The role of income distribution on civic community and institutional performance Rosa Valentim * Marilia Patta Ramos* Carlos Águedo Nagel Paiva *

Transcription

1 The role of income distribution on civic community and institutional performance Rosa Valentim * Marilia Patta Ramos* Carlos Águedo Nagel Paiva * ABSTRACT The objective of this article is to present and to evaluate critically the empirical study developed by Robert Putnam in Italy, proposing a new standard of analysis, incorporating information about income inequality on the determination of institutional and economic performance. The research question was: what is the relationship of the equitable distribution of income on the civic community, institutional performance and economic performance? The hypothesis tested in this article is the following: the equitable distribution of income leads to the development of a civic community which in turn leads to institutional performance and to the socioeconomic performance. The database of the World Value Survey - WVS in 2005 was used to test that hypothesis. The sample was composed of 49 countries that participated on that survey. This article built upon the work of Knack and Keefer (1997), but with the most updated information from WVS and their own assumptions that, although being in the same sense as the authors' hypotheses, have certain characteristics that differentiate them. Basically it was developed a critical analysis of the variables used by Putnam and multivariate regression models were estimated, through which the central hypothesis was tested. A key finding is that civism and active groups (people who belong to and participate actively in an organization / association), taken individually, have a negative effect on the determination of economic performance and consequently on institutional performance, contradicting the results achieved by Putnam in his study. The distribution of income had a negative and significant effect on economic performance in the sense that in countries where income inequality is greater, the lower is the economic performance. This article proposes a different standard theoretical analysis, a more economical that one proposed by Putnam. Key-words: civic community, institutional performance, economic performance, socioeconomic development, income inequality. PhD in Regional Development from Santa Cruz do Sul University, Brazil. Professor and researcher at the Department of Economics at Santa Cruz do Sul University, Unisc- Brazil. valentim@unisc.br Post doctorate at the Population Research Center - University of Texas, Ph.D. in sociology from Purdue University. Researcher and professor at the Department of Sociology and in the Sociology Graduate Program at Federal University of Rio Grande do Sul, Ufrgs- Brazil. Researcher 2 Cnpq. ramosm68@yahoo.com.br Ph.D. in Economics from the University of Campinas, researcher at the Economics and Statistics Foundation and professor in the Graduate Program in Regional Development, Santa Cruz do Sul University Unisc-Brazil.Researcher 2 Cnpq. carlosanpaiva@gmail.com 1

2 1 Introduction One of the pioneering work and a benchmark for many studies on civic community and economic development is the work entitled Community and Democracy: the experience of modern Italy by Robert Putnam (2005). Given the importance of this seminal work of Putnam and due to the fact its approach is critical in the ongoing debate, his study has become the object of analysis of this article. Putnam (2005), in his analysis developed in Italy between 1970 and 1989, aimed to understand the factors that generated major differences in the performance of 20 regional governments. According to the author, differences in government performance framework highlight the regional disparities, incorporating the Northeast and Central Italy to the group of developed regions, while the southern region, with lower performance, distanced itself further from the economic standard of the North. In the model proposed by Putnam all paths to explain the institutional performance and regional socioeconomic development begin with the civic community. According to the author, the more civic regions have better institutional performance and, consequently, higher socioeconomic development than the less civic regions of Italy. The empirical study conducted in Italy come to a result where the "culturalist" hypothesis of Plato is the strongest to explain institutional performance. Being the civic community the starting point of the model presented by Putnam, it is unclear in his study the civic community roots, i.e., what are the bases for the formation of a civic community. It is understood, therefore, that the civic community in its constitution, may be rooted in a more economic factors, such as more equitable distribution of income. Given this, we ask: what is the effect of equitable distribution of income on the civic community, institutional performance and consequently on the socioeconomic development? In a civic community the trust germinates more easily, providing greater integration among individuals focused on collective action. This occurs in a society in which all people share the same conditions, where there is equity in material terms, rules and values, provided by a lower income inequality. The object of this article is to test the impact of economic factors, specifically the distribution of income on the civic community and institutional performance. However, using the database given by Putnam¹, categorical problems were found in the composition of the variables civic community and socioeconomic development tested by the author. The variables tested by Putnam are not composed by indicators of equality and trust as theorized by Aristotle. As tested by Putnam trust stems from the civic traditions and not from income distribution. This article begins with a theoretical discussion of the main categories worked out by Putnam (civic community, socioeconomic development and institutional performance), focusing mainly on the origin of the civic community. Next, we present the results of statistical tests that identify the impact of the economic dimensions on the institutional performance. 2. Civic community, institutional performance and socioeconomic development Putnam 1 focuses on two possibilities to explain the difference in institutional 1 The author gave his dataset to the authors of this article and all the correlations and factor analysis that led to his main variables were redone 2

