Educational inequality and public policy preferences: Evidence from representative survey experiments☆
Introduction
Over the past decades, income and wealth inequality has increased in many industrialized countries (e.g., Piketty and Saez, 2014). The reasons for this trend are manifold, but increasing wage premia for higher education and cognitive skills seem to account for a large share of rising earnings inequality (Autor, 2014). At the same time, there is mounting evidence that factors outside of an individual's control determine educational achievement to a large extent. In particular, family background is a strong predictor of children's educational performance all over the world (e.g., Schuetz et al., 2008; Björklund and Salvanes, 2011; OECD, 2016b). Since educational inequality has important implications for economic inequality and the inequality of opportunity (e.g., Nickell, 2004; Corak, 2013), education policies that attenuate the influence of family background have taken center stage in the political debate. One often given reason for supporting stronger education policies is the belief that equalizing education outcomes will lead to reduced income disparities and greater opportunities for children in poor families.
This paper investigates determinants of public preferences for education policies aimed at fostering equality of opportunity. Traditionally, governments try to mitigate inequalities in income and other economic outcomes through redistribution. Policies aimed at equality of outcomes, such as progressive taxation or minimum wages, might yield economic inefficiencies since they can distort labor supply and human capital accumulation decisions (e.g., Bovenberg and Jacobs, 2005). Arguably, the trade-off between equity and efficiency is less severe for policies aimed at equality of opportunity, which have the goal to detach the opportunity to turn effort into economic success from individual circumstances such as family background.1,2 Consequently, economists have been advocating policies that equalize access to education in order to tackle income inequality (e.g., Alvaredo et al., 2018). But while a large strand of empirical literature has studied the public's preferences for policies aimed at equality of outcomes (e.g., Alesina and La Ferrara, 2005; Alesina and Giuliano, 2011; Kuziemko et al., 2015), the determinants of public preferences for – and thus, the political feasibility of – policies aimed at equality of opportunity are largely unexplored.
We study how the German public's concerns about educational inequality and its preferences for equity-oriented education policies are affected by information about the extent of educational inequality. Given that the public often holds biased beliefs about the extent of inequality in society (e.g., Norton and Ariely, 2011), we focus on how information on actual educational inequality shapes public policy preferences. To this end, we conduct survey experiments among representative samples of the German voting-age population (N = 7380). In the experiments, randomly selected treatment groups are informed about the actual extent of educational inequality before answering questions about concerns about inequality of opportunity in education and preferences for a series of equity-oriented education policies. The provided information refers to the association between parents' socioeconomic status and their children's educational achievement, which has been shown to translate into unequal opportunities later in life (Hanushek and Woessmann, 2008). The control group answers the same questions without receiving information.
We find that a majority of the German public is concerned about the extent of educational inequality and that providing factual information about educational inequality increases these concerns even further. In the uninformed control group, 55.4% view educational inequality as a serious or very serious problem (as opposed to a medium problem or less on a five-point scale). Even from this high baseline level, information provision strongly increases concerns by 12.4 percentage points to 67.8%. The information effect, which we replicate in two independent and representative samples, varies with respondents' prior beliefs about the extent of educational inequality: the treatment has the largest effect on respondents who initially underestimated the extent of educational inequality and decreases with higher belief accuracy. This pattern is particularly pronounced among respondents who are relatively confident that their beliefs are correct, suggesting that the treatment effect is driven primarily by genuine information updating, rather than priming or demand effects. Resurveying respondents in a follow-up survey about two weeks after the experiment, we find that the information effect on respondents' beliefs and concerns about educational inequality persists, further validating an interpretation of genuine information effects.
Going beyond concerns about educational inequality to preferences for equity-oriented education policies, we find that baseline support for many education policies aimed at reducing educational inequality is high. Focusing on policies that target equality of educational opportunity in the sense of preventing disadvantages that result from children's family circumstances (Coleman, 1975), we elicit preferences for eight equity-oriented education policies that had been visible in the German public debate: providing free preschool for children from low-income families, introducing compulsory preschool, increasing government spending for schools with many disadvantaged students, postponing ability tracking, providing bonuses for teachers who teach in schools with many disadvantaged students, introducing whole-day schooling for all students, teaching students with learning disabilities in regular classrooms, and increasing spending on need-based scholarships for disadvantaged university students. Among the control group, six of the eight policies have majority appeal, suggesting that policies aimed at equality of opportunity in the education sector are popular, even when the electorate holds biased beliefs about factual educational inequality.
