Corruption in turbulent times: a response to shocks? Joël Cariolle

1. Corruption in turbulent times: a response to shocks? JoëlCariolle University of Auvergne (UdA/CERDI) Foundation for Researches and Studies on International Development (FERDI)

2. MOTIVATIONS • The contribution of governance quality to output fluctuations is widely documented: • bad governance contributes to domestic fluctuations (Acemoglu et al. 2003; Mobarak, 2005); • good governance contributes to absorb external shocks (Rodrik, 2000; Arin et al, 2011). • Economic shocks are more likely to occur and to persist in countries with low governance quality

3. MOTIVATIONS • However, evidence on the effect of economic fluctuations on governance quality is scarcer and more recent: • Pro-cyclical effect (Voors et al., 2011); • Contra-cyclical effect(Borcan et al., 2012); • Nonlinear effect: depending on informational asymmetries between politicians and voters(Aidt and Dutta, 2008), or the opportunity cost of corrupt acts (Dalgaard and Olsson, 2008). • This paper proposes and tests an analytical framework for the effect of economic instability on corruption, reconciling these contributions

4. ANALYTICAL FRAMEWORK • Micro and macroeconomic literature on the effect of risk (Elberset al., 2007; Bardhan and Udry, 1999) separates: • the ex ante effect of economic instability, resulting from the perception of instability;from • the ex post effect of economic instability, resulting from the experience of instability. • Ex ante and ex post effects of economic instability on corruption?

5. ANALYTICAL FRAMEWORK • The ex ante effect of instability on corruption • When credit and insurance market are imperfect, high perceptions of instability may trigger ex antecorruption strategies aimed at reducing exposure to shocks. • Examples: • influencing procurement processes and winning long-term public contracts (Goldman et al., 2013); • building ex ante political connections to ensure financial support if hardships occur (Faccioet al, 2013); or • obtaining obliging regulations and protections (Grossman and Helpman, 1994; Cassing et al. 1986). • Positive ex ante effect of instability on corruption, resulting from “resource-locking” corruption strategies

6. ANALYTICAL FRAMEWORK • The ex post effect of instability on corruption • The experience of shocks may trigger two opposite ex post corruption responses: • Opportunistic corruption,pro-cyclical,following rises and falls in economic activity (Voors et al., 2010; Dalgaard and Olsson, 2008); • Survival corruption,contra-cyclical, arising from the necessity to compensate incomes losses (Borcanet al., 2012). • The direction of the net ex post effect is a priori uncertain, and depends on the marginal effect of shocks on corruption (Dalgaard and Olsson, 2008)

7. ANALYTICAL FRAMEWORK The ex post effect of instability on corruption Constant marginal effect of shocks Scenario 1: (Net) pro-cyclical effect Scenario 2: (Net) contra-cyclical effect Symmetric responses to shocks

8. ANALYTICAL FRAMEWORK The ex post effect of instability on corruption Non constant marginal effect of shocks Scenario 3:Symmetric deterrent effect of shocks Scenario 4: Symmetric positive effect of shocks Asymmetric responses to positive and negative shocks

9. ANALYTICAL FRAMEWORK • The ex post effect of instability on corruption • Institutional features affectthe marginal effect of shocks through the marginal cost of corrupt acts: • financial markets access  opportunity cost of corrupt acts (Wang and You, 2012); • the quality of democratic institutions probability of detection/sanction of corrupt acts (Ahlin et Pang, 2008; Lederman et al. 2005); and • the intensity of economic fluctuations (Dalgaard and Olson, 2008)  makes previous democratic and financial constraints binding. • Key channels for a nonlinear ex post effect of instability on corruption

10. EMPIRICAL FRAMEWORK • Corruption equation • Corruption = f(ex ante; ex post; controls) • Dynamic panel estimations (FE, sys-GMM): • Corruption perception data (ICRG): 1125 observations from 62 developed and developing countries. • Cross-section estimations (OLS): • data on bribery incidence (WBES): 22,000 firms’ aggregated bribe reports for 38 developing countries • Controls: government size, human capital, democracy, political regime durability, population size, natural resource endowments, openness, firms’ characteristics (cross section estimations)

11. EMPIRICAL FRAMEWORK • Variables of interest: instabilities in export volume • Major, universal and exogenous source of instability in both developed and developing countries (Bevan et al. 1993; Guillaumont, 2010): • Perception of instability (ex ante effect): rolling standard deviation of exports(in % of a rolling mixed trend), calculated over a long period (t; t-15) • Experience of instability (ex post effect): rolling skewness of exports (in % of the same mixed trend), calculated over a short period (t; t-5): • reflects the export distribution asymmetry  predominance of positive shocks if >0,negative shocks if <0 • reflects the intensity of fluctuations (Rancièreet al, 2008 QJE)  strong correlation with the kurtosis of a distribution at high absolute values

12. RESULTS Evidence on the ex post effect of instability Main result The ex post effect of instability is nonlinear, depending on institutional and financial constraints underlyingthe marginal effect of shocks on corruption

13. RESULTS Evidence on the ex post effect of instability 1.1. the intensity of fluctuations channel Corruption = f(stddev; skew>0, skew<0; [skew>0]2, [skew<0]2)

14. RESULTS • Evidence on the ex post effect of instability • Estimates of 1.1support that: • both positive and negative shocks reduce corruptionwhenfluctuations are normal (high frequency, low size) • both positive and negative shocks increase corruption whenfluctuations are intense(low frequency, large size)

15. RESULTS

16. RESULTS Evidence on the ex post effect of instability 1.2. The financial and institutional channels Corruption = f( stddev; skew>0, skew<0; [skew>0chan]; [skew<0chan])

17. RESULTS • Evidence on the ex post effect of instability • Estimates of 1.2support that: • both positive and negative shocks reduce corruption when access to credit is facilitated and when democratic institutions are effective • both positive and negative shocks increase corruption when access to credit is restricted andwhen democratic institutions are weak

18. RESULTS Access to creditchannel

19. RESULTS Democracychannel (ICRG)

20. RESULTS Democracychannel (WBES)

21. RESULTS Evidence on the ex ante effect of instability 2. the financial institution channel Corruption = f(stddev; [stddev × financial inst]; skew).

22. RESULTS • Evidence on the ex ante effect of instability • The ex ante effect of instability is also nonlinear, depending on financial market access: • Estimations of model 5 also support a positive ex ante effect of instability on corruption, especially when access to financial markets is restricted.

23. RESULTS

24. CONCLUSION • When economies are unstable and institutions are failing, economic agents are likely to engage in corruption to hedge against adverse fluctuations and to benefit from favorable ones. • As a result, both adverse and favorable shocksmayincrease corruption levels • Improving access to credit markets and supporting pillars of democracy, such as media independence, should dampen the adverse effects of instability on governance quality.

25. Thank you.