Policy Research Working Paper 10966 Perceptions of Economic Mobility and Support for Education Reforms Alexandru Cojocaru Michael Lokshin Iván Torre Europe and Central Asia Region & Poverty and Equity Global Practice November 2024 Policy Research Working Paper 10966 Abstract This paper investigates the relationship between the expec- support tax-financed education reforms. This correlation tations of economic mobility and support for tax-financed is robust to different formulations of mobility expecta- education reforms using data from the Life in Transition tions and persists over a decade, encompassing both stable Survey, which covers 39 countries in Europe, Central and post-crisis economic environments. The relationship Asia, and the Middle East and North Africa. The analysis is partially mediated by beliefs about the fairness of eco- demonstrates that individuals who expect themselves or nomic opportunities in society and individuals’ readiness their children to be upwardly mobile are more likely to to embrace risks. This paper is a product of the Office of the Chief Economist, Europe and Central Asia Region and the Poverty and Equity Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http:// www.worldbank.org/prwp. The authors may be contacted at acojocaru@worldbank.org, mlokshin@worldbank.org, and itorre@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Perceptions of Economic Mobility and Support for Education Reforms Alexandru Cojocaru, Michael Lokshin, and Iván Torre1 JEL: D3, J62, O1, O15, P2 Keywords: Economic mobility, equality of opportunity, support for education reform, Europe and Central Asia, Middle East and North Africa 1 The authors are at the World Bank. This paper’s findings, interpretations, and conclusions are entirely those of the authors and do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. Miguel Purroy provided excellent assistance in collecting, processing, and analyzing data for this paper. Send correspondence to acojocaru@worldbank.org. 1. Introduction Structural reforms can raise incomes, but the costs of reforms might fall disproportionately on some population groups, increasing inequality and fueling arguments opposing change. The risk of losses for large segments of society could be a powerful shaping force of economic and social policies. Identifying who will likely lose out from reforms and managing people’s perceptions about their future welfare is crucial for consolidating political support for costly measures and moving countries to higher growth trajectories. It might be particularly important to understand the factors that shape support for structural reforms for the middle-income countries in Europe, Central Asia, and elsewhere that are facing challenges in accelerating economic growth and raising their populations’ standards of living. This paper investigates whether support for education reforms is affected by expectations about future economic mobility. The focus on the subset of structural reforms related to education stems from the evidence that investments in human capital play a crucial role in driving economic growth (Hanushek and Woessmann 2008) and that education reforms are a powerful policy channel through which countries can improve the inclusiveness of growth by leveling the playing field and enhancing the upward economic mobility of disadvantaged groups (IMF 2019). The analysis is based on data on 39 countries in Europe, Central Asia, and the Middle East and North Africa from the 2010, 2016, and 2023 rounds of the Life in Transition Survey (EBRD 2023). This sample of countries is analytically interesting because, despite progress made over the last several decades, these countries lag the high-income countries in educational achievement (World Bank 2018). Reforms aimed at improving the quality of education are also more costly and politically challenging than efforts to expand access to education, which helped countries move from low-income to middle-income status (Aiyar and others 2013). Profound education reforms that take years to achieve lasting effects require a strong, broad-based, durable coalition at all levels of society. The paper finds that individuals who expect themselves or their children to be upwardly mobile are more likely to support tax-financed education reforms than individuals who do not. This relationship is robust to various formulations of mobility expectations and persists across several 2 survey rounds spanning both stable and post-crisis economic environments. This relationship is partially mediated by beliefs about the fairness of economic opportunities in society and individuals’ willingness to take risks. We use the instrumental variable approach to mitigate the possibility that our estimates are biased by the presence of unobserved individual traits correlated with expectations of mobility and individual willingness to support reforms. The rest of the paper is structured as follows. Section 2 reviews the literature. Section 3 presents our estimation framework. Section 4 describes the data and presents some descriptive results. Section 5 presents the main results of our multivariate analysis. Section 6 presents the results of our sensitivity analysis. Section 7 examines changes in support for education reform over time. Section 8 summarizes the paper’s main findings. 2. Literature review Economic theory suggests that self-interest drives policy preferences and voting behavior. The poor would be more likely to support higher levels of redistribution, for instance, whereas the rich would be more likely to oppose them (Meltzer and Richard 1981). Having experienced an adverse economic shock, such as job loss or insecure employment, heightens a person’s sense of vulnerability and leads to stronger preferences for more expansive safety nets (Margalit 2019; Heinemann et al. 2009; Boeri, Boersch-Supan, and Tabellini 2002). On the other hand, lottery windfalls are associated with support for more conservative and less egalitarian political views (Powdhavee and Oswald 2014). In transition economies, self-reported upward mobility was reported to be related to broader acceptance of the outcomes of privatization (Denisova et al. 2012) and greater support for the market economy (Cancho, Davalos, and Sanchez 2015). Earlier studies found that other factors—including the degree of understanding of what the reforms entail, political affinities, and normative beliefs, such as attitudes toward fairness—also shape preferences. Alesina and Glaeser (2004) argue that the difference in the size of the welfare state in the United States and Europe partly reflects differences in beliefs about the fairness underlying poverty and social mobility, as well as ethnic fractionalization, which influences preferences for redistribution. Luttmer and Singhal (2011) demonstrate that immigrants’ political preferences are influenced by beliefs about redistribution in their country of origin. 