Assessing Hardship and Happiness

CIAO DATE: 1/00

Assessing Hardship and Happiness: Trends in Mobility and Expectations in the New Market Economies

CSED Working Paper No. 7
October 1999

This paper was presented at an October 1999 MIT meeting of the MacArthur Foundation Research Network on Inequality and Economic Performance, and later versions will reflect comments from that workshop.

“... absolute income levels do matter, but... how much they matter is inversely related to their level, eg., the lower the level of per capita income in a country, the more absolute incomes matter to subjective assessments of well being.”

This paper ₁ explores a number of relationships which ultimately underlie the sustainability of market-based growth. An age-old puzzle is why some societies seem to be able to tolerate significant degrees of economic hardship and yet retain political and social stability, while others break out into violent protest as a result of much smaller economic declines or shocks. Related to this is the contrast between wide degrees of support for market policies in some societies that have high levels of income inequality, such as much of Latin America and the United States, and much harsher criticisms of the market process—and in particular of its distributive outcomes—in societies which have much more equality, such as some OECD and Eastern European countries.

In this paper we argue that the political sustainability of market oriented economic growth is as much determined by relative income trends as by absolute ones, and that opportunity and mobility over time are as important as current distributive outcomes. In addition, we posit that individuals’ subjective assessments of their past mobility, as well as their expectations about the future, are as important as objective trends are. ₂ The paper places a particular focus on individuals’ evaluations of their prospects of upward mobility (POUM) and how that affects their support for market policies. Not surprisingly, capturing these dynamics presents a number of measurement challenges, some but not all of which we have been able to overcome.

A central objective of this paper is to provide a conceptual framework for exploring the relationship between objective mobility trends, subjective assessments of those trends, and expectations for future progress. We then exploit new data that is available for Latin America, analyzing the relevant questions in a region-wide survey of public support for market reforms and for democratic institutions in 17 countries in 1997 and 1998. At our request, the survey also covered respondents’ expectations for future progress. Finally, we present the results of a pilot research project in Peru, which explores the relationship between objective and subjective trends through repeated interviews of a sub-set of households from a 1985-1997 nationally representative panel.

Three Propositions: A Framework

There are many relationships that must be taken into account when assessing the effects of perceived economic hardship and/or satisfaction, as well as actual mobility rates, on political support for market reforms and ultimately on social stability. We focus on three such relationships which we think are particularly critical.

First of all, relative income differences have important effects on how individuals weigh the importance of income versus other variables such as regime type or institutional performance in assessing their own well being or “happiness”. These assessments in turn affect the political and social sustainability of macroeconomic policies. We posit that absolute income levels do matter, but that how much they matter is inversely related to their level, eg., the lower the level of per capita income in a country, the more absolute incomes matter to subjective assessments of well being. Even upwardly mobile people may be dissatisfied if those around them are moving up more quickly or if large gaps persist between them and the highest income groups.

Our Peru pilot survey finds that it is precisely the most upwardly mobile people in the sample that are most negative in their self-assessments. At the same time, our regionwide data suggests, rather surprisingly, that those who live in countries where income distribution is most unequal assess their future prospects for upward mobility more positively than those in more equal countries. A plausible explanation is that the marginal room for advancement seems greater when inequality is high. Thus relative income differences matter, but the effects do not always run in the same direction.

The second proposition is that the level of macroeconomic volatility and the nature of social protection in particular countries result in individuals’ placing very different weight on the importance of job security, income, and social policies in their assessments of their satisfaction with the macroeconomic and political regimes under which they live. Our regional data show that those who live in countries with high levels of inflation are most negative in their assessments of their past economic progress. How recent the collective memory of high levels of volatility is — e.g. the timing of reform — also seems to affect the importance that citizens attach to macroeconomic stability and economic growth versus to redistribution.

Our data shows that citizens in countries that have only recently stabilized are more supportive of market policies such as privatization, and less in favor of redistribution, than those in countries where reforms are more established. This suggests that in volatile economic contexts, citizens place a premium on growth and stability, and turn their attention to distribution issues only later as reforms are consolidated. Our analysis also finds that citizens who live in countries with the lowest levels of social protection are the most supportive of a strong private sector role in the economy.

A final relationship, which we explore in less detail than the others, is that between “social capital” and mobility. We posit that the different objectives underlying civic participation can result in “social capital” having very different effects on individual mobility rates, on perceived well being, and on aggregate growth. Here we make an important distinction between participation that is driven by economic necessity, such as soup kitchens or group credit schemes in poor countries, and voluntary participation in civic organizations, such as Putnam’s famous choral groups. ₃

Our data from Peru suggests that more upwardly mobile people are the least likely to participate in civic associations, which runs counter to the findings of Putnam and others that highlight the contributions that social capital makes to growth. ₄ We explain this by making a distinction between kinds of organizations. The upwardly mobile are much less likely to be involved in group insurance schemes when they are seeking opportunities outside the neighborhood, and also presumably have less time to associate. And while autonomous neighborhood associations can play an important safety net role, they can also be poverty traps, as risk-averse people will be afraid to leave the security provided by group membership (such as meals for the family from a soup kitchen) in order to seek opportunities outside. ₅

Our analysis of the Latin America data is a first attempt to gauge the broader applicability of this framework. Prior to reporting the results of that analysis, we very briefly review some of the relevant literature on mobility, on the political economy of growth, and on social capital. We also discuss some of the measurement issues involved in assessing individuals’ assessments of their own well being or “happiness”.

What is and is not in the Literature

Mobility work has traditionally been the realm of sociologists. Thus there is no broad body of economics or political science literature that covers mobility issues. There certainly are some important works on the topic, such as those by Jere Behrman, Angus Deaton and Christina Paxson, Gary Fields, Isabel Sawhill, and Gary Solon among others, but they focus primarily on the United States and on macro and micro economic determinants of mobility, rather than on political economy. There is a wide body of literature which covers the political economy of market reform, but, at least until now, has focused on inequality issues in a static manner, rather than on the dynamics of mobility, opportunity, and political behavior. ₆ This paper is part of a broader collaborative effort—the genesis of which is described in Birdsall and Graham (1999)—to build from these existing literatures and to establish a new line of research on the economics and political economy of mobility in new market economies. ₇

Central to any exploration of the political economy of mobility is whether people are willing to accept more inequality (or the persistence of high levels of inequality) if economic change generates more opportunities and thus more mobility, including that which is downward. ₈ One plausible explanation for voters’ continued endorsement of market reforms in many emerging market countries despite the persistence or increase of inequality is that reforms create new opportunities. Voters may perceive that market signals reward hard work, productivity, and innovation more than previously state-dominated economies, thus making the move to the market acceptable in regions of high and increasing inequality such as Latin America and in regions of visible and painful downward mobility such as in Eastern Europe and the former Soviet Union.

Yet that is an optimistic interpretation, and it may be that increased inequality and insecurity reflect deep and persistent differences across individuals and households in their capacity to exploit markets or in their access to education, employment, or property rights. ₁₀ If inequality reflects discrimination against certain groups and historical handicaps that ensure the intergenerational transmission of poverty, then mobility, measured over lifetimes and even generations, will be low. Current acceptance of market reform could be the short-run outcome of the limited political voice of those excluded from new opportunities. This remains an open question. This paper seeks to help provide the methodological and empirical tools to answer it.

The existing political economy literature does not provide us with answers. Much of the literature has focused on the variables explaining the adoption and implementation of market reforms, and only recently has some work focused on the factors that make reforms politically sustainable among broad sectors of the population. ₁₁ What analysis there is is of an ex post nature, i.e. it examines how populations vote in the period after reforms have been implemented. ₁₂

In order to understand what underlies this voter behavior and if it may change in the future, it is necessary to understand two phenomena, neither of which we have a great deal of information about. The first of these is objective trends in mobility during reform, ie. who is moving up, who is moving down, and why. Recent work by Birdsall and Graham (1999) has attempted to collect what information there is on mobility in the emerging market countries, and is briefly reviewed below. The second is how people perceive their past mobility and their prospects of upward mobility in the future. This is clearly the most uncharted territory, which we explore in this paper.

There are several bodies of literature which can inform this effort. The first is that which attempts to measure overall economic well being or happiness. This effort was pioneered by the work of Richard Easterlin. ₁₃ The second is the political economy literature which examines the extent to which people’s evaluations of their prospects of upward mobility (POUM) determine their stance on redistributive voting. ₁₄ While this literature has focused primarily on the developed economies, it provides a useful framework for gauging the role of POUM in influencing voter support for market policies in the developing economies.

The third body of literature which informs our efforts is that which looks at the role of networks, community organizations, and other forms of organization in determining people’s mobility. These issues are often bundled into the catch-all concept of social capital, particularly since Putnam’s 1993 work. While not discounting the importance of the work that Putnam inspired, we find the work of sociologist Mark Granovetter more relevant to our analysis. ₁₅ We do not attempt an exhaustive review of these literatures but instead highlight work that has helped us in framing our thinking.

Mobility and Happiness

Richard Easterlin was a pioneer of the economics of “happiness”, which seeks to explain how individuals assess their own well being. ₁₆ In a cross country study using 30 surveys from 19 countries, including developing countries, he found that in all cultures the way that most people spend their time is similar: working and trying to provide for their families. Thus the concerns that they express when asked about happiness are similar. He finds that within countries there is a noticeable association between income and happiness, and in every survey those in the higher income bracket were happier than those in the lowest groups. However, whether any such positive association exists among countries at any given time is uncertain. Certainly there are not the differences in happiness among rich and poor countries that one would expect. Similarly, in the one national time series available at the time of his study (the United States since 1946), higher national income over time was not systematically associated with greater happiness. Health, meanwhile, is a demographic variable that has clear effects on happiness in all societies, a finding which later studies share. ₁₇

Easterlin’s findings suggest that absolute income levels matter up to a certain point, but after that it is relative income differences that matter. ₁₈ How these differences are evaluated depends on societal norms, which vary among societies. Due to such norms, some societies—such as the United States—are more willing to tolerate higher levels of inequality in exchange for benefits (real or perceived), such as greater freedom or opportunity, even if these benefits do not necessarily exist in practice. ₁₉ In general, though, Easterlin notes that while the aspirations of higher income people probably exceed those of lower income people, the dispersion in reference norms is less than in the actual incomes of the rich or poor. Thus those at the bottom tend to feel less well off. And as economic conditions improve over time, so do the reference norms, so that the positive correlation that shows up within countries appears only weakly, if at all, in comparisons among societies in time or space.

These findings about subjective assessments of well being are supported by the work of Robert Merton, who introduces the concept of reference groups in his 1957 analysis of Stouffer’s American Soldier. ₂₀ Merton finds that people’s aspirations—and therefore their satisfaction or happiness—are very much determined by the reference group that they compare themselves to. Infantrymen, whose cohorts were rarely promoted, reported higher scores of self-satisfaction than did their more upwardly mobile counterparts in the air force. Because promotion and upward mobility were the norm for air force men, and they assessed their own progress according to that of their peers, a higher percentage were dissatisfied with their own progress—even when they were upwardly mobile—than were infantrymen.

This relative definition of economic well being has also been used to explain social unrest and political violence in many countries. Ted Gurr, in his well known work on revolution, notes that: “Discontent arising from the perception of relative deprivation is the basic, instigating condition for participants in collective violence....Societal conditions that increase the average level or intensity of expectations without increasing capabilities increase the intensity of discontent. Among the general conditions that have such effects are the value gains of other groups and the promise of new opportunities.” ₂₁ The dramatic changes in incentives and rewards structures and in the nature of opportunities that have accompanied the turn to the market in many countries suggest that relative deprivation is a factor in the sustainability of those changes.

More recent studies of “happiness” confirm the basic thrust of Easterlin’s (and Merton’s) findings. Bruno Frey and Alois Stutzer, based on an analysis of 6000 residents in Switzerland’s cantons, explore the relationship between income and happiness. They find that at low and medium levels of income (for Switzerland), a higher equivalence income has no effect on happiness, while above a particular income level, it does have some effect. On the other hand, two variables: unemployment and poor health, have clear negative effects on happiness, and self-employed people are happier than employees. Inflation has a negative effect on happiness. Frey and Stutzer also explore the role of direct democracy. All residents in Swiss cantons receive public goods, but only Swiss nationals can participate politically. Controlling for differences in quality of public goods among the cantons, they find that happiness levels are higher among the Swiss nationals that take part in direct democracy than in the foreign residents that only benefit from the public goods it provides. ₂₂

Charles Kenny explores the links between happiness and growth. ₂₃ Like Easterlin, he notes the importance of relative rather than absolute income differences in people’s self assessments. He finds that, at least in wealthy countries, if there is a link between growth and happiness, it is from happiness to growth rather than the other way around. This linkage may be due to a social interactions effect: trust and social capital seem to be greater in “happier” societies, and a number of studies have found positive associations between these two variables and growth. Our analysis of the Latin America survey data also finds a positive association between trust in others and growth.

Kenny also notes that the nature of utility matters: indifference curves measure desire, not satisfaction, and the non-rational actor, like King Midas, may be moving up an indifference curve in pursuit of a desire that does not satisfy him. The measurement of “happiness” or subjective well being is far from straightforward, and entails all sorts of non-rational factors, as well as social norms, social interactions, neighborhood effects, and the economics of identity. ₂₄ The criteria that a young male member of a gang uses to assess his economic well being is probably quite different from that of a similar aged recent graduate of Yale, even though both live in the same country where the sort of norm-sharing that Easterlin refers to is in operation.

