Why It's Important to Examine Regional Differences When Measuring Bank Asset Quality


R. Alton Gilbert

When bankers measure how well they're doing, they often compare their banks against a national average. Average measures of bank asset quality, however, can mask substantial regional differences. For instance, during the late 1980s and early 1990s, the average asset quality of banks located in New England, the Southwest region and California was poor compared with banks located in other regions of the nation.

My method of measuring asset quality is to examine the percentage of commercial and industrial (C&I) loans that have been identified as nonperforming (either past due 90 days or more, or classified as non-accrual). I identify regions as the census divisions (groups of contiguous states). I focus on community banks because my measure of asset quality at community banks is more likely to reflect the economic conditions in the census divisions of their bank's headquarters than would the nonperforming loan ratios of very large banks that have offices located across several census divisions. A common definition of a community bank involves asset size of $1 billion or less. I identify relatively small community banks as those with total assets below $300,000.

chart 1

chart 2

The top chart illustrates the percentages of C&I loans that were nonperforming for relatively small community banks during the period between March 1988 and March 2003. The shaded area represents the range of average nonperforming ratios across nine census divisions, and the solid line represents the average ratio for these small banks. The decline in the solid line indicates that the banks' average nonperforming loan ratio was higher during the late 1980s and early '90s than during recent years. The range across the census divisions shrank substantially during the second half of the 1990s and did not widen during the recession of 2001 or the following economic expansion. The patterns are similar for the nonperforming C&I loan ratio for larger community banks and for the nonperforming ratio on the total loans of banks in each of these two size groups-above and below $300 million.

Why has the range of average nonperforming loan ratios across regions of the nation been so small in recent years? One reason appears to have been relatively small differences in the pace of economic activity across regions during recent years. The bottom chart presents the growth rate in non-farm employment between 1988 and 2003. The shaded area is the range of growth rates across the nine census divisions, and the solid line represents the national growth rate of employment. As one can see, the range of the growth rates for employment narrowed substantially about 1997. The average range of growth rates across census divisions was 4.5 percentage points during 1988-96 compared with 2 percentage points during 1997-2002. Thus, like the range of nonperforming loan ratios across regions in recent years, the range in growth rates of employment has been relatively small in recent years. More information about the condition of banks can be found on our web site, www.research.stlouisfed.org/fred2/categories/83.