3 Future expectations can also inform economic self-interest. Benabou and Ok (2001) show that expectations of socioeconomic mobility can affect preferences for redistribution. Individuals with below-average incomes who would benefit from redistributive policies financed by higher taxes may oppose such policies if they expect to be upwardly mobile. This prediction, known as the Prospects of Upward Mobility hypothesis, has been borne out by the data in experimental settings (Checchi and Filippin 2004; Agranov and Palfrey 2020) as well as in observational data (Alesina and La Ferrara 2005; Alesina and Fuchs-Schuendeln 2005; Cojocaru 2014; Rueda and Stegmueller 2019). The literature on preferences for redistribution points to the importance of self-interest, cultural factors, and social mobility, both past and future (Corneo and Gruner 2002; Alesina and Giuliano 2009; Alesina, Stancheva, and Teso 2018; Cavaille 2023; see also Mengel and Weidenholzer 2022 for a detailed survey of this literature). Less is known about the role these factors play in shaping preferences for policy choices unrelated to redistribution. This paper aims to contribute to closing this knowledge gap. Several earlier studies consider the drivers of support for market reforms at the beginning of the post-Soviet transition (Duch 1993; Denisova et al. 2021; Rovelli and Zaiceva 2013) or in Latin America (Kaufman and Zuckermann 1998). Denisova et al. (2012) find that economic hardships during the post-Soviet transition—such as the experience of downward mobility, the lack of human capital suitable for a market economy, and the lack of private assets— significantly increased the desire for revising the results of earlier privatization. Several earlier studies have examined the individual determinants of preferences for investments in education for a small subset of European countries (Busemeyer and Garritzmann 2017; Garritzmann, Busemeyer, and Neimanns 2018). Wiese and Eriksen (2024) investigate linkages between expectations of mobility and willingness to pay for improvements in public education and health care using data from the 2016 round of the LiTS, framing it as a test of the POUM hypothesis with respect to indirect rather than direct redistribution, which was the focus on the majority of earlier studies. However, willingness to pay additional taxes for investments in education or health care might not necessarily be driven by redistribution motives, as respondents can view such investments as benefiting themselves and their families; thus support 4 for such investments can be entirely consistent with self-interest. The latter is indeed the argument this study proposes to investigate empirically. 3. Theoretical framework and empirical strategy The effects of structural reforms take time—possibly as long as a generation—to materialize. Unlike low-income countries, where education policies focus mainly on expanding access to primary and/or secondary education, middle-income countries, which often already have universal access to basic education, require harder-to-implement, longer-term investments, such as upgrading the vocational and tertiary education that is critical for research and development (Doner and Schneider 2016). In deciding whether to support tax-financed investments in education, individuals compare the discounted value of future benefits from the reforms (for themselves or members of their families) to the losses in current consumption because of higher taxes. Perceptions about socioeconomic mobility are an important factor in this decision. Intergenerational mobility in education and income is characterized by a significant degree of persistence. Children of better- educated parents and/or parents with higher socioeconomic status are more likely than other children to have high levels of education and income as adults (van der Weide et al. 2024). One of the key factors driving this intergenerational persistence, in both education and incomes, is parental private investments in the education of their children (Restuccia and Urrutia 2004). Individuals who expect to improve their socioeconomic status (or maintain high status) would want to transfer that status to their children, partly by investing in educational attainment for them. Assume the educational outcomes are a function of two complimentary inputs – publicly- provided schooling and private parental investment. Then, individuals who anticipate their economic situation to improve in the future would expect higher returns from better public education because they can complement it with higher private investment. In contrast, people who believe their situation will deteriorate may see less value forgoing current consumption to finance improvements in education, because they may not be able to afford the private educational investments necessary to secure higher educational attainment for their children. 5 More risk-averse individuals will discount the future more heavily, reducing their support for reforms (which outcomes are uncertain). Belief in meritocracy, on the other hand, will be expected to increase willingness to finance investments in policies that promote meritocratic advancement. These considerations motivate our empirical model. The core hypothesis we test is that support for education reforms financed by higher taxes depends on individual expectations of future mobility. This relationship could be expressed as the following model: (∗ ′ > 0 ) = ( + + ), (1) where ∗ is the latent willingness to pay additional taxes for investments in education by individual i from country c (with the observed willingness to support tax-financed reforms = 1 if ∗ > 0 and 0 otherwise); F(.) is a standard normal cumulative distribution function of (− ); is a variable reflecting the expectation of future mobility; X is a vector of individual-level controls, including age, gender, education, area of residence, current income, household size, and presence of children in the household; are country fixed effects; and ~(0,1) is the idiosyncratic error term. Given the tradeoff between current consumption and the discounted value of future benefits from improved education, the control variables in our empirical specification are included to the extent that they affect the individual’s cost-benefit decisions. We expect individuals with higher education or income to support investments in human capital. Wealthier households would also have a greater capacity to support tax-financed education reforms. Younger people may have a different discount rate for future benefits than older ones. We control for the presence of children in a household to capture differences in benefits from improved education between households with and without children. Area of residence is included to account for the differences in economic opportunities — and thus different mobility prospects—between urban and rural residents. Socioeconomic mobility profiles also vary by gender (Narayan et al. 2018), which may affect expectations of mobility if such expectations are formed by learning from experience (Piketty 1995). Religious people were shown to discount the future less and may also have different preferences related to investments in the public 6 education system (Carter et al. 2012). More risk-averse individuals discount the future more because of their aversion to uncertainty. To the extent that moving up the income ladder can be seen as indicative of success in life, beliefs about what underpins such success—hard work, skills, and political connections—is important (Alesina, Stancheva, and Teso 2018). Support for policies promoting meritocratic advancement (advancements achieved through hard work and intelligence/skills) depends on whether human capital investments are seen to be instrumental for socioeconomic mobility. The model also includes country fixed effects to account for heterogeneity across countries that may be correlated with individual beliefs about future mobility and willingness to support tax-financed reforms. Unobserved factors may also affect our results. For instance, an individual with a pessimistic disposition may have more negative expectations of future socioeconomic mobility and hold pessimistic views about the reforms, and thus be less likely to support such reforms, biasing the coefficients on the mobility expectations variable. To address such endogeneity, we instrument individual expectations of future mobility with cohort-level estimates of past intergenerational mobility for the 10-year country-birth cohort corresponding to a given survey respondent, obtained from the estimates by Torre, Lokshin, and Foster (forthcoming). Our choice of past intergenerational mobility as an instrument for future mobility expectations draws on the literature showing that past mobility experiences inform people’s perceptions and expectations of the future (Piketty 1995; Kelley and Kelley 2009; Cojocaru 2023). Our exclusion restriction is based on the argument that there is no theoretical reason why measures of past intergenerational mobility at the birth-cohort level should affect individual support for education reforms directly beyond their effect on mobility expectations. We should note, however, that we cannot exclude the possibility of cohort shocks, namely that cohort measures of past mobility may still be correlated with other factors that may affect support for education spending. As such, the IV estimates should similarly be viewed with these caveats in mind. 4. Data We use data from the 2010, 2016, and 2023 rounds of the Life in Transition Survey (LiTS) conducted by the European Bank of Reconstruction and Development (EBRD) and the World Bank 7 (EBRD 2023). The survey covers the transition economies of Europe and Central Asia and several comparator countries in Western Europe, the Middle East, and North Africa. 