Date of Graduation
1997
Document Type
Dissertation/Thesis
Abstract
Bank lending can have an important influence on local economic development. Banks have competing obligations, however, which compel them to look carefully at the riskiness of particular assets. This research addresses both lending risk assessment and the link between risk and lending by deriving risk measures based on industry mixes and testing whether differences in risk in local markets affect asset allocation. The research is both theoretical and empirical. The theoretical model of lending is based upon imperfect markets, risk aversion, and cost differentials. The model is extended to group banking to provide insight into the effects of diversification on lending. The empirical work consists initially of calculating risk measures. A time series approach using business failure rates for broad sectors in five states is used to create proxy variables for the expected values and variances of loan default rates of firms in different economies. The sectoral failure rates show varying characteristics but reveal some fairly consistent shared trajectories among various sectors both within and between states. Predictive models for the failure rates generally have the expected coefficients for lagged values and other variables, but have varying degrees of fit depending upon the state and sector. The model residuals show much less correlation among sectors and states than do the failure rates. The lending risk measures based upon the state-level risk calculations, county industry mixes, and bank location and affiliation reveal substantial differences among banks and between states. Comparing risk measured at the bank level and at the holding company level reveals only small to modest diversification benefits. Given the risk variables, OLS regression is used to estimate the effect on lending of risk and size, market power, and loan demand variables. Regressions are estimated for both total and business lending amounts as a proportion of assets. The estimations deliver inconclusive results. While the risk measures are often significantly and negatively related to lending ratios, the results are not consistent. Other important findings include the need to fit state-specific models and insignificant or mixed results concerning the influence of each non-risk independent variable on the lending ratios.
Recommended Citation
Sorenson, David John, "Estimating local lending risk and its effect on bank asset allocation." (1997). Graduate Theses, Dissertations, and Problem Reports. 9803.
https://researchrepository.wvu.edu/etd/9803