Along these lines, William Foote Whyte examines the behavior of youths and the interactions within and among their groups in an Italian slum in Boston: clubs for upwardly mobile college boys, and gangs for the boys who remain on the street-corner in the slum. His findings show how norms can derive from social interactions established in early boyhood, when some reference groups form. Some of the boys who began excluded from the gangs were freer and more motivated to leave the slum and pursue higher education and successful careers than were gang members. Gangs, more than clubs, are tight networks of reciprocal obligations, which can end up being social traps. Yet reported satisfaction is not lower in the gangs than among the college boys, highlighting the role of reference groups in subjective assessments of well being. ₂₅

A related measurement issue in assessing subjective well being is the direction of causality: are people happy because of their economic conditions, or do happy people assess their economic conditions more favorably? In addition, there is clear evidence that respondents’ assessments are often affected by the momentary mood at the time of interview: the fate of the national football team or a recent election may sway a response as much as economic factors. The same factors can affect recall, and people often recall past events in a manner that supports their current assessments. ₂₆

Recognizing these limitations, we build on these approaches to assessing subjective well being in our analysis of public perceptions of trends in mobility and opportunity in Latin America. Most research has been done in developed economies. We hope that our exploration of these issues in the developing countries contributes to the empirical evidence that exists, as well as to the methods of assessment.

Mobility and Social Interactions

Social interactions have a role in determining mobility rates. While much has been written recently about the positive role of social capital on growth, our view is that the effects are far less straightforward than is typically assumed. The definition of social capital plays a critical role, and this definition hinges on the type of interaction at play. Much of the social capital literature assumes that social interactions are positive, while other research, such as that of Steven Durlauf and of Karla Hoff, show that some kinds of social interactions can result in poverty traps. ₂₇

Rather than focusing broadly on the relationship between social interactions and economic growth, we focus on mobility. The findings from our Peru survey show that the most upwardly mobile groups are the least likely to belong to neighborhood or other civil associations. There are many reasons for this, which include the opportunity costs of time spent associating; the kinds of linkages that these associations provide—or fail to provide—with the world outside the neighborhood; and the nature of the associations themselves. Many civic associations in developing countries arouse out of shared economic necessity: soup kitchens, mothers’ clubs, group credit schemes. Their purpose is to make up for the absence of adequate economic opportunities or safety nets. ₂₈ Leaving the group involves individuals risking losing the security benefits of membership in order to seek better opportunities outside. Those that choose to move on will self-select according to their education levels, their degree of risk aversion, and available information and opportunities.

In contrast to these kinds of civic organizations are those that Putnam and others refer to: the voluntary associational arrangements that foment trust, transmit information, and ultimately contribute to economic growth. These are distinct from the survival organizations of the poor in two ways. The first is that members choose to associate voluntarily rather than as a last resort. The second is the kinds of ties that the groups have with the rest of society. Granovetter distinguishes between strong ties, or friendships, which provide horizontal linkages within organizations or local groups, and weak ties or networks, which provide bridges to other groups and networks beyond the locale. ₂₉ His empirical work, based on interviews with U.S. blue and white collar workers, shows that weak ties are consistently the basis for upward mobility.

Other authors have made related arguments. David Krackhardt, building on Granovetter’s work, notes that weak ties provide access to information and resources beyond those available in their own social settings, while strong ties are better suited to providing assistance and adaptation to economic change and uncertainty. ₃₀ The survival organizations of the poor are bonded by strong ties, yet weak ties are more likely to precipitate upward mobility. Paul Collier defines social capital as social if it is an interaction that generates an externality, and as capital if its economic effects have persistence. ₃₁ Most organizations of the poor meet the former but not the latter criteria.

These distinctions are important when establishing linkages between “social capital” and upward mobility or economic growth. Many of the organizations of the poor play invaluable roles in providing safety nets or social cohesion, yet lack Granovetter’s “weak ties”, as well as Collier’s persistence effects.

In empirical work in Central America, Amber Seligson finds that of the many kinds of civil society organizations that the poor belong to, only one, community development organizations, foment weak ties. They provide channels for making demands: to the local mayor, to the legistature, or to a relevant central government agency. Other organizations, such as church or school-related associations, are inward looking and rarely need to look beyond their network to solve problems. ₃₂

Henry Dietz, in a longitudinal study of political behavior across poor neighborhoods in Peru, finds that as economic pressures build, the poor turn their organizational activities from social mobilization and political demand-making to inward-looking, neighborhood coping solutions. Their behavior becomes increasingly risk-averse as the need to preserve economic security or income generation capacity increases. ₃₃ Again, associational life is providing welfare externalities, but it is not encouraging upward mobility or economic growth; it may even in fact be deterring it.

Mobility and Voting

One way of gauging how people assess their current well being and their future prospects for advancement is how they vote. Roland Benabou and Efe Ok develop a hypothesis of the prospects of upward mobility (POUM). ₃₄ They posit that it is people’s perceived prospects of mobility that explains economic and political stability even when the median voter is well below the average in terms of income. ₃₅ Because the poor majority perceive that they—or at least their children—will be above average (mean) income in the future, they will not vote for redistribution, as higher taxes will hurt them later: tomorrow’s income is a concave function of today’s. The coalition in favor of laissez faire is larger the more concave the transition function, the longer the duration of the proposed tax scheme, and the more far-sighted the votes. They support their theoretical work with empirical evidence from the US, relying on the PSID. ₃₆

Benabou and Ok’s findings depart from those of Alberto Alesina and others, who have demonstrated formally that high inequality leads to political instability, populist economic policies, and ultimately lower rates of growth, as the median voter votes for high redistribution, which in turn undermines investor confidence. ₃₇ Yet most of these models have not accounted for mobility, and rely only on current income.

One exception is Thomas Piketty, who argues that individual mobility experiences are key to political attitudes, and that persistent differences in perceptions about social mobility can generate persistent differences in distribution patterns across countries. ₃₈ De Toqueville attributed different attitudes towards redistribution in the United States and Europe to their respective mobility rates. Piketty cites the importance of social origins and mobility experiences: voters with the exact same incomes but different social origins will vote differently on redistribution. These differences are particularly strong at the extreme tails of the distribution, i.e. stable low income and high income voters are very likely to maintain their political identities, while upwardly and downwardly mobile groups in the middle are more likely to shift identities. Research by Clifford and Health, based on empirical work in the U.K., applies a hypothesis of asymmetric mobility: those who are upwardly mobile usually adopt the political behavior (usually conservative) of the class they arrive in, while the downwardly mobile continue to associate with the class that they came from. ₃₉

Piketty shows how attitudes generated by past mobility experiences have persistent effects on future economic behavior, and can account for widening inequality. Even without redistribution, inequality for a given homogenous cohort can grow with age. When people are young and start out with the same beliefs, they put forward the same degree of effort, and the only inequality comes from shocks. But as time passes, people who have received negative shocks may get (rationally) discouraged and supply less effort, while more successful people keep putting out more effort. Eventually, inequality persists due to endogenous beliefs dynamics. ₄₀

In this paper we posit that both the effects of past mobility on political attitudes and prospects of upward mobility (POUM) are factors in the political economy of reform in Latin America. They may be significant variables explaining why voters in many countries have repeatedly voted for the continuation of “laissez faire” or neoliberal economic policies despite the persistence and even increase of inequality. Indeed, our empirical results show that POUM levels (self assessments of individual prospects of upward mobility) are actually higher in the more unequal countries early on in their reform programs (discussed below).

Objective Mobility Trends: What We Know

It is difficult to fully capture trends in mobility in the emerging market countries, largely due to the shortage of panel data. Yet it is possible to get a sense of these trends by relying on proxies for panels and focusing on shorter time periods (versus the intergenerational panel data that is available for the US and the OECD). There are also pockets of data from particular countries. Here we focus on broader indicators of what is happening on a regionwide level with the turn to the market, building largely on the work of the contributors to Birdsall and Graham (1999). ₄₁

Nancy Birdsall, Jere Behrman, and Miguel Székely have constructed indices of intergenerational mobility for countries of Latin America and use those indices to explore the effects of economic policies, macroeconomic conditions, and education programs on that mobility. They find that the depth of financial markets and an emphasis on basic schooling in public spending enhance intergenerational mobility. ₄₂ Though the immediate effects of market reforms and education policy reform on current income distribution are not evident, longer-run positive effects of greater mobility on distribution seem plausible.

In a similar effort to measure social mobility trends in the region, Momi Dahan and Alejandro Gaviria construct an index of mobility based on the correlation of schooling gaps between siblings: the between family variance of mean schooling gaps versus the overall variance of the gaps. ₄₃ With perfect mobility, family background would not matter, and siblings would be no more alike than two people taken at random (barring shared genetic traits). In an immobile society, family backgrounds would dominate and most siblings would fare alike. They compute their index based on household surveys from sixteen Latin American countries and find that social mobility is highly correlated with both average schooling and inequality of schooling. They also find a strong relationship between mobility and education expenditures, and only a weak relationship between mobility and GDP per capita and income distribution. ₄₄ More generally, they find that most countries in the region, with the exception of Mexico, experienced a slight increase in mobility in the early 1980’s and mid-1990’s.

Katherine Terrell examines worker mobility and winners and losers in the post-Communist economies. ₄₅ She defines winners and losers in terms of changes in relative earnings and employment status. She finds that the winners so far have been young, educated men whose skills enabled them to exploit new opportunities in the private sector. The growth in women’s returns to education has lagged behind men’s, and the skills of older workers are much less valued than before the transition.

David Hojman, in a study in Chile, focuses on “market-driven, medium term mobility”, i.e. the changes in mobility trends that are driven by policy change. ₄₆ After two decades of policy changes and structural reforms, Chile’s highly unequal income distribution remains very similar to the pre-reform period, despite major strides in reducing absolute poverty. Though income has increased across the board, by far the largest increases have gone to managerial (skilled) personnel. Hojman’s findings are also supported by evidence from annual, region-wide cross sections, which suggest that the rewards to skilled labor have far outpaced those to unskilled labor. The explanation for these trends is two-fold. First, trade liberalization has rewarded skilled rather than unskilled labor in the region, contrary to what classic economic theory would predict. ₄₇ Second, because education policy has not kept up with demand in the region, skilled labor is in relatively short supply, which has further increased its marginal gains relative to those of unskilled labor.

Younger, more educated, and skilled groups have been the relative winners in the transition to the market in Latin America and Eastern Europe, with subtle differences between regions. In Latin America, even the poor have made absolute gains in mobility, although relative gaps have widened. In Eastern Europe, in contrast, there has been significant downward mobility as the result of much more dramatic changes in the structure of economies and welfare systems. At the same time, many educated groups, whose labor was undervalued under state planning, have experienced upward mobility.

In the remainder of this paper, we explore two questions. The first is how individuals assess the effects of general changes in mobility on their own progress, and in turn how those assessments affect evaluations of future prospects. For this analysis, we primarily rely on the regionwide opinion surveys from Latin America. The second question is what factors influence these public perceptions and explain the variations between perceptions and actual reality. For this we rely on the results of our Peru perceptions survey. The three variables introduced in the introduction of the paper: the role of relative income differences, the level of macroeconomic volatility and the nature of social protection, and the role of social capital, are central to our analysis.

Data and Measurement Issues

Finding adequate and reliable data to assess mobility trends and expectations in new market economies is a challenge at best. Panel data is rare with the exception of a few countries. Thus assessing objective mobility trends requires exploiting existing data in new and innovative ways, such as by Berhman, Birdsall, and Székely (1999), as well as exploiting whatever panel data is available. Assessing trends in expectations is also difficult, as opinion polls differ in their reliability—and in their rural/urban representation—across countries. And as respondents in these polls are usually placed in socioeconomic categories according to limited objective criteria and to their own self assessments, there are a host of measurement errors. In particular, the reliance on self assessments tends to skew samples towards the middle income categories. ₄₈ Accepting these obstacles, we analyzed two new sets of data. The first is the region-wide Latinobarómetro survey, which has been conducted annually in seventeen countries since 1996. The survey is managed by a respected polling firm in Chile, Market Opinion Research International (MORI), which in turn identifies qualified firms in each of the countries. While the survey has an urban bias, and there are some differences in quality among the polling firms used, MORI’s reputation and its transparent management of the data give us confidence that the country surveys are at the least comparable. For 1997 and 1998, we were able to get MORI to include some additional questions about perceived mobility and expectations in the surveys.

We chose 109 questions from the survey, resulting in 17,839 observations upon which we have based our statistical analysis. Our most robust results come from analysis of the effects of micro-level factors on respondents’ answers to questions, for example, where we explored correlations between individuals’ perceived mobility and their prospects of upward mobility. We also attempted analysis of the effects of macro-level factors, such as inflation or economic growth, on individual attitudes, and here our statistical analysis is more limited, as we have had to rely on country averages.

We compiled indices for several of the variables that we explored, such as individuals’ prospects of upward mobility (POUM), confidence in institutions, and economic “happiness”. In each case the index was based on several relevant questions in the survey, which were weighted and then averaged. Each of the indices is described in detail in Appendix A of the paper. We also compiled country level indices for major macro level variables, such as recent growth trajectory, inflation, and the effectiveness of social welfare institutions, which are also described in the appendix. The results of statistical analysis based on these indices are in Appendix B.

Our Peru pilot survey is, thus far, the one case where we are actually able to compare objective mobility trends, as captured by a nationally representative 1985-1997 panel, with subjective self assessments. Richard Webb, in collaboration with Nancy Birdsall and Carol Graham, developed a questionnaire covering respondents’ assessment of recent trends in their own economic progress and of their expectations for their future progress as well as their children’s (and grandchildren’s). It also explored the effects of particular variables such as health shocks and memberships in community organizations. The complete questionnaire appears in Appendix C-3. The survey, which covered approximately 150 households in urban and rural areas, was conducted by Cuanto in May of 1998 and repeated in May 1999. ₄₉

The survey yielded results which have implications for survey research and methodology. In particular they exposed major differences in modes of answering questions among urban and rural respondents, differences which should be accounted for in the design of future surveys of household income and of subjective assessments of well being. The results of the pilot are discussed below. A next stage of this research will entail conducting the survey in other countries where panel data is available.