2 The 2010 round includes 35 countries, the 2016 round 34 countries, and the 2023 round 39 countries. The survey is based on a nationally representative sample of around 1,000 households per country in the second and fourth rounds and around 1,500 households per country in the third round. We capture support for tax-financed investments in education by examining responses to the question “would you be willing to give up part of your income or pay more taxes if you were sure that the extra money was spent on education?” This dummy variable is the main dependent variable of our analysis. Our measure of expectations of future mobility is based on the two social ladder questions: “Please imagine a 10-step ladder where on the bottom, the first step, stand the poorest 10 percent people in [COUNTRY], and on the highest step, the 10th, stand the richest 10 percent people in [COUNTRY]. On which step of the ladder is your household today?” and “Where on the ladder do you believe your household will be four years from now?” We take the difference between the expected future and current income ladder positions as our measure of expectations of future mobility, which ranges from –9 to +9. We also conduct a sensitivity analysis that relies only on the direction of the expectations by creating a variable with three categories (downwardly mobile, static, and upwardly mobile) based on the difference between the ladder position today and the ladder position four years from today. The LiTS data also allow for assessment of the expectation of intergenerational mobility, which is captured by responses to the following statement: “Children who are born now will have a better life than my generation,” with respondents answering on a five-step Likert scale ranging from strongly disagree to strongly agree. We construct a dummy variable that takes the value of 1 if the answer is either agree or strongly agree and zero otherwise. 2 We use the second, third, and fourth rounds of the LiTS, for which comparable information on parental education background is available. The classification of parental education in the first round of the LiTS is not comparable to that of the following rounds, which is why we excluded it from our study. 8 An individual’s risk aversion is captured by assessing responses to the following question: “Please rate your willingness to take risks in general, on a scale from 1 to 10, where 1 means that you are not willing to take risks at all and 10 means that you are very much willing to take risks.” (To measure risk aversion, we reverse the categories so that higher values imply higher risk aversion and lower values imply lower risk aversion.) To capture beliefs about the fairness of socioeconomic mobility, we rely on responses to the following question: “In your opinion, which of the following factors is the most important to succeed in life in [COUNTRY] now?” Answer options are effort and hard work, intelligence and skills, political connections, breaking the law, and others. We collapse these options into a binary variable that takes a value of 1 if respondents think that success stems from meritocratic reasons (effort/hard work or intelligence/skills) and zero if it relies on political connections, breaking the law, or other. Table 1 presents the summary statistics. 5. Main results In this section, we first examine the relationship between expectations of future mobility and willingness to pay for education investments across countries in the latest (2023) LiTS round. Respondents’ average expectations of their own mobility and that of children born today vary significantly across countries. In countries where the population is more optimistic about the future, the share of the population willing to pay part of its income or pay more taxes to support investments in education is also larger (panel a of Figure 1). Positive expectations about the prospects of children born today with respect to the current generation are also correlated with greater willingness to pay for education investments (panel b of Figure 1). We then estimate the relationship between support for tax-financed education reforms and expectations of one’s own future mobility conditional on the respondent’s characteristics. Table 2 reports the marginal effects from this baseline probit estimation in Specification 1. The results indicate that the respondent’s educational level is a strong predictor of support for investments in education, consistent with earlier findings (Busemeyer and Garritzmann 2017; Garritzmann, Busemeyer, and Neimanns 2018). The share of the population with primary education supporting education reforms is 23 percentage points lower than that of respondents 9 with post-graduate education; for people with upper-secondary education, support is 19 percentage points lower. The associations between willingness to pay for education reforms and age, gender, religious affiliation, area of residence, household composition, and income are not statistically significant. Turning to our primary variable of interest, more positive expectations of future mobility are associated with increased willingness to support education reforms: Moving up one step on the income ladder is associated with a 1.4 percentage point increase in the probability of supporting investments in education. 3 At first glance, a greater willingness to bear higher taxes for investments in education may seem to contradict the canonical POUM hypothesis, which posits that it will be in the self-interest of upwardly mobile individuals to support lower tax rates. However, as we have argued in section 3, if reaching high educational attainment requires private investments on the part of parents, then investments in improving education would actually be consistent with self-interest for those who expect upward mobility, because their children will be more likely to benefit from improved education. We further hypothesized that people who think success reflects hard work and skills (about 70 percent of the sample) are more willing to support reforms that reward meritocratic efforts than people who believe that success reflects other (nonmeritocratic) factors. The perception of fairness could be interpreted as the effectiveness of converting skills and education into future incomes. Beliefs about fairness are also associated with expectations of future mobility. The mean difference between future and current ladder positions in our sample is twice as high among those who perceive the path to success to be determined by meritocratic factors. People with a lower aversion to risk 4 may also have a lower discount rate than people with a higher risk aversion and may, therefore, be willing to invest more in reforms that bring payoffs in the future. Risk tolerance is also correlated with expectations of mobility – the mean difference 3 We also estimate specifications with polynomials of the expectations of mobility variables (quadratic and cubic). These estimations failed to produce significant coefficients on nonlinear parameters. 