3 performance between North and South of Italy: the Aristotelian school of thought based on the prevalence of socioeconomic factors related to the performance of democratic institutions and the Plato's thought, which highlights the prevalence of sociocultural factors. In his book Community and Democracy, the socioeconomic modernity synthesizes Aristotelian hypothesis of the influence of socioeconomic factors in the evaluation of institutional performance. This possibility is linked to the consequences of the Industrial Revolution, from which large crowds migrated from the countryside, where developed agricultural activity for the urban and suburban factories. To test the assertion of Aristotle, Putnam checked the correlation between the economic 2 modernity and institutional performance. The crossing of variables economic modernization and institutional performance result in a correlation equals to This result, according to Putnam (2005), shows that there is not a strong correlation between them. With this, the author dismisses the economic modernization as a key variable to explain the institutional performance and focuses the research on the civic community variable. That is, the author explains the institutional performance by focusing on civic community. According to the author, a civic community 3 is characterized by the interest and participation of individuals in their public issues and in the collective welfare rather than the purely individual and private interests (Putnam, 2005). And yet, the author tried to understand the origin of the civic community through history. According to Putnam (2005), the civic community originates in the traditions of civic participation 4 of Italian society, i.e. the continuity of civic values, building on their historical antecedents to contemporary Italy. Community involvement or involvement in programs and projects in support of regional development are closely linked to cultural characteristics of community such as trust, norms of civic cooperation, and, finally, to the accumulation of social capital (Putnam, 2005). The result of the correlation between civic community and institutional performance is equals to Comparing those results, Putnam (2005) asserts that the correlation between socioeconomic modernity and institutional performance is weaker than the correlation between civic community and institutional performance. Then, the empirical study in Italy come to a result in which the "culturalist" thesis of Plato is the strongest to explain institutional performance. Note that none of the variables mentioned above, namely, civic traditions and civic community, have in their composition indicators of equality, cooperation and trust. And yet, as defined by Putnam, trust is the foundation of the civic community. It is visible, therefore, the lack of a clear theoretical relationship between the concept of civic community, as well as its origin and the tested variables in his research. This hypothesis, which is understood by the author as "sociocultural", which rescues the culturalism of Plato, in contrast to the materialistic outlook on Aristotle, is understandable. However, as mentioned previously, the composition of the variables used in his statistical tests, and the associations made 2 The economic modernity indicator used by Putnam, is measured by a factor score based on per capita income and gross regional product, the plots of the workforce employed in agriculture and industry, and in the fields of value added for agriculture and industry, all in the period (Putnam, 2005, footnote 4, p. 222). 3 The variable civic community is composed by four indicators of regional civic life: sociability, civic, mass media (newspaper readers), index of turnout and the composite index of preferential voting. 4 This variable is based on five indicators of traditions of civic participation in the period , which are: membership in the societies of mutual assistance; number of cooperatives per capita, strength of socialist and popular parties; turnout and local associations founded before Therefore, the civic community, is a contemporary variable that stems from a historical variable: the civic traditions. 3

4 by Putnam between "socioeconomic factors" and "modernity" and between "sociocultural" factors and "civic community" are not clear and present some problems. Thus, according to our viewpoint, in this context of equality, cooperation and trust, the civic community in its constitution, may be rooted in a more economic factor such as more equitable distribution of income. It is worth remembering that these characteristics are present in Aristotle, Plato and Tocqueville, who served as the theoretical basis of Putnam s hypothesis. According to Plato (1976), good government is composed of wise and virtuous men who do not get carried away by ambition or greed. These men are conspicuous by equality and freedom and correspond, according to the philosopher, to democracy. In contrast, to Aristotle (1998), the best democracy and good governance occur when people, although numerous, have a small territory of which draw their livelihoods from agriculture or livestock, it is characterized as people of meager wealth, hard workers and very little ambitious. Men who govern should be good men, honest and do not have unlimited freedom, because they are under the yoke of others, are more concerned with equitably govern, making the responsibility something flawless, without threatening the superiority of the people. Thus, the degree of political participation of a society involves not only the presence but also the intensity in which some features are presented within each community, namely: trust, norms of cooperation - social capital - equity and equality of conditions. And yet, the intensity with which these features are presented or can be stimulated in a community is linked to income inequality established there. In a democratic society where there is no monopoly of power and property in the hands of a few people, the social distances are not so large and there are equal opportunities, it is expected that these communities are able to promote a network of ties and work towards potentialities of the region, which lead to the development (RAMOS; MARINO, 2004). According to Paiva (2004), social capital, resulting from social interaction, boosts the economic system (increasing production and productivity of the system), thereby becoming an economic resource, not privately appropriable. On this matter, the base of capital is in a solid trading system assuming that there is trust. Therefore, as a society develops, with a more equitable distribution of income, its ability to generate generalized trust also develops cooperative mechanisms - social capital - that reduce transaction costs and facilitate economic activities, thus promoting, the socioeconomic development. With the start of the regional socioeconomic development, increase the demands of economic and social agents related to the effectiveness of government in responding to their demands, thus determining the level of institutional performance. 5 Putnam (2005) states that having a good institutional performance means to solve problems and serve the community, using available resources. In his view, a civic community consists of educated individuals (newspapers readers) which interact politically through active participation in debates and discussions, with peers, about the problems that afflict them and deliberate how to address such issues. This conclusion seems redundant, because institutional performance is characterized by "meeting the demands of the community" and civic community is an organized community, which has expressed the collective interest. In other words, Putnam falls in a tautology saying 5 It is clear that the indicator of institutional performance used by Putnam is composed of cabinet stability, budget promptness, statistical and information services, reform legislation, legislative innovation, number of day care, number of familiar clinics, industrial policy instruments, ability to make expenditures in agriculture, local sanitary unit expenditures, housing and urban development and sensitivity of the bureaucracy 4

