What Democracy Does (and Doesn t do) for Basic Services School Fees, School Inputs, and African Elections Robin Harding and David Stasavage New York University May 4, 2012 Robin Harding and David Stasavage New York University () School Fees May 4, 2012 1 / 14
Question and Motivation What is the effect of democracy on basic service provision? obin Harding and David Stasavage New York University () School Fees May 4, 2012 2 / 14
Question and Motivation What is the effect of democracy on basic service provision? Two common approaches when it comes to poor developing countries Robin Harding and David Stasavage New York University () School Fees May 4, 2012 2 / 14
Question and Motivation What is the effect of democracy on basic service provision? Two common approaches when it comes to poor developing countries 1 Some examine broad correlations between democracy and outcomes but sidestep issue of mechanism Robin Harding and David Stasavage New York University () School Fees May 4, 2012 2 / 14
Question and Motivation What is the effect of democracy on basic service provision? Two common approaches when it comes to poor developing countries 1 Some examine broad correlations between democracy and outcomes but sidestep issue of mechanism 2 Others emphasize why democracy would fail to have an effect: information problems, weak capacity, weak parties, voting dominated by patronage and ethnicity Robin Harding and David Stasavage New York University () School Fees May 4, 2012 2 / 14
Question and Motivation What is the effect of democracy on basic service provision? Two common approaches when it comes to poor developing countries 1 Some examine broad correlations between democracy and outcomes but sidestep issue of mechanism 2 Others emphasize why democracy would fail to have an effect: information problems, weak capacity, weak parties, voting dominated by patronage and ethnicity Our objective - develop and test a new argument suggesting what democracy will and will not do in the context of a poor developing country Robin Harding and David Stasavage New York University () School Fees May 4, 2012 2 / 14
Question and Motivation What is the effect of democracy on basic service provision? Two common approaches when it comes to poor developing countries 1 Some examine broad correlations between democracy and outcomes but sidestep issue of mechanism 2 Others emphasize why democracy would fail to have an effect: information problems, weak capacity, weak parties, voting dominated by patronage and ethnicity Our objective - develop and test a new argument suggesting what democracy will and will not do in the context of a poor developing country Ties in with normative economic debate on user fees Robin Harding and David Stasavage New York University () School Fees May 4, 2012 2 / 14
A Problem of Attribution We know that ability to hold elected offi cials to account depends upon having information both about outcomes and actions obin Harding and David Stasavage New York University () School Fees May 4, 2012 3 / 14
A Problem of Attribution We know that ability to hold elected offi cials to account depends upon having information both about outcomes and actions Substantial literature on variation in levels of transparency (Besley and Burgess 2003, Stromberg 2004, Keefer and Khemani 2003, Reinikka and Svensson 2004) obin Harding and David Stasavage New York University () School Fees May 4, 2012 3 / 14
A Problem of Attribution We know that ability to hold elected offi cials to account depends upon having information both about outcomes and actions Substantial literature on variation in levels of transparency (Besley and Burgess 2003, Stromberg 2004, Keefer and Khemani 2003, Reinikka and Svensson 2004) Less effort to examine whether poor information environments might prompt politicians to emphasize some policies over others. obin Harding and David Stasavage New York University () School Fees May 4, 2012 3 / 14
A Problem of Attribution We know that ability to hold elected offi cials to account depends upon having information both about outcomes and actions Substantial literature on variation in levels of transparency (Besley and Burgess 2003, Stromberg 2004, Keefer and Khemani 2003, Reinikka and Svensson 2004) Less effort to examine whether poor information environments might prompt politicians to emphasize some policies over others. Our argument - elected offi cials likely to bias actions towards policies where their actions are observable obin Harding and David Stasavage New York University () School Fees May 4, 2012 3 / 14
Implications for African Primary Schooling Abolishing fees - A highly visible action where members of the public find it relatively easy to attribute the outcome to presidential action. obin Harding and David Stasavage New York University () School Fees May 4, 2012 4 / 14
Implications for African Primary Schooling Abolishing fees - A highly visible action where members of the public find it relatively easy to attribute the outcome to presidential action. Increased provision of inputs (schools, classrooms, teachers, materials) - Outcomes are visible but attribution for outcomes is more diffi cult. obin Harding and David Stasavage New York University () School Fees May 4, 2012 4 / 14
Three Step Empirical Analysis 1 Use DHS data to examine correlates of school attendance. obin Harding and David Stasavage New York University () School Fees May 4, 2012 5 / 14
Three Step Empirical Analysis 1 Use DHS data to examine correlates of school attendance. 2 Employ new data set to examine effect of elections on school fee abolition. obin Harding and David Stasavage New York University () School Fees May 4, 2012 5 / 14
