Do Community Banks Stabilize Housing Prices?

Abstract

This paper examines the influence that community bank residential mortgage lending has on regional housing prices and ultimately the housing price bubble that culminated with the Great Recession of 2007–09. Previous research suggests that community bank lending is influenced by financial intermediation frictions, such as difficulty attracting uninsured deposits and limited access to the bond market. Such frictions cause bank liquidity to play an important role in the ability of community banks to navigate swings in the local economy as well as real estate prices. We take these frictions as the starting point in our analysis and contribute to literature by focusing on the impact that community banks exert on housing prices. Specifically, we find that while community bank residential mortgage holdings are marginally related to housing price cycles, their mortgage lending does not contribute in an economically meaningful way to housing price bubbles. Even though bank liquidity, defined as the ratio of securities-to-assets, plays a role in the degree of community bank residential mortgage lending, it is not sufficiently strong enough to influence housing prices.

Keywords:

• Community banks
• mortgage lending
• housing prices
• bank liquidity

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Acknowledgements

For their helpful discussions and comments, we would like to thank seminar participants at the American Real Estate Society annual meetings, University of Central Florida Current Issues in Real Estate Conference, and the University of Southern Maine.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The FDIC (2020) defines community banks as those with assets under $1.65 billion and a strategic plan of serving the local community.

2 We use the terms “community bank” and “small bank” synonymously throughout this paper.

3 Over our sample period, Alaska and Hawaii had very few community banks.

4 Given the skewness of the distribution of bank assets, banks in the 95th percentile are considered small by Kashyap and Stein (2000).

5 We impose as the threshold for community banks the 85th percentile of total assets under the reasonable assumption that banks within that percentile are predominately involved in lending and deposit gathering within the state in which they are headquartered.

6 Securities represent the sum of held-to-maturity securities at amortized cost, available-for-sale securities at fair value, and equity securities with readily determinable fair values not held for trading on a consolidated basis.

7 In fact, our analysis indicates that securitized residential 1-4 family mortgages as a percentage of assets are quite small even for larger banks, those within the 95^th percentile of assets. In other words, the vast majority of bank residential mortgage securitization is performed by institutions with assets greater than the 95^th percentile. The FDIC first reported securitized mortgages and mortgage sold by bank in 2001:2.

8 The FHFA normalizes its Housing Price Indices so that all state housing price indices begin with an initial value of 100, thereby eliminating between-state-variation. Such variation is often needed in panel data methods. In order to recover the between-state-variation in housing prices over our sample period, we multiply each state’s housing price index by the state’s median house price as of 2000:2.

9 On an aggregate basis, it is difficult to infer why large bank securities portfolio balances are relatively uncorrelated with interest rates. On one hand, the value of a fixed-income portfolio moves inversely to interest rates, and on the other, we would expect a positive relationship with interest rates if the banks were primarily seeking interest income from securities while using them for short-term liquidity in managing loan demand.

10 To further justify shocking loans rates, in unreported results, we regress each state’s loan rate on a constant and the federal funds rate and obtain estimated r-squared values of 0.663 and 0.622 for low- and high-liquidity banks, respectively.

11 The full set of impulse responses from a shock to each variable in the system is reported in the appendix. We limit our presentation of the impulse responses here for brevity.

12 The shock to the first variable, the “pure shock,” absorbs much of the common movement in the residuals, with the second variable absorbing much of the remaining common movement, and so on. Since the correlation coefficients across the residuals are generally low, below 0.20, the order of our variables has little impact on our conclusions.

13 Equation (4)(4) $${\sum }_u=\mathrm{ }{\mathit{B}}_0^{-1}{\sum _v\mathit{B}}_{0.}^{-1\prime }$$(4) in conjunction with the method of maximum likelihood seeks to estimate contemporaneous linear relations from the VAR residuals. The estimated linear relations are interpreted in much the same way as OLS results. See Chapter 8 of Kilian and Lütkepohl (2017) and Chapter 6 of Enders (1995) for detailed developments of the model.

14 We also estimate these the same VARs with a variable composed of residential 1–4 family mortgages (balance sheet), securitized 1–4 family mortgages sold with recourse (off-balance sheet), and 1–4 family mortgages sold, but not securitized, with recourse (off-balance sheet). Combined, these amounts are very small as a percentage of assets, averaging 0.239% and 0.186% of assets for low-liquidity and high-liquidity banks, respectively, over the period 2001:2 to 2019:4. The VAR results with securitized loans and sold loans added to our mortgages are not reported and available upon request.

15 There is an alternative explanation provided by Chavaz (2017), who finds that banks hold securities to balance the risks associated with low geographic diversification. This finding is especially relevant to small community banks that likely endure low levels of geographic diversification because of their small footprint.