsistency of definitions and measurements (see Arndt, 2008, Arndt and Oman, 2006, Kurtz and Schrank, 2007, Iqbal and Shah, 2006, 2008, Langbein and Knack, 2008, Schrank and Kurtz, 2008, Thomas, 2006, Thompson and Shah, 2003). One of the most important limitations common to all available composite indexes of gover- nance is that they fail to capture how citizens perceive the governance environment and outcomes in their own countries. For governance assessments to be useful for policy purposes, they must concep- tualize governance and provide uniform and consistent criteria for measuring gov- ernance across countries and over time. Foremost concerns for such measurement should be citizens' evaluation of governance environment and outcomes in their own countries supplemented of course by objective indicators of the same. For develop- ment assistance purposes, these indicators could be supplemented by expert-based evaluations. There is some work available on objective indicators as done by the Doing Business indicators of the World Bank and on expert-based evaluations as done for the Global Integrity Index. The most important void in our knowledge is how citizens view governance environment and outcomes in their countries. This paper takes a first step to fill that void. The rest of the paper is organized as follows. Section 2 discusses conceptual issues in measuring governance, specifies a citizen-centric conceptual framework on mea- suring governance quality. Section 3 presents an empirical framework, data sources and aggregation techniques. Section 4 presents preliminary results. In Section 5 we discuss the robustness of our results, as well as the contributions and limitations of the empirical approach. A concluding section outlines an agenda for future research. 1 See Iqbal and Shah (2008) for examples of indefensible country ranking and questionable cross- country and time series comparisons by one of the more widely used indicators 2 2 Conceptualizing and measuring governance quality in a comparative context Governance is a fuzzy yet fashionable buzzword and its use in the literature has exploded in recent years. Dixit (2008) notes that there were only 4 citations in EconLit in the period 1970-1979 compared to 15455 in the most recent period of 2000-2007 and currently Google lists more than 152000 pages of this literature. According to American Heritage, Random House and Merriam Webster dictionaries, governance is equated with government and is defined as the "exercise of authority and control" or "a method or system of government and management" or "the act, process or power of governing". Huther and Shah (1998) defined governance as "a multi-faceted concept encompassing all aspects of the exercise of authority through formal and informal institutions in the management of the resource endowment of a state. The quality of governance is thus determined by the impact of this exercise of power on the quality of life enjoyed by its citizens" (p.2). The World Bank Governance and Anti-corruption (GAC) Strategy (World Bank, 2007) defines it as "the manner in which public officials and institutions acquire and exercise the authority to shape public policy and provide goods and services" (p.3). For our current purpose, none of the above definitions with the sole exception by Huther and Shah, is helpful in serving as an operational guide to carry out a comparative review of quality of governance across countries or even of one country over time. This is because of their singular focus on the processes/institutions which do not lend themselves to easy or fair comparability across countries and sometimes not even within one country without conducting deeper analytical studies. There can be little disagreement that same processes and institutions can lead to diver- gent governance outcomes just as dissimilar processes could yield similar outcomes in two different countries. For example, anti-corruption agencies in countries with fair governance helps curtail corruption but in countries with poor governance prove either to be ineffective or worse a tool for corrupt practices and victimization (Shah, 2007). As another example, budget secrecy prior to its presentation to the parlia- ment is just as important under parliamentary form of government as in Canada, UK, India, New Zealand, as open and participatory budget determination process is to presidential form of government as in the USA. There can be little disagreement that both types of processes have the potential to advance public interest but may succeed or fail in different country circumstances. During the past two decades, we have also seen that single party dominant political systems in China, Malaysia and Singapore have shown dramatic results in improving governance outcomes whereas pluralistic party systems have also shown positive results in other countries such as Brazil and India. Similarly monarchy has shown positive results in UK but un- welcome results in Nepal. Even similar electoral processes do not always lead to representative democracy and may instead yield aristocracy (elite capture) in some countries and corrupt oligarchies in others. In fact, Aristotle's main argument for elections was based upon the premise that these would produce aristocracy, a form of government he considered superior to median voter rule (see Azfar, 2008). An- drews (2008) argues that such "good governance picture of effective government ... constitutes a threat, promoting isomorphism, institutional dualism and "flailing 3 states" and imposing an inappropriate model of government that "kicks away the ladder" today's effective government climbed to reach their current state."(p.2) In any case, such comparisons of processes and institutions out of their context are almost always ideologically driven and value laden and could not be acceptable as unbiased professional (scientific) judgments. This also explains that while citizens of Bangladesh, China, India and Malaysia over the last decade have experienced remarkable improvement in governance outcomes, available primary indicators fail to capture these accomplishments due to their focus on processes at the neglect of outcomes. Even for the world as a whole, the information revolution by letting the sun shine on government operations, has brought about dramatic improvements in government accountability, but the WGIs with their on one-size-fit all vision of the world, have consistently failed to notice or recognize such a mega change. These indicators rank China in the lowest percentile on voice and accountability but ac- cording to the former Auditor General of Canada, China has the most effective public accounts committee anywhere which has a track record of holding government to account for malfeasance (Dye, 2007). Furthermore local governments in China have relatively much larger role in public service provision than most countries. Local governments below the provincial level account for about 54% of consolidated pub- lic expenditures in China compared to about 4% in India and about 27% in OECD countries (see Shah and Shah, 2006). Thus having the decision making closer to people, directly elected local governments, and party oversight of local government performance - all work to create a system of voice and accountability that is quite unique to China and not easily comparable to other countries (see Qiao and Shah, 2006). China has also demonstrated superior government effectiveness through its unique and unparalleled success in alleviating poverty and improving the quality of life of its citizens over the past two decades. About two decades ago, China had about 35% of its population below poverty level compared to less than 2% in 2006 (see Shah and Shen, 2007). In conclusions, comparisons of governance institutions requires deeper analytical work through in-depth comparative studies rather than aggregate indicators. Such indicators are more usefully used to compare governance outcomes and complementary analytical studies of institutions and process can be used to explain varying outcomes. Of course, governance outcomes also assume commonly shared values but it is relatively less problematic than one-size fit-all prescriptions on processes. To have meaningful governance comparisons across countries and over time, one needs to have concepts which are somewhat invariant to time and place and are focused on citizens' evaluations rather than interest groups' views. To this end, we define governance as an exercise of authority and control to preserve and protect public interest and enhance the quality of life enjoyed by citizens. Note that this definition encompasses both the governance environment (quality of institutions and processes) as well as governance outcomes. 