Three Step Empirical Analysis 1 Use DHS data to examine correlates of school attendance. 2 Employ new data set to examine effect of elections on school fee abolition. 3 Examine survey evidence from Kenya to investigate where voters assign responsibility. Robin Harding and David Stasavage New York University () School Fees May 4, 2012 5 / 14
Correlates of Education Outcomes - School Attendance Individual level data from 29 countries (all born since 1980). Estimate probability of ever having attended primary school. empirical strategy as in Kudamatsu (2012). Same S ihc = α + β 1 Multiparty ihc +β 2 Free ihc +γx ihc + µ h + ε ihc (1) Original data set for tuition fee abolition in Africa since 1990. Democracy Measure based on Executive Index of Electoral Competition (Beck et al. 2001). Robin Harding and David Stasavage New York University () School Fees May 4, 2012 6 / 14
Results for School Attendance (Linear Probability Model) (1) (2) (3) (4) (5) (6) (7) (8) All individuals Estimates by Wealth Quintile 1st 2nd 3rd 4th 5th Multiparty Democracy.040.029.010 -.006.001 -.019 -.023 (.021) (.022) (.027) (.028) (.026) (.027) (.016) Free schooling.055.042.023.056.049.057.008 (.021) (.021) (.027) (.028) (.027) (.026) (.017) Female -.060 -.060 -.060 -.065 -.064 -.060 -.055 -.052 (.004) (.004) (.004) (.006) (.005) (.004) (.004) (.004) Household fixed effects yes yes yes yes yes yes yes yes Birth year dummies yes yes yes yes yes yes yes yes Birth order dummies yes yes yes yes yes yes yes yes R 2 (within) 0.07 0.07 0.07 0.10 0.10 0.09 0.08 0.04 N= 522,914 522,914 522,914 90,331 87,094 92,713 96,018 110,063 Estimates of Probability of Individual Having Any Schooling - Linear probability model with standard errors clustered at the country-cohort level. Robin Harding and David Stasavage New York University () School Fees May 4, 2012 7 / 14
Correlates of School Inputs - Pupil Teacher Ratios Country-year data from WDI - 38 countries 1990-2007 OLS estimates of pupil-teacher ratios pupils teachers it = α + β 1 multiparty it+β 2 fees it + µ i + η t + ε it (2) Look at both enrolled and potential pupils Robin Harding and David Stasavage New York University () School Fees May 4, 2012 8 / 14
Results for Pupil Teacher Ratios (OLS Estimates) (1) (2) (3) (4) (5) (6) Enrolled Pupils Potential Pupils Multiparty Democracy 4.40 3.14 0.80-1.28 (2.16) (1.96) (14.4) (14.5) School Fees in place -8.14-7.85-24.7-22.7 (2.52) (2.70) (12.4) (12.3) Year fixed effects yes yes yes yes yes yes Country fixed effects yes yes yes yes yes yes R 2 (within).086.167.182.382.432.413 N= 341 364 330 329 351 318 OLS Estimates of Pupil Teacher Ratios with standard errors clustered at the country level. Robin Harding and David Stasavage New York University () School Fees May 4, 2012 9 / 14
When do Governments Abolish School Fees? Use newly collected data on school fee abolitions (since 1990) Examine whether this happens in immediate proximity to elections Pr(abolition) it = F (α + β 1 election it + µ i + η t + ε it ) (3) Instrument for election with scheduled election Robin Harding and David Stasavage New York University () School Fees May 4, 2012 10 / 14
Results for School Fee Abolitions (Linear Probability Model) 1 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Presidential election variable Linear Probability Model Instrumental Variables (2SLS) Election (any).045.052.075.080 (.019) (.020) (.023) (.026) Free and fair election.072.076.126.130 (.030) (.032) (.042) (.044) Election & winner share<80%.043.050.083.089 (.020).021 (.026) (.028) Year fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Country fixed effects no no no yes yes yes no no no yes yes yes R 2.051.064.050.135.149.132.045.048.039.130.136.123 First stage statistics F statistic on excluded instrument 243 80 217 210 67 191 Partial R 2 excluded instrument.574.362.505.579.366.510 N= 619 577 619 619 577 600 558 600 600 558 600 Estimates of Likelihood School Fees Will be Abolished. Dependent variable takes a value of 0 for all years with fees, 1 in year of abolition, and the country is dropped from data set for subsequent years. Standard errors clustered at the country level. In 2SLS estimates the excluded instrument is scheduled election. Robin Harding and David Stasavage New York University () School Fees May 4, 2012 11 / 14
Correlates of Voting Intentions Public opinion data from Kenya - estimate probability of intention to vote for incumbent Pr(kibaki = 1) ij = Φ(α + β 1 free ij + β 2 school problem ij +µ j + ε ij ) (4) Are those favorable to free schooling more likely to support Kibaki? Are those experiencing problems with local schools less likely to support Kibaki? Robin Harding and David Stasavage New York University () School Fees May 4, 2012 12 / 14
Results for Voting Intentions (1) (2) (3) (4) (5) (6) (7) Respondents prefer free schools.276.303.300.316 (.112) (.149) (.154) (.152) Experienced overcrowded classrooms -.065 -.073 (.070) (.069) Experienced poor classroom facilities -.089 -.075 (.082) (.083) Experienced problems with textbooks -.073 -.078 (.057) (.063) N= 936 554 549 559 524 520 529 Probit Estimates of the Probability of Expressing Intention to Vote for President Kibaki. All specifications include district level fixed effects, and standard errors are clustered at the district level. Robin Harding and David Stasavage New York University () School Fees May 4, 2012 13 / 14
Conclusion Democracy may have the biggest impact on policy actions where attribution is less of a problem Implications for access versus quality Ambiguous welfare implications Robin Harding and David Stasavage New York University () School Fees May 4, 2012 14 / 14