Who s to Blame? Punishing Poor Economic Performance in a Centralized Political System Quintin H. Beazer 1 Ora John Reuter 2 1 Florida State University qbeazer@fsu.edu 2 University of Wisconsin - Milwaukee Higher School of Economics - Moscow reutero@uwm.edu Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 1 / 23
Political Centralization s Poster Boy Under Putin, Russia s central govt. has increased control over subnational politics Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 2 / 23
Political Centralization & its Consequences Centralization benefits leaders by giving extra control. This is exactly what non-democratic leaders want. Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 3 / 23
Political Centralization & its Consequences Centralization benefits leaders by giving extra control. This is exactly what non-democratic leaders want. But what are the political costs of centralization? Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 3 / 23
Political Centralization & Blame Attribution important questions for performance-dependent regimes How does political centralization affect responsibility for local outcomes? In a centralized system, who gets blamed when local economic performance suffers? Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 4 / 23
Economic Voting in Centralized Systems a theoretical starting point Voters punish elected officials for poor economic performance. Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 5 / 23
Economic Voting in Centralized Systems a theoretical starting point Voters punish elected officials for poor economic performance. Much economic voting research asks: What do voters do when unclear which elected official to punish? ex: literature on clarity of responsibility. Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 5 / 23
Economic Voting in Centralized Systems a theoretical starting point Voters punish elected officials for poor economic performance. Much economic voting research asks: What do voters do when unclear which elected official to punish? ex: literature on clarity of responsibility. In Russia s case, we ask instead: What do voters do when can t punish local officials through elections? Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 5 / 23
Centralization & Blame Attribution in Non-Democracies competing expectations 1.) Institutions under authoritarianism are window dressing. Formal institutional changes make little difference. Responsibility is clear and obvious (...dictator!). Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 6 / 23
Centralization & Blame Attribution in Non-Democracies competing expectations 2.) Institutions under authoritarianism still affect behavior. Formal institutions shape how citizens relate to their leaders. Institutional changes can shift responsibility for outcomes. Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 7 / 23
Centralization & Blame Attribution in Non-Democracies competing expectations 2.) Institutions under authoritarianism still affect behavior. Formal institutions shape how citizens relate to their leaders. Institutional changes can shift responsibility for outcomes. This view implies that, under centralization, voters punish those who appointed poorly-performing local leaders. Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 7 / 23
Empirical Predictions Appointed local officials voters hold higher-level officials responsible for poor local economy Elected local officials voters do not hold higher-level officials responsible for poor local economy Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 8 / 23
Subnational Variation in Elected vs. Appointed Local Execs A research design based on Russian cities Studies comparing appointed vs elected systems have empirical challenges: Unobserved cross-national differences may confound results. National electoral institutions do not vary much over time. Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 9 / 23
Subnational Variation in Elected vs. Appointed Local Execs A research design based on Russian cities Studies comparing appointed vs elected systems have empirical challenges: Unobserved cross-national differences may confound results. National electoral institutions do not vary much over time. Within Russia, cities vary in whether or not citizens elect their chief executive. Variation over time and across cities (even in same region). Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 9 / 23
Russian Municipalities Data Data on roughly 200 of Russia s largest cities Dependent variable: vote share for United Russia in legislative elections Independent variables: centralization: appointed/elected mayor economic performance: change in unemployment centralization economic performance Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 10 / 23