4 In our sample, the average degree of risk tolerance (from a level of 1 being completely risk averse to a level of 10 being completely risk lover) is about 4.8, suggesting on average a mild risk aversion. Of the individuals in the sample, 12.3 percent have a risk tolerance above 7, and 36.1 percent have a risk tolerance below 4. 10 between the future and current income ladder positions is three times higher in our sample for those with risk tolerance above 7 compared to those with risk tolerance below 4. To test these hypotheses, we re-estimate the baseline model, adding controls capturing fairness (Specification 2 in Table 2), risk aversion (Specification 3), and both jointly (Specification 4). The estimations show that people who believe that success in their society is meritocratic (determined by hard work and intelligence/skills) are 9 percentage points more likely to give up part of their income or pay additional taxes to support investments in education than people who think that success in life reflects personal connections or other reasons unrelated to hard work or skills. Risk aversion is negatively correlated with support for education reforms. This result is consistent with the finding that risk-averse individuals discount the future more, making them less likely to make investments whose payoffs are uncertain. On average, a one-step increase on the risk aversion scale is associated with a 2 percentage point decrease in the willingness to pay for education reforms. The estimations also reveal that introducing fairness and risk aversion into the model reduces the size of the coefficient on the mobility expectations from 1.4 to 1.1 percentage points, equivalent to one-fifth of the baseline partial correlation – the result consistent with a mediating effect of these variables. Our focus on expectations of own future mobility stems from observed intergenerational persistence, where children of upwardly mobile parents may be more likely to benefit from an improved education system, increasing the probability of their parents supporting educational investments. We can also use expectations of intergenerational mobility directly by relying on the question “Children who are born now will have a better life than my generation.” 5 While the question refers to children in general, it is plausible to assume that respondents’ beliefs about children’s fortunes will be informed, to a larger extent, by the experiences of and expectations 5 Children from wealthy households may opt for private education, reducing their parents’ incentives to invest in public education. Private education is common in several Latin American countries. It is not widespread in Europe and Central Asia, which has a legacy of universal provision of education and well-established state universities tend to be more prestigious than the recently established private institutions. If the top of the socioeconomic ladder were to opt out of public education, the coefficient in our model would be attenuated. 11 for their own children and the children of relatives or friends who will likely share common socio- demographic characteristics with the respondent. Table 3 reports the estimates with an alternative, intergenerational measure of expected mobility. As in the case of own mobility, among respondents who agree that “children who are born now will have a better life than my generation,” support for education investments is 5.3 percentage points higher in the baseline specification (Specification 1). As before, we find evidence of partial mediation of the relationship between intergenerational mobility and support for education reforms through beliefs about fairness and attitudes toward risk. Accounting for both variables reduces the correlation between intergenerational mobility expectations and support for education reforms by 20 percent (Specification 3). How does the positive association between future upward mobility and willingness to pay for improved education square with evidence from previous studies that have found upward mobility expectations to be associated with a reduced preference for redistribution (Alesina and La Ferrara 2005; Alesina and Fuchs-Schuendeln 2005; Cojocaru 2014)? Upwardly-mobile parents are more likely to invest in education if they associate this mobility with a higher likelihood that their children will benefit from education, and, in the case of preferences for redistribution, upwardly mobile individuals will more likely be disadvantaged by higher taxes, thus making it in their self-interest to oppose such policies. But the returns on education are so high compared to any other investments, that they dominate the decrease in willingness to pay higher taxes. 6 6. Sensitivity and robustness analysis We test several alternative specifications of our main model to assess the sensitivity of the results to different definitions of our variables of interest. First, we assume that the difference in utility of respondents who move from the first to the second rung of the ladder is equal to that for someone who moves from the seventh to the eighth rung. To relax this cardinality assumption, we collapse the variable representing the expectations of future mobility into a variable with categories corresponding to possible directions of the mobility expectations: downward, static, 6 Deming (2022) demonstrates that each year of eduction yields about 10 percent rise in annual earnings. 12 and upward.7 Specification 1 in Table 4 shows that compared with the static (reference) category, upward mobility expectations are associated with a higher willingness to support education reforms and downward mobility expectations are associated with a lower willingness to do so. The magnitude of the effect is larger than that of the original variable (Specification 4 in Table 2); the size and statistical significance of other controls are similar in the two estimations. We also analyze the impact of respondents’ longer-term economic trajectories on their support for the reforms. These variables are constructed from the responses to questions about the respondent’s position on the economic ladder four years ago, now, and four years in the future. The rising income trajectory corresponds to a situation in which the expected ladder position four years in the future is higher than the ladder position four years prior to today, and there is no downward mobility in either of the two four-year windows (between four years earlier and today and between today and four years in the future). In other words, this measure documents an improvement in a respondent’s economic ladder position over the 8-year period. Such a longer- term upward mobility measure addresses the concern that 4 years is too short to record meaningful changes in mobility and guards against sharp short-term discontinuities, and in particular against the possibility that reported expected upward mobility could be simply an expectation of recovery after some negative economic shock. Specification 2 of Table 4 shows that respondents who expect their future income to improve in four years relative to four years earlier are more likely to support education reforms, and the conditional correlation between this mobility variable and support for reforms is even stronger than in Specification 1. In Specification 3, we conduct sensitivity analysis for the intergenerational mobility models reported in Table 3. We restrict our sample to respondents from households with children (roughly 30 percent of the full LiTS sample), to increase the salience of respondents’ own children when answering the question of whether children born today will fare better than the current 7 Individuals who expect to be on a lower step of the ladder in four years are categorized as having downward expectations of mobility. Individuals who expect to be on a higher step of the ladder four years from now are categorized as having upward expectations of mobility. Individuals who expect to be on the same step of the ladder four years from now are categorized as having static expectations. 