Banking structure and employment growth

Abstract

Recent studies have documented an association between financial structure and per capita income growth. The relationship between financial structure and job growth, by contrast, is an unexplored issue of independent interest. Here we find that US nonmetropolitan employment grew faster in 1973–1996 where there were fewer locally owned bank offices and a more concentrated initial banking market structure; these linkages were less stable in metropolitan areas. In addition, controlling for employment growth does not undermine the empirical linkage between initial bank structure and subsequent per capita income growth.

Acknowledgements

The authors thank Daniel Milkove for valuable comments and suggestions. This article was supported by cooperative agreement number 43-3-AEM-8-80119 between the Economic Research Service of the USDA and the University of Wyoming.

Notes

¹ DeYoung et al. (1999) disaggregate urban vs. rural lending markets and find that small business lending is positively related to concentration rates in urban areas but negatively related to concentration in rural areas. Avery et al. (1999) find no differences in the effect of within-ZIP-code banking consolidation on branch density between urban and rural markets.

² See, for example, Whalen, 1995; Berger and Udell, 1996; Peek and Rosengren, 1996; Avery et al., 1999; DeYoung et al., 1999; Jayaratne and Wolken et al., 1999.

³ While larger banks typically make a smaller percentage of small business loans than smaller banks (Berger and Udell, 1996; Peek and Rosengren, 1996; Jayaratne and Wolken, 1999), small businesses do not appear to be more credit constrained in the absence of smaller banks (Jayaratne and Wolken, 1999). Berger et al. (1998) find that banks overall originate fewer small business loans following their participation in a merger; however, if the merging banks are both small, their merger is associated with a subsequent increase in small business lending (Peek and Rosengren, 1998; Strahan and Weston, 1998).

⁴ Perhaps the most relevant prior study for these issues is Whalen (1995), who explores the impact on small business lending when an acquiring bank holding company is headquartered in a different state than the target bank.

⁵ Some previous studies, such as O hUallachain and Satterthwaite (1992) and Henderson and McDaniel (2005), have included different regressors such as measures of per capita income growth, localization economies, government fiscal policies and hedonic indices (climate, recreation or scenic amenities). We do not include per capita income growth here because it is considered endogenous, as rendered explicit in our income growth regressions. We omit localization economies (industry employment) because our model is market-wide rather than sector-specific. We omit government fiscal variables because O hUallachain and Satterthwaite (1992) found no significant linkage with employment growth in metropolitan areas, while Henderson and McDaniel (2005) similarly found no significant linkage with employment growth among rural counties.

⁶ The following markets were influential: Chicago; Los Angeles; New York City; Houston; Sioux Falls, SD and Summit County, CO. The following counties were outliers: Tunica, MS; Eureka, NV; Somervell, TX; Dillingham, AK.

⁷ This procedure has been previously used in a number of banking cost studies in adjusting zero quantities of one or more outputs to avoid undefined values when taking logarithms (Shaffer, 1993). It modifies the interpretation of the estimated coefficients but does not bias them. Here we use 0.01 rather than a smaller number to avoid the creation of large outlier values in the adjusted ratios.

⁸ Previous studies have found agglomeration economies in costs; see Eberts and McMillen (1998) for a review.

⁹ The corresponding values for the growth rates in employment are somewhat higher at 0.268 and 0.508, respectively.