5 that the civic community has a stronger relationship with the institutional performance than socioeconomic modernity, i.e., it is almost tautological to conclude that civic and democratic cultures, which are indicators of civic community, generate civilized and democratic practices, which are indicators of institutional performance. Concluding this discussion, Putnam, using sociological arguments, points out that horizontal systems of civic participation are the foundation of civic community and that "own trust is a property of the social system as much as a social attribute" (2005, p. 186) and further that this social context is anchored in history, but doesn t make explicit the origin of the civic community. Also, when using such arguments (with sociological roots), the author departs from the analysis about the economic aspects and the hypothesis of Aristotle: "the prospects for real democracy depends on social development and economic well-being" (Putnam, 2005, p. 26). We argue in this article that these forms of organization, the horizontal and vertical systems highlighted by Putnam, can be determined by factors such as income distribution, consisting of an alternative vision to it, to assess the civic community, institutional performance and socioeconomic development. 3. Methodology Regarding the methodology the work of Knack and Keefer (1997) was used as the reference in this article. However, it is noteworthy to say that the tests presented here are not depicted as a reply, in full, of that one made by the above authors because beyond the adjustments made to the data, the hypothesis tested in their study have some specifications. That is, Knack and Keefer (1997) highlight the relevance of ethnocultural homogeneity associated with the generation of trust and civic norms and in our article we highlight the origin of civic community from the income inequality. It is understood, however, that the equitable distribution of income produces trustworthy behavior. That is, a democratic society with fair procedural rules, with good governance and great equity in income develops standards of reliability and enhances social trust more than their ethnic and cultural homogeneity (YOU, 2005). We used an expanded sample and developed empirical tests 6 using the databases from the World Value Survey - WVS in the period and European Value Survey- EVS (2000) for 49 countries 8. 6 Basically correlation analysis and multivariate regressions with stepwise technique 7 World Value Survey contains data from 76,303 respondents in respect of 52 countries, 01 country observed in 2004, 13 countries observed in 2005, 21 countries observed in 2006, 15 countries observed in 2007 and 02 countries in The representativeness of the sample is verified by a patch made with a weighting variable that provides the weight value for the respondent in each country. However, it is interesting to note that some countries have the value of the weight equals to one, resulting in no correction for that country. The reason this happens has not been verified. 8 As mentioned, the WVS 2005/2006 contains information relating to 52 countries. Among these it was necessary to eliminate three countries, which are: Andorra, no data were obtained from the IMF and the World Bank per capita income at the beginning of the period, annual growth rates over the period and the income Gini at the beginning of the period, detrimental for inclusion in the same regression testing; Hong Kong: the questionnaire in Hong Kong is incomplete and there is not in the same information on any group participation; Iraq: no reliable data were obtained for the growth rate of GNP over the period and many questions about the degree of trust in public institutions were either not answered or were but a small number of respondents. The remaining 49 countries were classified according to the reliability of the information available using the following criteria: 1) number of years with available information on GNP. Only those for which complete information was obtained from 1980 to 2006 were considered pretty reliable (1), 2) the existence or otherwise of information on independent variables (trust, civic, gini, etc..) to the top period. Only those countries that relied on information stunted consistent (though not 5