13 generation. This restriction does not appear to affect the magnitude or the statistical significance of the children’s mobility variable. Finally, we explore the possibility that our estimates suffer from endogeneity bias stemming from unobserved individual traits (such as pessimism), which may be simultaneously correlated with expectations of mobility and individual willingness to support reforms. As noted in section 3, we employ 10-year cohort measures of mobility matched to the respondent’s country-age cohort as instruments for individual expectations of mobility. For each cohort, we derived shares of individuals with more or fewer years of schooling than their parents (i.e., the shares of upward and downward movers). We derive these shares based on both the father’s and mother’s education from Torre, Lokshin, and Foster (2024). To test the robustness of our instruments, we report IV estimates from two specifications that use different combinations of IVs. In Specification 4 of Table 4, expectations of future mobility are instrumented by the share of upward movers with respect to the father and downward movers with respect to the mother; Specification 5 shows the share of upward movers with respect to mothers and downward movers with respect to fathers. 8 The tests for weak instrument and overidentification tests pass at conventional levels, but endogeneity tests only marginally reject the null hypothesis of exogeneity in Specification 5 and fail to reject it in Specification 4.9 These findings suggest that the case for relying on the IV results over baseline probit estimates is not strong. Nevertheless, the second-stage IV estimates of the coefficient on the expectations of mobility variable remain statistically significant and increase. Given the above concerns, however, we do not want to overinterpret the larger magnitudes of the IV coefficients. 8 We use mixtures of father’s and mother’s education as overidentification tests, because when the shares of upward movers and downward movers are based on fathers, one of the instruments becomes insignificant in the first-stage regression. 9 Table 4 reports the Cragg and Donald (1993) tests of weak instruments and the Sargan (1975) overidentification tests. The p-value for the Wu-Hausman test (Hausman 1978) is 0.0417 for Specification 4 and 0.0585 for Specification 5; the p-value for the Durbin score (chi-square) (see Nakamura 1981) is 0.0414 for Specification 4 and 0.0582 for Specification 5. 14 7. Changes in support for education reforms over time We next investigate whether the regularities we uncover by analyzing the sample of LiTS 2023 existed in the earlier rounds of the survey.10 Expectations of future mobility or support for long- term reforms may have fallen in the wake of the COVID-19 pandemic, which had profound and long-lasting impacts on economic activity and household welfare (World Bank 2022). We use data from the 2010 and 2016 rounds of the LiTS to examine the stability of the linkages between expectations of mobility and support for education reforms over time. The 2010 and 2016 survey rounds offer two contrasting data points. Like the 2023 round, the 2010 round was conducted in the wake of the Global Financial Crisis and the European debt crisis of 2009/10. In contrast, the 2016 data were collected during the period of relative economic stability in the region. Table 5 presents estimations for three rounds of the LiTS (2010, 2016, and 2023). 11 The first three sets of results demonstrate that, as in 2023, respondents with more optimistic expectations about their economic mobility were more likely to support tax-financed education reforms in 2010 and 2016. The strength of this relationship doubled from 0.007 in 2010 to 0.014 in 2016, remaining virtually unchanged in 2023. The effect on support for reforms of expectations of intergenerational mobility increased from 0.054 in 2010 to 0.073 in 2016 and decreased to 0.041 in 2023. In all three survey rounds, beliefs about the fairness of the underlying economic environment and attitudes toward risk remain significant correlates of support for tax-financed education reforms. Two findings emerge from these intertemporal comparisons. First, the positive association between intra- and intergenerational expectations of upward mobility persists for over a decade and during periods of economic turmoil and relative calm. Second, the relationship between 10 The Chow (1960) test on structural break indicates that for all pairwise comparisons of three rounds of LiTS, estimates based on pooled samples are rejected in favor of separate estimates. 11 The results for the 2023 round of LiTS in Table 5 differ from the results shown in Table 2, because here we use the monthly consumption aggregate as a control variable whereas the results in Table 2 are based on a specification with monthly income as control variable. Information about monthly income was collected only in the 2023 round. 15 expectations of upward mobility—particularly intergenerational mobility—and support for tax- financed education reforms appears stronger during periods of relative economic stability. 8. Conclusions To move to steeper growth trajectories, middle-income countries of Europe and Central Asia must implement systemic structural reforms. The successful implementation of these often costly policies requires broad public support. Earlier studies, focusing primarily on preferences for redistribution, have found that support for reforms tends to be higher among people who expect to benefit from such reforms and people who believe the outcomes of these reforms to be fair. Misperceptions about the impacts of reforms add frictions by reducing the public’s willingness to adjust to new realities and can even jeopardize achieving the reform objectives. Managing people’s expectations about their future welfare is crucial to consolidating political support. This paper confirms the importance of self-interest and beliefs about fairness for support of tax- financed education reforms. People who expect themselves or their children to be upwardly mobile are more likely to support tax-financed education reforms than people who expect their position to deteriorate. We conjecture that this higher support stems partly from the economic self-interest of upwardly mobile individuals to invest in improved education that will benefit their children. This relationship is partially mediated by normative beliefs about the fairness of economic opportunities in society and individuals’ willingness to take on risks. These results are robust to different formulations of mobility expectations and persist across three survey rounds spanning both stable and post-crisis economic environments. Our results point to the critical role expectations of socioeconomic mobility and beliefs about fairness play in driving support for costly reforms. In addition, our findings highlight the challenges facing the countries of Europe and Central Asia, where beliefs that educational and professional advancement are not based on meritocracy remain widespread. Indeed, data from the 2016 round of the LiTS reveal that more than two-thirds of adults in the region report that having informal connections is important to secure a good government job, and almost 40 16 percent report that connections help obtain university admission. 