Do Banking Crises Improve Democracy Beni KoueviGath Universit

Do Banking Crises Improve Democracy? Beni Kouevi-Gath Université Libre de Bruxelles Pierre-Guillaume Méon Université Libre de Bruxelles Laurent Weill Université de Strasbourg ASSA 2019: Do Banking Crises Improve Democracy?

Motivation • Banking crises have: v Real consequences on the economy Ø Ø Ø v Ivashina and Scharfstein (2010): weaken bank lending Chodorow – Reich (2014): decrease Employment Hoggarth, Reis and Saporta (2002): Output losses But also political effects Ø Ø Ø Gries and Meierrieks (2013): Terrorism activity Gutmann, Pfaff and Voigt (2017): Deterioration of human rights Andersson (2016): Institutional reforms • … On Democracy? ? => First Attempt ASSA 2019: Do Banking Crises Improve Democracy?

Motivation: Why? • The Effect of BC on democracy may be Positive : because they open a widow of opportunity to contest the power by reducing the cost of fighting autocratic regimes - Acemoglu and Robinson (2001), Burke and Leigh (2010), Brückner and Ciccone (2011) Ø Negative : - Painful policies which are difficult to implement in democracies - Generate extremist political opinions because BC are perceived as policy failures: Schularick and Trebesch (2016) - Incentives to enhance repression to intimidate citizens: Gutmann and Pfaff (2013) Ø ASSA 2019: Do Banking Crises Improve Democracy?

BC and Democracy: An Empirical Matter Event Study Method • Event: Start of a banking crisis (BC) • Before/After the start of a BC ASSA 2019: Do Banking Crises Improve Democracy?

Method: Event Study • ASSA 2019: Do Banking Crises Improve Democracy?

Data • Democracy: Polity 2 variable from Polity IV database (Marshall, Gurr and Jaggers (2016)) v Code political regimes • BC: from Laeven and Valencia (2012) v Identify Systemic Banking Crises based on 2 criteria: Ø The financial and real sectors experience a financial distress Ø Significant policy measures are put in place in order to limit the real consequences of the banking crisis v Reinhart and Rogoff (2009) ASSA 2019: Do Banking Crises Improve Democracy?

Data • 129 countries, from 1975 to 2010 v v v 79 countries have experienced at least 1 BC 15 countries have experienced more than 2 BC (estimation with the whole sample and with a sample that excludes them) Total number of BC: 94 episodes ASSA 2019: Do Banking Crises Improve Democracy?

Results: Baseline Results Table 1: Banking crises and Democracy Dependent Variable: Democracy Before/After Whole Sample Multiple crises countries excluded Subprime crisis period excluded 0. 874 0. 871 0. 873 (74. 69)*** (66. 75)*** (73. 21)*** 0. 196 0. 209 0. 219 (2. 130)** (2. 083)** (2. 223)** -0. 431 -0. 311 -0. 424 (-2. 234)** (-1. 900)* (-2. 208)** 4, 132 3, 618 3, 800 Number of countries 129 114 128 Adjusted R-squared 0. 845 0. 837 0. 835 Country FE YES YES Year FE YES YES Demo (t-1) BC 10 years after the start Constant Observations ASSA 2019: Do Banking Crises Improve Democracy?

Results: Robustness Checks • Sample • Controls: Other Income Shocks (Large Income Shocks, Sovereign Debt crises), Regional Waves of democracy • Alternative Measure of Democracy: Varieties of democracy (V-Dem) • Alternative Estimation Technics: Dynanic Panel Data, Instrumental Variable (Capital flows bonanzas), Treatment Effects Models • Overall, our baseline results are confirms: democracy improves in the 10 years that follow the start of banking crises ASSA 2019: Do Banking Crises Improve Democracy?

Results: Extensions 1. Do all banking crises affect democracy in the same way? • We code “Moderate” and “Severe” banking crises separately. • A BC is Severe: If Duration ≥ 3 years : Banking crises that last a long time have more consequences on democracy. => Banking crises that last more than 3 years improve democracy while the others do not. Ø Ø If NPLs ≥ 40% : Both “severe” and “moderate” banking crises improve democracy, but in a similar way. => Both types of banking affect democracy in the same way. ASSA 2019: Do Banking Crises Improve Democracy?

Results: Extensions 2. Are all countries affected in the same way? We include an interaction between lagged-democracy and BC dummy in our baseline equation. The estimated coefficient of this variable is negative and statistically significant: the higher the level of democracy, the lower the impact of BC on democracy. ASSA 2019: Do Banking Crises Improve Democracy?

Extention 3: Timing of the effect • ASSA 2019: Do Banking Crises Improve Democracy?

Results: Timing of the impact Dependent Variable : Democracy Demo (t-1) BC 3 years before the start BC 0 to 2 years after the start BC 3 to 5 years after the start BC 6 to 8 years after the start BC 9 to 12 years after the start Constant Observations Number of countries Adjusted R-squared Country FE Year FE Whole Sample Periods around the start of a banking crisis Multiple crises countries excluded Subrpime crisis period excluded 0. 873 (72. 71)*** 0. 0331 (0. 214) 0. 0522 (0. 332) 0. 355 (2. 224)** 0. 265 (2. 374)** 0. 197 (2. 220)** -0. 431 (-2. 240)** 4, 132 129 0. 845 YES ASSA 2019: Do Banking Crises Improve Democracy? 0. 869 (65. 15)*** -0. 0235 (-0. 127) 0. 0831 (0. 405) 0. 341 (2. 021)** 0. 303 (2. 481)** 0. 231 (2. 113)** -0. 311 (-1. 904)* 3, 618 114 0. 837 YES 0. 872 (70. 80)*** 0. 0430 (0. 274) 0. 0716 (0. 439) 0. 380 (2. 278)** 0. 295 (2. 475)** 0. 222 (2. 203)** -0. 423 (-2. 218)** 3, 800 128 0. 836 YES

Conclusion • Banking crises leads to more democracy • This effect is independent of other income shocks • Banking crises that last a long time have more consequences on democracy • The bulk of the impact materializes from the 3 rd year after the end of the crisis. Thank you for your attention. ASSA 2019: Do Banking Crises Improve Democracy?