6 It is important to emphasize that the questionnaires used on those surveys, both the WVS, and the EVS have been modified over time. Thus, there was a variation not only in the countries where the surveys were implemented, but also their own questions and answer choices have changed. Moreover, the number of countries where these surveys are being applied has broadened. Due to the discontinuity of the research - marked by a strong tendency to expand the sample and increasing quality of informational questionnaires it was necessary to reconcile the information from the period, which allowed to operate with essentially the information from the World Value Survey of as a proxy for past years. 4. Study s variables As mentioned previously, problems were found, both theoretical and categorical, in the composition of the variables civic community and socioeconomic development tested by Putnam. The variables tested by the author are not composed of indicators of equality and trust as theorized by Aristotle. According to Putnam the trust stems from the civic traditions and not from income distribution. The following item shows the composition of the variables tested in our article, considering that underwent modifications that differentiate from the variables used by Putnam, Knack and Keefer. necessarily identical) to the contemporary values (which were used as proxies of these variables over the period) were full of reliable (1), 3) availability of all the values used the construction of those variables that are independent compound indexes by aggregation of different indicators or mean. Only those countries that relied on all the relevant information, reliability was computed as full (1). 9 It is worth to pay attention to the following cases: 1) Germany: The data for this country, referred to the period before unification (1990) relate exclusively to West Germany, 2) China was not characterized as a capitalist country that has conversion in so far that has not passed by the socially and politically disruptive processes that characterized the former USSR and the countries that adopted the collectivist regime after World War II in Eastern Europe, 3) Colombia: the measure Civic of this country suffered a intervention, not information was available for the item "does not pay public transportation." Not to exclude the country's research, since all other information was available, it was applied to this question the value of the arithmetic mean of the other three items that make up the Civic variable and were objects of computation in Colombia, 4) Egypt : the variable "mean trust in public institutions" is obtained by the arithmetic mean of trust in 10 institutions with public functions. In the case of Egypt, information was collected only for trust in the press, television, trade unions and public services (health, education, welfare, etc.).. Not to exclude this country, which has all other information, we calculated the "Mean Trust in Public Institutions" based on these four items, 5) Moldova: not included in the sample interviewed with no college degrees. It is believed that this fact is due to a peculiarity of the sample that is not strictly representative of the population. This is evidenced by the fact that the mean years of schooling of the population of Moldova (6.14) are superior to those of other countries with significant percentage of population with college degree, 5) Peru: none of the questions used to calculation of the variable "Civic" was computed in applied research in this country; 6) Rwanda: the questions "trust in the military" and "trust in government" were not subject to survey research. The indicator "Mean Trust in Public Institutions" was calculated, thus the mean of the other 8 items, 7) Trinidad and Tobago: the per capita income in the initial period is surprisingly high. Throughout the period the per capita income falls systematically until the year 1989, when it reaches the value of U.S. $ In 1993 is U.S. $ And at the end of the period in 2006 is U.S. $ 14, Apparently, the variations are highly influenced by the exchange rate and do not correspond strictly to the standard of living of the inhabitants of the country and 8) Vietnam: was not characterized as a capitalist country that has undergone conversion, as it has not gone by processes socially and politically disruptive that characterized the former USSR and the countries that adopted the collectivist system after World War II in Eastern Europe. 6

7 4.1 Institutional performance Indicators The trust in public institutions is also considered by Putnam a good measure of quality of government, i.e. according to the author "to have a good performance, a democratic institution has to be both sensitive and effective, responsive to the demands of their electorate and effective to use limited resources to solve those demands"(putnam, 2005, p. 25). Therefore, great trust in public institutions is a characteristic of a society satisfied with the performance of the Government. As Knack and Keefer did, we adopted the faith in government as a proxy variable of institutional performance used by Putnam. We identified the degree of trust in government to the percentage of respondents in the WVS who said "great trust" and "medium trust" in public institutions, according to items listed below. To achieve 100% was necessary to add the percentage of those who said "little trust" and "no trust" and "do not know." a) Armed forces, b) Print Media; c) TV Media, d) Unions; e) Police f) Judicial System; g) Government h) Political parties; i) Parliament; j) Public Services 4.2 Trust, civism and group Indicators The variable 'trust' 10 (trust in 2005) is measured by the percentage of people who agree more with the statement 'most people can be trusted' than with the statement 'you must be very careful when dealing with others' within the total of respondents, after removing the option 'not know' responses. This variable was composed of measures related to the first half of the year 1980, the first half of year 1990 and first half of As mentioned earlier, the design of the database World Value Survey - WVS originated in Europe in 1981 and was expanded from that period to countries worldwide. Due to this expansion project discontinuous WVS, it was observed that for the year 1980 there are information only for 12 countries, for the 1990 year the data are available only for 13 countries, that is, for these two periods mentioned there are no data for the totality of 49 countries of the sample. Due to the small number of cases for the years preceding the period and since trust is one aspect of cultural life of the people, and therefore does not change abruptly, we performed a correlation test for the measures 'trust 80' 'trust 90' and 'trust 05' which resulted in a high correlation demonstrating that this variable is stable, allowing the use of the measure 'trust in 2005' as a proxy for past years. The variable civism proposed by Knack and Keefer was reconstructed using the database of the World Value Survey - WVS in the period 2005 to The variable civism used in this article is measured by the sum of the averages of four items: a) never justified asking for government benefits without entitlement; b) never justified to avoid paying ticket on public transport; c) never justified no longer pay tax when they have a chance; d) never justified accepting a bribe to do his duty. The result is the average approval, measured by individual scores from 1 to 10. In the WVS questionnaire items with high scores are those related to the great 10 The variable 'trust' that is being presented at this time is a measure of generalized trust, i.e. trust in people is anonymous and not only in groups with kinship or friendship. This variable matches the variable trust used by Knack and Keefer (1997). 11 It is noteworthy that for the construction of civism and trust variables it was used WSV questionnaire applied in Brazil in

