Skip to main content

Advertisement

SpringerLink
Log in

Regulatory Change and Capital Adjustment of US Credit Unions

  • Published:
Journal of Financial Services Research Aims and scope Submit manuscript

Abstract

We investigate the determinants of US credit union capital-to-assets ratios, before and after the implementation of the current capital adequacy regulatory framework in 2000. Capitalization varies pro-cyclically, and until the financial crisis credit unions classified as adequately capitalized or below followed a faster adjustment path than well capitalized credit unions. This pattern was reversed, however, in the aftermath of the crisis. The introduction of the PCA regulatory regime achieved a reduction in the proportion of credit unions classified as adequately capitalized or below that continued until the onset of the crisis. Since the crisis, the speed of recovery of credit unions in this category following an adverse capitalization shock was sharply reduced.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2

Similar content being viewed by others

Notes

  1. The tax exemption affords credit unions a competitive advantage over banks and other mutual financial services providers, and has assisted the credit union industry in maintaining capitalization substantially higher than the minimum regulatory level (US Government Accountability Office 2005). Large capital buffers might be maintained as a means of signalling strength to depositors or the regulator. The tax-exempt status of US credit unions has been justified by its proponents as a policy tool to tackle financial exclusion. An August 2010 report (and several previous reports) on tax reform suggests it might be appropriate for credit unions to be subject to corporate taxation (US Government Accountability Office 2005; The President’s Economic Recovery Advisory Board 2010).

  2. Historically, losses imposed on insurance funds have been lower for credit unions than for banks (Kane and Hendershott 1996; Wilcox 2005a).

  3. Insured credit unions are more highly capitalized, more liquid and take fewer risks than their non-insured counterparts (Karels and McClatchey 1999). By contrast, increased access to deposit insurance provides perverse incentives for banks to undertake riskier investments, increasing the probability of bank failure (O’Hara and Shaw 1990).

  4. In addition to establishing minimum capital-to-assets requirements, CUMAA 1998 grandfathered the fields of membership expansions that the NCUA had previously granted to credit unions, introduced restrictions on credit union member business lending and permitted conversion of credit unions into mutual savings banks. See Walter (2006) and Wilcox (2011) for an extended discussion.

  5. The CUMAA specifies mandatory actions for credit unions that do not meet capital adequacy standards. These include: annual earnings retentions of at least 0.4 % of total assets; the submission and adherence to a net worth restoration plan (NWRP); lending restrictions; and the prohibition of increases in assets until net worth is restored. The CUMAA allows the NCUA to use 14 supervisory actions to supplement the mandatory actions.

  6. Wilcox (2011), analysing the recent crisis period, argues that as a result of capital pressures many credit unions have been forced to limit their traditional role as safe harbours for depositors, accommodating less deposit growth than during and after past recessions, and reducing their lending substantially. In addition some credit unions have been forced to ask large depositors to temporarily shift their deposits elsewhere.

  7. US Government Accountability Office (2011) notes that the effectiveness of PCA for banks was limited because of its reliance on capital, which can lag behind other indicators of financial health. Problems with a bank’s assets, earnings or management typically manifest before these problems affect capital levels. Consequently, once an institution falls below PCA capital standards, it may be unable to recover regardless of the regulatory action imposed.

  8. The influence of firm-specific, industry, legal and institutional factors on the capital structure of non-financial firms in the US and elsewhere is examined by Rajan and Zingales (1995), MacKay and Phillips (2005), Flannery and Rangan (2006), Lemmon et al. (2008), Cook and Tang (2010), Hovakimian and Li (2011), and Oztekin and Flannery (2012).

  9. Outside the US, Brewer et al. (2008) find that the capitalization of the largest international banks during 1992–2005 was highest in countries with effective corporate governance laws, and where PCA was enforced. Gropp and Heider (2010) report that the market-to-book ratio, profitability and asset size are important in determining the capital structure of large US and European banks during the period 1991–2004. Regulatory constraints are of less importance than market variables in determining capital structure.

  10. According to Cook and Tang (2010), the impact of macroeconomic conditions on the dynamics of capital adjustment has been neglected in the empirical literature.

  11. Bogan (2012) examines the link between capital structure and operational self-sufficiency for microfinance institutions located in Africa, East Asia, Eastern Europe, Latin America, the Middle East and South Asia for the period 2003–2006.

  12. This approach addresses a shortcoming of a recent analysis of the capital dynamics of non-financial firms: “A more thorough investigation of the survivorship issue would entail a model of firm exit and an appropriate identification strategy, i.e. instrument(s), to disentangle the exit decision from the capital structure decision … a potentially fruitful area for future research” (Lemmon et al. 2008, p1581).