Perceived Mobility, POUM, and Support for Markets in Latin America

As the research by Benabou and others suggests, how people evaluate their future prospects of upward mobility can have a major role in determining their attitudes towards markets, taxes, and a host of other issues related to the structure of the economy. These attitudes are in turn reflected in how citizens vote and behave economically. Benabou posits that if the majority believes that their income will be above the mean in the future, they will not vote for redistribution, even if that same majority is well below the mean. In Latin America, where the majority of the population is far below the mean income, support for markets, at least as reflected in continued voting for market policies, remains high in most countries. ₅₀ We attempted to gauge respondents’ prospects of upward mobility as measured by the Latinobarómetro questionnaire, as well as how respondents evaluated their recent economic progress. We explored how these attitudes varied by country, by age cohort, and by occupational categories. We then explored the correlations of these attitudes with a number of micro and macro variables, both through bivariate analysis of the entire Latinobarómetro sample, and through analysis of the data aggregated at the country level. As noted above, we relied on indices to capture individual attitudes, such as about POUM and perceived mobility (IPM), and country-wide variables (Appendix A). The correlation coefficients among the micro level individual variables appear in Appendix B-1, and those among these attitudes and macro-level variables in B-2.

An important caveat is that the macro-level analysis was limited in terms of statistical significance. While the micro-level analysis involved 17,839 observations, the macro-level analysis entailed creating country level measures — summarizing individual observations pertaining to attitudes — in order to analyze them in light of macro-level variables such as GNP per capita. This exercise was inherently limited by the aggregation process and the number of observations, e.g.17 countries.

Many of our findings in the resulting statistical analysis are intuitive; some are not.

Our most significant findings pertain to the relationship between the POUM, IPM, and pro-market indices. These three indices respectively attempted to capture individual attitudes about future prospects for mobility, evaluations of past mobility, and degree of support for market policies.

Figure 1. POUM Index, 1997-1998

The countries with the highest POUM ratings over the two years surveyed are Brazil, Bolivia, Guatemala, Paraguay, and Chile. (Figure 1) The country with by far the lowest POUM—indeed an outlier in the study—is Mexico. This may reflect Dahan and Gaviria’s findings that Mexico was the only country in their sixteen country sample where mobility did not increase in the 1980’s and 1990’s. ₅₁ Mexico also had the largest drop in its POUM rating, falling 34% from 1997 to 1998. One plausible explanation for the drop could be public apprehension of the potential spill-over effects of the Brazil currency crisis. While the effects were not that strong for Mexico in the end, the crisis was a source of public concern in a population that only recently recovered from its own major currency crisis in December 1994. ₅₂

Colombia, meanwhile, was the country with the highest POUM ranking in the region in 1997, and its ranking fell by 22.7% in 1998. This is not surprising, as during the period Colombia entered the worst recession it had experienced in several decades. This certainly was noticed in a country with a history of prudent economic management and steady levels of economic growth.

As the Peru results will show, people’s evaluations of their future prospects, which are even more subjective than their evaluations of past progress, are susceptible to changes in overall macroeconomic conditions and in the national political mood. These effects are possibly greater in emerging market economies, where both macroeconomic performance and politics are more volatile than they are in contexts where authors such as Benabou and Piketty have explored the effects of mobility on voting.

Within particular countries, respondents’ evaluations of their future prospects were affected by the same demographic and occupational variables that determine objective mobility trends. Students and private employees had much higher POUM levels than any other occupational category (see Table 1). The temporary unemployed had the next highest POUM rankings. A plausible explanation is that many of those who belong to this group have left a job with the hopes of finding a higher paying one. Among the self-employed, professionals (e.g. lawyers and doctors) had much higher POUM levels than any other self-employed occupational category. Owners of stores or business followed professionals in POUM rankings, while farmers had the lowest rankings of all. The retired and house-keepers had the lowest rankings of the occupations. ₅₃

Table 1. Relative levels of Prospect of Upward Mobility (POUM), by different categories
Category	High POUM	Moderately high POUM	Moderately low POUM	Low POUM
Country	Brazil, Bolivia, Paraguay, Guatemala	Honduras, Venezuela, El Salvador, Chile, Panama	Costa Rica, Uruguay, Nicaragua, Argentina, Peru, Ecuador	Mexico
Age	less than 20	20s	30-50	over 50
Education	University and Secondary completed	Incomplete secondary	Primary	Illiterate
Occupation	Student private employee	Temporary unemployed	Self-employed, public employee	Retired, housekeeper
Wealth group	Richest	Rich and Medium-Rich	Medium-Poor	Poor
Source: authors’ calculations from Latinobarometro 1998

There was a clear regionwide correlation between age and POUM rankings, with younger groups being much more optimistic about their future economic prospects. While one would expect this relationship between age and expectations in most societies, the effect may be stronger in new market economies, where rewards go to more educated and more adaptable workers, and in Latin America younger age cohorts are far more educated than older ones. ₅₄ Not surprisingly, there is a similar correlation between education and POUM rankings, although with some outliers. In Argentina, illiterates had much higher POUM rankings than any other group. One explanation may be Argentina’s relatively good education record: as most of the population has been well educated for a long time, the differences between rewards to the educated and non-educated are most obvious to the illiterate group.

There is, not surprisingly, a relatively strong regionwide correlation between POUM rankings and those for perceived past mobility (Appendix B-1). One outlier was Paraguay, with a very high POUM ranking, but a very low IPM. Both IPM and POUM rankings, meanwhile, were positively correlated with confidence in institutions. ₅₅ Interestingly enough, confidence in institutions was weakly and positively correlated with inequality in the Latinobarómetro survey. Trust in other individuals, on the other hand, was strongly negatively correlated with inequality (Appendix B-2).

There was also a significant but less strong correlation between POUM rankings and the pro-market reform index (Appendix B-1). Those people who supported market reforms also tended to be positive about their prospects for upward mobility. At a country level, some of the highest pro-market rankings were in Central America, even in countries with low POUM rankings. This may be explained by the timing of reforms in Central America, where progress is well behind that of South America.

Across the board, we found that pro-market views are stronger in countries earlier on in their reform processes. Generally, market reforms produce increased inequality in the short term. But they also create new opportunities for mobility and opportunity. In addition, there are tangible benefits from reforms in the early stages, such as the reduction of inflation, which contribute to favorable public opinion. And the reforms often take place during the new leaders’ political honeymoons, which are enhanced by the demonstration of political will necessary to implement difficult reforms.

Yet high levels of public approval are difficult to support over time without a deepening of reforms. Second stage reforms are often more difficult to implement, as they challenge entrenched interest groups in public sector institutions. ₅₆ And with time, the new niches for advancement created by reforms are filled by workers with skills and education, narrowing the margin for upward mobility. In our analysis, the fast late-comers to reform, who were still in the early phases when the surveys were conducted, have higher POUM levels than the “slow pioneers”(Table 2).

Table 2. POUM and Country groups (defined by pace and timing of reform)
	POUM level
Country group	High	Average	Low
Slow Pioneer	Honduras	Chile, Uruguay	Colombia, Mexico
Fast Late-comer	Bolivia, Brazil, Paraguay	Costa Rica	Peru
Note: Darker shading implies more support for the conclusion made by the authors, according to which slow-pioneers and fast late-comers present respectively low and high levels of optimism Source: Morley et al. (1999) and authors’ calculations based on Latinobarómetro 1998.

Our timing/maturity of reforms hypothesis is supported by the mixed relationship we see between POUM rankings and support for productivity (versus redistribution). Indeed, Chile, one of the countries with the most extensive reform trajectory and an average POUM ranking, had the lowest percentage of respondents in the region in favor of productivity. Support for productivity enhancing measures seems higher in countries earlier on in their reform process, where the collective memory of crisis and/or high inflation is more recent, while support for redistribution is higher in countries where the reform process is more complete and a record of macroeconomic stability has been established. Public attention can then turn to distribution issues.

Cultural and institutional differences can also play a role in determining people’s attitudes towards redistribution. However, in this case, the two countries with the strongest tradition of good social welfare services in the region, Chile and Costa Rica, have completely different outcomes: while Chile had the lowest percentage of productivity supporters, Costa Rica had one of the highest. Costa Rica, in contrast to Chile, has made much less progress implementing structural reforms such as privatization, which supports our timing of reforms hypothesis. These findings also support our initial emphasis on relative versus absolute income levels: in poorer countries in the region such as in Central America and Bolivia, issues of absolute income—e.g. growth—are accorded more importance than those of distribution, even when inequality levels are high. ₅₇ In wealthier countries such as Chile, Argentina, and Venezuela, where per capita incomes are higher, then issues of inequality and distribution may be accorded more importance in the public debate. We found that GDP per capita is highly correlated with the level of support for redistributive policies, with support for redistribution increasing with per capita GDP levels. (Figure 2. See Appendix B-2 for correlation tables). ₅₈

A nuance here may stem from the dynamics explored in Benabou’s POUM research: while inequality issues may be accorded importance in the public debate, it is not clear that it will translate into votes for higher taxes, particularly in countries with high POUM rankings. A next stage of this research will explore the relationship between POUM rankings and pro-reform votes in several countries in the region.

Economic “happiness”, meanwhile, was strongly and positively correlated to POUM and IPM rankings, as well as to political happiness. Political “happiness” was less directly related to POUM rankings. (See Appendix B-1) Political happiness was highest and POUM rankings were lowest in Costa Rica and Uruguay. These countries represent two cases of consolidated democracies with (relatively) low levels of inequality and well established social welfare systems.

The lower POUM rankings in this context support the thesis that people evaluate their upward prospects more moderately when income differences are smaller and the marginal room for advancement is not as great. If Easterlin’s findings about the role of relative versus absolute growth apply, the overall “happiness” assessments of citizens in more equal countries will be higher precisely because they are less concerned with narrowing the gaps between their position and that of wealthier groups in society, i.e. they are happier in general because they are less concerned with POUM!

There was a weak positive relationship between participation levels and POUM, with the overall scores skewed by some clear outliers: Chile with high POUM rankings and low participation ones, and Ecuador with low POUM and very high participation scores. The existence of at least a positive relationship—suggesting that societies with higher rates of participation are more optimistic/happier—is supported by Frey and Stutzer’s research work on Swiss cantons. On the other hand, for Latin America, participation was also positively correlated with a propensity to violence. This may reflect how respondents with higher participation rates answer questions about propensity to violence as much as actual violence rates, however.

The propensity to violence in the region was strongly and negatively correlated with POUM (see Appendix B-2). This supports Collier’s research, which finds that civil wars are much more likely to occur where the opportunity costs are low and poverty is high. The propensity to violence among respondents for Latin America was also positively correlated with poverty levels. The POUM ranking might be a proxy for respondent’s perceived opportunity costs.

In terms of macro-level variables, several key relationships are clear, while a few are counter-intuitive. Inflation and unemployment are negatively and weakly correlated with both POUM and IPM rankings. Not surprisingly, respondents were less positive about their past and expected progress in countries where inflation/and or unemployment was high. Inequality, meanwhile, was positively correlated with POUM and IPM rankings. While this seems surprising, it is supported by Benabou’s empirical findings that even when a majority of people are well below the mean income, there can be high prospects of upward mobility. This may also reflect a sort of regression towards the mean: in countries where there is high inequality, people have to assess their future prospects positively if they expect to approach the mean, while in countries with lower inequality, respondents may feel that they are already close to the mean and have less room for improvement.

Another finding which supports the relative versus absolute income differences hypothesis is that POUM levels are lower in countries with more effective social welfare institutions, again suggesting that poor respondents in countries with high inequality and low levels of social protection see greater marginal room for advancement than do low-income respondents in countries where incomes are closer to the mean. There is a strong negative correlation between productivity supporters and the social welfare effectiveness index (Appendix B-2).

In may be that in countries with stronger social welfare institutions people weigh the importance of personal upward mobility along with the support they expect from the state and from redistribution, while in countries where these institutions are weak, people see individual effort and economic growth as the only means to advancement. Related to this, the lower levels of trust reported in unequal countries suggest that their publics are less likely to have faith in the state and in redistributive policies, and accord more importance to individual achievement.

Another factor, identified by Piketty, may also be at play here. Political attitudes about inequality, which are formed by past mobility experience, can persist over time and across generations. Many of the countries that developed strong social welfare systems, such as Chile, Costa Rica, and Argentina, did so in response to political pressures related to public concerns about inequality. Those concerns were institutionalized via welfare policies, but also through political attitudes and platforms. Even under significantly changed economic conditions, these attitudes may still play a role in public perceptions about the role of individual mobility versus that of the state in furthering economic and social development.

The Accuracy and Significance of Subjective Assessments:
Results from a Pilot Survey in Peru

Peru is one of the few countries for which we have panel data, compiled by the Instituto Cuanto. The panel contains 676 households, with four observations at the national level, 1991, 1994, 1996, and 1997, and with a smaller subset of households from Lima in the panel since 1985, a time period in Peru which encompassed both unprecedented levels of macroeconomic volatility and dramatic policy reform.

In addition, we compared respondents’ subjective assessments with actual mobility trends, based on a sub-set of households in the 1985-1997 panel. A pilot survey of 152 households was conducted in May 1998 and again in May 1999, and provides important insights into the issue of perceptions and mobility. The full questionnaire appears in Appendix C-2. Among other things, the Peru results allow us to get a better sense of what the regionwide subjective assessments of well being signify.

