DP4651 Money Market Pressure and the Determinants of Banking Crises
|Author(s):||Tai-Kuang Ho, Jürgen von Hagen|
|Publication Date:||October 2004|
|Keyword(s):||conditional logit model, events method, identification of banking crises, index of money market pressure|
|JEL(s):||C43, E44, G21|
|Programme Areas:||International Macroeconomics|
|Link to this Page:||cepr.org/active/publications/discussion_papers/dp.php?dpno=4651|
Identifying banking crises is the first step in the research on determinants of banking crises. The prevailing practice is to employ market events to identify a banking crisis. Researchers justify the usage of this method on the grounds that either direct and reliable indicators of banks? assets quality are not available, or that withdrawals of bank deposits are no longer a part of financial crises in a modern financial system with deposits insurance. Meanwhile, most researchers also admit that there are inherent inconsistency and arbitrariness associated with the events method. This paper develops an index of money market pressure to identify banking crises. We define banking crises as periods in which there is excessive demand for liquidity in the money market. We begin with the theoretical foundation of this new method and show that it is desirable, and also possible, to depend on a more objective index of money market pressure rather than market events to identify banking crises. This approach allows one to employ high frequency data in regression, and avoid the ambiguity problem in interpreting the direction of causality that most banking literature suffers. Comparing the crises dates with existing research indicates that the new method is able to identify banking crises more accurately than the events method. The two components of the index, changes in central bank funds to bank deposits ratio and changes in short-term real interest rate, are equally important in the identification of banking crises. Bank deposits, combined with central bank funds, provide valuable information on banking distress. With the newly defined crisis episodes, we examine the determinants of banking crises using data complied from 47 countries. We estimate conditional logit models that include macroeconomic, financial, and institutional variables in the explanatory variables. The results display similarities to and differences with existing research. We find that slowdown of real GDP, lower real interest rates, extremely high inflation, large fiscal deficits, and over-valued exchange rates tend to precede banking crises. The effects of monetary base growth on the probability of banking crises are negligible.