4 2.1 Towards a simple framework for assessing country governance quality Considering a neo-institutional perspective, various orders of government (agents) are created to serve, preserve, protect and promote public interest based upon the values and expectations of the citizens of a state (principals). Underlying assumption is that there is a widely shared notion of the public interest. In return, governments are given coercive powers to carry out their mandates. A stylized view of this public interest can be characterized by four dimensions of governance outcomes. · Responsive Governance. The fundamental task of governing is to promote and pursue collective interest while respecting formal (rule of law) and informal norms. This is done by government creating an enabling environment to do the right things - that is it promotes and delivers services consistent with citizen preferences. Further, the government carries out only the tasks that it is authorized to do that is it follows the compact authorized by citizens at large. · Fair (equitable) Governance. For peace, order and good government, the gov- ernment mediates conflicting interests, is focused on consensus building and inclusiveness and ensures a sense of participation by all and protection of the poor, minorities and disadvantaged members of the society. · Responsible Governance. The government does it right i.e. governmental authority is carried out following due process with integrity (absence of cor- ruption), with fiscal prudence, with concern for providing the best value for money and with a view to earning trust of the people. · Accountable Governance. Citizens can hold the government to account for all its actions. This requires that the government lets sunshine in on its operations and works to strengthen voice and exit options for principals. It also means that government truly respects the role of countervailing formal and informal institutions of accountability in governance. Given the focus on governance outcomes, Table 1 presents some preliminary ideas for discussion on how to operationalize these concepts in individual country assessments. The above simple framework captures most aspects of governance outcomes es- pecially those relevant for development policy dialogue and can serve as a useful starting point for a consensus framework to be developed. In any event, there can be little disagreement that one cannot embark on measuring governance quality without first defining and defending an appropriate framework that measures gover- nance - a point also emphasized by Thomas (2006) and the European Commission (see Nardo et al., 2005). Once a consensus framework is developed then one needs to focus on only a few key indicators that represent citizens' evaluations and could be measurable with some degree of confidence in most countries of the world and could be defended for their transparency and reasonable degree of comparability 5 Table 1: Governance outcomes and relevant considerations Governance outcome Relevant considerations · public services consistent with citizen preferences; · direct possibly interactive democracy; · safety of life, liberty and property; · peace, order, rule of law; Responsive · freedom of choice and expression; governance · improvements in economic and social outcomes; · improvements in quantity, quality and access of public services; · improvements in quality of life; · fulfillment of citizens' values and expectations in relation to participation, social justice, and due process; · access of the poor, minorities and disadvantaged Fair governance groups to basic public services; · non-discriminatory laws and enforcement; · egalitarian income distribution; · equal opportunity for all; · open, transparent and prudent economic, fiscal and financial management; · working better and costing less; Responsible · ensuring integrity of its operations; governance · earning trust; · managing risks; · competitive service delivery; · focus on results; · justice-able rights and due process; · access to justice, information; · judicial integrity and independence; Accountable · effective legislature and civil society oversight; governance · recall of officials and rollbacks of program possible; · effective limits to government intervention; · effective restraints to special interest capture. Source: Shah (2008) 6 and objectivity.2 Having an enormous number of indicators, which could not be scrutinized, is nothing but a distinct disadvantage for a measure that aims for wider acceptance and confidence. Implementation of the above framework requires a worldwide survey with uni- form questionnaire honing on the four dimensions of governance identified above across countries. Given that such a survey is not available and costly to commis- sion, in the following section, we take a pragmatic approach based upon available survey data to develop rough indexes of governance quality. 3 Citizen-centric governance: Empirical framework Following Table 1, public interest is characterized by four dimensions of governance outcomes - responsive governance, fair governance, responsible governance, and ac- countable governance. Each of these categories is split further on sub-categories in order to characterize a concrete governance outcome (such as improvements in qual- ity of life, safety, peace, etc.). Public opinion survey, with the questions assigned to each subcategory, should be used for the assessment of governance. The procedure of the assessment consists of the two main steps. First, data source - the raw data from inter-country public opinion survey - is chosen. The responses on questions in the survey, which characterize governance outcomes, are recorded. Second, the responses are aggregated in order to achieve governance index for each country from the sample. In what follows, we consider both steps in detail. 3.1 Data Reliable, comprehensive and consistent through time and space source of data is essential for qualitative estimation of citizen-centric governance indicators (CGIs). With an additional requirement of being publicly accessible and, preferably, free of charge, such data source hardly exists at present. There is a database of governance- related questions included into different surveys across the world (Governance Sur- veys Database published by the World Bank). In principle, each of these questions could be included into our estimation (questions taken separately from different polls) if the data is available. However, as the experiments in the construction of surveys suggest (see Bertrand and Mullainathan, 2001, for examples), even the small difference in the formulation of a question (assigned to the same sub-criterion) or the sequence of questions in a survey may bring significant discrepancies in the responses for the same country and same sub-criterion. Therefore, we decided to use only one data source, which covers sufficient amount of countries. Effectively it means, that almost the same questionnaire is used in all participating countries. 2 see Andrews and Shah (2005) for details and relevant indicators of an approach that emphasizes citizen-centric governance and Shah and Shah (2006) for citizen-centered local governance and relevant indicators 7 The principal data source for our further analysis is the World Values Survey (WVS) project, conducted by WVS Association (see WVS, 2008). Table A2 shows its characteristics in comparison with other potential data sources. WVS provides an acceptable compromise of consistency and coverage for showing an initial picture of citizen-centric governance indicators. On the one hand, WVS publishes quite outdated information (with the time lag of 2-3 years after actual survey was taken), and only a few questions from this survey are relevant for our purposes (since the survey is about cultural values, not governance). On the other hand, WVS provides quite comprehensive geographical coverage (97 countries with all major economies included) combined with acceptable time coverage and questionnaire. The coding (which is used further in text and in the dataset) and questions assigned to each sub-criterion of governance are presented in the Table A1 of Ap- pendix. As one can see, for a few sub-criteria, specified in the Table 1 of the paper, no survey questions are available. This is a drawback of WVS, as this survey was not constructed to evaluate governance. However, each governance outcome has a sufficient representation by questions in order to get reasonable estimates. Based on the data from WVS (questions from the Table A1 of Appendix), as well as from the other freely available data sources (AFR, ASB, TI GCB - see Table A2 for notation), a unique dataset was constructed, which can be used for the evaluation of citizen-centric governance indicators by any researcher. 421994 people's responses (256152 of them by WVS) on 74 different questions (20 from WVS) are recorded in this dataset. 125 countries are covered, 97 of them by WVS. The records in the dataset can be sorted by the gender, income, education of a respondent, as well as by the sub national administrative unit of his/her residency. For the reasons explained above our main estimation procedure is based on 3 waves of the World Values Surveys depending on the year when the surveys were taken. Wave 1 includes countries surveyed from 1994 to 1998, wave 2 - from 1999 to 2004, and wave 3 - from 2004 to 2008. In addition to questions from WVS, in the wave 3 we also use one question about corruption from Transparency International Global Corruption Barometer (see TI, 2005). As an alternative to the WVS, we apply additional data sources in our estimation of citizen-centric governance indicators. In particular, in this paper we report the results when using Gallup World Poll data points, which are available freely from the Worldwide Governance Indicators (WGI) project (see WBI, 2008).3 4 questions from GWP are used in WGI. While this coverage is quite limited, yet it allows us to estimate 3 governance outcomes for a wide range of countries. 