Context for the Survey and Objective Mobility Trends

Peru is one of the poorest countries in South America. During the period under study, the country experienced hyperinflation and the deepest recession in its history, with GDP dropping by 25% from 1988 to 1990 and inflation reaching an annual rate of 8000%. In 1990, the newly elected government of Alberto Fujimori implemented a dramatic reform program, which entailed stabilization followed by extensive structural reforms ranging from trade, tax, and social security reform to wide-scale privatization. By 1994, Peru was the fastest growing economy in the world. The government also targeted social welfare expenditures to the rural poor, and made some important gains in reducing extreme poverty, although poverty levels remain high, even by regional standards. ₅₉ Not surprisingly, this set of economic conditions and policy changes was reflected in both changing mobility trends and in public expectations and voting patterns.

Mobility trends during this period are captured by the nation-wide panel. ₆₀ Transitions are defined here as transitions either into or out of poverty, or from or to extreme poverty from poverty. For the period studied, 39.8% of the total had some mobility, and mobility was higher in rural than in urban areas: 34.8% of urban households made a transition, while 44.9% of rural ones did (See Appendix C-1). ₆₁

Upward mobility was higher in 1991-94 than in 1996, no doubt reflecting the effects of stabilization and high growth rates in the first period, and then the minor economic adjustment of the latter period. The percent of the sample that moved up in 1991-94 was 26.8, while in 1994-96, that figure was 15.7. The percent that moved down in the first period was 10.8, and in the second period was 14.6. In the latter period, upward mobility was only 1.1 points greater than downward movement, and of the total poor, the same percent moved up as down (11.4%). In Lima, more people moved down than up in the latter period: 13.4% moved down while 11.9% moved up. In rural areas, in contrast, 19.6% moved up, while 16.0% moved down. The largest movements—across two poverty lines, either up or down—were in the rural areas (See Appendix C-1).

Part of this story reflects the positive benefits of stabilization and the elimination of substantial distortions in the policy framework for both poor rural and urban groups. The downward trends for Lima in 1994-96 reflect the 1995 economic adjustment’s more direct impact on urban groups. Another part of this story, which explains the more positive trends for the extreme rural poor during the second period, lies in dramatic changes in public expenditure patterns and transfers to the poorest. ₆₂

After an October 1993 constitutional referendum, which President Fujimori won by a narrow margin, but lost in most rural areas, discretionary public expenditures were dramatically re-oriented to poor rural areas. ₆₃ Prior to 1993, municipal fund expenditures went disproportionately to Lima, which took 54% of the total. After 1993, that percentage was reduced to 17.4%. The expenditures of the social fund, Foncodes, were directed to departments that voted “no” in 1993. While this was clearly a politically driven allocation of funds, and resulted in the resignation of the head of the social fund, it also resulted in the funds being re-directed to the departments with the worst social indicators in the country. Food aid, which prior to 1993 was regressively allocated and concentrated in Lima, was also re-directed to poor rural areas. ₆₄ While it is an insignificant share of total income for most quintiles, it is a large proportion of the total for the poorest groups.

These trends are reflected in voting patterns. Anti-Fujimori sentiment was very strong in the poorest rural areas in 1993, due to the public perception that rural areas were not benefiting from reform. With the notable post-1993 improvement in the economic position of the rural poor, as well as the rhetorical attention they received from the government, political support for Fujimori in subsequent elections increased markedly in poor rural areas and declined in Lima. While the pro-reform vote was driven by macroeconomic trends and the defeat of the Shining Path, it also reflects the increased mobility of the poorest groups (many moving out of extreme poverty).

The implications of this for future political behavior, however, are far from clear. Evaluations of past progress and future expectations are influenced by a complex array of variables. Indeed, many upwardly mobile respondents in Peru do not evaluate their progress positively, as the following results demonstrate.

Perceptions

The pilot survey was of 152 households, rural (40) and urban (112). The answers were affected by respondents’ location, socioeconomic level, and by expectations themselves. The perceptions questionnaire addressed the following topics: perceptions of and satisfaction with changes in the household’s economic welfare over the last 10-15 years; perceptions and changes in the availability and quality of public services used by the household (health, schools, security, water, sanitation, municipal government); perception of future economic prospects; presence and participation in community organizations; and family health history—especially occurrence of and effects of major problems. One reason for including a separate section on health is the role of health or other stochastic shocks in determining mobility patterns over the life cycle. ₆₅ Another reasons is that health questions were of particular importance given that a number of studies have isolated poor health, along with unemployment, as having a negative effect on “happiness”. ₆₆

The Index of Perceived Mobility (IPM) for Peru was constructed by Cuanto using a slightly different methodology than the one we used for our regionwide data. ₆₇ Five questions about economic trends in the past five years were used to construct the index. These were: compared to 10-15 years ago, is the economic situation of your household...much worse, worse, same, better, much better; compared with 10-15 years ago, is your family’s job situation...much worse, etc; compared with yourself, did your parents live...much worse, etc; compared with 10-15 years ago is the purchasing power of your household...less, same, better; with respect to your current standard of living, is your degree of satisfaction...very poor, poor, acceptable, good, very good? All questions were given equal weight in the index except for the first two, which were assigned double weight, as they most directly express economic mobility.

The majority of households in the pilot panel—61%—had income increases of 30% or more from 1985-1990. Twenty-five percent were relatively unchanged, and 14% had income drops of 30% or more. Using the larger panel as a benchmark, households in the pilot fared slightly better than the average Peruvian household during this period. In order to better capture these mobility patterns and their effects on different income groups, we use a Markov transition matrix. In this matrix, the population in the panel is divided into income quintiles, with the rows being the quintile of origin in 1991 and the columns being the quintile of destination in 1997. The figures are in percentages; thus 100% in a same row and column would imply complete immobility and 20% would be complete mobility.

Table 3. Markov Matrices

a. No Income Mobility
	Quintile in T₁
Quintile in T₀	1	2	3	4	5	Total
1	100	0	0	0	0	100
2	0	100	0	0	0	100
3	0	0	100	0	0	100
4	0	0	0	100	0	100
5	0	0	0	0	100	100
Total	100	100	100	100	100	100

b. Perfect Income Mobility
	Quintile in T₁
Quintile in T₀	1	2	3	4	5	Total
1	20	20	20	20	20	100
2	20	20	20	20	20	100
3	20	20	20	20	20	100
4	20	20	20	20	20	100
5	20	20	20	20	20	100
Total	100	100	100	100	100	100

Income mobility in Peru, 1991-97
	Quintile 1997
Quintile 1991	1	2	3	4	5	Total
1	41	30	19	11	0	100
2	26	33	15	19	7	100
3	22	15	30	22	11	100
4	11	19	22	26	22	100
5	0	4	15	22	59	100
Total	100	100	100	100	100	100

As one can see from the matrix in Table 3c, there was a fair amount of mobility — both upward and downward. Those in the fourth quintile clearly experienced the most downward mobility, with 52% moving to lower quintiles between 1991 and 1997. Those that experienced the most and most intense upward mobility were in quintiles 1 and 2 (the poorest), with 60 and 41% respectively moving up, and a significant percent of these moving up two and even three quintiles.

In contrast to these positive objective results, however, there was a negative skew on self assessments. Fifty-eight percent of households had very negative or negative views of their own economic experiences, while 28% were indifferent and 12% were positive. In contrast, the majority of households (65%) were confident that their children would do better than they; only 13% thought their children would do worse.

The negative skew on economic assessments contrasts with a fairly positive one on self-assessments of housing improvements: 47% said that housing quality was “better” while only 5% said “worse”. This may reflect the contrast between the ease in identifying housing improvements and the difficulty in making accurate economic assessments over time, particularly for the self-employed who do not earn regular wages. Earnings are also affected by external circumstances such as luck, overall economic conditions, and sectoral shifts. Housing changes are more clearly determined by individual effort and savings. The generally optimistic assessments for respondent’s children, meanwhile, reflects non-economic variables such as hope and determination, which are not necessarily determined by socioeconomic levels or education.

Table 4. Long term Perceived Mobility* vs. 1985-97 Income mobility

Source: Webb in Birdsall and Graham (1999)

The correlates for the IPM were gender, education, area of residence, and income status. There was a striking absence of correlation between IPM values (perceived mobility) and actual mobility. Of the highest performers in the sample (those with per capita improvements of 100% or more from 1985-97), 64% said they were worse off and only 11% said better. Of the worst performers (with declines of 30% or more), 65% stated, accurately, that they were worse off, yet 29% said that their situation had not changed and 7% saw themselves as better off (see Table 4).

Women were more negative than men (63% of female headed households had negative IPM’s versus 57% of male ones). Rural respondents were slightly less negative (53%) than urban ones (60%), and were much less likely to use stronger “much worse” statements: 28% of urban households responded “much worse” while only 3% of rural ones did. Superior, post-secondary education seemed to produce a similar, disinhibiting effect: 35% of respondents with higher education made the strong negative statement, as did 36% and 33% of the top two income quintiles. By contrast, none of those in the bottom quintile said “much worse”, although 47% said that they were “worse” off.

Relative income difference effects are no doubt influencing these assessments of well being, which supports Easterlin’s conclusions. The differences in responses are also the result of cultural and class differences, as well as higher expectations and more experience answering surveys among urban respondents. This introduces a methodological issue which applies when interpreting the more general and regionwide Latinobarómetro study: location specific differences in answers to the same survey questions, which could result in inaccurate or incomparable survey results.

Gaps between actual and perceived mobility were larger when answers were subjective and entailed a long recall period than when they were easily verifiable, as in the case of housing. In addition, as the most significant positive gains in mobility were made in 1991-94 rather than in 1994-96, and the first pilot survey was conducted in May 1998, the recall period for positive progress was long. Not surprisingly, responses about expectations for children were the most subjective. Finally, the survey was conducted during a period of substantial economic instability and change, changes which were felt more strongly by the self-employed than by wage laborers, as well as shifts in public expenditures and transfers and several rounds of elections (discussed above). These, no doubt, had some influence on national mood at the time the assessments were made.

National mood changes had effects on the public’s evaluations of the overall national economic situation during this period. From May 1995 to May 1998, the percent of households reporting an improvement in their economic situation dropped two-thirds, from 31 to 10%, while the proportion reporting a deterioration doubled, from 22 to 47%. ₆₈ This shift in perceptions far exceeded changes in real incomes over those years, and likely reflects general changes in optimism about the national state of affairs, including declining support for President Fujimori.

An important result was the effect of participation in community organizations. There was a negative relationship with income levels: 60% of the households in the poorest quintile were involved in five or more community organizations, while only 10% of the richest quintile were. The most upwardly mobile households (100% income increase or more) were less likely to belong to organizations than were the lowest achievers. Perceived mobility becomes more positive as organizational density increases, but the relationship is not strong, and may reflect the same differences in expectations and culture that explain differences among income and education cohorts.

We separated organizations into two groups, voluntary and survival, to reflect the motivation for their participation. Those respondents with positive IPM’s were more likely to belong to voluntary organizations than those with negative or indifferent IPM’s, who were more likely to belong to survival organizations (e.g. soup kitchens). These results support our emphasis on the need to distinguish between different kinds of social capital. Community organizations in Peru are important survival and safety net strategies for the poorest groups, yet can also become poverty traps which discourage them from accepting or seeking better opportunities. ₆₉ They do not provide the “weak ties” that Granovetter identifies as key to upward mobility. Their raison d’etre is to strengthen strong ties—kinship and other close relations—as a coping mechanism which compensates for the absence of ties beyond the neighborhood that can result in new jobs or public goods. ₇₀

The upwardly mobile in our sample were more likely to belong to the kinds of groups that provide “weak ties” or else not to associate at all.

Results for the repeat survey in 1999 largely confirm those from the 1998 survey, displaying a very strong negative skew among the most upwardly mobile groups. Only two households were dropped from the 1998 survey sample. As there was no living standards survey in 1999 to assess objective mobility trends, additional questions were added to the pilot survey to assess the impact of any major economic changes with potential impact on respondents’ mobility patterns (See Appendix C-3).

To a large extent, there was economic stability, with some downward movement. Fifty nine percent of urban respondents and 44% of rural ones reported no change in their economic situation, while 31% of the sample reported a deterioration in conditions. This group was 22.5% of the urban sample and 53.8% of the rural one (Table 5).

Table 5. Economic Situation in 1999 compared to 1997
	Urban	Rural	Total
Better	18.0	2.6	14.0
Same	59.5	43.6	55.3
Worse	22.5	53.8	30.7
Total	100.0	100.0	100.0
Source: Cuanto (1999)

Education levels among respondents in the sample increased slightly. While 20.4% had completed higher education in late 1997 (when the LSMS survey was conducted), 21.1 had it in 1999. The mean years of education for respondents over age 3 was 7.99 in 1997 and 8.07 in 1999. For respondents over age 15, the mean increased from 9.43 to 9.48. The means for urban areas remained higher than in rural areas, with means of 10.25 and 10.27 for those over 15 versus means of 6.04 and 6.49 in rural areas.

Organizational activity also increased slightly by the time of the repeat survey. While in the first survey 19% of respondents did not belong to any organization, in the second only 12% fell into that category. The percent of respondents that belonged to one or two organizations increased from 21 to 37%. Of those with negative IPM’s, the percent belonging to an organization increased from 9 to 24%, and by 1999 all those with a positive IPM belonged to at least one organization.