Russian Municipalities Data Data on roughly 200 of Russia s largest cities Control variables: press freedoms, regional political climate, working age population, average income, past UR electoral success Empirical strategy: generalized diff-in-diff Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 11 / 23
Regional Elections: UR Punished More for Poor Economic Performance in Centralized Cities DV: United Russia Vote Share % of vote in regional legislative elections (1) (2) Political Centralization 0.016 0.324 dummy; 1 = appointed mayor (2.158) (2.136) 0.994 0.880 Unemployment 0.884 1.601 annual change in unemp. rate (0.615) (0.561) 0.152 0.005 Centralization -3.845-4.062 Unemployment (1.570) (1.476) 0.015 0.007 Number of Observations 363 322 All Control Variables No Yes Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 12 / 23
Marginal Effects of Poor Economic Performance Conditional on Centralization 5 Marginal Effect of Unemployment 0 5 10 Elected Appointed Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 13 / 23
Types of Centralization: Appointments vs. Party Affiliation Investigating a rival mechanism DV: United Russia Vote Share Political Centralization -2.487 dummy; 1 = appointed mayor (2.812) 0.378 Unemployment 4.040 annual change in unemp. rate (1.592) 0.012 Centralization -4.217 Unemployment (1.541) 0.007 Centralization -2.842 UR member (1.733) 0.103 Number of Observations 285 All Control Variables Yes Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 14 / 23
National Officials Not Blamed More in Centralized Cities Probing the extent of blame attribution DV: United Russia Vote Share % of vote in Duma elections (1) (2) Political Centralization -6.566-4.032 dummy; 1 = appointed mayor (2.045) (2.267) 0.002 0.077 Unemployment -2.403-1.944 annual change in unemp. rate (0.771) (0.788) 0.002 0.015 Centralization 1.452-0.106 Unemployment (1.333) (1.518) 0.277 0.945 Number of Observations 476 444 All Control Variables No Yes Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 15 / 23
Conclusion Political centralization creates a trade-off for the ruling group. Consolidates responsibility for local performance as well as control over local politics. Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 16 / 23
Conclusion Political centralization creates a trade-off for the ruling group. Consolidates responsibility for local performance as well as control over local politics. There are limits to blame attribution. In the Russian case, we observe: Voters punish officials who actually appoint local leaders (regional) No punishment for officials farther from appointment process (national) Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 16 / 23
Thank you in advance for comments and suggestions. Quintin H. Beazer qbeazer@fsu.edu Ora John Reuter reutero@uwm.edu Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 17 / 23
Centralization of Russian Cities Increasing Over Time Proportion of Cities with Appointed Mayors Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 18 / 23
Estimated Effects of Centralization on UR Vote Share Conditional on Unemployment 20 Marginal Effects of Appointment (dy/dx) 10 0 10 20 3 2 1 0 1 2 3 Unemployment t Unemployment t 1 Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 19 / 23
Evidence Supports Parallel Trends Assumption A placebo test of differences prior to reforms DV: United Russia Vote Share % of vote in regional elections Eventual Centralization -2.653 dummy; 1 = appointed mayor in future (1.751) 0.132 Unemployment 0.987 annual change in unemp. rate (0.807) 0.223 Eventual Centralization -3.655 Unemployment (3.302) 0.270 Number of Observations 238 City Fixed Effects No All Control Variables Yes Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 20 / 23
Differences in Electoral History, by Cities Electoral Institution Electoral History in % vote share Group Means Elected Appointed n 1, n 2 p-value Past Support (regional) 39.68 43.97 (124, 73) 0.04 UR s regional vote share (lagged DV) (1.27) (1.68) Past Support (national) 34.28 43.07 (123, 61) 0.00 UR s vote share in 2003 Duma elections (0.93) (2.10) Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 21 / 23
Differences in Political Openness, by Cities Electoral Institution Political Openness subjective expert measures Group Means Elected Appointed n 1, n 2 p-value Strength of Civil Society (1991-93) 2.87 2.64 (272, 109) 0.01 ordinal; min=1, max=4 (0.05) (0.07) Regional Democracy 16.12 14.03 (282, 113) 0.00 composite score; min=0, max=29 (0.37) (0.62) Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 22 / 23
Results Robust to IV Analysis Using mayor s age as instrument DV: United Russia Vote Share % of vote in regional elections (Reduced) (2SLS) Centralization (IV) -1.852-8.768 dummy; 1 = mayor 47 (1.956) (9.374) 0.345 0.350 Unemployment 4.857 2.193 annual change in unemp. rate (2.278) (1.431) 0.034 0.125 Centralization (IV) -5.880-12.242 Unemployment (2.632) (5.463) 0.027 0.025 Number of Observations 245 245 All Control Variables Yes Yes Beazer & Reuter (FSU & UWM/HSE) Who s to Blame? Feb 2016 23 / 23