3.2 Aggregation The underlying assumption of our empirical investigation is that the quality of governance in a given country directly affects governance outcome, which is being analyzed in a certain survey question. Thus, the answers of survey respondents - citizens of this country - are better for each question the higher is the quality of governance in the country. At the same time, answers of the respondents are random 3 Gallup World Poll, described in the Table A2, is itself very expensive (28 thousands US Dollars per year), and therefore cannot be used as a base for a rigorous, replicable research 8 variables, which are subject to personal errors: 1 1 sijk = k gi + ijk gi = sijk - ijk , (1) k k where i = 1, .., M is the index of a country, j = 1, .., Ni is the index of a respondent (total number of respondents, obviously, changes from country to country), and k = 1, .., K is the index of a question in a survey (thus of a particular governance outcome). sijk is the answer on question k of the respondent j in the country i. Each response was normalized by us on a scale from 0 to 1, with 0 being the worst answer, and 1 being the best answer. gi is the quality of governance in the country i. It does not depend neither on concrete respondent, nor on specific question. Coefficient k reflects a degree, to which governance affects the answer of a respondent. Note that 2 it does not depend on country or respondent. Finally, ijk N (0, ik is the personal random error of the respondent j in the country i, which may also depend on a specific question. Each error is independently normally distributed with zero mean 2 and the variance ik , which may depend on country and specific question. The expression for gi can be rewritten: gi = wk sijk - wk ijk , (2) where wk = 1k - are the question-specific weights assigned to each question. The weights are normalized to add up to one - K = 1 - so that gi is between 0 and 1 k=1 for each country. For our main estimation, and for further comparative analysis, the weights are exogenously chosen and are reported in the Table A1 of the Appendix. They reflect the relative importance of every question in assessment of governance (i.e. "satisfaction with life in general" is clearly more comprehensive than "satisfac- tion with health" or "satisfaction with environment"), as well as alleviate certain data deficiencies (i.e. European countries were not asked some questions in the sec- ond wave of WVS, so these questions received lower weight). At the same time, the weights can be easily changed to tailor one's specific research agenda or check the robustness of the results. Given our assumptions, the most efficient, unbiased, and consistent estimator for the governance in country i is just the sample mean of weighted averages of citizens' responses, the estimator for the governance's variance is adjusted sample variation: Ni K 2 1 K Ni Ni gi = ^ ^ wk sijk , var(gi ) = wk Ni1 2 sijk - 1 sijk . (3) -1 Ni Ni j=1 k=1 k=1 j=1 j=1 We gave up more sophisticated data mining approaches (e.g. principal component analysis, canonical analysis or random projections) for the sake of transparency and simplicity. The choice of weights or aggregate procedure does not significantly change the appearing governance picture (see Section 5). Our procedure is maxi- mally open and simple in order to allow for a further research and analysis. Besides, in addition to the governance scores we report and analyze the aggregate responses on each question, which makes our indicators "actionable", and allows drawing the conclusions, which are completely independent of weights and aggregation proce- 9 dure. 4 Citizen-centric governance: Preliminary rankings Based on the estimation procedure described above we report our results in this section. First, we analyze citizen-centric indicators (CGIs) as well as responses on separate questions in all countries in 3 waves of World Values Surveys and Gallup World Poll. Then we compare the indexes by groups of countries, through time (across 3 waves), and with other governance indicators (in particular, Worldwide Governance Indicators). In the last subsection, we give examples of sub-national CGIs in several countries. 4.1 Country rankings: Waves 1 to 3 The countries' citizen-centric governance indicators (CGIs) are presented in Figure 1 and Figure 2. On the first figure we show the estimations based on the data from World Values Survey, for the second figure we use the data from Gallup World Poll (see Section 3.1 for details about data sources). All 3 waves of WVS surveys are shown in Figure 1: (a) Wave 1 - for surveys taken between 1994 and 1998 (53 countries), (b) Wave 2 - for surveys taken between 1999 and 2004 (71 countries), (c) Wave 3 - for surveys taken between 2005 and 2008 (51 countries). The maps of citizen-centric governance evaluations are, in our opinion, more con- venient tool for analysis than the tables with more than 100 records, though those are also available from authors at the request. In Figure 1 we split our sample of countries into 3 broad categories (6 categories in Figure 2): from dark-green high- governance-quality countries to light-green low-governance-quality countries. While developed countries (especially Scandinavian countries and Switzerland) show sta- ble and high grades, it is rather unexpected that East Asian countries (especially, Vietnam, China) are relatively high rated. In some countries of the Middle East (Jordan, Saudi Arabia) the popular support of the government is also "unexpect- edly" high. At the same time, countries of Central and Eastern Europe are always in the lowest percentiles of the samples. In Figure 3 we compare citizen-centric governance indicators with correspond- ing Worldwide Governance Indicators (WBI, 2008), which are considered to be the "gold standard" of governance assessment by the media. The scale changes from dark-green for countries, which were severely underestimated by WGIs, to dark- red for countries, which were greatly overestimated. 27 out of 82 countries in our sample were over- or underestimated at a significance level less than 25% (9 at a level less than 5%) by WGIs in comparison to our assessments. The pattern de- scribed in the paragraph above is supported: Middle East and East Asian countries are mostly underestimated (with China, Vietnam, Iran and Saudi Arabia being the leading outliers), while Central and Eastern European countries are too praised by WGI (Latvia, Lithuania, Moldova and Hungary being the leading outliers). Appar- ently, our indicators reflect last decade's obvious successes of East Asian and Middle 10 Figure 1: Citizen-centric governance indicators (data source - WVS, waves 1-3) 11 Figure 2: Citizen-centric governance indicators (data source - GWP) Note: u. X-Y% means that the country was underestimated by WGI in comparison to CGI at the significance level between X and Y%; o. X-Y% means that the country was overestimated by WGI in comparison to CGI at the significance level between X and Y%. The time period considered is 1994-2005, aggregate CGIs are taken, WGIs are averaged over all 6 components Figure 3: CGI vs. WGI (Worldwide Governance Indicators) 12 Note: Averages on each governance outcome (as is defined in the Table A1) in the selected groups of countries: World - the whole sample, EU-15 - countries from European Union before the extension of 2004, CEE - Central and Eastern European countries, East Asia - East Asian countries (China, Taiwan, India, Indonesia, Korea, Malaysia, Vietnam, Thailand) Figure 4: WVS wave 3: governance outcomes by groups of countries East countries in economic outcomes. At the same time, WGIs rely more on the Anglo-Saxon institutional design of a government, which does not always lead to desired governance outcomes given local historical and institutional contexts (see our discussion in the Introduction). The disaggregated data are analyzed in Figure 4. Here we depict regional av- erages by each governance outcome (based on the data from the third wave of WVS). It can be seen that the curve of the EU-15 group - "old" members of the European Union - is almost always above other curves in the dimension of Respon- sive Governance (till the "happiness" point on the X-axis). When it comes to the questions about Responsive and Accountable Governance (confidence in parliament, government, press, TV, courts) the curve steeps down. The curve of the East Asian countries, while mostly above the world's average, rises above the curve of EU-15 only in trust-related dimensions. Similar properties (though with somewhat lower averages) have the curves of Middle East and African countries (the curves are not depicted in the figure to keep at least some tractability). The curve of Central and Eastern European countries (CEE) is always below East Asian curve, as well as the world's average. Particularly low (relative to others) citizens of CEE countries evaluate their confidence in police ("safety" on X-axis) and respect for human rights in their respective countries ("human rights" on the X-axis). The fact that people in the East Asia, Middle East and Africa trust their govern- ments more than the people in developed countries of Western Europe and North America may not only reflect the overall public satisfaction (or dissatisfaction) with governance outcomes. In depressed countries, it may also be the result of people's 13 fear to disclose their true opinion about government. Alternatively, when mass me- dia in a country are controlled by the government, people in this country may be indoctrinated to believe and trust those on the top. In the Section 5.2 we analyze these possible effects and their magnitude for the countries from our sample. 