The negative skew in the sample remained strong, particularly for upwardly mobile respondents. While 65% of the upwardly mobile reported a negative IPM in 1998, 58.3% did so in 1999 (58% of the total sample reported a negative IPM in 1998, while 57.3% did so in 1999). In both cases, this percentage was much stronger than those who actually experienced negative movements. Urban respondents were still much more negative than rural ones and became more so in 1999: 60% of urban answers were negative in 1998 and 65% in 1999. Rural respondents, meanwhile, became more positive: 53.8% were negative in 1998 and only 35.9% were negative in 1999.

The negative skew by education group also increased, with more educated respondents becoming more negative in 1999. Among those with primary education, negative IPM’s declined from 63% to 54%. Among those with secondary and higher education, however, negative answers increased, from 52 to 59% among the secondary category and from 50 to 60% in the higher category.

The education results may be linked to the most notable change between the two years, which is a marked increase in optimism among the poorest and the most wealthy groups, and a marked increase in pessimism among those that constitute the “middle class”. Of the respondents in the poorest two quintiles in the sample, 54.1% had a negative IPM in 1998, while 47.5% did in 1999. Of those in quintiles 3 and 4 (roughly the middle class), 58.3% had a negative IPM in 1998 and 65% did in 1999. These results are even more marked if one looks at upwardly mobile groups. Of the upwardly mobile in quintiles 1 and 2, 71.4% had negative IPMs in 1998 and only 42.9% had negative IPM’s in 1999. Of those in the middle (quintiles 3 and 4) with upward mobility in the 1991-97 period, negative IPM’s increased from 58.8% in 1998 to 70.6% in 1999. Negative IPM’s among the upwardly mobile in the wealthiest quintile decreased substantially, meanwhile, with only 47.4% reporting negative IPM’s in 1999 versus 57.9% in 1998 (see Figure 3).

Figure 3. IPM among the upwardly mobile.
Percentage response by income groups (1998 and 1999)
Source: Table in Appendix C-2

An important caveat in interpreting these results is that objective mobility trends and the definition of quintiles are based on information for 1991-1997, while respondents were asked about their perceptions in 1998 and 1999. There were national mood changes during this period, some related to the anticipated and real spill-over affects of the Brazil crisis. Those who were more optimistic (higher IPM) in 99 vis-a-vis 98 are those who in 1997 were either in the poorest two quintiles or in the wealthiest one and had experienced upward mobility. Those who are the most negative are the ones that in 1997 belonged to the middle quintiles (three and four) and had experienced upward mobility. As is noted above, most of the upward mobility that occurred was in the 1991-94 period, with a slowing down in 1994-96, excepting poor rural groups, making the recall period for positive mobility even longer for middle income groups.

These trends played out in different ways among various income groups. The wealthy, for example, have and will continue to benefit from the rewards that the market is yielding for skills and education, while the poor have seen a significant expansion of transfers and public expenditures, an improvement in public services, and a new focus of government. ₇₁ Those in the middle probably had more differential rewards, depending on their skill and education levels. They are also more likely to rely on the wealthy as a reference group than are the very poor, and thus even some upward mobility could still result in frustration or “unhappiness,” an unhappiness which was more evident in 1999 than in 1998. ₇₂ In contrast, absolute income increases among the poorest sectors increased “happiness”.

Looking more closely at the traits of what we term the “frustrated middle class”, defined as those in quintiles 3 and 4 that were upwardly mobile from 1991-97, yet reported negative IPM’s, we found that they were, on average, less educated than the “non-frustrated” members of the middle class, and also slightly more rural in composition (only 88% were urban, versus 97% of the non-frustrated group). ₇₃ The “frustrated” group actually reported less objective deterioration or changes in income between 1997 and 1999 than did the non-frustrated group, confirming the extent to which their negative IPM’s are perception rather than reality based. (Appendix C-2). The distinct education gap between the frustrated and non-frustrated groups, meanwhile, suggests that even the less educated public may perceive that best opportunities are increasingly going to those with more education and skills.

In the 1999 survey, an additional set of questions was asked about respondents’ expectations for their children and grandchildren. The optimistic tilt of these responses stands in contrast to the negative skew of the IPM’s. Sixty-nine percent of respondents believed that their children would attain a higher standard of living than they had; only 13% answered negatively. Expectations were even higher for grandchildren, with 74% expecting their grandchildren to live better than they. Sixty-one percent felt that their children would attain higher education, and 67% believed that their grandchildren would. These results highlight the extent to which expectations, even more than subjective assessments, are affected by non-economic factors such as hope and determination.

Implications

The Peru results support Easterlin’s hypothesis that relative income levels matter as much as absolute ones. They also suggest that macroeconomic volatility and the insecurity it generates plays a role in public perceptions. Indeed, it may be one of the primary variables influencing the overall negative skew of the self assessments of the upwardly mobile. The results also show that individual assessments are affected by national mood swings and changes. They suggest that the relationship between social capital and growth or mobility is far from straightforward, and that such social interactions are not always a positive force for upward mobility. Finally, the most clear result is how far public perceptions about who is getting ahead can be from actual trends. Yet people vote according to perceptions, and perceptions can also have significant and lasting effects on individual effort and economic behavior. ₇₄

Conclusion

This paper has been as much an attempt to establish a new research agenda as it has to find definitive answers. Yet we were able to draw some conclusions about objective and subjective mobility trends, some of which challenge established assumptions. These relate to the three issues that we raised at the beginning: relative versus absolute income differences; the extent of macroeconomic volatility and the nature of social protection; and the role of social capital.

We found that relative incomes matter as much if not more than absolute ones in developing economies. However, it is not clear that they always confirm the standard view of relative deprivation: a surprising finding was that expectations for future upward mobility were higher in countries with more inequality. It may be that people assess their prospects for upward mobility more positively when the margin for absolute advancement is greater. ₇₅ A related finding was more in keeping with standard assumptions about relative deprivation: more upwardly mobile people were more critical in their self assessments than were less mobile people, no doubt because the former compare themselves to wealthier people rather than to their original cohorts.

The extent of macroeconomic volatility—and how recent the collective memory of it was—played a definitive role in public assessments of the market process and their prospects under it. Those countries with the most recent (and often the fastest paced) reforms scored highest on the pro-market index and had the highest percent of respondents that favored productivity over redistribution.

Individuals in countries with the most effective social welfare systems, meanwhile, tended to be less pro-market and to favor redistribution as a means of advancement for the country. In part these results reflect the maturity of the reform process: citizens in countries where that process was more mature and where macroeconomic stability was consolidated were probably more comfortable turning attention to distribution issues. Yet they also reflect cultural differences and social norms: welfare institutions in particular countries reflect public consensus about levels of redistribution that has developed over time. These attitudes both reflect and in turn have effects on citizens evaluations of past mobility trends and future expectations.

A third finding was that involvement in civic associations—or social capital—was negatively correlated to upward mobility. We stress the need to disaggregate the different kinds of civic associations to explain their diverse effects on growth.

A fourth issue, which we have not yet explored, is how or if the patterns we see affect voting behavior. The results suggest that support for market reforms may be high in some of the most unexpected circumstances, i.e. very unequal countries in the most difficult parts of their reform programs, and weaker or at least more mixed in countries that have consolidated reforms and attained sustained patterns of growth. At the least, distribution issues are likely to be more front and center in the latter set of countries, with publics expecting a role for the state as well as for markets, opportunities, and individual initiatives in determining their future prospects and those of their children. At the same time, it seems that some degree of optimism for individual future advancement or mobility remains key to sustainable market growth.

Recent research in post-communist economies suggests that perceptions of mobility and opportunity are far more important in influencing voter behavior than are actual trends. Key to developing a political economy framework that can account for the dynamics of mobility and opportunity will be better understanding the relationship between subjective assessments, actual trends, and prospects for upward mobility.

A very different but also important concluding issue pertains to method. The pilot study of perceptions in Peru suggests that rural people and the very poor may answer questions about their well being and expectations differently from other income groups, precisely because their expectations are so much lower. How the different expectations of various socioeconomic groups affect their evaluations of their past and potential progress has implications for the design of living standards measurement surveys, for evaluations of subjective well being, and for analysis of political as well as economic behavior.

Bibliography

Akerlof, George, “Social Distance and Social Decisions”, Econometrica 65 (September 1997), pp.1005-1027.

Akerlof, George and Rachel Kranton, “Economics and Identity”, Paper presented to Brookings Center on Social and Economic Dynamics/MacArthur Foundation Social Interactions Networks Meeting, Washington,D.C., 9-10 April 1999.

Alesina, Alberto and Roberto Perotti, “The Political Economy of Growth: A Critical Review of the Recent Literature”, World Bank Economic Review, Vol.8, No.3, 1994.

Behrman, Jere, “Social Mobility: Concepts and Measurement Issues in Latin America and the Caribbean,” in Birdsall and Graham (1999)

Behrman, Jere, Nancy Birdsall, and Miguel Székely, “Intergenerational Mobility in Latin America: Deeper Markets and Better Schools Make a Difference” in Birdsall and Graham (1999).

Benabou, Roland and Efe Ok, “Social Mobility and the Demand for Redistribution: The POUM Hypothesis”, NBER Working Paper #6795, 1998.

Birdsall, Nancy and Carol Graham, New Markets, New Opportunities? Economic and Social Mobility in a Changing World (Washington, D.C.: Brookings and Carnegie, 1999)

Birdsall, Nancy, Carol Graham, and Richard Sabot, eds. Beyond Tradeoffs: Market Reforms and Equitable Growth in Latin America. (Washington, D.C.: The Brookings Institution and the InterAmerican Development Bank, 1998).

Burtless, Gary, “Effects of Growing Wage Disparities and Changing Family Composition on the U.S. Income Distribution”, Center on Social and Economic Dynamics Working Paper No.4, The Brookings Institution, Washington, D.C., July 1999

Clifford, P. and A.F. Heath, “The Political Consequences of Social Mobility”, Journal of the Royal Statistical Society, Vol.156, Part 1, 1993, pp.51-61.

Collier, Paul, “Social Capital and Poverty”, Paper presented to World Bank workshop on Social Capital, Washington, D.C., June 22-24, 1999.

Cuanto, S.A., “Percepción de la Movilidad Social en el Perú, 1999,” Mimeo, Lima, Peru.

Dahan, Momi and Alejandro Gaviria, “Sibling Correlations and Social Mobility in Latin America”, Mimeo, Office of the Chief Economist, InterAmerican Development Bank, Washington, D.C., February 1999

Deaton, Angus and Christina Paxson, “Intertemporal Choice and Inequality”, Journal of Political Economy, Vol.102, no.3, 1994.

Diener, Ed, “Subjective Well Being”, Psychological Bulletin, Vol.95, No.3, 1984, pp.542-575.

Dietz, Henry, Urban Poverty, Political Participation, and the State: Lima 1970-1990 (Pittsburgh: University of Pittsburgh Press, 1998).

Durlauf, Steven, “Neighborhood Feedback Effects, Endogenous Stratification, and Income Inequality”, forthcoming in Barnett, Gandolfo, and Hillinger, eds., Disequilibrium Dynamics: Theory and Applications (Cambridge: Cambridge University Press, forthcoming).

Duryea, Suzanne and Miguel Székely, “Labor Markets in Latin America: A Supply Side Story”, InterAmerican Development Bank, OCE Working Paper, No. 374, September 1998.

Easterlin, Richard A., “Does Economic Growth Improve the Human Lot?”, in Paul A. David and Melvin W. Reder, Nations and Households in Economic Growth (New York: Academic Press, 1974

________. “Will Raising the Incomes of All Increase the Happiness of All?”, Journal of Economic Behavior and Organization, Vol.27, No.1, June 1995.

ECLAC, Anuario Estadístico de America Latina y El Caribe, Santiago, Chile, 1999a.

_______. Panorama Social de América Latina, 1998, Santiago, Chile, 1999b.

Esping-Andersen,Gosta, Three Worlds of Welfare Capitalism (Princeton: Princeton University Press, 1990

Fields, Gary, “Income Mobility: Meaning, Measurement, and Some Evidence from the Developing World,” in Birdsall and Graham (1999)

Foote Whyte, William, Street Corner Society (Chicago and London: University of Chicago Press, Fourth Edition, 1993)

Frey, Bruno and Alois Stutzer, “Happiness, Economics, and Institutions”, Mimeo, Univeristy of Zurich, 1999

Geddes, Barbara, “The Politics of Economic Liberalization, Latin American Research Review, Vol. 30, No.2, 1995

Glewwe, Paul and Gillette Hall, “Are Some Groups More Vulnerable to Macroeconomic Shocks than Others? Hypothesis Tests Based On Panel Data from Peru”, Journal of Development Economics, Vol.56, 1998, pp.181-206.

Graham, Carol, Private Markets for Public Goods: Raising the Stakes in Economic Reform (Washington, D.C.: The Brookings Institution, 1998).

_______. Safety Nets, Politics, and the Poor: Transitions to Market Economies (Washington, D.C.: The Brookings Institution, 1994).

Graham, Carol and Moises Naim, “The Political Economy of Institutional Reform in Latin America” in Birdsall, Graham, and Sabot (1998).

Graham, Carol and Cheikh Kane, “Opportunistic Government or Sustaining Reform: Electoral Trends and Public Expenditure Patterns in Peru, 1990-95,” Latin American Research Review, vol. 33, no.1 (1998), pp.71-111.

Granovetter, Mark, “The Strength of Weak Ties”, American Journal of Sociology, Vol.78, May 1973, pp.1360-1379.

Gurr, Ted Robert, Why Men Rebel (Princeton: Princeton University Press, 1970) p.13.

Haggard, Stephan and Steven B.Webb, Voting for Reform: Democracy, Political Liberalization, and Economic Adjustment (New York: Oxford University Press and the World Bank, 1994)

Hoff, Karla, “Market Failures and the Distribution of Wealth: A Perspective from the Economics of Information”, Politics and Society, Vol.24, No.4, Dec. 1996, pp.411-432.

