The impact of education on political ideology: Evidence from European compulsory education reforms Andrew G. Meyer A presentation by Alessia De Santo and Camilla Fiorina 2 INTRODUCTION ▹ Evidence from previous studies (Dunn, 2011; Weakliem, 2002; Weil, 1985) → correlation between educational attainment and political ideology Results: education pushes individuals to political left. ▹ Problem: endogeneity of education was not taken into account → Omitted Variable Bias 3 1. Which measure can be used to for eudcational attainment? Compulsory education laws Unobserved variables (personality or parents income/ideology) may affect both educational attainment and ideology. Changes in compulsory education laws in 20th century Europe → exogenous shift in educational attainment, not related to ideology. 2. Which indicator for political ideology? Self identified left-right (L-R) placement on a scale Conceptual associations of “left” and “right” is correlated with education and self-placement on the scale” (Bauer et al., 2015). Educational attainment estimated with RDD: compare individuals born just prior/after a reform → conceptual associations similar across these two groups. 4 POLITICAL SCIENCE LITERATURE Self-interest theory: Socialization theory: Higher levels of education → higher incomes → right-wing identity out of self-interest Political attitudes are transmitted to students through social learning processes. Powdthavee and Oswald (2014) Weakliem’s (2002): core values thesis. Development theory: More education → more liberal ideology → more left-wing political identity (Phelan, Link, Stueve, & Moore, 1995). 5 DATA Left-right political ideology ▸ ▸ 25 waves of Eurobarometer surveys from October 2004 to March 2012 with a question about political identification. Face-to-face interviews on ~ 1000 individuals (per wave) from each country Compulsory education laws ▹ Recent papers: Borgonovi et al. (2010), Brunello, Fabbri, and Fort (2013), Gathmann et al. (2015). ▹ 18 reforms (from 1928 to 1983) in 11 Western EU Countries that increases the min education level by 1 to 4 yrs. 6 DATA – The Sample ▹ Individuals affected: born between 1917 and 1969 ▹ Initial sample: 94,694 observation (bandwidth of ±10 years, then restricted) ▸ ▸ Individuals with k ∈ [−10, −1] : pre- reform sample Individuals with k ∈ [1, 10] : post- reform sample. Where 𝑘 = years between a certain cohort and the pivot cohort (first cohort potentially affected → must be excluded from the sample). ▹ Distribution of the data is symmetric, spike around “5” on the L-R scale (centrist placement); on average no significant differences among countries 7 EMPIRICAL STRATEGY - 1 Goal Estimate the impact of educational attainment on political ideology Problem Omitted variable bias ▹ ▹ Unobserved factors may impact at the same time level of education and ideology Education attainment is endogenous Solution Instrumental variable (IV) strategy ▹ Instrument: compulsory education laws (in Europe) 8 IMPACT OF COMPULSORY EDUCATION LAWS ON AVERAGE EDUCATIONAL ATTEINMENT Fig. 1. Education age vs. interval (k), bandwidth of 10 Fig. 2. Education age vs. interval (k), narrowed bandwidths ▹ Evidence of discontinuity in average educational attainment in the year of reform 9 EMPIRICAL STRATEGY – 2 ▹ Fuzzy Regression Discontinuity design using two-equations model specification for individual i from reform c First stage Second stage ▹ The pivot cohort must be excluded → “donut-hole” RD estimator 10 RESULTS ▹ Overall, statistically significant local average treatment effect (LATE) of educational attainment on left-right ideology; ▹ Results are consistent to: ▸ different methods of estimation ▸ different bandwidhts ▸ both linear and quadratic controls Beneficiaries of compulsory education laws are mostly those located in the lower end of the education attainment distribution → results cannot be applied to the general population 11 RESULTS – 2SLS METHODOLOGY ▹ Outcome: continuous 10-point scale ▹ On average, an additional year of educational attainment leads to an approximate 0.25-point movement to the right ▹ ~ 5% movement to the right compared to the sample mean ▹ Similar results obtained for different bandwidths → ± (2,3,4,5,6,8,9) 12 RESULTS – PROBIT METHODOLOGY ▹ Outcome: latent index → dependent variable coded into 3 (IVOP1) and 5 (IVOP2) categories ▹ Baseline results confirmed: an extra year of educational attainment increases the probability of falling into the right category by ~𝟒% 13 WHICH THEORY IS SUPPORTED BY THE DATA? Socialization theory: Self-interest theory: Weak support → it may dampen or intensify the self-interest movement to the right Consistent with the findings ▹ UK, Netherlands, Italy : education reforms increased income → consistently positive estimates from the 2nd stage regression ▹ ▹ Mixed results when grouping countries by welfare regimes (liberal, conservative, social democratic) France, Sweden: education reforms didn’t significantly affect income → mixed results ▹ ▹ But large standard errors But results support the socialization theory in countries where the self-interest theory is supported Development theory: not supported by the data ROBUSTNESS CHECK 14 For all checks: bandwidth at ±7 years ; linear cohort controls. ▹ Sample restictions: ▸ include only countries with strongest evidence of the reforms increasing educational attainment → LATE significant at 5% ; ▸ include pivot cohort → LATE significant at 10% ; ▹ Include survey wave fixed effect; Cluster SE at the reform level; Include regional reforms in Germany → LATE significant at 5% ; ▹ Is a single country the most responsible for the results? No significant evidence. 15 CONCLUSIONS ▹ Evidence: 1 extra year of education → average reform complier moves to the right of the political continuum by approximately 5%; ▹ Causal mechanism goes from education to right ideology through increased earnings (self-interest thesis); ▹ Omitted variables problems is crucial, ignoring it may give opposite and contrasting results; ▹ Caveat: results apply for the specific instrument utilized and for the specific subgroup of individuals affected by the instrument → these results cannot be used to make inference in a different contexts. 16 REFERENCES ▹ Meyer, Andrew G., 2017. "The impact of education on political ideology: Evidence from European compulsory education reforms" Economics of Education Review, Elsevier, vol. 56(C), pages 9-23.