4.2 Intertemporal comparison The consistent through time questionnaires of the WVS and repeated surveys during three waves allow us to assess the progress of the governance in certain countries. In particular, citizens of 41 countries were surveyed both during the first wave of WVS (1994-98) and during the second wave (1999-2004). Surveys both from the second wave and the third wave (2005-2008) are available for 33 countries. In Table 2 we report the countries which achieved the biggest progress in each governance outcome (both from Wave 1 to Wave 2, and from Wave 2 to Wave 3). Not surprisingly, the list is dominated by developing countries and the countries in transition - of 110 positions (10 governance outcomes plus CGIs themselves) only 14 are taken by developed countries (Spain and Germany between waves 1 and 2, and Japan between waves 2 and 3). These numbers clearly reflect increase in the standard of living and stable economic growth in certain parts of the world. Especially it concerns the speedy economic recovery of CEE countries after the horrible post-communist "hangover" of the 1990s. The most commonly mentioned countries are Nigeria, Venezuela, Latvia, Bangladesh, Moldova between waves 1 and 2, and Turkey, Russian Federation, Jordan, India and South Africa between waves 2 and 3. The governance in the world (over the sample of countries surveyed by WVS) statistically significantly (at the level of less than 1%) increased from wave 1 to wave 2 (see Figure 5) - in contrast to the WGI's world of unchanging governance quality - but practically did not change from wave 2 to wave 3. As it can be seen from the figure the main driver of the growth in world's quality of governance was increasing (in practically all regions) satisfaction of the citizens with their financial situation. This trend was kept from wave 2 to wave 3 as well, but the overall progress was apparently mitigated by the fall of confidence in governments, courts, army, etc. in developing and countries in transition (though CEE countries still ended up progressing from wave 2 to wave 3). 4.3 Subnational CGIs Our estimation procedure as well as dataset collected allows us to extend citizen- centric governance indicators from countries to their subnational units. The idea is to aggregate the citizens' responses not over the whole country, but over its juris- dictions. For the Wave 3 of WVS there are 1121 of them in the sample - usually the second tier of a country's administrative structure (in some countries - groups of second tier jurisdictions). The examples of some countries are given in Figure 6. On the left we depict Germany, and on the right - Italy. Both countries were surveyed in 2006. In Germany 14 Table 2: CGI (WVS): top performers by the progress in time Governance outcome Top-performers: Wave Top-performers: Wave 1 to Wave 2 2 to Wave 3 Total CGI Nigeria, Germany, Turkey, Russian Fed- Venezuela, Latvia, eration, Jordan, South Finland Africa, India Responsive gover- nance safety of life, order, rule Macedonia, India, Morocco, Japan, of law Bangladesh, Nige- China, Korea ria, Venezuela, Latvia improvements in eco- Venezuela, Moldova, Turkey, Jordan, Ar- nomic and social out- Spain, Nigeria, Ar- gentina, Korea, South comes gentina Africa improvements in quality Estonia, Bulgaria, Turkey, Jordan, of life: general Moldova, Venezuela, Russian Federation Slovenia Ukraine, Moldova improvements in quality Nigeria, South Africa, Moldova, Jordan, Ar- of life: health Mexico, Bangladesh, gentina, Indonesia, Mo- BiH rocco peace Bangladesh, Latvia, Bulgaria, Italy, South India, New Zealand, Africa, Chile, Mexico Macedonia Responsible gover- nance earning trust: executive Venezuela, Nigeria, Turkey, Iraq, South branch New Zealand, Spain, Africa, Argentina, Ko- Albania rea earning trust: legislative Nigeria, New Zealand, Morocco, Turkey, branch Venezuela, Spain, Ger- South Africa, Korea, many India Accountable gover- nance access to information, in- Bangladesh, Germany, Bulgaria, Morocco, dependent mass media - Slovenia, Sweden, India Vietnam, Jordan, India press access to information, in- Albania, India, Morocco, Iraq, Viet- dependent mass media - Bangladesh, Nige- nam, Jordan, Egypt television ria, Venezuela judicial integrity and in- Macedonia, India, Japan, Morocco, dependence Bangladesh, Nige- China, Turkey ria, Venezuela, Latvia Note: Top performers - in each governance outcome (as defined in the Table A1) 5 countries with the biggest mean difference between corresponding waves 15 Note: Progress in time for some governance outcomes and CGI in 4 regions. First 2 columns for each outcome compare wave 1 and wave 2 over common sample of countries, columns 3 and 4 compare wave 2 and wave 3 over common sample of countries. Governance outcomes included are: "satisfaction with financial situation in the household", "peace" (confidence in the army), "confidence in government", and "confidence in courts". The regions: World - all countries in the samples, EU-15 - European Union members before the extension of 2004, CEE countries - Central and Eastern European Countries, East Asia - East Asian countries. Figure 5: CGI (WVS) waves 1-3: progress over time by regions 16 Note: left side - Germany, survey of 2006; right side - Italy, survey of 2006. The scale is common to both countries. Figure 6: Subnational CGI (WVS): examples rich industrial lands4 of Hessen, Nordrhein-Westfalen and Saarland together with independent cities of Bremen, Hamburg and Berlin are the most satisfied with their governments. At the same time, the scores are much lower in the poorer eastern part of the country - only in Sachsen-Anhalt citizen's gave their government more than 0.55 (the score of the land is 0.56). Surprising are the average scores received u by the governments of rich southern states - Baden-W¨rtemberg and Bayern. The relative correspondence between richness of a jurisdiction and it's govern- ment's score is also kept in Italy. Most regions of the rich country's North score more than 0.55. At the same time, most of the poorer South - with the exception of Abruzzo, Molise, and Basilicata regions - is below 0.55. Subnational CGIs is, to our knowledge, the first attempt to assess governance at less aggregate than the country level. Analyzing these may prove to be helpful in empirical research on decentralization and governance, decentralization and welfare, difference between capital and non-capital regions, industrialized and rural regions, etc. 5 Robustness Combination of survey data with the simple aggregation procedure raises quite a few questions about the validity and reliability of our results. In this section we try to 4 a L¨nder in German - second tier jurisdictions in the country 17 resolve some of them. First, we provide some arguments in favor of our aggregation procedure and overall analysis of the data. Second, we make a critical assessment of the data we have available. 5.1 Alternative aggregation techniques Transparency, simplicity and possibility to tailor the assessment procedure for one's research agenda are the main reasons behind adopting our aggregation procedure - taking weighted averages of citizens' responses. Besides, some questions are rel- atively more important and comprehensive for assessing governance, which cannot be detected by mechanized data mining algorithms. In addition, many of our find- ings and conclusions concern directly separate governance outcomes (responses on a separate question), which does not depend on aggregation procedure. Nevertheless, we use alternative aggregation techniques to test the robustness of our results. In particular, we apply uniform weights to our data, as well as we use averaging over percentile rankings (the way it is done in the Doing Business project - Djankov, 2007). Naturally, both methods produce slightly different rank- ings comparing to our main methodology. In particular, European countries lose some positions and East Asian countries gain - the result of increased reliance on the governance outcomes, which are related to trust and confidence in governmental institutions. However, only 11 of 51 countries in case of uniform weights (10 out of 51 in case of averaged percentile rankings) significantly change their standing (according to classification provided in the Figure 1, wave 3 - when country changes one of three categories). 5.2 Adjusting the data In our estimation we use survey data from countries around the world, and the public opinion in a country - especially about the issues related to the government - might be influenced by factors, which we would definitely like to account for. One of the factors is so-called "intimidation" effect, when people are afraid to express their true - negative - opinion about their government, because they think they could be punished for that. Another factor, frequently mentioned in the literature, is the "indoctrination" effect, when mass media in a country praise the government so much, that it has a significant positive impact on public opinion. Another factor is the degree of citizen activism and perceived role of government in a country. In particular, Norris (1999) argues about the emergence in the 70s in developed countries of the class of so called "critical citizens" - people, who were becoming more and more critical and demanding towards their governments despite their obvious successes. Taking into account 3 factors mentioned above ("intimidation", "indoctrination", "critical citizenship") we conclude that in general a response on a question about governance outcome of an individual might be affected not only by the quality of governance in a country. The true model can be rewritten in the following way: sijk = ik + k gi + ik intij + ikindij + µik cr citik + ijk , (4) 18 where similarly to the notation from Section , sijk is a response of an individual j in a country i on a question k, gi is the quality of governance in a country i, , and ijk is a citizen-, country- and question-specific error. intij , indij , cr citij are the degrees of intimidation, indoctrination and critical citizenship of an individual j in a country i. ik , ik and µik - depending on country and question - are the coefficients of our interest. The estimation of ik , ik and µik is not possible from the model above, since we do not observe governance gi (this is in fact what we are trying to assess). However, the problem can be resolved if we note, that for some questions (governance out- comes) there are no effects of