COUNTER-CYCLICAL PROVISIONS, ANAGERIAL DISCRETION AND LOAN GROWTH: THE CASE OF SPAIN by S. Carbó-Valverde and F.

1. COUNTER-CYCLICAL PROVISIONS, ANAGERIAL DISCRETION AND LOAN GROWTH: THE CASE OF SPAINbyS. Carbó-Valverde and F. Rodríguez-Fernández João A.C. Santos Federal Reserve Bank of New York The views expressed here are those of the authors and not necessarily those of the Federal Reserve Bank of New York or the Federal Reserve System.

2. Motivation • There is an ongoing debate on the need to introduce countercyclical capital rules in banking. • A second area of concern is how to mitigate the procyclical effects of the current loan-loss provisions schemes. • The focus of the present paper is on this latter issue.

3. Objective • “to evaluate the effectiveness of the Spanish counter-cyclical provisions system by analyzing the extent to which this scheme has in fact reduced procyclicality, restricted managerial discretion and curbed excessive lending growth.”

4. Hypotheses Procyclicality of loan-loss provisions: Following the implementation of statistical provisions, the expected negative relationship between loan-loss provisions and GDP growth should decrease over time. Managerial discretion: Since loan-loss provisions depend upon the “true” credit risk of loans and not upon net operating income, the significance of any statistical relationship between loan-loss provisions and net operating income should diminish over time. Excessive loan growth: In order for counter-cyclical provisions to be fully effective, they should provide incentives for banks to grant loans more carefully. So, lagged loan growth rates should be unable to explain current loan default rates.

5. Methodology: Testing hypotheses 1 and 2 LLP = a*GDPG +b*NOI + CONTROLS LLP is the ratio of loan-loss provisions to total assets GDPG is the GDP growth NOI is net operating income Hyp 1: If the coefficient a decreases following the implementation of the counter-cyclical provisions, this will contribute to reducing procyclicality, as suggested in hypothesis 1. Hyp 2: Consistent with hypothesis 2, we would expect profit smoothing behavior if the coefficient of NOI is positive and significant.

6. Methodology: Testing hypothesis 3 NPLt = a1*LGt-1 + a2*LGt-2 + …. a1*LGt-12 + CONTROLS NPL is the ratio of non performing loans to total loans LG is the quarterly growth rate of loans Hypo 3: If the high-order lags of the NPL variable are statistically significant and the low-order lags are not, this would suggest that bank managers relax credit quality as the time from the last downturn increases.

7. Results • The coefficient on the GDPG is negative and significant: “There appears to be evidence of procyclicality in loan loss provisioning”. • The coefficient on NOI is positive and significant, which “suggests that banks employ income smoothing”. • The coefficients on LGt-1 and LGt-4 are not significant, but those on LGt-8 and LGt-12 are positive and significant, which “suggests that excessive lending eventually produces poorer credit quality, …., indicating a lack of institutional memory”.

8. Conclusion • “Empirical findings suggest that some of the primary and defining objective of reducing procyclicality have been achieved, via a system of counter-cyclical provisions. However, these provisions do not appear, by themselves, to have effectively reduced earnings management and excessive loan growth.”

9. Comments: Procyclicality • Presumption that all procyclicality is negative, but is that so? • “….This behavior may exacerbate fluctuations in lending cycles. Concretely, in good times an accumulation of potential risk (expected losses) is built up, while this risk emerges in bad times as a result of previous excessive lending and declining credit quality.” • Assuming that banks should be “used” to lower procyclicality, should this be accomplished via capital standards, loss provisioning, or other instruments/policies? • Theories that justify bank capital regulation build on the need to “control” bank risk of failure. Can one instrument be used to achieve two objectives?

10. Comments: Paper objective • “to evaluate the effectiveness of the Spanish counter-cyclical provisions system by analyzing the extent to which this scheme has in fact reduced procyclicality, restricted managerial discretion and curbed excessive lending growth.” • According to the authors “procyclicality is the phenomenon of amplifying feedbacks within the financial system and between the financial system and the macroeconomy.” • How does a reduction of managerial discretion lowers that phenomenon of feedbacks amplification? • What is excessive lending growth? • Where are the attempts to separate supply from demand effects?

11. Comments: Methodology • Given the stated objective of investigating the effectiveness of the Spanish system of loan-loss provisioning, then the exercise needs to compare its metrics before and after the introduction of that system.

12. Comments: Methodology (cont.) • Hypothesis 2: Since loan-loss provisions depend upon the “true” credit risk of loans and not upon net operating income, the significance of any statistical relationship between loan-loss provisions and net operating income should diminish over time. • Consistent with hypothesis 2, we would expect profit smoothing behavior if the coefficient of NOI is positive and significant.

13. Comments: Methodology (cont.) • Hypothesis 3: In order for counter-cyclical provisions to be fully effective, they should provide incentives for banks to grant loans more carefully. So, lagged loan growth rates should be unable to explain current loan default rates. If the high-order lags of the NPL variable are statistically significant and the low-order lags are not, this would suggest that bank managers relax credit quality as the time from the last downturn increases.

14. Comments: Conclusions • “The counter-cyclical system has significantly reduced the procyclicality of loan-loss provisions over time.” ???? • “Counter-cyclical provisions have not prevented Spanish banks from employing mechanisms for earnings management.” • “Evidence of a lack of institutional memory, leading many banks to excessive lending and a deterioration of credit quality.”???? • “Importantly, income smoothing, …. and the procyclicality of loan-loss provisioning are significantly larger in banks showing the higher loan growth rates before the crisis and in those which eventually received government funding under the restructuring scheme implemented in Spain during the crisis.”????