8 acceptance of bribes and tolerances with no undue appropriation rights. In this case, the scale was reversed, because to measure the degree of civic values should be contrary to those presented in the database. That is, in countries where individuals are more tolerant to non-civic actions, and with opportunism, smaller is the imputed value. After an average of four questions, we obtained a measure of civism to each country. The indicators that comprise the civism variable ( civic ) used in this study do not coincide with the component indicators of the variable civism used by Knack and Keefer (1997), because the questionnaire administered in the version of the WVS survey has changed. But these changes follow the same direction as the WVS database , towards achieving the proposed objective, causing no damage to the reconstitution of this variable. Given this, it is interesting to illustrate in detail the difference in the composition of the two variables. The variable civic used in our work is composed of four indicators, as mentioned above. The civic variable used by Knack and Keefer (1997) is composed of five indicators of civism, and the first three are identical. Therefore, the difference in the measures of this variable is in: (d) keep the money that you find and (e) fail to inform if you hit a parked car and caused some damage, which in our study were replaced by item (d) is never justifiable to accept a bribe to do his duty. Another difference between these two variables is that Knack and Keefer use a scale of one to ten to five response options, which make up the variable civic. Thus the maximum value for this scale reaches 50 points and the minimum 05 points and the variable civism that has been reconstructed in our study uses the average of the four questions presented. Likewise the variable trust, to the variable civism in the early 80s there are information only to ten countries and in the early 90s this number increases to 21 countries, but is still considered a small number of cases. They were then tested for correlation with existing data on civic for 1980,1990 and 2005 to determine whether the measure Civic 2005 could be used as a proxy for previous years. Faced with a very low result of this test, we concluded that the Civic 2005 is not a good proxy measure to be used for variable civic. It is noteworthy that this result is not surprising, because the variable is defined by civic attitudes toward cooperating with others (strangers), or how people should behave. The authors call attention to the results of their tests in which the variable civic and the variable trust had a low correlation 12. According to the authors this result is probably due to the fact that the data of civism are not reliable, since one can not know whether people do what they say, that is, if what it is said corresponds to the real behavior. Nevertheless, the variable civism - Civic measured by the sum of the averages of the four categories previously referred to the period of 2005 is being maintained as a proxy for civic in previous years. Following Knack and Keefer (1997) we constructed variables groups, P- groups and O-groups for the WVS / To compose the variables Gruperten05 and Grupativo05 we used as reference the results of the questionnaires from the World Value Survey - WVS ( ) 14, 12 According to the authors the correlation between civism and trust resulted in a correlation coefficient 'r' equal to 0.39 (KNACK; KEEFER, 1997). 13 It should be noted that these variables were constructed according to the requirements that are present in the 2005/2006 version of the questionnaire WVS-2005/2006. As these questions are not exactly the same questionnaire that serves as a reference to the work of Keefer and Knack, the variables P- groups and O-groups used in this work are not strictly identical. Nevertheless, we sought to preserve the criteria used by Knack and Keefer in differentiating the groups, as it seeks to demonstrate just ahead. 8