intimidation, indoctrination or critical citizenship, and for some there are. For instance, when an individual is asked about the satisfaction with her/his health, it is likely that she/he will not be intimidated to say true. At the same time, questions like "Do you have confidence in your government?" are most probably subject to all above mentioned effects. Therefore, by taking the dif- ference between the answers on these questions we can get rid of the governance on the right-hand side while intimidation, indoctrination and critical citizenship effects remain. The estimation model than become: K1 K2 1 1 dif fij = sijk - sijk = i +i intij +µi indij +i cr citij +ij , (5) K1 k=1 K2 - K1 k=K1 +1 where sijk , k = 1, .., K1 are the citizens' answers on the questions, which are exposed to the biasing effects (intimidation, indoctrination, critical citizenship), sijk , k = K1 + 1, .., K2 are the answers on the questions with no role for above mentioned effects. Therefore, the left-hand side of our model is the difference between the averages of the two groups of questions (governance outcomes). Assuming that these groups of governance outcomes explain governance to the same degree (average k 's are the same) we get rid of the quality of governance in the right-hand side, and can test for ik , ik and µik directly. After taking into account these effects the estimator for the quality of governance can then be expressed as: Ni K K Ni 1 1 gi = wk sijk - wk (i intij + µi indij + i cr citij ) (6) Ni j=1 k=1 k=1 Ni j=1 gi is now the weighted average of people's responses (the formula we adopted in the main body of the paper) less the effects of intimidation, indoctrination and critical citizenship - averaged over all residents of a country surveyed and multiplied by the weight of the questions in the survey, which are exposed to these effects. We assume the following questions (governance outcomes) to be independent from the bias effects: · How satisfied are you with the financial situation of your household? (im- provements in economic and social outcomes) · All things considered, how satisfied are you with your life as a whole these days? (improvements in quality of life: general) · All in all, how would you describe your state of health today? (health) 19 · How serious do you consider poor water quality, air quality, sewage and sani- tation to be here in your own community? (environment) · Taking all things together would you say you are [happy, unhappy]? (happi- ness) On the opposite, the following questions (governance outcomes) are assumed to be exposed to bias effects: · How much confidence do you have in government? (trust: executive branch) · How much confidence do you have in parliament? (trust: legislative branch) · How much confidence do you have in press? (trust: press) · How much confidence do you have in television? (trust: television) · How much confidence do you have in courts? (trust: courts) 5.2.1 Testing for the intimidation, indoctrination and "critical citizenship" effects We use 2 types of estimation procedures to extract i , i and µi - effects of intim- idation, indoctrination and "critical citizenship" in a country i. First, we test for indoctrination (i ) on an individual level, since there can hardly be any proxy for biasedness of mass-media (indoctrination) on a country-level. On a contrary, it is hard to come up with the proxies for personal intimidation and "critical citizenship" (this effect was in fact defined only for countries as a whole). That is why we use country-level regressions to identify these effects. As the proxy for indoctrination we take the frequency, with which an individual exposes her- or himself to media - TV and press. Specifically, we use questions "Did you watch TV during the last week?" and "Did you read newspapers last week?" from the World Values Survey. The more people watch TV or read newspaper the more they are exposed to possible indoctrination (or excessive criticism of mass- media). The exact estimation model then becomes: dif fij = i + 1i tvij + 2i pressij + i demogrij + ij , (7) where tvij , pressij are the dummies for watching TV and reading newspapers last week (as it was posed in the questions of the survey), demogrij is a set of individual demographic variables (we take respondent's education, income, age, marital status, political activism - participation in demonstrations, boycotts, signing petitions). We report the results in the Table 3. The main conclusion from it is that even though developing countries, especially those in Middle East and East Asia, seem to be indoctrinated, the mass media bias is also present in many developed countries - Japan, Sweden, Switzerland, USA, France. This might be the outcome not of state monopoly (or dictate) on mass media, but of too optimistic or patriotic news coverage in these countries. The magnitude of the indoctrination effect ranges from 0.02 (except for Ukraine and Rwanda, where those who watch TV are actually more 20 Table 3: Mass media bias in public opinion Media bias, magnitude TV Press (1i , 2i ) 0.08 - 0.12 Japan, Mexico, India, Slovenia, Cyprus, Thailand, Cyprus Ethiopia 0.04 - 0.08 Sweden, Switzerland, Brazil, Turkey, Peru, Jordan, Malaysia Moldova, Indonesia, Vietnam, Serbia, Egypt, Andorra, Burkina Faso, Zambia, France 0.02-0.04 China USA, Mexico, Brazil, Romania, Egypt Argentina, Australia, Bulgaria, Chile, Taiwan, Colombia, Finland, Germany, Ghana, Italy, Republic of Korea, Mali, Morocco, Nether- 0 lands, Poland, Russian Federation, South Africa, Spain, Trinidad and Tobago, United Kingdom -0.08 - -0.02 Ukraine, Rwanda Indonesia Note: First column - ranges for point OLS estimates are reported. For each range, only the countries, for which coefficients are different from 0 at a significance level less than 5%, are reported. " 0" range - countries with no significant TV or press bias. Sample of the countries used - WVS wave 3 (except Iran, Iraq, Hong Kong, New Zealand, where questions about mass media were not asked) critical towards the government) to 0.12, which combined with on average 75% of respondents watching TV or reading newspaper, may lead for some countries to a decrease in our estimates of governance by 0.005-0.03 points.5 Intimidation and "critical citizenship" effects are estimated on a country level. Specifically, as a proxy for the intimidation level in a country we use the average score of the country in the "Freedom in the World" ranking - an annual publication of the Freedom House, where political and civil rights of the citizens are assessed. As for the "critical citizenship" effect, we follow Pippa Norris (Norris, 1999) in her definition of a "critical citizen", and define the country to be in the stage of "critical citizenship" if it had been classified "free" by the Freedom House for at least ten years before the survey was conducted (long period of stable democracy), and the GDP per capita in this country (taken from IMF) was more than 10000 US dollars (wealthy population). Most OECD countries together with Slovenia and Chile enter the group. The estimation model than becomes: dif fi = + f reedomi + µcr citi + demogri + i , (8) 5 Note that our estimates of governance are assessed on a scale from 0 to 1. 21 Table 4: Effects of indoctrination and "critical citizenship" Dependent vari- Coef. Std. P>t 95% conf. int. able - diff Err. freedom -0.03 0.007 0.000 -0.05 -0.02 cr cit 0.09 0.025 0.001 0.04 0.14 F(6,157) 17.65 Prob¿F 0.00 R-squared 0.4 Adj. R-squared 0.38 No. of observa- 164 tions Note: *** - significant at less than 1% level. Method of estimation - OLS. Sample - countries surveyed by World Values Survey during all 3 waves. where f reedomi is an index of Freedom House, cr citi is the "critical citizenship" dummy defined above, and demogri is a set of demographical country-specific vari- ables (average level of education, share of married population, share of males, average age). The estimation results are presented in the Table 4. As one can see from the table, both freedom of the county and its being in the stage of "critical citizenship" are highly statistically significant in explaining biases on responses on trust-related questions in the WVS surveys. The directions of the effects are what would be intu- itively foreseen. In the Freedom House ranking a country has the higher score the less civil and political rights it's citizens have: 1 is the best score, 7 is the worst. There- fore, negative in our estimation means that the intimidation effect plays a greater role in less free countries. 