Hojman, David, “Inequality, Growth, and Political Stability: Can Income Mobility Really Provide the Answers?” in Birdsall and Graham (1999).

InterAmerican Development Bank. “Facing Up to Inequality in Latin America.” Economic and Social Progress in Latin America, 1998-99 Report. Washington, DC, 1999.

International Monetary Fund. International Financial Statistics, 1999.

JPMorgan, Latin American Economic Outlook, May 1999

Kenny, Charles, “Does Growth Cause Happiness, or Does Happiness Cause Growth?”, Kyklos, Vol.52, No.1, 1999, pp.3-26.

Knack, Stephen and Philip Keefer, “Does Social Capital Have an Economic Payoff? A Cross-Country Investigation”, The Quarterly Journal of Economics, November 1997, pp.1251-1288.

Krackhardt, David, “The Strength of Strong Ties: The Importance of Phylos” in Nohria, Nitin and Robert G. Eccles, eds., Networks and Organizations: Structure, Form, and Action (Cambridge: Harvard Business School Press, 1992).

Lam, David and Robert F. Schoeni, “Effects of Family Background on Earnings and Returns to Education in Brazil”, Journal of Political Economy, Vol.101, No.4, 1993.

Londoño, Juan Luis, Antonio Spilimbergo, and Migeul Székely, “Income Distribution, Factor Endowments, and Trade Openness,” InterAmerican Development Bank, Office of the Chief Economist Working Paper Series, No.356, October 1997.

Lora, Eduardo and Juan Luis Londoño, “Structural Reforms and Equity in Latin America”, in Birdsall, Graham, and Sabot (1998).

Mateju, Petr, “Mobility and Perceived Change in Life Chances in Post-Communist Countries” in Birdsall and Graham (1999).

McMurrer, Daniel and Isabel Sawhill, Getting Ahead: Economic and Social Mobility in America (Washington, D.C.: Urban Institute Press, 1988)

Merton, Robert K., Social Theory and Social Structure (Glencoe: The Free Press of Glencoe, 1957).

Mesa Lago, Carmelo, “Old-Age Security and Health Care for the Poor in Latin America and the Caribbean,” Paper persented to InterAmerican Development Bank Conference on Social Protection and Poverty, Washington, D.C., February 5, 1999.

Morley, Samuel, Roberto Machado, and Stefano Pettinato, “Indexes of Structural Reform in Latin America,” Serie Reformas Económicas de la CEPAL, LC/L.1166, January 1999.

Namazie, Ceema and Peter Sanfey, “Happiness in Transition: The Case of Kyrgyzstan”, Discussion Paper No.DARP 40, Distributional Analysis Research Programme, London School of Economics, July 1998.

Okun, Arthur, Equality and Efficiency, The Big Tradeoff (Washington, DC: The Brookings Institution, 1975).

Oswald, Andrew, “Happiness and Economic Performance”, Economic Journal, November 1997.

Pastor, Manuel Jr., and Carol Wise, “The Politics of Second Generation Reform,” Journal of Democracy, Vol.10, No.3, 1999.

Piketty, Thomas, “Social Mobility and Redistributive Politics”, The Quarterly Journal of Economics, Vol.CX, No.3, August 1995.

Putnam, Robert, Making Democracy Work: Civic Traditions in Modern Italy (Princeton: Princeton University Press, 1993).

Ravallion, Martin and Michael Lokshin, “Who Wants to Redistribute? Russia’s Tunnel Effect in the 1990’s”, Policy Research Working Paper #2150, The World Bank, Washington, D.C., July 1999.

Robbins, Donald, “HOS Hits Facts: Facts Win. Evidence on Trade and Wage Inequality in the Developing World”, Harvard Institute for International Development, Development Discussion Paper No.557, October 1996.

Roberts, Kenneth, and Moises Arce, “Neoliberalism and Lower-Class Voting Behavior in Peru” Comparative Political Studies 31(2), 1998, pp.393-407.

Rodrik, Dani, “Understanding Economic Policy Reform”, Journal of Economic Literature, Vol. XXXIV, March 1996, pp.9-41

Seligson, Amber, “Civic Association and Democratic Participation in Central America: A Test of the Putnam Thesis”, Comparative Political Studies, Vol.32, No.3, May 1999, pp.342-362.

Sen, Amartya, “Poor, Relatively Speaking”, Oxford Economic Papers, Vol.35, 1983, pp.153-169.

___________, “The Political Economy of Targeting”, in Dominique Van de Walle and Kimberly Nead, eds., Public Spending and the Poor: Theory and Evidence (Baltimore: The Johns Hopkins University Press and the World Bank, 1995), pp.11-24.

Solon,Gary, “Intergenerational Mobility in the United States”, American Economic Review, Vol.82, 1992, pp. 393-408.

Stokes, Susan, “Public Opinion and Market Reforms: The Limits of Economic Voting”, Comparative Political Studies, Vol.29, No.5, October 1996, pp.499-519

Terrell, Katherine, “Mobility in Eastern Europe: Empirical Evidence” in Birdsall and Graham (1999).

United Nations Development Programme. Human Development Report 1999. New York, NY.

Veenhoven, R., “Is Happiness Relative?”, Social Indicators Research, Vol.24, 1991, pp.1-24

Webb, Richard, “Household Perceptions of Mobility in a New Market Economy: Peru 1998,” in Birdsall and Graham (1999)

Weyland, Kurt, “The Political Fate of Market Reforms in Latin America, Africa, and Eastern Europe”, International Studies Quarterly, Vol.42, 1998, pp.645-674.

World Bank, Poverty and Social Developments in Peru, 1994-1997 (Washington, D.C.: The World Bank, 1999).

World Bank, World Development Indicators, 1999.

Young, Peyton, “Diffusion in Social Networks”, Center on Social and Economic Dynamics Working Paper Series, No.2, Washington, D.C., The Brookings Institution, June 1999.

Surveys

MORI, S.A., Latinobarómetro, Santiago, Chile,1997 and 1998.

Cuanto, S.A., Peru Pilot Survey, Lima, Peru, 1998 and 1999.

_______. ENNIV Survey, (Lima, Peru, 1997).

Appendix A. Indices: Methodology and Rationale

Individual Indices ¹

Prospect of Upward Mobility (POUM) Index . This index reflects the degree of mobility that the respondent expects in the near future, as well as that in the long-term for his or her own children. We used the data from the Latinobarómetro surveys to construct this index. The index is based on two questions. The first gauged the respondents’ expectations for their personal and family economic situation in one year compared to present (short-term POUM). The second gauged the respondents’ perception of their children’s future standard of living compared to their own situation at present (long-term POUM). Two sub-indexes were constructed from the two questions, assigning the values indicated in the tables. The total POUM index was built as a weighted-average of the two, with a weight distribution of 75% (for the short-term POUM) and 25% (for the longer-term POUM, which is more speculative in nature).

Economic situation in one year	value
Better	1
Same	0.6
Worse	0
Don’t know/don’t answer	0.3

Economic prospects for respondents’ children in the future	value
Better	1
Same	0.6
Worse	0
Don’t know/don’t answer	0.3

Index of Perceived Mobility (IPM). This indicator attempts to measure respondents’ evaluations of their past mobility. It is based on an evaluation of individual mobility trends in the previous year, and those relative to the respondent’s parents. The first question asks the respondent to assess the present economic situation with respect to the previous year. The latter asks the respondent to compare their current standard of living relative to that of his/her parents. We constructed two sub-indexes using the values listed in the tables below. The two sub-indexes were averaged and weighted: 75% for short-term mobility, and 25% for parents’ relative standard of living.

Present personal economic situation relative to one year ago	value
Better	1
Same	0.6
Worse	0
Don’t know/don’t answer	0.3

Parents’ standard of living relative to present own	value
Better	0
Same	0.6
Worse	1
Don’t know/don’t answer	0.3

Pro Market Index (Pro_mkt) . The pro-market index was based on respondents’ answers to seven questions pertaining to attitudes about government control of the economy, private enterprise, and foreign trade and investment. This indicator is based on seven questions in the Latinobarómetro. The respondent was asked to state his/her agreement with the following:

the government should leave productive activity to the private sector
privatizations were beneficial for the country
prices should be freely determined by competition
market economy is the most convenient for the country
foreign investment should be fostered
private enterprise is beneficial for the country
the government has to cooperate with international financial organizations

Each question allowed five default answers, with the corresponding values listed in the table below. The index was calculated as a simple average of the responses to the seven questions.

Answers	value
Fully Agree	1
Agree	0.7
Disagree	0.2
Strongly Disagree	0
Don’t know/don’t answer	0.3

Participation Index (Particip). This index attempts to gauge the degree of respondents’ civic participation. The types of organizations considered in the Latinobarometro survey were: religious; community-based; artistic; labor; ecological; professional; youth; sports; and political. Values assigned to answers were 1 for active participation, and 0 for no participation, and then averaged. This index is particularly limited as the questionnaire’s coverage of the neighborhood organizations typically frequented by the poor was limited.

Political Happiness Index (Polhap). This indicator is on one question in the Latinobarómetro , which gauges the degree of satisfaction with democracy. There were five possible answers, with the corresponding values listed in the table below.

Satisfaction with democracy	Value
Very satisfied	1
Moderately satisfied	0.8
Not very satisfied	0.2
Not satisfied at all	0
Don’t know/don’t answer	0.3

Economic Happiness Index (Ecohap). This index is based on two questions. The first attempts to assess the respondent’s present economic situation (individual and household). The second asks the respondent to assess the present economic situation for the country as a whole. The values for each response are listed in the table below. The overall Economic Happiness index is calculated as an average of the values from the two questions, assigning a weight of 75% to the individual asessment, and 25% to the country assessment. The rationale behind these weights is that individuals are both more concerned with and more accurate in assessing their own well being than they are with the more general economic conditions of their country.

Economic situation	Value
Very good	1
Good	0.8
Regular	0.5
Bad	0.2
Very bad	0
Don’t know/don’t answer	0.3

Confidence Index (Confid). This index aims to quantify the degree of confidence in political institutions in the region. We based it on five questions in which the respondent was asked to assess his/her degree of confidence in the following institutions: political parties; the judiciary, the police; the Congress; the Presidency. The table below shows the possible answers and the values assigned to each response.

How much confidence...	Value
A lot	1
Some	0.6
Little	0.3
None	0
Don’t know/don’t answer	0.2

Trust Index . This index tries to measure for the degree of reliance on other individuals, social cohesion, and, implicitly, views about society in general. This index was constructed based on a question in the Latinobarómetro survey regarding respondents’ level of trust in others, with the values listed below assigned to responses. The respondent was asked to choose one out of two statements:

Generally speaking, would you say that one can...	value
trust the majority of the people	1
never be cautious enough in dealing with others	0
Don’t know/don’t answer	0.3

Tolerance Index . This index gauges the degree of tolerance that respondents had for individuals that are different from him/her. Respondents were asked to highlight those groups of people that they would dislike to have as neighbors. The groups to choose from were: homosexuals, drug-addicts, Asians, Africans, poor, Muslims, Jewish, political extremists. Each check mark was subtracted from 1 by an amount of 1/10. The index gave a final value ranging from 0 to 1, with the latter reflecting maximum tolerance.

Violence Index. This index captures the degree to which the respondent is willing to act against the law. Three questions were utilized: (a) participation by the individual in protests; (b) blocking of traffic; (c) occupation of property. Possible answers to these questions and corresponding values assigned are listed in the table below. The index was built as a weighted-average, assigning 20%, 30%, and 50% to answers a, b, and c respectively, assuming that willingness to occupy property reflected more of a transgression than did participation in protests.

Answers	Value
Have done it	1
Would do it	0.7
Would never do it	0
Don’t know/don’t answer	0.3

Macro Indices ²

Productivity Supporters (ProdSup). This is based on a question in the survey pertaining to support for productivity-based reforms versus redistributive measures. The respondent was asked to declare which statement he or she agreed the most with: “the government should enhance its output and productivity” or “the government should improve the distribution of wealth.” For each country we calculated the percentage of the sample in favor of the former statement.

Social Welfare Effectiveness Index (SWEI) . Five components were used to construct this index: (a) illiteracy rate; (b) life expectancy at birth; (c) infant mortality rate; (d) social protection (health coverage); (e) education efficiency and equity; (f) general system reputation. Raw data for components (a), through (c) was obtained from the most recent World Bank data (World Development Indicators, 1999). Component (d) is in Mesa Lago (1998). The educational component (e) was calculated using data from Duryea and Székely (1998) for mean and variance of schooling. The final component (f) was derived from the historical track record, the literature, and the degree of institutionalization and consolidation of the social welfare system in the country. Each array of data—from (a) to (f)—was used to construct a cross-country index that assigned value 0 to the worst performer in terms of social welfare effectiveness and 100 to the best. The resulting series were averaged to obtain the SWEI. Our SWEI is methodologically and conceptually related to the Human Development Indicator, to which it is correlated by a coefficient of 0.95.³

GDP per capita. The data for this variable is obtained from the World Bank (1999), and refers to the latest observation available—1997 for the entire region, with the exception of Nicaragua and Panama, for which we had to use 1996 figures. Note that the values used are calculated with purchasing power parity (ppp) conversion rates. This reduces measurement distortions since some Latin American countries still have double digit inflation rates and/or overvalued exchange-rates.

