Personnel Politics: Elections, Clientelistic Competition, and Teacher Hiring in Indonesia Jan H. Pierskalla and Audrey Sacks Department of Political Science, The Ohio State University GPSURR, World Bank June 25, 2018 Pierskalla and Sacks Personnel Politics 1 / 21
Motivation Source: Jakarta Post Pierskalla and Sacks Personnel Politics 2 / 21
Motivation 20.0 Realized Exp % of Total 17.5 15.0 12.5 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Year Source: World Bank 2013 Pierskalla and Sacks Personnel Politics 3 / 21
Motivation Contract PNS 1000000 8% 750000 Number 500000 75% Year 2006 2008 2010 13% 250000 39% 2% 25% 0.7% 60% 0 TK SD SMP SMA TK School Type SD SMP SMA Source: World Bank Teacher Census Pierskalla and Sacks Personnel Politics 4 / 21
Motivation Cambodia Philippines Lao PD Mongolia Myanmar Middle income World Low & middle income Lower middle income Korea, Rep. Vietnam United Kingdom Japan Singapore China Indonesia Thailand High income Malaysia United States Philippines Myanmar Cambodia Lower middle income Low & middle income Middle income Lao PD World Thailand Vietnam Mongolia Korea, Rep. China United States Singapore United Kingdom Malaysia High income Indonesia Japan 0 10 20 30 40 Student Teacher Ratio, Primary School 50 0 10 20 30 Student Teacher Ratio, Secondary School Pierskalla and Sacks Personnel Politics 5 / 21
Motivation Pierskalla and Sacks Personnel Politics 6 / 21
Motivation Failures in the Education Sector Pierskalla and Sacks Personnel Politics 7 / 21
Motivation Failures in the Education Sector - Weak student learning outcomes (e.g., PISA scores) - Inefficient spending, especially on teachers (salaries and certification) - Teacher absenteeism (10-19%) - Rampant cheating in national exams Pierskalla and Sacks Personnel Politics 7 / 21
Motivation (Small) Research Question What is the effect of electoral competition on teacher hiring in Indonesia? Pierskalla and Sacks Personnel Politics 8 / 21
Motivation (Bigger) Research Questions 1 What is the effect of democratization on public goods provision? 2 What role does electoral competition play for bureaucratic quality? 3 What happens to clientelism when competition between elites intensifies? Pierskalla and Sacks Personnel Politics 9 / 21
Argument Elections and the Bureaucracy The positive story: Pierskalla and Sacks Personnel Politics 10 / 21
Argument Elections and the Bureaucracy The positive story: - Voters demand the delivery of high quality public goods and services - Elections discipline politicians in charge of the civil service - Meritocracy in the civil service is essential for effective service delivery Elections ought to improve governance of the civil service Pierskalla and Sacks Personnel Politics 10 / 21
Argument Budget Cycles and Clientelism...but: Pierskalla and Sacks Personnel Politics 11 / 21
Argument Budget Cycles and Clientelism...but: - Elections in a post authoritarian, developing country setting are often different Autocratic elites Low information environment Weak rule of law Low credibility of partisan platforms Clientelism is prevalent (Hanusch & Keefer 2013, 2014) Elections lead to competition between clientelistic elites - Bureaucrats are essential cogs in the clientelistic machine Direct vote and turnout buying via targeted exchange Colonizing the civil service to gain control over discretionary resources Control over the election process - Form of Geddes Politician s Dilemma Pierskalla and Sacks Personnel Politics 11 / 21
Argument Observable Implications - Patronage jobs are an important currency in clientelistic exchanges H1: An increase in electoral competitiveness will increase hiring in the civil service. - Patronage hiring will be particularly pronounced in election years (Hanusch & Keefer 2013, 2014) H2: There will be an increase in civil service hiring and financial rewards in election years. Pierskalla and Sacks Personnel Politics 12 / 21
Empirical Analysis Empirical Strategy - An analysis of direct district elections in Indonesia - Why? 1 Relevant case more 2 (Plausibly) exogenous and staggered phasing-in of elections in 2005 Causal identification more Balance 3 Detailed census data for the education sector (2+ million teachers, 2006, 2008, 2010): - Hiring in the education sector - Certification rates for teachers - Panel data analysis with parallel trends assumption more Pierskalla and Sacks Personnel Politics 13 / 21
Empirical Analysis Direct Election Effect, Hiring Indirect Indirect Election Type Direct Election Type Direct 0 1000 2000 3000 Number of Contract Teachers 4000 0 1000 2000 3000 4000 Number of PNS Teachers 5000 Pierskalla and Sacks Personnel Politics 14 / 21
Empirical Analysis Election Year Effect, Hiring Non Election Non Election Type of Year Type of Year Election Election 0 1000 2000 3000 4000 Number of Contract Teachers 0 1000 2000 3000 Number of PNS Teachers 4000 More Results Pierskalla and Sacks Personnel Politics 15 / 21