1 score up in the Freedom House ranking of a country makes the citizens of this country to be more cautious in answering government- related questions in a survey, and consequently overestimate their governments in trust-related questions by 0.03 points. For a completely depressed country (with the score 7) the effect on our governance estimate would be -0.07 points. From the other side, residents of the countries, which are in a stage of "critical citizenship", do have significantly less confidence in their governments then they should have had. If not too "critical", residents of these countries would give their governments score 0.09 points higher, which would be reflected in the increase of citizen-centric indicator on about 0.03. Even though we find statistically significant effects of indoctrination, intimida- tion and "critical citizenship" in some countries, the magnitude of these effects is not particularly immense. For example, Vietnam with our score of 0.72 is not a free country based on criteria of Freedom House (it had rank 6 in 2005), and there is a moderate (0.05) effect of indoctrination on television. Together these effects would cut citizen-centric governance indicator in Vietnam by 0.07 points. New indicator would be 0.65 - still in the highest 20th percentile of the sample. Apparently, there are other reasons for some governments to score so high in the public opinion polls. In case of East Asia the main of them is probably last decade's stable economic 22 growth and development in the region (as it is argued for China by Wang, 2005). At the same time, poor economic performance, political conflicts and corruption in the 90s (and for many countries up until today) in Central and Eastern European countries keep the scores the governments in this regions extremely low. 6 Concluding remarks This paper has provided a conceptual framework for measuring governance quality using citizens' evaluations consistently across countries and over time. It further provided empirical illustration - using data from the World Values Survey Asso- ciation - of the usefulness of the methodology by developing governance quality rankings for 120 countries. These rankings significantly differ from those provided by available indicators that mostly capture foreigners' (mostly interest groups) or arm-chair experts' opinions. The surveys of the WVS project are certainly subject to important limitations. They are not conducted in the same year for all countries, and the questionnaires may slightly differ from country to country, which may produce significant departures from objective estimation. It is also possible that in spite of the claims to the contrary by the survey organization, the survey may not be based on stratified random sampling for some countries due to practical difficulties (for instance, WVS for Vietnam). Notwithstanding these limitations, the dataset constructed by us has important merits. The governance-related questions and answers are reported on the level of individual respondents in our dataset, which gives researchers a great flexibility in composing the rankings. In particular, it is possible to compose rankings among the people with higher education, different genders, income, etc. Most importantly and contrary to many other indicators, the data used in our estimation are freely accessible, and can be easily used by other researchers to replicate or modify our estimation procedure. Ideally, our theoretical framework should be implemented using a world poll with stratified random sampling employing a uniform questionnaire across countries and over time. The World Gallup Poll or a similar instrument might offer such an opportunity in the near future. 23 References Andrews, M. (2008). Are One-Best-Way models of effective government suitable for developing countries?, J. F. Kennedy School of Government Faculty Research working paper. Andrews, M. and A. Shah (2005). Citizen-centered governance: A new approach to public sector reform. In A. Shah (ed.), Public Expenditure Analysis, Washington, DC: World Bank, chapter 6, 153­182. Arndt, C. (2008). The politics of governance ratings. International Public Manage- ment Journal 11 (3): 275­297. Arndt, C. and C. Oman (2006). Uses and abuses of governance indicators. Paris: OECD Development Center. Azfar, O. (2008). Power to the people, article in daily newspaper "The Dawn", Karachi, Pakistan. Bertrand, M. and S. Mullainathan (2001). Do people mean what they say? Impli- cations for subjective survey data. American Economic Review 91 (2): 67­72. Dixit, A. (2008). Governance institutions and development, presentation, PREM seminar, World Bank, Washington, DC. Djankov, S. (2007). Ease of Doing Business, mimeo, Doing Business Project, World Bank. Dye, K. (2007). Corruption and fraud detection by supreme audit institutions. In A. Shah (ed.), Performance Accountability and Combating Corruption, Washing- ton, DC: World Bank, 303­322. Huther, J. and A. Shah (1998). Applying a simple measure of good governance to the debate on fiscal decentralization, World Bank Policy Research paper 1894. Iqbal, K. and A. Shah (2006). Critical appraisal of governance perception indi- cators and their uses, World Bank Institute, unpublished paper, available at http;//www.worldbank.org/wbi/publicfinance. Iqbal, K. and A. Shah (2008). Truth in advertisement: How do Worldwide Gover- nance Indicators stack up?, World Bank Institute, unpublished paper, available at http;//www.worldbank.org/wbi/publicfinance. Kurtz, M. and A. Schrank (2007). Growth and governance: Models, measures and mechanism. Journal of Politics 69 (2): 538­554. Langbein, L. and S. Knack (2008). The Worldwide Governance Indicators and tau- tology, World Bank Policy Research working paper. Nardo, M., M. Saisana, A. Saltelli, and S. Tarantola (2005). Tools for composite indicator building, European Commission's Joint Research Center working paper. 24 Norris, P. (1999). Critical Citizens: Global Support for Democratic Governance. New York: Oxford University Press. Qiao, B. and A. Shah (2006). Local government organization and finance: China. In A. Shah (ed.), Local Governance in Developing Countries, Washington, DC: World Bank, 137­168. Schrank, A. and M. Kurtz (2008). Conceptualizing and measuring institutions: A view from Political Science, mimeo, Ohio State University. Shah, A. (2007). Tailoring the fight against corruption to country circumstances. In A. Shah (ed.), Performance Accountability and Combating Corruption, Washing- ton, DC: World Bank, 233­254. Shah, A. (2008). Demanding to be served: On holding government to account for service delivery. In J. de Jong and G. Rivzi (eds.), The State of the Access, Wash- ington, DC: Brooking Institution Press. Shah, A. and S. Shah (2006). The new vision of local governance and the evolving role of local governments. In A. Shah (ed.), Local Governance in Developing Countries, Washington, DC: World Bank, 1­46. Shah, A. and C. Shen (2007). Fine tuning of the intergovernmental transfer system to create a harmonious society and a level playing field for regional development. In J. Lou and S. Wang (eds.), Public Finance in China: Reform and Growth for a Harmonious Society, Washington, DC: World Bank, 129­154. Thomas, M. (2006). What do the Worldwide Governance Indicators measure?, mimeo, John Hopkins University. Thompson, T. and A. Shah (2003). Transparency International's Corrup- tion Perception Index: Whose perceptions are they anyway?, available at: http;//www.worldbank.org/wbi/publicfinance. Transparency International (TI) (2005). Report on Transparency International Global Corruption Barometer 2005. URL http://www.transparency.org Wang, Z. (2005). Before emergence of Critical Citizens: Economic development and political trust in China. International Review of Sociology 15 (1): 155­171. World Bank (2007). Strengthening Bank Group Engagement in Governance and Anticorruption, World Bank Group strategy paper. World Bank Institute (WBI) (2008). Worldwide Governance Indicators. URL info.worldbank.org/governance/wgi/index.asp World Values Survey (WVS) (2008). 2005 Official Datafile v. 20081015. World Val- ues Survey Association. URL http://www.worldvaluessurvey.org 25 A Appendix Table A1: Governance outcomes: weights and questions assigned Governance Weights used Code Questions assigned criteria 1 2 3 comp. A Responsive 0.6 0.6 0.6 0.6 governance 11 public services How satisfied are you with the 0.25 0.15 0 0 consistent with way the people in national of- citizen prefer- fice are handling the country's ences affairs? 21 safety of life, or- How much confidence do you 0.05 0.05 0.03 0.1 der, rule of law have in police? 31 freedom of How satisfied are you with the 0.15 0.15 0 0 choice and way the democracy is develop- expression ing in your country? 32 How democratically is your 0 0 0.1 0 country being governed today? 41 improvements How satisfied are you with 0.2 0.2 0.2 0.3 in economic and the financial situation of your social outcomes household? 51 improvements in All things considered, how sat- 0.25 0.35 0.25 0.4 quality of life: isfied are you with your life as general a whole these days? 61 improvements in All in all, how would you de- 0.05 0.05 0.05 0.1 quality of life: scribe your state of health to- health day? 71 improvements in How serious you consider poor 0 0 0.03 0 quality of life: water quality to be here in your environment own community? 72 How serious you consider poor 0 0 0.03 0 air quality to be here in your own community? 73 How serious you consider poor 0 0 0.03 0 sewage and sanitation to be here in your own community? 81 peace How much confidence do you 0.05 0.05 0.03 0.1 have in armed forces? 91 inmprovements Taking all things together 0 0 0.25 0 in quality of life: would you say you are [happy, happiness unhappy]? B Fair gover- 0.1 0.1 0.1 0.1 nance 26 Table A1: (continued) Governance Weights used Code Questions assigned criteria 1 2 3 comp. 11 social justice, re- How much respect is there for 0.8 0.8 0.8 0.8 spect for human individual human rights nowa- rights days in the country? 21 government How proud are you to be your 0.2 0.2 0.2 0.2 represents the nationality? whole country C Responsible 0.15 0.15 0.15 0.15 governance 11 earning trust: How much confidence do you 0.2 0.2 0.3 0.5 executive branch have in government? 19 earning trust: How much confidence do you 0.2 0.2 0.3 0.5 legislative have in parliament? branch 21 corruption Would you say that this coun- 0.3 0.3 0 0 try is run by a few big interests looking out for themselves, or that it is run for the benefit of all people? 22 In your view, does corruption 0 0 0.4 0 affect your personal and family life, business environment, po- litical life not at all, to a small extent, to a moderate extent, or to a large extent? 