In contrast to the effects on concerns about educational inequality, however, information treatment effects on these preferences for equity-oriented education policies are small on average. Informing participants about the extent of educational inequality raises a policy index that combines all eight policies by 2.4 percentage points (from a baseline support of 63.0%). While reaching statistical significance for the index (in particular when exploiting the full range of preferences from strong opposition to strong support), information treatment effects on the separate policy proposals are quantitatively small and mostly insignificant. The effect of information provision on support for introducing compulsory preschool is more pronounced, but it cannot be significantly distinguished from the smaller positive treatment effects on other policy items. The pronounced effect on compulsory preschool is also found in a second, independent and representative sample, which makes it unlikely that the effect is a chance finding. Our pattern of results resembles the earlier findings on public preferences for policies aimed at equality of outcomes by Kuziemko et al. (2015) who find that correcting biased beliefs about income inequality through information provision has large effects on concerns about inequality, but only little effect on preferences for redistribution policies in the overall population. However, instrumental-variable estimates that use the information treatment as an instrument for concerns about educational inequality indicate substantial effects of concerns on policy preferences: being induced to consider educational inequality as a serious problem increases support for the education policies by 21.1%age points.
We investigate several possible explanations why the information treatment does not translate into education policy preferences to a larger extent. In a first additional experiment, information that the education policies are meant to reduce educational inequality has no additional effect on respondents' policy support, suggesting that disconnect between concerns and the actual education policies meant to address them does not account for the small treatment effects on policy preferences. In a second additional experiment, support for compulsory preschool increases by 7.2 percentage points when respondents are informed about educational inequality and by 5.1 percentage points when informed about scientific evidence that preschool can effectively reduce educational inequality. Importantly, providing both pieces of information increases support by roughly the sum of the separate effects (12.6 percentage points), suggesting that the effect of information about educational inequality is not mediated by additional information on policy effectiveness. Subgroup analysis reveals that the information treatment does not have differential effects on supporters of the governing parties or on an oversample of teachers (N = 713), suggesting that the extent of (dis)trust in the government or in educational institutions is also unlikely to account for the small information effects on policy preferences. If anything, treatment effects are larger for those respondents who do not prefer public school spending to increase, speaking against a role for aversion against increased government spending in explaining the small treatment effects on policy preferences.
Overall, our results suggest that effects of correcting biased beliefs about the current extent of educational inequality are positive, but are smaller for preferences for education policies than for concerns about educational inequality.
We contribute to two strands of economics research. A large literature studies the determinants of public preferences for redistribution (see Clark and D'Ambrosio, 2015, for a recent overview). Among others, historical experience, culture, prospects of upward mobility, and socioeconomic background have been identified to shape redistributive preferences (e.g., Alesina and La Ferrara, 2005; Alesina and Guiliano, 2011; Luttmer and Singhal, 2011; Roth and Wohlfart, 2018). More recently, several papers have used large-scale survey experiments to investigate whether factual information about the extent of inequality affects preferences for redistribution (e.g., Cruces et al., 2013; Kuziemko et al., 2015; Bublitz, 2016; Karadja et al., 2017). These studies generally investigate policies aimed at equality of outcomes. Our focus on preferences for equity-oriented education policies extends this growing experimental literature to the dimension of policies aimed at equality of opportunity.3 We are aware of only one other experimental and representative study, conducted contemporaneously to and independently of ours, that investigates preferences for policies aimed at equality of opportunity, focusing on beliefs about intergenerational mobility: Alesina et al. (2018a) find that a pessimistic perception treatment on intergenerational mobility tends to increase support for policies aimed at equality of opportunity among left-wing respondents, but not among right-wing respondents. To the best of our knowledge, ours is the first paper that provides causal evidence on how information on factual educational inequality affects public concerns and preferences for various education policies, the very policies aimed at increasing equality of opportunity.4
More generally, our analysis is related to the literature that studies the effects of education policies on educational inequality (for reviews of the literature, see, e.g., Woessmann, 2008; Björklund and Salvanes, 2011). For example, international evidence suggests that the extent of educational inequality is particularly large in Germany, our country of investigation, and that reduced educational inequality is associated with more extensive preschool education and with postponed between-school ability tracking (Schuetz et al., 2008; OECD, 2016a). We add a political-economy dimension to this literature by studying the determinants of the electorate's support for these and other policies that might mitigate educational inequality.