Growth Trajectory Index (Growth). Three factors were considered for the construction of this index: (a) long-term growth trajectory (1970-98); (b) medium-term growth trajectory (1988-98); (c) projected growth (1999, 2000). Raw data for components a, was obtained from the most recent World Bank data (World Bank, 1999). Component b was obtained from ECLAC (1999a). Data for component c comes from JPMorgan (1999). The results were indexed across countries assigning value 0 to the worst performer and 100 to the best. The resulting three sub-indices were averaged, assigning a weight of 30% to the long-term trajectory, 60% to the medium-term, and 10% to the projected.

Change in Reforms (ChRef). This variable measures the degree of progress in the implementation of structural reforms for 1985-1995. It is based on a structural reform index that summarizes progress in privatization, trade reform, capital account liberalization, financial reform, and tax reform for 17 Latin American countries from 1970 to 95.⁴ This index (roughly) gauges which countries in the region have progressed the most with market reforms. A caveat, however, is that it is slightly biased towards the fast, later reformers such as Peru and Bolivia, versus those which reformed more slowly over time. Chile, for example, which implemented many changes prior to 1985, does not score particularly high on this index, even though it is one of the countries that has proceeded the furthest with reforms. Thus in this paper we attempted to interpret the rate of change controlling for the initial level of reform (see Table 2 in text). We used the index to assess the impact of structural reforms on our micro level attitude indicators and on aggregated (country-level) perceptions indices.

Inequality (Ineq). The distribution of income (rather than consumption or wealth) has been estimated with national Gini coefficients taken from ECLAC (1999b), supplemented with IDB (1999) data available.

Poverty . Poverty levels were calculated based on the percentage of households living below national poverty lines, from ECLAC (1999b).

Inflation . The most recent observable inflation rate figures were obtained from International Monetary Fund (1999).

Unemployment (Unempl). The most recent unemployment rate figures were taken from the Statistical Appendix of ECLAC (1999b). Data cover national territories, with the exception of Argentina and Paraguay that provide data only for their capital cities.

1. Calculated from the data obtained from the 1997 and 1998 Latinobarometro Surveys.
2. Calculated from country aggregate data.
3. See UNDP (1999).
4. For a more detailed description of the methodology for the construction of the structural reform index see Morley, Machado, and Pettinato (1999).

Appendix B. Correlation Coefficients for Individual Indices and Correlations Across Country-average Indices and Macro Variables

The tables for Appendix B

Appendix C. Mobility in Peru

Appendix C-1. Objective Mobility in Peru

Table 1. Mobility of Households with Respect to Poverty. Peru, 1991-1996

Table 2. Mobility of Households (1991-1996)

Source: Cuanto, S.A. data, compiled by Martin Cumpa of the IDB.

Appendix C-2. Subjective Mobility in Peru

Table 1. Index of Perceived Mobility (IPM) in 1998 v. 1999, by Quintiles and by objective income mobility for the 1991-97 period
Total Sample

Poorest (1st and 2nd quintile)			Middle Income (3rd and 4th q.)			Richest (5th quintile)
	ipm98	ipm99		ipm98	ipm99		ipm98	ipm99
Negative	54.1	47.5	Negative	58.3	65.0	Negative	65.5	62.1
Indifferent	31.1	34.4	Indifferent	23.3	21.7	Indifferent	27.6	20.7
Positive	14.8	18.0	Positive	18.3	13.3	Positive	6.9	17.2
Total	100.0	100.0	Total	100.0	100.0	Total	100.0	100.0


With Upward Income Mobility

Poorest (1st and 2nd quintile)			Middle Income (3rd and 4th q.)			Richest (5th quintile)
	ipm98	ipm99		ipm98	ipm99		ipm98	ipm99
Negative	71.4	42.9	Negative	58.8	70.6	Negative	57.9	47.4
Indifferent	14.3	35.7	Indifferent	23.5	17.6	Indifferent	36.8	26.3
Positive	14.3	21.4	Positive	17.6	11.8	Positive	5.3	26.3
Total	100.0	100.0	Total	100.0	100.0	Total	100.0	100.0


With No Income Mobility

Poorest (1st and 2nd quintile)			Middle Income (3rd and 4th q.)			Richest (5th quintile)
	ipm98	ipm99		ipm98	ipm99		ipm98	ipm99
Negative	42.9	52.4	Negative	56.5	60.9	Negative	71.4	85.7
Indifferent	38.1	28.6	Indifferent	26.1	26.1	Indifferent	14.3	14.3
Positive	19.0	19.0	Positive	17.4	13.0	Positive	14.3	0.0
Total	100.0	100.0	Total	100.0	100.0	Total	100.0	100.0


With Downward Income Mobility

Poorest (1st and 2nd quintile)			Middle Income (3rd and 4th q.)			Richest (5th quintile)
	ipm98	ipm99		ipm98	ipm99		ipm98	ipm99
Negative	38.5	46.2	Negative	100.0	100.0	Negative	100.0	100.0
Indifferent	46.2	38.5	Indifferent	0.0	0.0	Indifferent	0.0	0.0
Positive	15.4	15.4	Positive	0.0	0.0	Positive	0.0	0.0
Total	100.0	100.0	Total	100.0	100.0	Total	100.0	100.0
Source: authors’ calculations from Peru Pilot Survey

Portrait of the Frustrated Middle Class, 1998 vs. 1999

We identified the “frustrated middle class” as the group of households who: a) belonged to the 3rd and 4th quintile in 1997, and experienced upward mobility between 1991 and 1997, and b) gave negative assessments of their long term mobility (negative IPM) in 1998 (1999). This table compares them with the “non frustrated middle class” for the respective years:

1998	1999
Size
20 households 13% of total sample 33% of total Middle Class (3rd and 4th quintiles)	24 households 16% of total sample 40% of total Middle Class (3rd and 4th quintiles)
Educational level
(relatively low) on a scale 0-3 = 1.35, well below non-frustrated middle class average (1.62)	(relatively low) On a scale 0-3 = 1.49, below non-frustrated middle class average (1.69)
Recent income variation 1997-99
15% of the 1998 frustrated middle class experienced an improvement, and 15% a deterioration. For the non-frustrated middle class the corresponding figures are 14% and 33%.	13% of the 1999 frustrated middle class experienced an improvement, and 25% a deterioration. For the non-frustrated middle class the corresponding figures are 14% and 32%.
Area of residence
Only 85% of the households in this group are urban (vs. 97% for the non frustrated middle class)	88% percent of the households in this group are urban (vs. 97% for the non frustrated middle class)

Appendix C-3

Cuanto, 1999

Survey of Perceived Economic Progress

Part One: Present well-being and expectations

Good morning/afternoon, my name is... I am a representative of Cuanto S.A., a company which specializes in public opinion polls and inquires, and we’re taking the opportunity to carry-out interviews to understand the public opinion about some aspects of the well-being of your family and the population in general.

I. Evolution of Present Well-Being

1.1 The economic situation of your family, in relation to 10-15 years ago is... (choose one)

much worse
worse
equal
better
much better

1.2 The work/employment situation of you and your family members, with respect to 10 - 15 years ago is...

much worse
worse
equal
better
much better

1.3 In comparison to you, your parents lived...

much worse
worse
equal
bette
much better

1.4 Would you say that the present access you and your family has to health services, in relation to 10-15 years ago is...

much worse
worse
equal
better
much better

1.5 The access to educational services for you and your family, in relation to 10-15 years ago is...

much worse
worse
equal
better
much better

1.6 Your present access to basic everyday services such as water, electricity, and sewer(plumbing), when compared to your access 10 - 15 years ago is...

poor
equal
better

1.7 The state of your life presently, with respect to your life 10 - 15 years ago is...

much worse
worse
equal
better
much better

1.8 The purchase-power of your family, in relation to 10-15 years ago is..

worse
equal
better

1.9 The level of security that currently exists in your region (violence, delinquency), in comparison to the level 10-15 years ago is..

worse
equal
better

1.10 The management of your city government, in comparison to 10-15 years ago is...

much worse
worse
equal
better
much better

1.11 I will tell you some services that are offered to your community, please tell me how these services have changed in the last 10-15 years.

	Worse	No Change	Better	N/A
Schools	1	2	3	4
Sewer	1	2	3	4
Water	1	2	3	4
Electricity	1	2	3	4
Community Police	1	2	3	4
Sanitation	1	2	3	4
Roads	1	2	3	4

II. Perspective of the Future

2.1 The economic situation of your family in the future, in comparison with the present, will be...

much worse
worse
equal
better
much better

2.2 The standard of living for your children in the future, in relation to your present level, will be...

much worse
worse
equal
better
much better

2.3 The standard of living for your grandchildren (if you have them or for those that you expect to have) in the future, in relation to your present level, will be...

much worse
worse
equal
better
much better

2.4 How would you qualify your opportunity to have a higher standard of living in the future?

very bad
bad
equal
good
very good

2.5 How long do you think it will take you to reach a satisfactory standard of living?

1 one to two years
2 three to five years
3 six to ten years
4 more than ten years
5 never

2.6 What educational level do you expect your children to reach in the future (if you have them or for those that you expect to have)?

1 No level
2 Primary incomplete
3 Primary complete
4 Secondary incomplete
5 Secondary complete
6 Higher

2.7 What educational level do you expect your grandchildren to reach in the future (if you have them or for those that you expect to have)?

No level
Primary incomplete
Primary complete
Secondary incomplete
Secondary complete
Higher

III. Degree of Present Satisfaction and Perspective of Future Progress

3.1 With regard to your present standard of living, your degree of satisfaction is

very bad
bad
the same
good
very good

3.2 Would you say that your present opportunity to improve your standard of living is

very bad
bad
the same
good
very good

3.3 Would you say that the opportunity of your parents to access a better standard of living in comparison to your own opportunity was

worse
equal
better

3.4 Your opportunity to have a better standard of living than that of your parents has been

worse
equal
better

3.5 Would you say that the opportunity of your children to have a better standard of living than you will be

1 worse
2 equal
3 better

3.6 Would you say that the opportunity for your grandchildren (if you have them or you for those that you expect to have) to reach a better standard of living than yours will be

1 lower
2 same
3 higher

IV. Organizations and Participation

4.1 Do the following organizations exist in your community?

4.2 Do you belong to, have some connection to, or have received some benefit from any of these organizations within your community?

4.3 ... and outside the community?

	4.1		4.2			4.3
	Yes	No	Yes	No		Yes		No
Parents’ Association (Asociación de padres de familia)	1	2	1	2		1		2
Christian/Parish Community (Communidad Cristiana/Parroquia)	1	2	1	2		1		2
Clubs and Assoications (Clubes y Asociaciones)	1	2	1	2		1		2
Mothers Clubs (Club de Madres)	1	2	1	2		1		2
Community Organizations (Organizaciones Comunales)	1	2	1	2	1		2
Asociaciones de Professionales (Professional Associations)	1	2	1	2	1		2
Labor Unions (Sindicatos de trabajadores)	1	2	1	2	1		2
Political Parties (Partidos politicos/Frente Civico)	1	2	1	2	1		2
Town Councils (Municipios)	1	2	1	2	1		2
Soup Kitchens (Comodores Populares)	1	2	1	2	1		2
Milk Program (Programa del Vaso de Leche)	1	2	1	2	1		2
Nucleos Ejecutores	1	2	1	2	1		2
Development Committes (Comités de Desarrollo)	1	2	1	2	1		2
Brotherhood Fraternities (Hermandades Cofradias)	1	2	1	2	1		2
Community Police Watch (Serenazgo)	1	2	1	2	1		2
Organization of Self-defense (Organización de autodefensa)	1	2	1	2	1		2
Community Enterprise (Empresas Comunales)	1	2	1	2	1		2
Farmers Cooperative (Campesina Comunidad)	1	2	1	2	1		2
Committee of Indigenous Peoples (Comunidad Indígena)	1	2	1	2	1		2
Immigrants Association (Asociaciones de Inmigrantes)	1	2	1	2	1		2

Part Two: Health Survey

Has your family experienced any health problems?
1. Yes
2. 2 No (end of survey)

How has your family been affected...

By the cost of the treatment?
1. Yes
2. No
Not being able to work?
1. Yes
2. No

(If you responded “yes” in Question 2)

How have you coped with this health problem?
1. Own savings
2. Bank loans
3. Mortgage
4. Sale of vehicles or machinery
5. Family or friends’ contributions
Does the ill member receive some type of medical or family insurance?
1. Yes, public insurance
2. Yes, private, personal or family insurance
3. Yes, a public and private mix of insurance
4. No
Did the illness and treatment cause you to forgo your plans for the following: improvements/upgrading your home, purchasing goods and appliances, education, travel, etc?
1. Yes
2. No
Are there any pending debts in relation to the funds or external aid required for the treatment of the ill member, like: repayment of debts, interest, rescheduling debt, purchasing or repairing appliances/fixtures, gifts, as well as compensating for the help of family and friends?
1. Yes
2. No (if “no” continue to question 10)
Is the above-mentioned debt affecting the regular incomes of the household?
1. Yes
2. No (if “no” continue to question 10)
Is there any permanent or long-term expense related to the treatment of the illness or accident that affects the family income, like: follow-up exams, special diets, rehabilitation, etc?
1. Yes
2. No
The experience made you drain your savings or other monetary resources that were targeted for education, appliances, cars, land, houses, home improvements, etc....
1. Yes
2. No
Has the illness and its ramifications caused any change in the structure of the household, like: separation, divorce, departure or arrival of family members, etc...?
1. Yes
2. No
In general, do you consider that the illness or accident of a member of your family adversely affected the permanent form of your family’s standard of living?
1. Yes
2. No

(If you responded “yes” in Question 3)

As a result of the accident or illness, has the ill member not been able to work on a full or part-time basis?
1. Yes, part-time
2. Yes, full-time
3. No
Has the illness/accident of the ill member caused the permanent loss of the job: of the ill member OR another member of the household OR have they experienced permanent work problems like the reduction of job status OR a cut in pay/wage?
1. Yes, loss of job (ill member / family member)
2. Yes, have had labor problems (ill member / family member)
3. No

Part Three: For the Household Head

Questions to be addressed exclusively to the household head. That is, the person that contributes the highest amount of income. The objective of this section is to capture any major employment changes which could have precipitated upward or downward mobility in the 1997-99 interval.