12 Estimates of intergenerational mobility across age cohorts point to deteriorating prospects of socioeconomic mobility in Europe and Central Asia for people who reached adulthood after the fall of the Berlin Wall (Cojocaru 2021). Governments can increase support for structural reforms by managing expectations of socioeconomic mobility and investing in institutional reforms aimed at addressing inequality of opportunity. Clearly defining the future course of the reforms, building a narrative around them, and signaling the government’s determination to follow through might reduce individuals’ uncertainties and improve their perceptions of their prospects, easing the adoption of the new policy regime. The political legitimacy of reforms critically depends on the government’s ability to demonstrate that they will be carried out fairly and inclusively, providing a level playing field for all segments of the population. 12 Questions about the importance of informal connections were not asked in the 2023 LiTS round. 17 References Agranov, M., T.R. Palfrey. 2020. “The effects of income mobility and tax persistence on income redistribution and inequality.” European Economic Review, 123. Aiyar, S., R. Duval, D. Puy, Y. Wu, and L. Zhang. 2013. “Growth Slowdowns and the Middle-Income Trap.” IMF Working Paper 13-71. Washington, DC: International Monetary Fund. Alesina, A., and E. La Ferrara. 2005. “Preferences for Redistribution in the Land of Opportunities.” Journal of Public Economics 89 (5–6): 897–931. Alesina, A., and N. Fuchs-Schuendeln. 2005. “Goodbye Lenin. Or Not? The Effect of Communism on People’s Preferences.” NBER Working Paper 11700, National Bureau of Economic Research, Cambridge, MA. Alesina, A., and P. Giuliano. 2009. “Preferences for Redistribution.” NBER Working Paper 14825, National Bureau of Economic Research, Cambridge, MA. Alesina, A., and E. Glaeser. 2004. Fighting Poverty in the US and Europe: A World of Difference. Oxford: Oxford University Press. Alesina, A., S. Stancheva, and E. Teso. 2018. “Intergenerational Mobility and Preferences for Redistribution.” American Economic Review 108 (2): 521–54. Benabou, R., and E. Ok. 2001. “Social Mobility and the Demand for Redistribution: The POUM Hypothesis.” Quarterly Journal of Economics 116: 447–87. Boeri, T., A. Boersch-Supan, and G. Tabellini. 2002. “Pension Reforms and the Opinions of European Citizens.” American Economic Review 92 (2): 396–401. Busemeyer, M.R., and J.L. Garritzmann. 2017. “Public opinion on policy and budgetary tradeoffs in European welfare states: evidence from a new comparative survey”, Journal of European Public Policy, 24:6, 871-889. Cancho, C., M. Davalos, and C. Sanchez. 2015. “Why So Gloomy? Perceptions of Economic Mobility in Europe and Central Asia.” Policy Research Working Paper 7519, World Bank, Washington, DC. Carter, E.C., M.E. McCullough, J. Kim-Spoon, C. Corrales, and A. Blake. 2012. “Religious People Discount the Future Less.” Evolution and Human Behavior 33 (3): 224–31. Cavaille, C. 2023. Fair Enough? Support for Redistribution in the Age of Inequality, Cambridge University Press. Checchi, D., and A. Filippin. 2004. “An Experimental Study of the POUM Hypothesis.” Research in Economic Inequality 11: 115–36. Chow, G. 1960. “Tests of Equality between Sets of Coefficients in Two Linear Regressions.” Econometrica 28 (3): 591–605. Cojocaru, A. 2014. “Prospects of Upward Mobility and Preferences for Redistribution: Evidence from the Life In Transition Survey.” European Journal of Political Economy 34: 300–14. ———. 2021. “Inequality and Well-Being in Transition: Linking Experience and Perception to Policy Preferences.” In The Palgrave Handbook of Comparative Economics, ed. E. Douarin and O. Havrylyshin. London: Palgrave Macmillan. ———. 2023. “Flaunt Them If You’ve Got Them? Informal Connections and Beliefs about Prospects of Upward Mobility in Transition Economies.” Comparative Economic Studies 65 (3): 416–41. 18 Corneo, G., and H.P. Gruner. 2002. “Individual Preferences for Political Redistribution.” Journal of Public Economics 83: 83–107. Cragg, J.G., and S.G. Donald. 1993. “Testing Identifiability and Specification in Instrumental Variable Models.” Econometric Theory 9: 222–40. Deming, D. (2022). “Four Facts about Human Capital.” Journal of Economic Perspectives, 36(3): 75-102. Denisova, I., M. Eller, T. Frye, and E. Zhuravskaya. 2012. “Everyone Hates Privatization, But Why? Survey Evidence from 28 Post-Communist Countries.” Journal of Comparative Economics 40 (1): 44−61. Doner, R.F., and B.R. Schneider. 2016. “The Middle-Income Trap: More Politics Than Economics.” World Politics 68 (4): 608–44. Duch, R.M. 1993. “Tolerating Economic Reform: Popular Support for Transition to a Free Market in the Former Soviet Union.” American Political Science Review 87 (3): 590–608. EBRD (European Bank for Reconstruction and Development). 2023. Life in Transition Survey. London. Flechtner, S., and S. Panther. 2016. “Economic Inequality, Political Power, and Political Decision-Making: The Case of the ‘Middle-Income Trap.’” In Varieties of Economic Inequality, ed. S. Fadda and P. Tridico. London: Routledge. Fisher, S. 1986. “Issues in Medium-Term Macroeconomic Adjustment.” World Bank Research Observer 1 (2): 163–82. Garritzmann, J.L., Busemeyer, M.R., and E. Neimanns. 2018. “Public demand for social investment: new supporting coalitions for welfare state reform in Western Europe?”, Journal of European Public Policy, 25:6, 844-861. Hanushek, E.A., and L. Woessmann. 2008. “The Role of Cognitive Skills in Economic Development.” Journal of Economic Literature 46 (3): 607–68. Hausman, J.A. 1978. “Specification Tests in Econometrics.” Econometrica 46 (6): 1251–71. Heinemann, F., I. Bischoff, G. Hennighausen, and T. Hennighausen. 2009. “Choosing from the Reform Menu Card: Individual Determinants of Labour Market Policy Preferences.” In Labour Economics, Jahrbücher für Nationalökonomie und Statistik, ed. Bernd Fitzenberger, Werner Smolny, and Peter Winker, 180–97. Berlin: De Gruyter Oldenbourg, https://doi.org/10.1515/9783110508284- 006. IMF (International Monetary Fund). 2019. G-20 Report on Strong, Sustainable, Balanced, and Inclusive Growth. Washington DC. Kaufman, R.R., and L. Zuckermann. 1998. “Attitudes toward Economic Reform in Mexico: The Role of Political Orientations.” American Political Science Review 92 (2): 359–75. Kelley, S.M.C., and C.G.E.K. Kelley. 2009. “Subjective Social Mobility: Data from 30 Nations.” In Charting the Globe: The International Social Survey Programme 1984–2004, ed. Max Hallier, Roger Jowell, and Tom W. Smith, 106–24. London: Routledge. Luttmer, E., and M. Singhal. 2011. “Culture, Context, and the Taste for Redistribution.” American Economic Journal: Economic Policy 3 (1): 157–79. Margalit, Y. 2019. “Political Responses to Economic Shocks.” Annual Review of Political Science 22: 277– 95. Meltzer, A.H., and S.F. Richard. 1981. “A Rational Theory of the Size of Government.” Journal of Political Economy 89: 914–27. 19 Mengel, F., and E. Weidenholzer. 2023. “Preferences for Redistribution.” Journal of Economic Surveys 37: 1660–77. Nakamura, A. 1981. “On the Relationships among Several Specification Error Tests Presented by Durbin, Wu, and Hausman.” Econometrica. 49 (6): 1583–88. Narayan, Ambar, Roy Van der Weide, Alexandru Cojocaru, Christoph Lakner, Silvia Redaelli, Daniel Gerszon Mahler, Rakesh Gupta N. Ramasubbaiah, and Stefan Thewissen. 2018. Fair Progress? Economic Mobility across Generations around the World. Equity and Development. Washington, DC: World Bank. OECD (Organisation for Economic Co-operation and Development). 2023. PISA 2022 Results. Volume I: The State of Learning and Equity in Education. Paris. Piketty, T. 1995. “Social Mobility and Redistributive Politics.” Quarterly Journal of Economics 110 (3): 551–84. Powdthavee, N., and A.J. Oswald. 2014. “Does Money Make People Right-Wing And Inegalitarian? A Longitudinal Study of Lottery Winners.” IZA Discussion Paper 7934, Institute of Labor Economics, Bonn, Germany. Restuccia, D., C. Urrutia. 2004. “Intergenerational persistence in earnings: The role of early and college education.”, American Economic Review, 94(5): 1354-1378. Rovelli, R., and A. Zaiceva. 