The remainder of the paper is structured as follows. Section 2 introduces the opinion survey and the experimental design. Section 3 presents and discusses the results. Section 4 concludes.
Section snippets
Data and empirical strategy
This section describes the opinion survey, the experiments, and the econometric model.
Results
We present three sets of results. First, we analyze how information on educational inequality affects the public's concerns about the issue. Second, we discuss how the information shapes public support for equity-oriented education policies. Third, we provide analyses of potential explanations for the small average information treatment effect on policy preferences.
Conclusion
Unequal educational opportunity for children from different social backgrounds is a key determinant of persistent economic inequality in society. But in contrast to public preferences for redistribution through policies aimed at equality of outcomes, little is known about the determinants of preferences for equity-oriented education policies. We administered representative survey experiments in Germany, a country with substantial inequality of educational opportunity, to study the public's
References (71)
- et al.
Preferences for redistribution
- et al.
Preferences for redistribution in the land of opportunities
J. Public Econ.
(2005) - et al.
Education and family background: Mechanisms and policies
- et al.
Pre-school, day care, and after-school care: who’s minding the kids?
- et al.
Information provision and preferences for education spending: evidence from representative survey experiments in three countries
Eur. J. Polit. Econ.
(2020) - et al.
Attitudes to income inequality: experimental and survey evidence
- et al.
Biased perceptions of income distribution and preferences for redistribution: evidence from a survey experiment
J. Public Econ.
(2013) - et al.
Interpreting the evidence on life cycle skill formation
- et al.
Does early child care affect children’s development?
J. Public Econ.
(2018) - et al.
Returns to skills around the world: evidence from PIAAC
Eur. Econ. Rev.
(2015)
How information affects support for education spending: evidence from survey experiments in Germany and the United States
J. Public Econ.
The effects of early tracking on student performance: evidence from a school reform in Bavaria
Econ. Educ. Rev.
Equality of opportunity
Experienced inequality and preferences for redistribution
J. Public Econ.
Intergenerational mobility and support for redistribution
Am. Econ. Rev.
Immigration and Redistribution. NBER Working Paper 24733
World Inequality Report 2018
Multiple inference and gender differences in the effects of early intervention: a reevaluation of the Abecedarian, Perry Preschool, and Early Training Projects
J. Am. Stat. Assoc.
Evaluating Post-secondary Aid: Enrollment, Persistence, and Projected Completion Effects. NBER Working Paper 23015
Skills, education, and the rise of earnings inequality among the “other 99 percent”
Science
The contribution of the minimum wage to US wage inequality over three decades: a reassessment
Am. Econ. J. Appl. Econ.
Homer gets a tax cut: inequality and public policy in the American mind
Perspect. Polit.
Incentives and prosocial behavior
Am. Econ. Rev.
Redistribution and education subsidies are Siamese twins
J. Public Econ.
Misperceptions of income distributions: cross-country evidence from a randomized survey experiment
The pluralism of fairness ideals: an experimental approach
Am. Econ. Rev.
Inflation expectations, learning and supermarket prices: evidence from field experiments
Am. Econ. J. Macroecon.
How does your kindergarten classroom affect your earnings? Evidence from project STAR
Q. J. Econ.
The size of the LGBT population and the magnitude of antigay sentiments are substantially underestimated
Manag. Sci.
What is meant by “an equal educational opportunity”?
Oxf. Rev. Educ.