Empirical Analysis Election Year Effect, Certification District Level Individual Level Pre Election Year Pre Election Year Election Year Election Year Post Election Year Post Election Year Direct Election Direct Election 3 2 1 0 1 2 3 4 5 6 7 8 Regression Coefficient 0.07 0.04 0.01 0.02 0.05 0.08 0.11 0.14 Regression Coefficient Pierskalla and Sacks Personnel Politics 16 / 21
Empirical Analysis Distortions? - No aggregate improvement in student learning - Contract teachers in election years have lower levels of educational attainment 3.35 vs. 3.41 out of 7, difference of 0.057 significant below the 0.01% level No difference for civil servant teachers - We know that contract teachers have higher absenteeism rates - Contract teachers often pressure for conversion to PNS status huge fiscal implications - Randomized evaluation finds no effect of certification (de Ree et al. 2016) Pierskalla and Sacks Personnel Politics 17 / 21
Empirical Analysis Distortions - RCT designed to estimate the effects of the teacher certification program collected student-level test score data in math, sciences, Indonesian and English language skills for over 80,000 students in 20 districts in 2011 and 2012. - We match these individual-level learning data to our data estimate and find: 1 Districts with more contract teachers score worse on math, sciences, and English language scores (2011 data only) 2 Election years have no average effect but out of 36 district-subject areas with elections, 10 had a statistically significant and negative effect on student learning, while only five had positive effects and the remaining recorded non-significant effects (2011-12 data with student fixed effects). - This suggests that elections and contract teacher hiring is not beneficial and potentially disruptive. Pierskalla and Sacks Personnel Politics 18 / 21
Empirical Analysis Heterogeneous Effects Election Year, Non Golkar Election Year, Golkar 1.0 0.8 0.6 0.4 0.2 0.0 0.2 0.4 0.6 0.8 1.0 Effect Size Pierskalla and Sacks Personnel Politics 19 / 21
Conclusion Conclusion - Clear presence of election-related distortions in the education sector - Hiring follows a political logic - Effect varies with context factors - This matters for service delivery, democratic accountability, democratic consolidation - Clientelism is not only about vote buying, but also building machines inside the bureaucracy - Competitive elections might not lead to the selection of leaders that push for a meritocratic civil service. Pierskalla and Sacks Personnel Politics 20 / 21
Conclusion Open Questions 1 Other context factors that matter? 2 Do parties matter? 3 Does this extend to other parts of the bureaucracy? Pierskalla and Sacks Personnel Politics 21 / 21
Appendix Why the Indonesian Education Sector? - General context fits: Before democratization, centralized form of clientelism After democratization competition between clientelistic elites intensifies District governments gain control over staff and expenditures - Clientelist practices are commonplace in the education sector: District governments manage schools and teachers Teaching positions are used as political rewards Teachers are used as vote canvassers, man polling stations Teachers themselves are an important voting bloc Teachers are centrally embedded in social networks Teachers are rent generators via school fees they act as organizational brokers (Holland & Palmer-Rubin 2015) Back to Research Design Pierskalla and Sacks Personnel Politics 1 / 5
Appendix The Election Schedule - 1999-2004 indirect Candidates only need narrow elite support - 2005- direct (plurality and 30%) Candidates also need to win (some) mass support - (Plausibly) exogenous and staggered phasing-in of elections (Skoufias et al. 2014) - Districts with and without elections are balanced on covariates Back to provides good counterfactuals Research Design Pierskalla and Sacks Personnel Politics 2 / 5
Appendix Balance Statistics Table: Balance Statistics for Elections 2005 Variable Mean Treatment Mean Control T-test p-value KS Bootstrap p-value Golkar Share 0.24 0.25 0.57 0.60 PDI-P Share 0.19 0.19 0.86 0.29 Services Provision 0.23 0.03 0.04 0.06 Natural Resource Revenue pc 52383 113022 0.15 0.008 Inequality 25.6 25.6 0.91 0.62 Total Revenue pc 789360 937316 0.09 0.32 log Population 13 12.85 0.19 0.15 Poverty Share 0.18 0.17 0.64 0.44 log GDP pc 1.58 1.59 0.94 0.54 ELF 0.43 0.41 0.42 0.348 Back to Research Design Pierskalla and Sacks Personnel Politics 3 / 5
Appendix Model Specification y it = α i + γ t + τ D it 1 + δ t 1 E it 1 + δ E it + δ t+1 E it+1 + β x it 1 + ɛ it - y it our outcome measures (teacher data for 2006, 2008, and 2010) - α i and γ t are fixed effects - τ is the effect for the introduction of direct elections - δ s capture election cycles - Controls: incumbency, Golkar and PDI-P vote shares, quality of public services, total revenue pc, nat resource rev pc, Gini index, poverty, GDP pc, population size Back Pierskalla and Sacks Personnel Politics 4 / 5
Appendix Election Year Effect, Hiring, Civil Servant Share Pre Election Year Election Year Post Election Year Direct Election 0.20 0.15 0.10 0.05 0.00 0.05 0.10 0.15 Regression Coefficient Back Pierskalla and Sacks Personnel Politics 5 / 5