9 in which it was shown a list of organizations / voluntary associations and asked: "I wonder if Mr (s) belongs to and participates in each one of them, one belongs, but does not participate or does not belong to these organizations / associations (read the items and check one answer for each): a) church or organization / group religion; b) sports or recreation organization / association 15 ; c) artistic, musical or educational organization / association; d) trade union; e) political party; f) environment organization / association; g) professional association; h) humanitarian or charity organization / association 16 ; i) consumer organization / association 17 ; j) others. (Open question, codified a posteriori) If the respondent belongs to and participates in some organization / association is assigned the value two, if the respondent belongs to, but not participate in any organization / association is assigned the value one, and if it does not belong to these organizations / associations is assigned the value zero ( Gruperten05 ). It should be noted that the instrument used in the WVS - 90 the possible answers to this question are as follows: " Belongs" and " do not belong". In the instrument used in the WVS 2005/2006-answer options are "owned but not participate" and "belonging and participating" ( Grupativo05 ) and "does not belong." Due to this change occurred in the WVS survey instrument /2006 and to make compatible with the information used by Knack and Keefer for the WVS - 90 the following response options were used: "belongs, but does not participate" and "belonging and participating" constitute the item "belongs" in order to have equivalence with the previous year 18, i.e., this item corresponds to those who belong to independent groups for their involvement. From this, we put together the number of respondents who belong to any organization for all items that match the construction of the indicator and divided that number by the total respondents. Comparing with the study made by Knack and Keefer (1997) presented before and looking at the data from WVS 2005/2006, it was observed that the ten items collected in this question, corresponding to that used by Knack and Keefer, are different because, as mentioned before, the instruments of data collection changed during the period. And yet, as the purpose of this article is to perform analysis using data for the years 1981, 1990 and 2005/2006 was necessary to reconcile the information over the 14 It should be remembered that the questions used in this article refer to the instrument applied to Brazil by the Center for Public Opinion Research at the University of Brasilia DATAUnB - Applied Social Research, Item (b) sporting or recreation organization / association was not used by Knack and Keefer as reported in the footnote (24) on page 1272, because there are few countries respondents to this item in WVS However, due to the importance of that item in the study by Putnam, it was reconciled with the item (h) Youth, the Knack and Keefer, and maintained in this work. 16 Item (h) humanitarian or charity organization / association, in this study was considered the union of the items (a) social welfare services for elderly, handicapped or needy persons, and (g) third world development or human rights, from the study by Knack and Keefer (1997). 17 Item (i) consumers organization / association is not in the instrument used in years 81 and 90, so it was excluded in the construction of variable Groups Because the WVS - 81 is equivalent to WVS - 90, the compatibility of information permeate the whole studied period. 9

10 period. 19 Before to do that, the items from both years, namely 1990 and 2005/2006 were matched for the construction of the variable Groups The comparative table below illustrates the reconciliation of these items. YEAR 2005/2006 YEAR 1990 Church or organization / religious group Recreational or sporting organization / association Artistic, musical or educational organization / association Temple or religious organizations Youth (eg, scouts, guides, youth clubs, etc.). Education, arts, music or cultural activities Trade unions Political party Environment organization / association Professional association Humanitarian or charity organization / association Trade unions Political party or group Conservation, environment, ecology Professional associations Human rights and social welfare services for elderly, handicapped or needy people TABLE 1 Compatibility of the items of the questionnaires used in the WVS /2006 and WVS Source: Prepared by authors based on information from WVS /2006, WVS and Knack and Keefer (1997). In a second step was to replicate the construction of indicators for WVS and WVS and harmonized with the items WVS / This compatibilization is illustrated in the table below. YEAR 1981 YEAR 1990 YEAR 2005/2006 Temple or religious organizations Youth (e.g, scouts, guides, youth clubs, etc.). Temple or religious organizations Youth (e.g, scouts, guides, youth clubs, etc.). Trade unions Trade unions Trade unions Political party or group Political Party or group Political Party Conservation, environment, ecology Conservation, ecology environment, Church or organization / religious group Recreational or sporting organization / association Environment organization / association Professional associations Professional associations Professional associations Human rights and social Human rights and social Humanitarian or charity 19 It is important to emphasize also that a correlation test was performed for the variables Groups, P- groups and O-groups and the results of these tests are below the expectations due to changes in these variables during the period. 20 This process of compatibilization related to the three periods result in the presence or absence of some indicators. For example: item (f) community local actions towards issues such as poverty, employment, housing and racial equality, were excluded because they were not present on WVS /