2013. “Did Support for Economic and Political Reforms Increase During the Post-Communist Transition, and If So, Why?” Economics of Transition 21 (2): 193−240. Rueda, D., and D. Stegmueller. 2019. Who Wants What? Redistribution Preferences in Comparative Perspective, Cambridge University Press. Sargan, J.D. 1975. “Testing for Misspecification after Estimating Using Instrumental Variables.” Contributions to Econometrics. New York: Cambridge University Press. Torre, I., M. Lokshin, and J. Foster. Forthcoming. “Does Social Mobility Affect Economic Outcomes? Cross-Country Analysis Using Different Mobility Measures.” Policy Research Working Paper, World Bank, Washington, DC. Van der Weide, R., C. Lakner, D. Gerszon Mahler, A. Narayan, and R. Gupta. 2024. “Intergenerational Mobility around the World: A New Database.” Journal of Development Economics 166: 103–67. Wiese, R., S. Eriksen. 2024. “Willingness to pay for improved public education and public healthcare systems: the role of income mobility prospects.” Fiscal Studies, 45: 55-76. World Bank. 2018. World Development Report 2018: Learning to Realize Education’s Promise. Washington, DC. doi:10.1596/978-1-4648-1096-1. ———. 2022. World Development Report 2022: Finance for an Equitable Recovery. Washington, DC. doi:10.1596/978-1-4648-1730-4. 20 Figure 1 Expectations of intra- and intergenerational mobility and willingness to pay for investments in education a. Intragenerational expectations b. Intergenerational expectations .8 .8 Share willing to pay more tax to support investmetns in education Share willing to pay more tax to support investmetns in education .6 .6 .4 .4 .2 .2 -1 0 1 2 .2 .4 .6 .8 1 Expected future mobility Share agree that children will have a better life vis-a-vis current generation Linear fit Linear fit Note: Authors’ estimates, based on data from the 2023 Life in Transition Survey. 21 Table 1 Summary statistics for the full sample Number of Variable Mean Std. dev. Minimum Maximum observations Willingness to pay for tax-financed investments in 0.479 0.500 0 1.0 25,268 education (yes/no) Expectations of own mobility (income ladder) 0.397 1.627 –9.0 9.0 25,268 Expectations of upward intergenerational mobility 0.490 0.500 0 1.0 25,268 (children will live better than parents [yes/no]) Belief about fairness 0.705 0.456 0 1.0 24,820 Upward trend in income trajectory 0.306 0.461 0 1.0 25,256 Downward expectations of mobility 0.236 0.425 0 1.0 25,268 Static expectations of mobility 0.385 0.486 0 1.0 25,268 Upward expectations of mobility 0.380 0.485 0 1.0 25,268 Risk aversion (categorical) 4.800 2.864 1.0 10.0 25,151 Gender and age Female 0.580 0.494 0 1.0 25,268 Age of respondent 47.723 17.117 18.0 95.0 25,268 Education No degree/no education 0.029 0.167 0 1.0 25,268 Primary education 0.048 0.213 0 1.0 25,268 Lower-secondary education 0.144 0.351 0 1.0 25,268 Upper-secondary education 0.438 0.496 0 1.0 25,268 Post-secondary non-tertiary education 0.077 0.267 0 1.0 25,268 Tertiary education (not a university diploma) 0.047 0.212 0 1.0 25,268 Bachelor’s degree or more 0.163 0.370 0 1.0 25,268 Master’s degree or PhD 0.054 0.227 0 1.0 25,268 Marital status Single (never married) 0.223 0.417 0 1.0 25,200 Married 0.533 0.499 0 1.0 25,200 Widowed 0.131 0.337 0 1.0 25,200 Divorced 0.097 0.296 0 1.0 25,200 Separated 0.016 0.125 0 1.0 25,200 Religion Atheist, agnostic, or none 0.105 0.307 0 1.0 24,100 Christian 0.525 0.499 0 1.0 24,100 Buddhist 0.019 0.138 0 1.0 24,100 Jewish 0.001 0.027 0 1.0 24,100 Muslim 0.340 0.474 0 1.0 24,100 Other 0.011 0.103 0 1.0 24,100 Respondent lives in urban area 0.611 0.488 0 1.0 25,268 Household size 2.815 1.702 1.0 20.0 25,268 Share of children in household under 15 0.135 0.224 0 0.875 25,268 Monthly household income (000s USD at PPP 2017) 0.227 0.299 0 10.514 20,944 Monthly household expenditure (000s USD at PPP 2017) 0.097 0.141 0 4.002 22,676 Note: The household expenditure variable includes the following categories: food and consumption staples, transportation, education, health, clothing, and durable goods. 22 Table 2 Willingness to pay for investments in education as a function of expectations about economic mobility and respondent characteristics. Probit marginal effect (ME). Specification 1 Specification 2 Specification 3 Specification 4 Variable ME Std. err. ME Std. err. ME Std. err. ME Std. err. Expected own mobility 0.014 *** 0.004 0.013 *** 0.004 0.011 *** 0.004 0.011 0.004 ** Beliefs about fairness 0.086*** 0.017 0.087*** 0.017 Risk aversion –0.019*** 0.003 –0.019*** 0.003 Gender and age Female –0.007 0.008 –0.009 0.008 0.005 0.008 0.003 0.008 Age/100 –0.210 0.200 –0.171 0.193 -0.182 0.199 –0.151 0.194 Age squared –0.028 0.204 –0.064 0.197 0.010 0.205 –0.021 0.198 Education reference category: master’s degree or PhD No degree/no education –0.230*** 0.053 –0.233*** 0.055 –0.216*** 0.055 –0.219*** 0.057 Primary education –0.231 *** 0.035 –0.231 *** 0.033 –0.216 *** 0.035 –0.215*** 0.033 Lower-secondary education –0.256*** 0.033 –0.252*** 0.032 –0.244*** 0.033 –0.241*** 0.033 Upper-secondary education –0.187*** 0.025 –0.181*** 0.025 –0.180*** 0.026 –0.173*** 0.026 Bachelor’s degree or more –0.091 *** 0.026 –0.087 *** 0.025 –0.086 *** 0.026 –0.082*** 0.026 Marital status reference category: single [never married] Married 0.014 0.017 0.012 0.017 0.017 0.017 0.016 0.017 Widowed –0.008 0.018 –0.009 0.018 –0.003 0.018 –0.003 0.018 Divorced –0.007 0.019 –0.007 0.019 –0.007 0.018 –0.006 0.018 Separated 0.009 0.033 0.005 0.033 0.011 0.033 0.008 0.033 Religion reference category: atheist, agnostic, or none Jewish 0.346*** 0.045 0.343*** 0.044 0.335*** 0.047 0.336*** 0.045 Christian –0.022 0.033 –0.024 0.034 –0.019 0.033 –0.021 0.034 Muslim –0.018 0.052 –0.017 0.052 –0.019 0.051 –0.017 0.052 Other –0.013 0.043 –0.003 0.049 –0.017 0.044 –0.009 0.051 Other variables Household size 0.015*** 0.005 0.017*** 0.005 0.016*** 0.005 0.017*** 0.005 Share of children 0.030 0.031 0.022 0.031 0.026 0.031 0.019 0.031 Urban –0.025 0.020 –0.023 0.020 –0.025 0.020 –0.023 0.020 Monthly income 0.063 0.047 0.059 0.045 0.053 0.046 0.050 0.044 Number of observations 21,160 20,800 21,048 20,704 Note: Marginal effects of probit estimations are reported in this table. Marginal effects on 38 county dummies are omitted. Standard errors are clustered at the country level. *** Significant at the 1 percent level, ** significant at the 5 percent level, * significant at the 10 percent level. ♣ Age and age-squared are jointly significant at the 1 percent level. 23 Table 3 Willingness to pay for investments in education as a function of expectations about intergenerational mobility and respondent characteristics. Probit marginal effect (ME). Specification 1 Specification 2 Specification 3 Specification 4 Variable ME Std. err. ME Std. err. ME Std. err. ME Std. err. Children will live better (yes/no) 0.053*** 0.015 0.046*** 0.014 0.049*** 0.014 0.042*** 0.013 Beliefs about fairness 0.092*** 0.016 0.093*** 0.015 Risk aversion –0.020 *** 0.003 –0.019*** 0.003 Gender and age Female –0.006 0.008 –0.008 0.009 0.007 0.008 0.004 0.009 ♣ ♣ ♣ Age/100 –0.110 0.215 –0.086 0.213 –0.101 0.212 –0.081♣ 0.210 Age squared –0.150 0.217 –0.170 0.214 –0.085 0.215 –0.104 0.212 Education reference category: master’s degree or PhD No degree/no education –0.222 *** 0.047 –0.229*** 0.049 –0.206*** 0.050 –0.215*** 0.052 Primary education –0.222*** 0.032 –0.223*** 0.031 –0.202*** 0.032 –0.203*** 0.032 Lower-secondary education –0.249*** 0.032 –0.246*** 0.032 –0.234*** 0.032 –0.232*** 0.032 Upper-secondary education –0.182*** 0.025 –0.176*** 0.025 –0.173*** 0.026 –0.167*** 0.026 Bachelor’s degree or more –0.083*** 0.027 –0.077*** 0.026 –0.077*** 0.028 –0.071** 0.028 Marital status reference category: single [never married Married 0.010 0.017 0.008 0.017 0.013 0.018 0.013 0.018 Widowed –0.011 0.017 –0.011 0.018 –0.005 0.017 –0.005 0.018 Divorced –0.010 0.017 –0.010 0.017 –0.009 0.017 –0.008 0.017 Separated 0.018 0.030 0.015 0.031 0.015 0.030 0.013 0.031 Religion reference category: atheist, agnostic, or none Jewish 0.322*** 0.044 0.323*** 0.042 0.310*** 0.045 0.315*** 0.043 Christian –0.035 0.035 –0.036 0.036 –0.033 0.036 –0.033 0.036 Muslim –0.025 0.056 –0.025 0.056 –0.028 0.055 –0.027 0.055 Other –0.095 0.089 –0.096 0.097 –0.090 0.081 –0.091 0.088 Other variables Household size 0.016*** 0.005 0.017*** 0.005 0.016*** 0.005 0.017*** 0.005 Share of children 0.022 0.029 0.017 0.030 0.021 0.030 0.017 0.030 Urban –0.021 0.019 –0.019 0.019 –0.023 0.019 –0.020 0.019 Monthly income 0.066 0.047 0.062 0.045 0.056 0.046 0.052 0.045 Number of observations 22,000 21,569 21,876 21,461 Note: Marginal effects of probit estimations are reported in this table. Marginal effects on 38 county dummies are omitted. Standard errors are clustered at the country level. *** Significant at the 1 percent level, ** significant at the 5 percent level, * significant at the 10 percent level. ♣ age and age-squared are jointly significant at the 1 percent level. 