31 open, transpar- How satisfied are you with the 0.3 0.3 0 0 ent and prudent way the people in national of- economic, fiscal fice are handling the country's and financial affairs? management D Accountable 0.15 0.15 0.15 0.15 governance 11 access to in- How much confidence do you 0.25 0.25 0.25 0.25 formation, have in press? independent mass media - press 27 Table A1: (continued) Governance Weights used Code Questions assigned criteria 1 2 3 comp. 18 access to in- How much confidence do you 0.25 0.25 0.25 0.25 formation, have in television? independent mass media - television 21 judicial integrity How much confidence do you 0.5 0.5 0.5 0.5 and indepen- have in courts? dence Note: The data source for all (but C24) questions is World Values Survey (WVS, 2008). Question C24 was taken from Transparency International Global Corruption Barometer (TI, 2005). The coding corresponds to the coding used in our dataset. Weights used : 1 - for wave 1 (1994-98) of WVS, 2 - for wave 2 (1999-2004), 3 - or wave 3 (2004-08), comp. - for comparison between these 3 waves. Weights of sub-categories are given within the category (A, B, C, or D) 28 Table A2: Existing sources of data and their main features Geographical coverage Data access Name Code Years Freq., y. Relevancy Num. Region Free Lag, y. World Values Survey WVS 97 worldwide 1994- 3-6 yes 2-3 average 2008 Afrobarometer AFR 20 Sub-Saharan 2001- 3 yes 1-2 high Africa 2008 Asiabarometer ASB 25 East Asia 2003- 2 yes 1-2 high 2006 Business Environment BEEPS 26 Central and 1999- 3 yes 1-2 low and Enterprise Perfor- Eastern 2005 mance Survey Europe Transparency Interna- TI GCB 62 worldwide 2004- 1 yes <1 very low tional Global Corruption 2008 29 Barometer Latinobarometro LBO 18 Latin Amer- 2004- 1 no 1 high ica 2007 Eurobarometer EUB 30 Europe 1973- 0.5 yes <1 very high 2008 Gallup World Poll GWP 130 worldwide 2007- 1 no n.a. n.a. 2008 GWP - datapoints from GWP 119 worldwide 2007 1 yes 0 low World Bank Institute WGI (WBI) (2008) Note: Number - the total number of countries, which participated in all waves of survey; Freq. - average time period in years, in which a country is surveyed; Lag - the time period in years between taking a survey and posting data; Relevancy - correspondence of questions in a questionnaire to the subcriteria of governance from the Table 1, given on the scale: very low-low-average-high-very high. Table A3: Citizen-centric governance indicators: aggre- gate and disaggregate data by country, waves 1-3 A B C D country year N prec CGI var 11 21 31 37 41 51 61 74 81 91 11 21 11 19 21 24 31 11 18 21 WAVE 1 Albania 1998 999 38 65 .. .. 40 42 75 .. 56 .. .. 81 46 54 21 .. 38 33 39 65 83 45 0.6 Azerbaijan 1997 2002 42 46 52 .. 40 49 66 .. 53 .. 58 86 77 64 22 .. 42 36 40 46 100 48 0.6 Argentina 1995 1079 35 32 .. .. 44 66 68 .. 32 .. .. 81 33 26 12 .. 35 41 36 32 83 42 0.8 Australia 1995 2048 43 63 .. .. 60 73 77 .. 59 .. .. 90 36 40 32 .. 43 32 38 63 83 55 0.7 Bangladesh 1996 1525 74 42 .. .. 56 60 62 .. 56 .. .. 92 70 72 60 .. 74 61 59 42 83 62 0.7 Bosnia and 1998 1200 48 68 .. .. 40 50 66 .. 77 .. .. 80 63 53 43 .. 48 50 54 68 83 53 0.8 Herzegovina Brazil 1997 1149 49 40 .. .. 50 68 73 .. 63 .. .. 82 43 31 25 .. 49 53 49 40 83 52 1.1 Bulgaria 1997 1072 36 49 .. .. 29 41 64 .. 72 .. .. 77 54 45 27 .. 36 46 60 49 83 43 0.8 30 Belarus 1996 2092 22 40 29 .. 25 37 51 .. 65 .. 34 68 50 35 17 .. 22 44 47 40 100 34 0.5 Chile 1996 1000 51 49 .. .. 55 66 67 .. 53 .. .. 81 50 40 32 .. 51 48 51 49 83 54 0.8 China 1995 1500 .. .. .. .. 57 65 74 .. .. .. .. 76 .. .. .. .. .. .. .. .. 32 63 2.7 Colombia 1997 6025 31 48 .. .. 78 81 75 .. 57 .. .. 94 39 30 21 .. 31 46 49 48 83 54 0.8 Croatia 1996 1196 44 56 .. .. 40 58 63 .. 67 .. .. 75 51 46 34 .. 44 36 36 56 83 49 0.8 Czech rep. 1998 1147 35 45 .. .. 46 60 63 .. 44 .. .. 73 37 30 18 .. 35 45 48 45 83 45 0.7 Dominican 1996 417 17 28 .. .. 53 68 73 .. 41 .. .. 89 27 27 8 .. 17 43 46 28 83 40 0.8 rep. Estonia 1996 1021 30 47 43 .. 33 44 57 .. 46 .. 43 63 48 44 15 .. 30 51 58 47 100 41 0.5 Finland 1996 987 42 69 .. .. 63 75 74 .. 68 .. .. 78 40 40 28 .. 42 40 50 69 83 57 0.6 Georgia 1996 2008 30 37 31 .. 23 41 62 .. 48 .. 32 86 45 39 6 .. 30 52 53 37 100 36 0.6 Germany 1997 2026 38 54 52 .. 58 66 66 .. 45 .. 53 53 32 35 29 .. 38 31 35 54 100 49 0.5 Hungary 1998 650 40 52 .. .. 44 54 60 .. 54 .. .. 80 44 42 18 .. 40 37 44 52 83 46 0.7 India 1995 2040 41 43 .. .. 57 61 67 .. 73 .. .. 88 52 56 29 .. 41 57 53 43 83 52 1.0 Table A3: (continued) A B C D country year N prec CGI var 11 21 31 37 41 51 61 74 81 91 11 21 11 19 21 24 31 11 18 21 Japan 1995 1054 28 63 .. .. 59 62 65 .. 56 .. .. 62 40 37 23 .. 28 59 58 63 83 50 0.5 Korea, rep. 1996 1249 42 49 .. .. 52 .. 73 .. 61 .. .. .. 47 39 17 .. 42 57 55 49 66 47 0.8 Latvia 1996 1200 30 37 32 .. 29 43 56 .. 36 .. 36 59 40 33 4 .. 30 48 52 37 100 36 0.5 Lithuania 1997 1009 29 34 38 .. 34 44 59 .. 45 .. 35 60 44 39 10 .. 29 58 60 34 100 38 0.5 Macedonia 1998 995 28 36 .. .. 41 52 71 .. 46 .. .. 86 28 25 26 .. 28 33 36 36 83 39 0.9 Mexico 1996 2364 33 35 .. .. 69 73 65 .. 54 .. .. 87 42 44 29 .. 33 49 48 35 83 51 0.8 Moldova 1996 984 27 37 26 .. 23 30 51 .. 53 .. 30 70 43 41 17 .. 27 41 47 37 100 32 0.6 New 1998 1201 31 68 .. .. 61 74 78 .. 56 .. .. 87 30 30 22 .. 31 41 44 68 83 52 0.7 Zealand Nigeria 1995 1996 29 32 .. .. 52 62 76 .. 46 .. .. 81 33 32 11 .. 29 56 58 32 83 44 1.1 Norway 1996 1127 64 67 .. .. 64 74 78 .. 60 .. .. 80 57 58 72 .. 64 42 49 67 83 65 0.5 31 Pakistan 1997 733 .. 33 .. .. 41 .. 69 .. 92 .. .. 94 .. .. .. .. .. 54 59 33 38 51 1.3 Peru 1996 1211 49 34 .. .. 46 60 64 .. 50 .. .. 92 46 28 57 .. 49 42 45 34 83 49 1.0 Philippines 1996 1200 47 54 .. .. 56 65 66 .. 62 .. .. 89 55 56 41 .. 47 65 64 54 83 57 0.9 Poland 1997 1153 40 51 .. .. 37 60 56 .. 67 .. .. 89 43 40 20 .. 40 48 49 51 83 47 0.8 Puerto Rico 1995 1164 48 55 .. .. 66 79 72 .. 59 .. .. 95 52 37 39 .. 48 52 45 55 83 59 0.9 Romania 1998 1239 27 43 .. .. 32 43 64 .. 72 .. .. 76 32 31 20 .. 27 41 49 43 83 38 0.8 Russian fed- 1995 2040 17 36 .. .. 26 38 50 .. 63 .. .. 65 32 31 7 .. 17 43 47 36 83 31 0.7 eration Serbia and 1996 1520 36 46 .. .. 34 52 63 .. 58 .. .. 71 41 39 31 .. 36 35 36 46 83 43 0.9 Montenegro Slovakia 1998 1095 41 43 .. .. 40 56 62 .. 58 .. .. 77 44 37 34 .. 41 46 49 43 83 46 0.7 Slovenia 1995 1007 40 49 .. .. 48 61 59 .. 47 .. .. 84 45 35 22 .. 40 46 52 49 83 48 0.7 South 1996 2935 48 65 .. .. 42 56 75 .. 52 .. .. 92 59 58 56 .. 48 52 58 65 83 55 1.1 Africa Table A3: (continued) A B C D country year N prec CGI var 11 21 31 37 41 51 61 74 81 91 11 21 11 19 21 24 31 11 18 21 Spain 1995 1211 29 54 .. .. 52 62 70 .. 44 .. .. 85 37 40 33 .. 29 46 44 54 83 47 0.7 Sweden 1996 1009 45 65 .. .. 58 75 78 .. 52 .. .. 78 45 47 41 .. 45 39 50 65 83 57 0.6 Switzerland 1996 1212 54 58 .. .. 70 78 79 .. 47 .. .. 67 49 45 39 .. 54 35 40 58 83 59 0.6 Taiwan 1994 780 44 54 .. .. 57 62 64 .. 62 .. .. 60 58 48 48 .. 44 46 50 54 83 54 0.6 Turkey 1996 1907 34 61 .. .. 47 58 68 .. 86 .. .. 90 43 45 20 .. 34 49 48 61 83 49 0.9 Ukraine 1996 2811 21 39 25 .. 22 33 50 .. 60 .. 27 60 43 39 12 .. 21 44 47 39 100 31 0.5 UK 1998 1093 .. .. .. .. .. 73 .. .. .. .. .. .. .. .. .. .. .. .. .. .. 15 73 4.7 USA 1995 1542 45 61 .. .. 62 74 78 .. 72 .. .. 92 41 40 27 .. 45 39 39 61 83 56 0.7 Uruguay 1996 1000 35 49 .. .. 64 68 74 .. 37 .. .. 89 41 41 23 .. 35 53 51 49 83 51 0.9 Venezuela 1996 1200 19 34 .. .. 44 64 76 .. 59 .. .. 97 31 28 16 .. 19 57 53 34 83 42 1.1 WAVE 2 32 Albania 2002 1000 26 58 34 .. 42 46 74 .. 51 .. 41 89 54 45 35 .. 26 40 52 58 100 44 0.7 Algeria 2002 1282 32 60 41 .. 55 52 62 .. 63 .. 38 89 49 34 13 .. 32 47 45 60 100 47 0.9 Argentina 1999 1280 33 32 44 .. 50 70 71 .. 35 .. 34 85 28 23 10 .. 33 44 40 32 100 45 0.7 Austria 1999 1522 .. 64 60 .. .. 78 .. .. 45 .. 63 81 .. 46 .. .. .. 41 .. 64 60 66 0.9 Bangladesh 2002 1500 62 51 62 .. 51 53 66 .. 68 .. 61 90 76 78 44 .. 62 75 69 51 100 59 0.6 Belgium 1999 1912 .. 50 44 .. .. 71 .. .. 41 .. 56 64 .. 41 .. .. .. 41 .. 50 60 56 1.1 Bosnia and 2001 1200 35 57 39 .. 43 53 71 .. 58 .. 39 66 39 34 19 .. 35 38 42 57 100 45 0.7 herzegovina Bulgaria 1999 1000 .. 47 37 .. .. 50 .. .. 54 .. 40 67 .. 36 .. .. .. 37 .. 47 60 45 1.5 Belarus 2000 1000 .. 43 37 .. .. 42 .. .. 61 .. 41 63 .. 40 .. .. .. 44 .. 43 60 43 1.3 Canada 2000 1931 53 68 57 .. 65 76 80 .. 59 .. 68 87 44 43 47 .. 53 42 44 68 100 63 0.6 Chile 2000 1200 55 53 53 .. 52 68 71 .. 48 .. 54 87 53 39 35 .. 55 47 51 53 100 56 0.7 China 2001 1000 59 60 65 .. 52 61 70 .. 80 .. 73 68 79 76 83 .. 59 59 62 60 100 64 0.6 Croatia 1999 1003 .. 47 31 .. .. 63 .. .. 56 .. 51 74 .. 33 .. .. .. 31 .. 47 60 50 1.2 Table A3: (continued) A B C D country year N prec CGI var 11 21 31 37 41 51 61 74 81 91 11 21 11 19 21 24 31 11 18 21 Czech rep. 1999 1908 .. 43 42 .. .. 67 .. .. 39 .. 56 69 .. 28 .. .. .. 44 .. 43 60 53 1.0 Denmark 1999 1023 .. 72 59 .. .. 80 .. .. 55 .. 78 80 .. 49 .. .. .. 41 .. 72 60 70 0.8 Egypt 2000 3000 77 78 77 .. 47 48 70 .. 59 .. 63 94 55 62 31 .. 77 62 61 78 100 62 1.1 El Salvador 1999 1254 .. 51 .. .. 59 72 71 .. 49 .. .. 93 43 35 26 .. .. 48 52 51 70 58 1.3 Estonia 1999 1005 .. 41 42 .. .. 55 .. .. 42 .. 52 60 .. 37 .. .. .. 45 .. 41 60 48 1.1 Finland 2000 1038 .. 73 53 .. .. 76 .. .. 69 .. 75 83 .. 46 .. .. .. 43 .. 73 60 68 0.7 France 1999 1615 .. 57 48 .. .. 67 .. .. 55 .. 54 75 .. 40 .. .. .. 38 .. 57 60 57 1.1 Germany 1999 2036 .. 59 59 .. .. 71 .. .. 49 .. 62 63 .. 41 .. .. .. 42 .. 59 60 61 0.9 Greece 1999 1142 .. 36 51 .. .. 63 .. .. 59 .. 58 80 .. 33 .. .. .. 37 .. 36 60 53 1.1 Hungary 1999 1000 .. 44 40 .. .. 53 .. .. 45 .. 52 79 .. 38 .. .. .. 36 .. 44 60 48 1.3 Iceland 1999 968 .. 68 55 .. .. 78 .. .. 42 .. 72 88 .. 61 .. .. .. 44 .. 68 60 68 0.7 33 India 2001 2002 52 42 56 .. 44 46 68 .. 84 .. 65 87 53 52 34 .. 52 64 65 42 100 52 0.7 Indonesia 2001 1004 36 52 40 .. 61 66 70 .. 63 .. 59 80 52 46 30 .. 36 53 56 52 100 54 0.6 Iran 2000 2532 59 56 55 .. 53 60 75 .. .. .. 61 95 62 63 51 .. 59 44 50 56 97 58 0.8 Iraq 2004 2325 .. .. .. .. 