11 welfare services for elderly, handicapped or needy people welfare services for elderly, handicapped or needy people organization / association TABLE 2 - Compatibility of the items of the questionnaires used in the WVS , WVS and WVS /2006 Source: Prepared by authors based on information from WVS /2006 and WVS , WVS Knack and Keefer (1997). To compose variables P- groups (Putnam groups) and O-groups (Olson groups) we used the same indicators used by Knack and Keefer. The variable P-GrupKK05 contains sports or recreation organization / association, artistic, musical or educational organization / association, and others. (Open question, codified a posteriori ) which were identified as those groups least likely to act as "distributional coalitions" but which involve social inter-actions that can build trust and cooperative habits. The variable O- GrupKK05 includes trade union; political party, and consumer organization / association, which were considered most representative of groups with redistributive goals. 4.3 Trust in the system Indicators The proxies of the putnamianas variables developed by Knack and Keefer, from available information in the WVS, seem relevant and, therefore it is worthy to reproduce them in the tests conducted on this study. However, one can not deny that they are not strictly equivalent to the variables operated by Putnam. This author used a number of years and a volume of primary interviews and surveys that enabled the gradually criticism and increasing refinement of the variables entering into his system. While being critical to the construction of some variables in Putnam s work, through factor analysis, there is no doubt that they synthesize a set of cultural elements and trust s patterns and sociability much broader than simple variables articulated by Knack and Keefer (1997), from WVS questionnaires, which were updated here. Moreover, we must consider that, in particular, the civic variable, which was obtained from the civic WVS based only in the self-reported behavior stated by the respondent - is particularly unreliable. The same argument can be imposed to the relevance of the variable (s) group (s). Especially in countries where non-participation in communal and religious activities is particularly a bad sign (as in Muslim countries, for example), it is expected that a non-negligible number of respondents who declare themselves to be members - active or not of such organizations are, in fact, non-members. Finally, we must see that the trust variable, as constructed by Knack and Keefer (1997), retrieves only the trust of respondents in other individuals, but not the trust in social institutions. In particular, it is silent with regard to trust in the ethical and socially consistent pattern of relationships established within the civil society that are not strictly individual, but permeated by the action of institutions focused on commercial gain. Seeking to address this set of limitations, it established a new indicator from available information in the WVS which is called "Trust in the System." This indicator consists of the following variables: TrComp = the variable work pays, it is measured by the percentage of those who said agree more with the words "long-term work pays" than the phrase "get rich is a matter of relationships and luck." ExGanGan = win, win expression. This variable refers to the percentage who reported more agreement with the words "wealth can grow for all" than the term "people can only get rich at the expense of others." 11

12 ConfGrandEmpr = the variable trust in large companies is the sum of the percentages of respondents who reported "a great deal" and "quite a lot" as a measure of trust in big business (to get 100% you have to add the percentages of those who said " not very much "and" not at all "). 4.4 Economic performance Indicators Putnam uses the variable socioeconomic development, measured by economic modernity, recognizing that it is a static measure of end point, that is, analyzed at the end of the period. Therefore in the tests in this study, we chose to use a dynamic variable that is the economic performance measured by variation rate of gross national product during the studied period. So economic performance is measured by TxvarPNB: This variable is the geometric average annual variation of GNP in national currency in constant prices. 21 As Knack and Keefer (1997), the GDP80 variable was used to measure income per capita at the beginning of the studied period. The authors emphasize that the impact of trust on growth should be higher in poorer countries, that is, to trust is essential where the contracts are not properly executed by the legal system and whose access to formal credit is more limited due to the presence an underdeveloped financial sector. Then, the variable used here to measure per capita income in the early period was PNBpercapinper - measured by mean gross national product of the first four years of period, in U.S. dollars and calculated at current exchange Income Indicators The variable distribution of income was measured by the GINI index which measures income inequality in several countries. This variable presents some peculiarities, as the Gini indexes of each country are not internationally comparable. Those characteristics are the indicators used in research to measure the income of the population in several countries. In Brazil, for example, a survey of census data with regard to family monetary income has the reference the month of the survey. This income indicator used in population censuses reduces relatively the effective real income, because beyond their monetary income, these families have agricultural production to self-consume, which is expressive but is not counted. This accounting standard is the same adopted by the developed capitalist countries, but in these last countries the percentage of rural population is significantly low, so the underestimation of the real grower income is significantly low. By contrast, some countries with a high percentage of rural population adopt standards of accounting for familiar income not only based on money, which will imply different standards of accounting of the value of rural non-monetary income, in a gradient that goes from the value that similar goods (grains, milk, fresh produce, etc..) are in the urban environment, to the monetary value that could be achieved by the sale of such products in rural areas. Objectively, what matters here is that, depending on the standard of accounting for non-monetary rural income and the expression of this portion of product in overall national income, one can obtain more distinct levels of concentration of income in the same country. 21 Source: World Economic Outlook, 2009, FMI. 22 Source: World Economic Outlook,