24 Table 4 Willingness to pay for investments in education in five model specifications. Probit marginal effects (ME). Probit (marginal effects) 2SLS (second stage) Specification 1 Specification 2 Specification 3 Specification 4 Specification 5 Variable ME Std. err. ME Std. err. ME. Std. err. ME. Std. err. ME. Std. err. Expected mobility (reference category: static) Downward –0.028** 0.014 Upward 0.042*** 0.012 Upward trend in own mobility 0.052*** 0.011 Children will live better than parents (yes/no) 0.040** 0.019 Expected own mobility 0.096** 0.043 0.103** 0.050 Belief in fairness 0.086*** 0.017 0.088*** 0.016 0.096*** 0.018 0.070*** 0.008 0.070*** 0.009 Risk aversion –0.018*** 0.003 –0.019*** 0.003 –0.015*** 0.004 –0.014*** 0.002 –0.013*** 0.002 Gender and age Female 0.003 0.008 0.004 0.008 0.012 0.013 –0.003 0.007 –0.003 0.007 Age/100 –0.144 0.193 –0.053 0.191 0.799* 0.441 0.413** 0.190 0.432** 0.204 Age squared –0.013 0.198 –0.111 0.195 –0.990** 0.493 –0.380** 0.151 –0.389** 0.155 Education level (reference category: master’s degree or PhD) No degree/no education –0.220*** 0.057 –0.217*** 0.054 –0.161** 0.075 –0.244*** 0.031 –0.246*** 0.032 Primary education –0.215*** 0.033 –0.211*** 0.032 –0.186*** 0.040 –0.214*** 0.024 –0.216*** 0.024 Lower-secondary education –0.240*** 0.033 –0.239*** 0.032 –0.149*** 0.039 –0.226*** 0.018 –0.226*** 0.018 Upper-secondary education –0.172*** 0.026 –0.171*** 0.026 –0.107*** 0.033 –0.168*** 0.016 –0.168*** 0.016 Bachelor’s degree or more –0.081*** 0.026 –0.077*** 0.026 –0.039 0.034 –0.138*** 0.020 –0.138*** 0.020 Marital status reference category: Single (never married) Married 0.017 0.017 0.015 0.017 –0.002 0.034 0.033*** 0.012 0.034*** 0.013 Widowed –0.003 0.018 –0.002 0.017 0.088** 0.042 0.009 0.015 0.009 0.015 Divorced –0.005 0.018 –0.002 0.018 –0.047 0.036 0.007 0.015 0.007 0.015 Separated 0.009 0.033 0.022 0.032 –0.033 0.042 0.027 0.030 0.028 0.030 Religion reference category: atheist, agnostic, or none Buddhist –0.019 0.019 –0.011 0.023 0.019 0.014 0.005 0.036 0.006 0.037 Jewish 0.334*** 0.046 0.339*** 0.041 0.366*** 0.135 0.369*** 0.136 Christian –0.020 0.034 –0.022 0.035 –0.059* 0.034 –0.028** 0.014 –0.029** 0.014 Muslim –0.018 0.051 –0.020 0.055 –0.077 0.071 –0.058** 0.024 –0.060** 0.025 Other –0.011 0.051 –0.085 0.080 –0.179 0.120 –0.002 0.035 –0.002 0.035 Other variables Household size 0.017*** 0.005 0.017*** 0.005 0.019** 0.008 0.015*** 0.003 0.015*** 0.003 Share of children in household 0.020 0.031 0.017 0.030 –0.119** 0.055 0.014 0.022 0.012 0.023 Urban –0.024 0.020 –0.021 0.020 –0.017 0.026 –0.026*** 0.008 –0.026*** 0.008 Monthly income (thousand USD at PPP 2017) 0.049 0.044 0.049 0.045 0.012 0.037 0.039*** 0.013 0.039*** 0.013 Sargan (score) chi2 0.001 (p = 0.9704) 0.608 (p = 0.4355) Cragg and Donald F(2; 19,438) 28.635 (p = 0.0000) 21.287 (p = 0.0000) Number of observations 20,704 22,095 6,216 19,495 19,495 Note: Marginal effects in specifications 1, 2, and 3 are calculated at the means of independent variables and controls. Marginal effects on 30 county dummies in specifications 1, 2, and 3 are omitted. Standard errors are clustered at the country level. *** Significant at the 1 percent level, ** significant at the 5 percent level, * significant at the 10 percent level. ♣ Age and age-squared are jointly significant at the 1 percent level. Table 5 Intertemporal comparisons of willingness to pay for investments in education as a function of expectations about intra- and intergenerational mobility and respondent characteristic. Probit marginal effect (ME). 2010 2016 2023 2010 2016 2023 Variable ME Std. err. ME Std. err. ME. Std. err. ME Std. err. ME. Std. err. ME Std. err. Expected mobility 0.007** 0.003 0.014*** 0.004 0.013*** 0.004 Children will live better (yes/no) 0.054*** 0.016 0.073*** 0.014 0.041*** 0.012 Belief in fairness 0.067*** 0.019 0.108*** 0.011 0.083*** 0.015 0.062*** 0.018 0.106*** 0.010 0.088*** 0.015 Risk aversion –0.014*** 0.002 –0.019*** 0.002 –0.018*** 0.003 0.016*** 0.002 –0.019*** 0.002 –0.018*** 0.003 Gender and age Female 0.013 0.014 –0.002 0.008 0.003 0.008 0.013 0.013 –0.005 0.008 0.005 0.009 Age/100 0.227♣ 0.144 0.054♣ 0.162 –0.217♣ 0.196 –0.314** 0.139 0.085♣ 0.143 –0.245♣ 0.199 Age squared –0.300** 0.145 –0.199 0.172 –0.036 0.192 –0.382*** 0.140 –0.253 0.156 0.037 0.193 Education (reference category: master’s degree or PhD) No degree/no education –0.247*** 0.034 –0.235*** 0.032 –0.234*** 0.058 0.255*** 0.034 –0.232*** 0.028 –0.234*** 0.052 Primary education –0.169*** 0.035 –0.205*** 0.024 –0.239*** 0.033 –0.172*** 0.034 –0.194*** 0.024 –0.230*** 0.032 Lower-secondary education –0.183*** 0.030 –0.195*** 0.025 –0.257*** 0.032 –0.178*** 0.026 –0.192*** 0.023 –0.250*** 0.034 Upper-secondary education –0.131*** 0.027 –0.129*** 0.020 –0.186 *** 0.026 –0.132*** 0.024 –0.123*** 0.019 –0.178*** 0.027 Tertiary (non-university) –0.106*** 0.029 –0.117*** 0.020 –0.179 *** 0.030 –0.111*** 0.025 –0.112*** 0.018 –0.165*** 0.029 Bachelor’s degree or more –0.065*** 0.024 –0.112 *** 0.033 –0.064*** 0.022 –0.111*** 0.036 Marital status (reference category: single [never married]) Married 0.020* 0.011 0.031* 0.016 0.023 0.016 0.012 0.012 0.026* 0.015 0.023 0.016 Widowed –0.033* 0.017 –0.000 0.020 0.002 0.015 –0.035** 0.017 –0.003 0.019 0.001 0.015 Divorced –0.004 0.015 –0.029 0.018 –0.002 0.019 –0.011 0.018 –0.027 0.017 –0.003 0.018 Separated –0.047* 0.026 0.019 0.025 0.011 0.033 –0.028 0.026 0.015 0.024 0.016 0.028 Religion (reference category: atheist, agnostic, or none) Buddhist –0.012 0.042 0.118*** 0.015 –0.000 0.021 0.008 0.051 0.116*** 0.015 0.028 0.021 Jewish 0.100 0.091 –0.041 0.073 0.294*** 0.047 0.115 0.095 –0.011 0.072 0.273*** 0.040 Christian –0.008 0.027 –0.022 0.020 –0.030 0.031 –0.011 0.027 –0.021 0.020 –0.039 0.031 Muslim –0.050 0.038 0.013 0.046 –0.011 0.047 –0.048 0.036 0.018 0.049 –0.009 0.048 Other 0.009 0.038 –0.012 0.028 0.009 0.048 0.023 0.036 –0.005 0.028 0.002 0.034 Other variables Household size 0.021*** 0.005 0.003 0.005 0.015*** 0.005 0.020*** 0.005 0.004 0.005 0.015*** 0.005 Share of children in household 0.025 0.029 0.031 0.028 0.013 0.030 0.043 0.030 0.032 0.027 0.019 0.028 Urban 0.031 0.022 –0.016 0.013 –0.022 0.020 0.023 0.021 -0.023* 0.012 -0.021 0.019 Monthly expenditure (000s USD) 0.000*** 0.000 0.550*** 0.147 0.084 0.082 0.000*** 0.000 0.551*** 0.129 0.072 0.081 Number of observations 15,135 33,459 22,415 15,732 35,413 23,387 Note: Marginal effects of probit estimations are reported in this table. Marginal effects for 30 county dummies are omitted. Standard errors are clustered at the country level. *** Significant at the 1 percent level, ** significant at the 5 percent level, * significant at the 10 percent level. ♣ Age and age-squared are jointly significant at the 1 percent level. 26