Only for household heads that are wage earners—dependent or independent—as well as those who are independent businesspeople

2.1 With respect to 1997, has your employment situation changed?

2.2 Between 1997 and today did you change your main job?

Yes
No (go to 2.4)

2.3 Why did you change your main job?

For economic improvement
Due to layoff/liquidation
Due to personal/family reasons
Due to travel
Other ______________ (specify)

2.4 Between 1997 and today, have you worked part-time?

2.5 Between 1997 and today, have you experienced an advance or economic improvement in your job or business?

2.6 Is your job permanent?

2.7 With respect to 1997, your present income is...

Much lower
Somewhat lower
Same (end of survey)
Somewhat higher
Much higher

2.8 With respect to 1997, how much has your present income increased (or decreased)?

Less than 20%
Between 20% and 50%
More than 50%

Only for agricultural landowners

2.9 With respect to 1997, the present production of your small farm (chacra) has...

increased considerably
increased
stayed the same
decreased
decreased considerably

2.10 Why?

2.11 With respect to 1997, do you think that in general the present prices for the products of your small farm (chacra) have...

increased considerably
increased
stayed the same
decreased
decreased considerably

2.12 With respect to 1997, the money that you get from the sale of the products of your small farm (chacra) is...

A lot more
More
Same
Less
A lot less

2.13 With respect to 1997, the profit margin that you get for the sale of the products of your small farm (chacra) is...

A lot more
More
Same
Less
A lot less

Endnotes

Note 1: The authors acknowledge the financial support for various surveys from the MacArthur Foundation and the InterAmerican Development Bank, and more general research support from the MacArthur and Tinker foundations. They also gratefully acknowledge the collaboration of Nancy Birdsall and Richard Webb in several components of this research, as well as helpful comments from George Akerlof. Back.

Note 2: Indeed, recent research in Eastern Europe finds that voting patterns are influenced much more by subjective assessments than they are by objective mobility trends. See the chapter by Petr Mateju in Birdsall and Graham (1999). Back.

Note 3: The distinction that Granovetter makes between strong and weak ties in determining individual mobility rates is also relevant here. Granovetter posits that strong ties, eg friendships, are important to close knit groups and local civil society, but do little to build the kinds of bridges that explain upward mobility. Weak ties, eg acquaintances, which establish bridges beyond individuals’ close knit group of family and friends, play a much more important role in determining individuals’ new opportunities and upward mobility. He supports his assertions with empirical studies of white and blue collar workers job promotion experience in the United States. See Granovetter (1973). Back.

Note 4: See Putnam (1993). See also Knack and Keefer (1997). Back.

Note 5: For detail on the safety net role that these groups can play, see Graham (1994). For these groups as poverty traps, see Hoff (1996). Back.

Note 6: See for example, McMurrer and Sawhill (1988) for trends in the United States. Deaton and Paxson (1994) compare intragenerational mobility trends in the United States and Taiwan. For conceptual and methodological frameworks for measuring mobility, see the chapters by Fields and Behrman in Birdsall and Graham (1999). For detailed research on U.S. intergenerational mobility, see Solon (1992). Back.

Note 7: See, among others, Haggard and Webb (1994), Barbara Geddes (1995), Rodrik (1996), and Carol Graham (1998). Back.

Note 8: See the introductory chapter to Birdsall and Graham (1999). Back.

Note 10: Sen’s classic definition of poverty, for example, centers on peoples’ capabilities to participate as productive members of society, rather than just on their level of income. See Sen (1995). Back.

Note 11: See Graham (1998). Back.

Note 12: An excellent analysis of voting during reform episodes is in Stokes (1996). Also Weyland (1998). Back.

Note 13: See Easterlin (1974). See also Easterlin (1995). Back.

Note 14: See Benabou and Ok (1998). Back.

Note 15: Granovetter (1973). Back.

Note 16: See Easterlin (1974) and Easterlin (1995). Back.

Note 17: See, for example, Diener (1984). See also the study of happiness in Switzerland discussed below. Deaton and Paxson (1994) have highlighted the role of negative shocks—which include poor health and bad luck—in determining lifetime mobility patterns. Such shocks, no doubt, also effect subejective assessments of well being, as Diener notes. Back.

Note 18: Veenhoven, for example, finds that the correlation between income and happiness is much greater in poor countries. Recent work in the transition economy of Kyrgyzstan confirms this. See Veenhoven (1991); and Namazie and Sanfey (1998). Back.

Note 19: For a thoughtful review of different societies tolerance for inequality, see Esping-Andersen (1990). For an excellent overview of trends in mobility and opportunity in the United States, see McMurrer and Sawhill (1998). Back.

Note 20: The authors are grateful to George Akerlof for helping them develop this line of analysis. See the chapter on The American Soldier in Merton (1957). Back.

Note 21: Gurr (1970). Back.

Note 22: They also find that married people are happier than single people, and that couples without children are happier than those with them; and women are happier than men. See Frey and Stutzer (1999). See also, Oswald (1997). For the same issues in the transition economies, see Namazie and Sanfey (1998). Back.

Note 23: Kenny (1999). Back.

Note 24: For work on social interactions and neighborhood effects, see, for example, Durlauf (forthcoming). For the role of identity in influencing economic behavior, see Akerlof and Kranton (1999). Back.

Note 25: See Foote Whyte (1993) Back.

Note 26: For detail on measurement issues, see Diener (1984). Back.

Note 27: Durlauf (forthcoming); Hoff (1996). Peyton Young, meanwhile, finds that in many cases both the structure and strength of interactions is endogenous to the context, such as in bilateral trading relationships, and can generate stochastically stable states which take a long time to undo. See Young (1999). One implication of this is that inequality generated by unequal bargaining positions and outcomes can persist between the networks of the rich and those of the poor. Back.

Note 28: For the central role played by such organizations in providing safety nets worldwide, see Graham (1994). Back.

Note 29: See Granovetter (1973). We would like to thank Judith Tendler for raising this point in this context. Back.

Note 30: Krackhardt (1992). Back.

Note 31: Collier (1999). Back.

Note 32: Seligson (1999). Back.

Note 33: Dietz (1998). Back.

Note 34: Rather ironically, POUM is also the acronym for the Catalonian Marxist Party, the Partido Obrero Unificado Marxista, a party that George Orwell joined during the Spanish Civil War! The authors would like to thank Alan Angell for pointing this out. Back.

Note 35: Benabou and Ok (1998). For a more philosophical discussion of why U.S. citizens support high levels of inequality and do not vote for more taxation, which supports the basic premises of Benabou and Ok, see Okun (1975). Back.

Note 36: An interesting empirical constrast is highlighted by Martin Ravallion’s recent research in Russia, which finds that because most Russians (accurately) expect declining living standards and mobility in the future, there is strong demand for redistribution. See Ravallion and Lokshin (1999). Back.

Note 37: Alesina and Perotti (1994). Back.

Note 38: Piketty (1995). Back.

Note 39: Clifford and Heath (1993). Back.

Note 40: Piketty (1995). Back.

Note 41: For a review of the panel data that is available for developing countries, see the paper by Gary Fields in this volume. Back.

Note 42: In contrast, expenditures on higher education were inversely related to enhanced intergenerational mobility, as in the Latin American context, higher education expenditures overwhelmingly favor the highest income deciles. See Behrman, Birdsall, and Sz*kely (1999). Back.

Note 43: See Dahan and Gaviria (1999). For a similar approach and findings for Brazil, see Lam and Schoeni (1993). Back.

Note 44: The authors do find a high correlation between inequality and assortative mating in the region. This is in keeping with the findings of Gary Burtless for the United States, which point to changing family composition as one of the key explanatory variables for increasing inequality in the United States. See, for example, Burtless (1999). Back.

Note 45: See Terrell (1999). For a discussion of changes in occupational mobility, see Mateju (1999). Back.

Note 46: See Hojman (1999). Back.

Note 47: For a discussion of the effects of trade opening on differential rewards to labor, see Robbins (1996). For a discussion of empirical evidence of differential returns to labor in Latin America, see Lora and Londoño (1998). See also Londoño, Spilimbergo, and Sz*kely (1997). Back.

Note 48: This occurs because some of those in the highest and lowest income categories are reluctant to report their situation accurately (the rationale of the former may be fears of increased taxation; and the latter due to pride). In Peru for example, in surveys conducted by APOYO, the well-respected polling firm that conducted the Latinobar—metro survey, only 7% of those in the highest socioeconomic category (A) placed themselves there when asked, while 81% of respondents in the A category placed themselves in the second or B category. Of respondents in the lowest or D category, 12% placed themselves in the second (B) category, 44% placed themselves in the third (C) category, and 42% in the D category. [Data from a survey conducted by Apoyo Opinion y Mercado S.A. in Lima and environs, Peru, July 1997] Back.

Note 49: For a discussion of the initial 1998 results see Webb’s chapter in Birdsall and Graham (1999). Back.

Note 50: See Graham and Kane (1998); Stokes (1996); and Weyland (1998). Back.

Note 51: See Dahan and Gaviria (1999). Back.

Note 52: The survey was taken at a time—November 1998—of much speculation about the possible effects of a devaluation in Brazil. The effects of the January 15 measure were, in the end, less severe than anticipated. Back.

Note 53: Other occupational rankings were: self-employed and public employees (ranked third). Back.

Note 54: See Duryea and Sz*kely (1998). Back.

Note 55: This positive correlation only appeared in the analysis of micro-level variables. Back.

Note 56: For second stage reforms, see Graham and Naim (1998). See also Pastor and Wise (1999). Back.

Note 57: It is worth noting here the linkages that Sen (1983) draws between absolute and relative income, and how these linkages depend on one’s relative position. Not owning leather shoes, for example, is hardly deprivation in an absolute sense. But in many societies, it could make one ineligible for a number of jobs. Back.

Note 58: This was further supported by a simple cross-country empirical exercise for the region in which we regressed the pro-market index on the level of reform in 1985, controlling for the change in reform in 1985-95, growth, and GDP per capita in 1997. The results supported the negative impact of the level of reform in 1985 on the average pro-reform index. With the exception of Peru, the POUM scores for the fast and late reformers also tended to be higher compared to those of the slow and early reformers. Back.

Note 59: Even with a significant decline of several percentage points, the poverty ratio remained high, at 49% of the population, in 1997. The consumption based Gini was.348 in 1997, the income Gini was.484, and the wealth Gini was.726. For detail see The World Bank (1999). Back.

Note 60: The living standards measurement surveys were conducted by Cuanto, S.A., and the panel data was analyzed by Martin Cumpa of the IDB with guidance from Richard Webb, who jointly presented the data at a Brookings-IDB workshop on mobility in June 1998. Back.

Note 61: This transition analysis does not include the 1997 observations, which were not ready at the time the analysis was done. Trends in 1997 show a continuity of progress for the extreme rural poor. Back.

Note 62: This is also supported by the preliminary results of the 1997 ENNIV/Cuanto survey. The authors are grateful to Richard Webb for providing this data. Back.

Note 63: See Graham and Kane (1998), p.89. While disbursements in 1991-93 correlated negatively with GDP per capita, suggesting some progressiveness, this did not guarantee targeting of the poorest. Funds were concentrated in more populated areas and in better educated departments, and allocations were not affected by deficits in public services. As of 1994 this changed, and more funds were allocated to departments where Peruvians voted "no". These were also the departments with the worst social indicators. For a slightly different interpretation of these results, see Roberts and Arce (1998). Back.

Note 64: In 1993, the proportion of families that used food aid was high only in Lima: 47% of low income families used food aid, while 9% of those in the highest strata in the sample did (the wealthiest 20% of the distribution were excluded from the sample), a percentage which exceeded that of even the lowest socioeconomic levels in other cities. See Graham (1994), Chapter 3. Back.

Note 65: See Deaton and Paxson (1994). Back.

Note 66: See, for example, Diener (1984) and Frey and Stutzer (1999). Back.

Note 67: This index and the results of the survey are described in greater detail in the chapter by Richard Webb in Birdsall and Graham (1999). Much of the following section is based on that chapter. Back.

Note 68: These figures are based on a survey conducted by Apoyo and cited by Webb in his chapter in Birdsall and Graham (1999). Back.

Note 69: For detail on these kinds of effects, see Hoff (1996); and Akerlof (1997). Back.

Note 70: In fact, a study by Glewwe and Hall of vulnerable groups in the same panel in Peru found that the only households that were able to preserve their income levels during the hyperinflation period were those that had ties to households or family members living abroad, which would constitute very weak ties in Granovetter’s terminology. See Glewwe and Hall (1998). Back.

Note 71: See Graham and Kane (1998), and World Bank (1999). Back.

Note 72: Hirschman’s "tunnel effect" may also be at play here. Back.

Note 73: Caution is necessary when drawing generalizations from this description, as sample size (the upwardly mobile in quintiles 3 and 4 reporting negative IPM’s is very small, although statistically significant (n=24 in 1999 and n=20 in 1998). Back.

Note 74: See the above discussion of Piketty’s endogenous beliefs dynamics. Piketty (1995). Back.

Note 75: This is supported by Benabou’s empirical findings for the United States, which suggest that even when people are well below the median income, they can expect to find themselves (or their children) above it in the future. See Benabou and Ok (1998). Back.