Income inequality, equality of opportunity, and intergenerational mobility
J. Econ. Perspect.
Who benefits from universal child care? Estimating marginal returns to early child care attendance
J. Polit. Econ.
How to get valid answers from survey questions: what we learned from asking about sexual behavior and the measurement of sexuality
Measuring and bounding experimenter demand effects
Am. Econ. Rev.
Does aid matter? Measuring the effect of student aid on college attendance and completion
Am. Econ. Rev.
Cited by (26)
Earnings information and public preferences for university tuition: Evidence from representative experiments
2023, Journal of Public EconomicsCan internet surveys represent the entire population? A practitioners’ analysis
2023, European Journal of Political EconomyHow do people trade off resources between quick and slow learners?
2022, European Economic ReviewCitation Excerpt :The literature has largely focused on the support for public education spending as a whole, or relative to other government spending areas (Bursztyn, 2016; Busemeyer et al., 2018). There are also survey experiments on the effect of information provision on public support for various education policies: Lergetporer et al. (2018a) find that information about existing spending levels reduces support for increased spending levels and teacher salaries, while Lergetporer et al. (2020) study the effect of information about educational inequality on concerns for educational inequality, and the support for equity-oriented education policies to alleviate the inequality. Lergetporer et al. (2020) find that concern for educational inequality, and to some degree the support for equity-oriented policies, increases with information.
“Against all odds” Does awareness of the risk of failure matter for educational choices?
2022, Economics of Education ReviewCitation Excerpt :First, it adds to the literature on educational choices amongst different options or pathways, which studies areas such as choice of college major (e.g., Altonji, Arcidiacono & Maurel, 2016; Wiswall & Zafar, 2015 (Baker et al., 2018)) private versus state school (Anders, Green, Henderson & Henseke, 2020; Green, Anders, Henderson & Henseke, 2017) and very closely related to our study question, the randomized control trial study of choice between vocational and general education as analysed in Goux, Gurgand and Maurin (2015) in France,10 or an experimental study with study counsellors in Germany (Fitzenberger, Hillerich-Sigg & Sprietsma, 2020). More specifically, it contributes to the literature that examines the effects of information provision on educational choices (e.g., Baker et al., 2019; McGuigan, McNally & Wyness, 2016; Peter & Zambre, 2017; Pekkala et al., S. 2020; Hoxby & Turner, 2013; Bettinger, Long, Oreopoulos & Sanbonmatsu, 2012) and the effects of providing information in surveys on preferences for education, such as Hastings, Neilson and Zimmerman (2015) on college and major choice, Bleemer and Zafar (2018) and Lergetporer, Werner and Wößmann (2021), Oreopoulos and Dunn (2013) on wage expectations and the demand for college education, Roth et al. (2021) on the demand for government spending, and finally Cattaneo et al. (2020) and Lergetporer, Schwerdt, West, Wößmann and Werner (2018 & P. 2020) on preferences for education policies. Second, it adds to the literature that exploits the uniqueness of the Swiss case with its different linguistic regions to study the effect of cultural differences on preferences in various domains, such as taxes (Eugster & Parchet, 2019), work attitudes and job search (Eugster et al., 2017), and social insurance (Eugster et al., 2011).
COVID-19 and educational inequality: How school closures affect low- and high-achieving students
2021, European Economic ReviewDoes the education level of refugees affect natives’ attitudes?
2021, European Economic Review
- ☆
For helpful comments, we would like to thank the editor Johannes Spinnewijn, three anonymous referees, Peter Bergman, Elisabeth Bublitz, Jonathan Davis, Paul Hufe, Emmanuel Saez, Stefanie Stantcheva, Joachim Winter, and seminar participants at Harvard, the ECINEQ meeting in Paris, the CESifo education meeting in Munich, the European Society for Population Economics in Glasgow, the German Economic Association in Vienna, its economics of education group in Hannover, and the CRC retreat in Schwanenwerder. We are also most grateful to Franziska Kugler and Elisabeth Grewenig for their help in preparing the surveys. Financial support by the Leibniz Competition (SAW-2014-ifo-2) and the German Science Foundation (CRC TRR 190) is gratefully acknowledged.