Debt Settlements Newsroom

The Effect Of The Business Cycle On The Resolution Costs Of Failed Banks

Banks play an important role that both affects and is affected by changes in the economic environment. Starting with Bernanke (1983), there has been a great deal of work on the economic impact of bank closures. This credit channel literature examines the effect of banks exacerbating an economic downturn by contracting credit at the same time that the economy in general is contracting. There has also been some recent work looking at the effect on banks’ probability of failure when the economy turns down. A failure usually represents a loss to the stockholders, debt holders, uninsured depositors, and the deposit insurance system. An important related question to the effect of macroeconomics on the probability of failure is the effect of those same cyclical forces on the loss rate when a bank fails.

There is an increasing interest in the relationship between business cycles and banks. The new Basel Capital accord is considering allowing credit risk modeling to be incorporated into the capital requirements. Since most of these models assume independent distributions on the probability of default and the loss given default, the question has come up as to whether these two important factors could be correlated and whether economic trends play a part. If economic
trends affect both the probability of default and the recovery rate given default, then the credit risk models may be underestimating the necessary level of capital. In addition, if default rates and recovery rates are both tied to economic trends, then capital requirements may rise at a time when the economy is contracting, intensifying concerns over the procyclicality of capital requirements if the effect turns out to be large.

This study examines the effect of business cycles on the cost of resolving a failed bank. While the primary focus is on estimating the expected loss from a bank failure, we have made an effort to formulate a model that could be used predictively  a model which, for example, could be used to calculate an aggregate expected loss rate on a category of banks. As there is no previous literature specifically examining the effect of the business cycle on the resolution costs
of failed banks, the following section covers previous work that has been done on topics of estimating losses from bank failures, the correlation between default probabilities and recovery rates, and the relationship between the business cycle and bank failures. The third section on empirical analysis contains information about the data and the empirical results on the loss rates for failed banks. Section IV highlights the results and conclusions.

Altman et al. (2002) looked at the cyclicality of recovery rates and defaults on bonds. The authors were interested in how a negative relationship between recovery rates and default rates would affect credit risk models and, therefore, on the required capital levels. They found a clear relationship between default rates and recovery rates for bonds, with bond default rates
influencing bond recovery rates. They had less success predicting recovery rates based on macroeconomic variables rather than bond default rates.

In a related exercise, Frye (2000) also found a relationship between the probability of default and the losses given default on bonds. Using a data set that included the probabilities of default for all of the Moody’s bond rating categories and the recovery rates stratified by seniority between 1983 and 1997, he estimated and parameterized a systemic risk factor. This systematic risk factor had an effect on both the default rates and recovery rates on bonds. The default
probabilities and losses given default for the years 1998-1999 were used as an out-of-sample check on the model, and the predicted negative relationship between default probability and recovery rate was consistent for this sample as well. From the perspective of managing risk to the deposit insurance fund s, any relationship between the probability of bank failures and the
loss given failure would be an important area to explore.

Also, estimating the loss that occurs when a bank has failed is an issue that comes up repeatedly within the broader topics of fair pricing for deposit insurance, depositor discipline, and the “least cost resolution” test that the FDIC uses to determine resolution method. There have been several previous studies that have attempted to explain the differences across banks in the costs of resolving failed banks. The literature on estimating loss rates started with Bovenzi and Murton (1988). They had access to data on the actual and expected recoveries and liquidatio n costs for 218 small banks that failed in 1985 and 1986; they then combined this data with examination data that measured the quality of the assets of the failed bank to estimate
losses.1 The resolution costs for these failed banks were approximately 30 percent of the banks’ assets. They modeled resolution costs as a function of the measures of asset quality, controls for bank size, location, and fraud, which improved the estimation results. They were able to explain about 56% of the variation of resolution costs using the asset quality and other controls for their sample.

Christopher James (1991) modeled the dollar value of the loss on assets of failed banks in the 1985 to 1988 period. He also used the classification of assets by the examiners to measure the quality of the assets in failed banks in the sample. However, James focused on using the data on the disposition of assets classified as loss, substandard, and non-classified and whether the riskiest assets were passed to the acquirer to examine the premium paid by acquiring banks in terms of a risk premium and charter value. He was able to find a “going concern value”2 that is lost in bank failures and that these losses decreased as more assets were passed to the acquiring bank.