49 47 74 .. 55 .. 39 90 40 .. 30 .. .. .. 54 .. 60 48 1.6 Ireland 1999 1012 .. 73 56 .. .. 80 .. .. 58 .. 67 91 .. 41 .. .. .. 44 .. 73 60 69 0.9 Israel 2001 1199 .. .. .. .. .. 67 .. .. .. .. .. 78 .. .. .. .. .. .. .. .. 23 68 4.9 Italy 1999 2000 .. 59 42 .. .. 69 .. .. 51 .. 56 75 .. 41 .. .. .. 42 .. 59 60 58 1.0 Japan 2000 1362 28 49 45 .. 57 61 65 .. 57 .. 54 59 37 34 16 .. 28 59 58 49 100 49 0.5 Jordan 2001 1223 63 83 59 .. 44 51 76 .. 85 .. 62 89 78 62 31 .. 63 59 57 83 100 60 0.7 Korea, rep. 2001 1200 39 49 42 .. 53 58 73 .. 57 .. 47 64 40 24 12 .. 39 56 56 49 100 48 0.6 Kyrgyzstan 2003 1043 38 29 39 .. 52 61 67 .. 53 .. 38 74 38 38 17 .. 38 46 51 29 100 45 0.8 Latvia 1999 1013 .. 42 40 .. .. 47 .. .. 47 .. 50 73 .. 35 .. .. .. 46 .. 42 60 46 1.3 Lithuania 1999 1018 .. 37 35 .. .. 47 .. .. 48 .. 31 55 .. 27 .. .. .. 60 .. 37 60 41 1.4 Luxembourg 1999 1211 .. 60 64 .. .. 76 .. .. 50 .. 73 77 .. 54 .. .. .. 46 .. 60 60 67 0.9 Table A3: (continued) A B C D country year N prec CGI var 11 21 31 37 41 51 61 74 81 91 11 21 11 19 21 24 31 11 18 21 Macedonia 2001 1055 26 48 27 .. 38 46 72 .. 51 .. 36 78 20 17 7 .. 26 33 35 48 100 37 0.9 Malta 1999 1002 .. 59 64 .. .. 80 .. .. 62 .. 62 91 .. 49 .. .. .. 40 .. 59 60 67 0.8 Mexico 2000 1535 44 34 42 .. 63 79 70 .. 53 .. 48 91 39 28 27 .. 44 45 47 34 100 53 0.8 Moldova 2002 1008 31 38 27 .. 34 40 50 .. 54 .. 31 60 39 38 9 .. 31 46 49 38 100 36 0.6 Morocco 2001 2264 46 51 44 .. 49 56 77 .. 66 .. 42 95 54 25 23 .. 46 41 36 51 100 49 0.9 Netherlands 1999 1003 .. 57 59 .. .. 76 .. .. 44 .. 70 65 .. 51 .. .. .. 53 .. 57 60 65 0.6 New 2004 954 .. 63 .. .. 63 77 72 .. 62 79 69 89 45 42 .. .. .. 37 43 48 73 63 0.7 Zealand Nigeria 2000 2022 59 39 57 .. 59 65 87 .. 49 .. 56 87 49 47 28 .. 59 62 68 39 100 57 0.8 Pakistan 2001 2000 43 35 27 .. 28 43 69 .. 79 .. 53 93 42 68 11 .. 43 55 55 35 100 43 0.5 Peru 2001 1501 45 33 45 .. 46 60 64 .. 37 .. 46 90 35 28 43 .. 45 39 40 33 100 47 0.7 34 Philippines 2001 1200 49 58 47 .. 53 63 67 .. 65 .. 71 94 51 57 39 .. 49 63 65 58 100 58 0.8 Poland 1999 1095 .. 55 44 .. .. 58 .. .. 62 .. 51 89 .. 40 .. .. .. 50 .. 55 60 54 1.5 Portugal 1999 1000 .. 58 62 .. .. 67 .. .. 61 .. 57 91 .. 47 .. .. .. 57 .. 58 60 62 1.0 Puerto Rico 2001 720 47 57 54 .. 72 83 75 .. 55 .. 53 98 49 39 48 .. 47 48 39 57 100 62 0.7 Romania 1999 1146 .. 47 32 .. .. 47 .. .. 72 .. 36 77 .. 28 .. .. .. 45 .. 47 60 44 1.6 Russian 1999 2500 .. 34 19 .. .. 41 .. .. 61 .. 25 65 .. 27 .. .. .. 36 .. 34 60 35 1.4 Federation Saudi Ara- 2003 1502 .. .. .. .. 69 70 84 .. .. .. 62 89 .. .. 41 .. .. 60 63 .. 58 67 1.5 bia Serbia and 2001 2260 38 43 41 .. 33 51 65 .. 58 .. 48 65 36 33 29 .. 38 36 39 43 100 43 0.7 Montenegro Singapore 2002 1512 71 .. .. .. 63 69 .. .. .. .. .. 82 .. .. 77 .. 71 .. .. .. 53 69 1.2 Slovakia 1999 1331 .. 45 33 .. .. 56 .. .. 62 .. 53 65 .. 42 .. .. .. 47 .. 45 60 50 1.2 Slovenia 1999 1006 .. 50 45 .. .. 69 .. .. 45 .. 45 81 .. 36 .. .. .. 57 .. 50 60 56 1.1 Table A3: (continued) A B C D country year N prec CGI var 11 21 31 37 41 51 61 74 81 91 11 21 11 19 21 24 31 11 18 21 South 2001 3000 44 56 48 .. 45 59 81 .. 51 .. 51 86 51 49 32 .. 44 53 61 56 100 53 0.9 Africa Spain 1999.52409 46 53 56 .. 58 67 72 .. 44 .. 58 81 46 48 40 .. 46 45 43 53 100 56 0.6 Sweden 1999 1015 44 62 52 .. .. 74 .. .. 47 .. 63 76 .. 50 .. .. 44 48 .. 62 74 60 0.7 Tanzania 2001 1171 53 63 63 .. 28 32 70 .. 86 .. 67 91 78 74 52 .. 53 70 72 63 100 54 1.1 Turkey 2001 4607 34 62 25 .. 37 51 68 .. 80 .. 28 82 43 39 17 .. 34 34 37 62 100 43 0.9 Uganda 2001 1002 55 56 58 .. 43 52 73 .. 71 .. 60 85 72 69 50 .. 55 63 62 56 100 57 0.8 UK 1999 1000 .. 60 50 .. .. 71 .. .. 69 .. 59 79 .. 42 .. .. .. 26 .. 60 60 60 1.0 Ukraine 1999 1195 .. 36 27 .. .. 40 .. .. 61 .. 31 57 .. 33 .. .. .. 46 .. 36 60 38 1.5 USA 1999 1200 55 62 56 .. 61 74 81 .. 71 .. 62 89 44 44 37 .. 55 38 38 62 100 60 0.6 Venezuela 2000 1200 54 41 57 .. 58 72 .. .. 59 .. 49 97 53 36 63 .. 54 59 58 41 97 58 0.9 35 Vietnam 2001 1000 80 82 86 .. 55 61 66 .. 88 .. 86 92 91 91 91 .. 80 72 78 82 100 75 0.5 Zimbabwe 2001 1002 36 61 37 .. 24 33 72 .. 58 .. 36 88 52 50 18 .. 36 55 57 61 100 41 0.9 WAVE 3 Andorra 2005 1003 .. 53 .. 49 59 68 72 44 .. 73 60 75 41 .. .. .. .. 43 41 41 88 58 1.4 Argentina 2006 1002 .. 31 .. 67 61 75 70 16 38 73 44 87 41 25 .. 20 .. 41 39 30 100 52 1.1 Australia 2005 1421 .. 69 .. 68 59 70 66 45 69 76 64 88 44 42 .. .. .. 30 35 51 94 60 0.8 Brazil 2006 1500 .. 43 .. 58 54 74 67 37 62 75 47 72 45 29 .. .. .. 43 41 47 94 57 0.8 Bulgaria 2006 1001 .. 51 .. 37 34 47 52 22 64 53 35 73 38 29 .. 14 .. 48 58 40 100 41 0.7 Burkina 2007 1534 .. 51 .. 52 41 51 65 10 61 67 54 94 48 41 .. .. .. 52 55 47 94 51 0.7 Faso Chile 2005 1000 .. 54 .. 66 52 68 57 39 55 69 49 84 46 32 .. .. .. 45 47 35 94 55 1.8 China 2007 2015 .. 67 .. 64 55 64 59 56 75 65 71 65 77 77 .. 22 .. 62 63 68 100 62 0.6 Colombia 2005 3025 .. 48 .. 59 .. 81 64 .. 58 78 45 96 49 31 .. 25 .. 44 47 39 82 58 0.7 Cyprus 2006 1050 .. 58 .. 64 62 71 71 28 63 74 58 80 51 49 .. .. .. 40 42 61 94 61 1.5 Table A3: (continued) A B C D country year N prec CGI var 11 21 31 37 41 51 61 74 81 91 11 21 11 19 21 24 31 11 18 21 Egypt 2008 3051 .. .. .. .. 43 53 58 3 .. 64 .. 91 .. .. .. .. .. 56 67 .. 61 51 1.7 Ethiopia 2007 1500 .. 40 .. 36 43 44 60 14 47 63 38 88 36 35 .. 28 .. 35 36 37 100 42 0.6 Finland 2005 1014 .. 75 .. 71 67 76 62 69 71 74 81 83 56 52 .. 47 .. 42 50 67 100 67 0.6 France 2006 1001 .. 59 .. 62 57 66 66 .. 57 75 .. 72 34 39 .. 21 .. 40 38 40 86 55 0.8 Germany 2006 2064 .. 61 .. 61 56 68 64 69 49 67 61 62 33 33 .. 29 .. 39 41 53 100 56 1.8 Ghana 2007 1534 .. 53 .. 83 46 57 71 29 69 75 72 97 65 60 .. 29 .. 56 65 60 100 60 0.7 Hong Kong 2005 1252 .. 65 .. .. 57 60 55 .. 52 63 64 54 53 50 .. 29 .. 55 59 .. 81 57 0.5 India 2006 2001 .. 60 .. 61 48 53 61 27 76 67 73 89 54 60 .. 27 .. 69 67 65 100 57 0.7 Indonesia 2006 2015 .. 50 .. 61 58 66 64 27 64 73 63 79 54 43 .. 22 .. 52 57 51 100 57 0.5 Iran 2007 2667 .. 57 .. 47 56 60 60 14 59 65 43 84 53 49 .. .. .. 44 51 51 94 53 0.6 Iraq 2006 2701 .. .. .. .. 41 38 57 .. 60 47 35 93 56 .. .. .. .. .. 65 .. 65 46 1.3 36 Italy 2005 1012 .. 63 .. 53 61 65 63 55 59 69 52 77 36 39 .. 17 .. 37 32 50 100 54 0.4 Japan 2005 1096 .. 57 .. 65 57 67 53 49 61 73 51 60 38 34 .. .. .. 60 59 66 94 59 0.4 Jordan 2007 1200 .. 85 .. 75 60 68 76 19 88 72 65 90 81 61 .. .. .. 64 65 83 94 67 2.4 Korea, rep. 2005 1200 .. 53 .. 60 51 60 64 49 51 66 58 69 46 35 .. 12 .. 56 57 48 100 53 3.2 Malaysia 2006 1201 .. 64 .. 67 61 65 72 37 71 77 64 88 67 60 .. 31 .. 58 62 68 100 63 0.4 Mali 2007 1534 .. 64 .. 67 53 57 62 16 76 73 72 96 65 55 .. .. .. 56 62 61 94 61 0.8 Mexico 2005 1560 .. 36 .. 62 68 80 61 35 59 83 53 92 45 31 .. 21 .. 48 47 40 100 59 0.7 Moldova 2006 1046 .. 32 .. 45 42 49 51 21 41 49 33 60 37 34 .. 29 .. 43 49 35 100 41 0.6 Morocco 2007 1200 .. 58 .. 44 44 47 70 13 63 68 56 83 54 47 .. .. .. 51 55 60 94 52 0.5 Netherlands 2006 1050 .. 53 .. 62 65 75 65 .. 45 78 .. 69 36 38 .. 49 .. 39 42 46 86 60 1.6 Peru 2008 1500 .. 29 .. 51 52 67 50 15 35 65 33 .. 26 22 .. 12 .. 33 34 21 98 43 0.7 Poland 2005 1000 .. 48 .. 52 46 67 54 29 60 71 55 86 31 27 .. 14 .. 45 46 40 100 50 0.7 Romania 2005 1776 .. 43 .. 53 42 53 50 44 70 52 40 73 33 28 .. .. .. 47 51 36 94 46 0.7 Table A3: (continued) A B C D country year N prec CGI var 11 21 31 37 41 51 61 74 81 91 11 21 11 19 21 24 31 11 18 21 Russian 2006 2033 .. 37 .. 37 41 57 44 .. 60 58 .. 75 44 34 .. 26 .. 40 45 40 86 46 1.1 Federation Rwanda 2007 1507 .. 76 .. .. 38 44 40 32 .. 65 .. 92 .. 69 .. .. .. 62 55 70 74 54 0.7 Serbia 2006 1220 .. 39 .. 46 42 56 54 18 49 56 33 78 34 30 .. .. .. 33 33 35 94 43 0.6 Slovenia 2005 1037 .. 43 .. 54 59 69 55 59 42 66 49 82 36 31 .. .. .. 39 43 39 94 55 0.5 South 2007 2988 .. 57 .. 71 52 67 70 26 58 72 63 91 64 60 .. 27 .. 56 65 60 100 60 4.2 Africa Spain 2007 1200 .. 55 .. 71 54 70 65 .. 50 68 56 84 46 49 .. 48 .. 45 41 52 94 59 0.5 Sweden 2006 1003 .. 63 .. 73 67 75 70 83 48 80 67 76 45 52 .. .. .. 41 51 62 94 68 1.0 Switzerland 2007 1241 .. 66 .. 74 76 78 72 59 49 79 74 74 57 51 .. 43 .. 42 41 63 100 67 1.5 Taiwan 2006 1227 .. 43 .. 66 56 62 68 57 44 68 58 54 38 25 .. 7 .. 30 28 42 100 51 1.3 37 Thailand 2007 1534 .. 46 .. 67 62 69 65 50 52 77 68 95 45 43 .. 26 .. 49 51 64 100 61 0.5 Trinidad 2006 1002 .. 38 .. 57 57 70 68 56 46 79 39 96 37 31 .. .. .. 35 38 41 94 57 1.4 and Tobago Turkey 2007 1346 .. 66 .. 55 55 72 59 18 82 73 41 93 59 56 .. 23 .. 36 38 68 100 56 0.9 UK 2006 1041 .. 62 .. 61 64 73 67 .. 67 81 .. 81 39 41 .. 33 .. 28 39 55 86 61 1.3 Ukraine 2006 1000 .. 38 .. 35 40 52 46 19 52 61 37 67 35 30 .. 23 .. 46 49 37 100 43 1.2 USA 2006 1249 .. 61 .. 59 54 70 69 37 71 76 59 85 44 36 .. 25 .. 37 38 53 100 56 2.3 Vietnam 2006 1495 .. 85 .. 77 59 68 54 41 93 72 79 93 93 92 .. .. .. 81 87 84 94 73 0.4 Zambia 2007 1500 .. 49 .. 63 49 56 64 37 55 59 51 83 47 44 .. .. .. 51 55 52 94 54 0.8 Note: The table presents citizen-centric governance indicators for all countries and waves of surveys as well as mean responses by each question used in estimation. The data source for all (but C24) questions is World Values Survey (WVS, 2008). Question C24 was taken from Transparency International Global Corruption Barometer (TI, 2005). year - year of the survey. N - number of respondents. Columns 4 to 23 - mean responses to each question used in our estimation, the coding corresponds to the coding used in our dataset. prec - weights-adjusted amount of questions actually asked in a country during a survey (some questions were not asked in some countries), weights for each question are given in the Table A1. CGI - citizen-centric governance indicators, point estimates. var - estimates of variance of CGIs. All numbers are given in percentages (including variance).