13 Due to this fact, to achieve the rate of inequality income and consumption for several countries, researchers at the University of the UN - United Nations University, UNU - generated different indicators of income distribution to the most distinguished UN affiliated countries calculated from the criteria adopted officially in various parts of the world. This initial movement was developed by a researcher at the Department of Political Science, University of North Carolina at Chapel Hill, Frederick Solt, who used in his research two types of the Gini index, namely, a Gini index of gross with values before Government taxation, i.e. the amount that people earn before the tax payment and a Gini index based on a net of available income, corresponding to the amount of people s income after taxation. This distinction in the rates causes variations because in countries where most of the taxes levied directly on income (e.g. income tax) the amount received in terms of income is different from the amount that people enjoy, i.e. available income. 23. So, for this two measures of income inequality were used in our article: - GiniNetSolt80-85 = mean (for the years 1980 and 1985) of the Gini indexes of the countries based on net income calculated by Frederick Solt within the program "Standardized World Income Inequality Database"; - GiniGrossSolt = mean (for the years 1980 and 1985) of the Gini index of gross income countries by Frederick Solt calculated within the program "Standardized World Income Inequality Database." 4.6 Education indicators The variable education (Analf15years80) is measured as the percentage of illiterates among the total population over 15 years of age in the early 80s. This indicator was obtained for the year 1980 or the closest year for which census information was available in the sample countries. The sources of information were obtained from Unesco and the Central Intelligence Agency CIA. 4.7 Capitalist conversion The capitalist conversion is a dummy variable that specifies whether the country was a socialist and underwent conversion to capitalism throughout the period for which we have information about the average performance of the sample countries. This variable was introduced to prevent the collapse of the economic performance of countries that went through the turmoil of the revolution, which put an end to socialism, do not mask the positive contribution of the income distribution and property for the development of trust and economic growth (central hypothesis to this article). This masking could be imposed as: 1) the countries with the socialist past have - even today, despite the concentration of ownership associated with conversion to capitalism - relatively democratic distributional patterns, 2) the political turbulence during the process of capitalist restructuring strongly depressed economic growth rates over several years, and 3) this period of low growth (or even negative rates) have a great impact on the average rate calculated for these countries in the extent to which the conversion processes were associated to earlier dismemberment of large political units (especially the USSR and Yugoslavia). For this, the number of years for which there is information on the economic performance of the new market s economies is lower than 23 For this reason, see Frederick Solt at 13

14 the number of years for which there is available information for countries that did not experience that process. In face of that, there were two alternatives regarding the use of information for these countries: to exclude those countries completely from the database and would lose a representative portion of the sample or to include those countries, regarding that transitional period in which some countries had, for some years, very low economic growth rates compared to others. From this clarification, the following item shows the results of the analysis performed in our study. 5. Results This section illustrates the empirical findings of the research about the impact of economic dimensions on the civic community and institutional performance. To determine which factors explain the institutional performance, we used the statistical technique of linear regression with stepwise method for the identification of effective independent variables (relevant and significant) among a wide set of variables in this study, named " potential independent "; i.e., variables that are part of different theory models (and even some degree, competitors like the one we advocated in this article and the Putnam s hypothesis) of explanation to the institutional performance. The aim of this method is to allow the statistical software (SPSS 14.0, in this case) to select the variables that enter into the equation, gradually, from its significance and representativeness in the determination of the dependent variable without researcher s interference and induction. The aim is to clearly determine whether the putnamianas variables (civic and groups) or variables of the "Aristotelian" hypothesis (in the sense given in our article) are privileged by the system solely due to statistical determinations. The variables entering in the stepwise regression test are: - Dependent Variable: ConfMédiaInstPúb (mean trust in public institutions) - Potential independent variables: TxvarPNB, PNBpercapinper, GiniNetSolt80-85, GiniGrosSolt80-85, ReconvKista, ConfGrandEmp, TrComp, ExGanGan, Trust (2005), Civic (2005), Gruperten05, Grupativo05, P-GrupKK05, O-GrupKK05, Analf15anos80 The results are translated into the equation given below, the adjusted coefficient of determination (R ²) is equal to (1) ConfMédiaInstPúb = ConfGrandEmp TxvarPNB Trust (2005) + e 24 Note that there is a functional relationship of institutional performance, as measured by average trust in the institutions, with the remaining variables in the equation: trust in large companies, the growth rate of gross national product and generalized trust. The other independent variables that were also present in the regression test were excluded by the stepwise method. The interpretation of the above equation is that trust in public institutions is related to trust in large companies (private sector), with a regression coefficient equals 24 The values of the variables correspond to standardized regression coefficients (β). 14