Financial Crisis and the Great Depression: A Regime Switching Approach

Article excerpt

RECENT EVENTS IN SOUTHEAST ASIA have focused much attention on financial crises. Despite this attention, there is no consensus view as to what causes financial crises, what are the appropriate policies for ending a crisis, and what are the real effects of such crises. History provides several examples of financial crisis from which one can draw inference.

Perhaps the most notable and most studied of these occurred in the United States during the 1930s. However, even in this case, there are a number of possible causes and solutions. Potential causes include a widespread contagion of fear among depositors after the failure of a few large banks and the stock market crash of 1929. Candidate solutions include the bank holiday and Roosevelt's fireside chats, the suspension of the gold standard, and the introduction of federal deposit insurance. Finally, it has been argued that this financial crisis had effects on the real side of the economy, both through a collapse in the money supply (Friedman and Schwartz 1963) and through a collapse in credit intermediation (Bernanke 1983).

The purpose of this paper is twofold; to make inferences about the timing of the 1930s' U.S. financial crisis and to test whether or not it had real-side effects. In order to do this I assume that, at any point in time, the financial sector is in one of two regimes, crisis or calm. Using a bivariate version of Hamilton's (1989) Markov switching model, I then estimate a conditional probability of financial crisis for each date in my sample. To test for real effects I include this time series of estimated probabilities in a reduced-form equation for output growth.

This gives me some insight into the areas suggested above. First, by estimating a time series of conditional probabilities over the state of the financial sector I can draw inferences about the timing of the financial crisis. This allows me to see which events coincide with the onset of financial crisis and whether any policy reforms coincide with its end. My evidence suggests that the financial sector did not move into a prolonged state of crisis until the first wave of banking panics in October 1930. This is consistent with Mishkin's (1991) view that a stock market crash, in this case, the 1929 crash, does not necessarily imply a financial crisis. Perhaps the more interesting result regarding the timing of the financial crisis pertains to its end. The traditional view has the crisis ending in the Spring of 1933 with the bank holiday, Roosevelt's fireside chats, and the abandonment of the gold standard. (1) The results in this paper, however, suggest the financial crisis persists until the introduction of federal deposit insurance in early 1934.

Second, by using the estimated probabilities over the state of the financial sector as explanatory variables in output equations, I am able to test whether the financial crisis had significant effects on the real side of the economy. I find that these probabilities do have additional explanatory power over an autoregressive process for manufacturing production. Consistent with Bernanke's view that the effects of the financial crisis were not solely through the monetary channel, I find that this explanatory power remains when the growth rate of M2 is added to the equation. Recently Cooper and Ejarque (1995), and Cooper and Corbae (1999) have suggested that the interwar financial sector can be characterized as shifting between two states that are indexed by agents' confidence in the intermediation process. The evidence of regime switches I find in this paper is consistent with such an approach. Also consistent with this approach's ability to further explain interwar output fluctuations, I find that the probabilities of financial crisis retain their explanatory power once linear measures of financial crisis are added to the model.

In the next section I outline the channels via which it has been argued that the disruptions to the financial sector had real effects. …