Does education increase political participation? Evidence from Indonesia

ABSTRACT

Studies show educated citizens are more likely to vote in elections but few papers look at the relationship in developing countries and even fewer analyze whether the relationship is causal. I examine whether education increases voter turnout and makes better-informed voters in Indonesia using an exogenous variation in education induced by an extension of Indonesia's school term length, which fits a fuzzy regression discontinuity design. The longer school year increases education, but I do not find education increases voter turnout; it does not seem to affect voters’ views of political candidates’ religion, ethnicity, or gender when they vote either.

KEYWORDS:

• Education
• political participation
• regression discontinuity design
• Asia
• Indonesia

JEL CLASSIFICATION:

• D72
• I26

Acknowledgements

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Disclosure statement

No potential conflict of interest was reported by the author.

ORCID

Rasyad A. Parinduri http://orcid.org/0000-0001-7540-8906

Notes

1 In the US, for example, the relationship that voter turnout increases in education holds for white or black Americans, local or national elections, and self-reported or public recoded votes (see Sondheimer and Green (2010) and the papers cited herein).

2 In the literature on voter turnout, education is considered one of the determinants (see, for example, Fumagalli and Narciso [2012]). See Persson (2015) for a review of the literature.

3 Another theory is education increases social status and expands social network in society, which increase political participation (Nie, Junn, and Stehlik-Barry 1996).

4 However, Berinsky and Lenz (2011), using the Vietnam draft as an instrumental variable, do not find education increases voter turnout in the US using compulsory schooling laws in 15 European countries as instrumental variables, Borgonovi, d’Hombresy, and Hoskins (2010) do not find education increases voter turnout either, but it makes better-informed voters. The other papers use compulsory schooling laws, child labor laws, or school expansions as instrumental variables. Papers that use matching strategies also find mixed results (Kam and Palmer 2008; Mayer 2011; Persson 2014).

5 See Tempo (1978), Government of Indonesia (1985), and MPKRI (1978).

6 Inpres stands for Instruksi Presiden or Presidential Instruction. Using this decree, the then President Suharto expanded access to primary schools in Indonesia by building schools across the Indonesian archipelago.

7 See Frankenberg and Thomas (2000), Strauss et al. (2009a), and Strauss et al. (2009b) for the details of the survey.

8 Administratively, Indonesia is divided into provinces; each province is then divided into districts. The heads of provinces are governors; the heads of districts are called bupati or walikota (mayors).

9 DPD stands for Dewan Perwakilan Daerah (Regional Representative Council), the upper house of the People's Consultative Assembly; DPR Dewan Perwakilan Rakyat (People's Representative Council), the lower house. DPRD I and II are the local representative councils at the provincial- and district levels.

10 It was customary for political candidates to give money to voters; the gifts were called ‘transport money’ though voters did not necessarily use it to pay transportation costs to attend political rallies.

11 Zero on the horizontal axis indicates the end of 1971 and beginning of 1972. Each dot in the graphs is a bin-specific average or proportion; the two bars indicate its 95 percent confidence interval. The number of bins is calculated using Calonico, Cattaneo, and Titiunik’s (2015) data-driven algorithm. The trend line is the RD plot—a quadratic polynomial function of the assignment variable (quarter of birth) that may jump at zero. The x-axis is the number of years relative to the cutoff.

12 Nevertheless, the second-stage results of regressions in which I use educational attainment as a measure of education are similar to the basic results: none of the RD estimates are statistically significant.

13 I present the RD plots of only four outcome variables but the trend lines of the others are similar; the graphs are available upon request.

14 The trends lines in the RD plots of other outcome variables are similar.

15 I do not present the estimates on all outcome variables for the sake of brevity; all are statistically insignificant.

16 See Parinduri (2014) for the details of the falsification tests.