  13. Goddard et al. (2002) and Wilcox (2005a,b; 2006) report evidence that larger credit unions are more efficient and more robust than their smaller counterparts. Goddard et al. (2009) examine the determinants of credit union acquisition. Wheelock and Wilson (2011) suggest that future deregulation permitting credit unions to expand their range of activities will lead to further growth and efficiency improvements.

  14. In the present case, the ability of cross-sectional probit regressions to explain survival or non-survival is variable, and the estimated coefficients are unstable, owing to the relatively small proportion of disappearances (around 1.5 % of the total population) in each 6-monthly period (see Table 5). A pooled probit regression with time-period dummies allows a sufficient number of exits for reliable estimation. It seems plausible that there might well be a cyclical pattern in the rate of credit union acquisition, by far the largest exit category. The time-period dummies in the probit regression are highly significant.

  15. The role of bank-specific, regulatory and regional economic conditions as determinants of bank failure is examined by Wheelock and Wilson (2000) and Cole and White (2012). The limited evidence on credit union failure suggests that young and small credit unions, and those that are weakly capitalized, are most likely to fail (Wilcox 2005a; Goddard et al. 2009).

  16. The number of lags of the dependent variable, and the number of lags of the predetermined, endogenous or strictly exogenous variables, that are available for inclusion in the instrument matrix (either in levels or in first differences) naturally increases as the time dimension of the panel increases.

  17. Flannery and Hankins (2013) report Monte Carlo simulation evidence for the order of magnitude of the bias in the fixed-effects estimator of the coefficient on the lagged dependent variable, as follows: 0.144 for T = 6, 0.070 for T = 12 and 0.026 for T = 30.

  18. If both the covariates and the disturbances of (2) are autocorrelated, simultaneity bias may arise despite the use of lagged covariates. Owing to an absence of suitable instruments capable of satisfying the instrument relevance and instrument exogeneity conditions among the Call Report data, an instrumental variables treatment of the possible simultaneity problem affecting the estimation of (2) is not possible in this study.

  19. Recent research has focused on the role of capital buffers in absorbing shocks to a bank’s capital base. The results of studies that examine the link between bank capital buffers and economic growth are mixed. On balance, however, it appears that commercial banks’ capital is negatively related to economic growth (Ayuso et al. 2004; Lindquist 2004). This is not the case for savings and cooperatives banks, which appear to build capital during periods of economic growth (Stolz and Wedow 2011). Smith and Woodbury’s (2010) comparison of the loan performance of banks and credit unions over the period 1986–2009 suggests that credit union loan delinquencies and net charge-offs are less affected by macroeconomic fluctuations (unemployment rate) than those of banks.

  20. For the remaining 40 acquisitions, we were unable to identify the acquiring credit union. Most of these acquisitions were small. We consider the effects of this omission on the estimation results to be negligible.

  21. For the semi-annual observation immediately following an acquisition, k i,t is the capital-to-assets ratio for the acquirer, incorporating the capital and the assets of the acquired credit union. The constructed value of Δk i,t for the acquirer is k i,t –k ∗ i,t − 1 , where k ∗ i,t − 1 is the synthetic capital-assets ratio of the acquirer and the acquired before the acquisition, [(NWacquirer + NWacquired)/(ASSETacquirer + ASSETacquired)] t–1. The constructed values of k i,t–1, Δk i,t–1 and Δk i,t–2 for the acquirer are k ∗ i,t − 1 , k ∗ i,t − 1  − k ∗ i,t − 2 and k ∗ i,t − 2  − k ∗ i,t − 3 , respectively. For the next semi-annual observation, Δk i,t and k i,t–1 are the actual values for the acquirer incorporating the acquired, and the constructed values of Δk i,t–1 and Δk i,t–2 are k i,t–1–k * i,t–2 and k ∗ i,t–2  − k ∗ i,t–3 , respectively. For the following observation, Δk i,t , k i,t–1 and Δk i,t–1 are the actual values, and the constructed value of Δk i,t–2 is k i,t–2–k ∗ i,t–3 . The other covariates are treated in the same manner.

  22. The analysis is undertaken using semi-annual data. Before 2001 credit unions with assets of less than $50 million were not formally required to file quarterly reports. In 2001, the NCUA instituted a new Examination Scheduling Program, which required all federally-insured credit unions to file a quarterly Call Report (5300 report).

  23. Most credit unions that exited did so as a consequence of having been acquired. Credit unions that failed or were subject to a purchase and assumption order account for a very small proportion of total exits (see Table 4). For the estimation of the probit regression of survival or disappearance we combine all disappearances into one exit category. This suggests that the estimation results are driven mainly by the preponderance of mergers. In a study of the corporate demography of the credit union industry, Goddard et al. (2014) identify some differences in the relative importance of factors driving disappearance through acquisition and disappearance through failure.

  24. That older credit unions are at greater risk of disappearance has been found elsewhere in the credit union literature. For example Barron et al. (1994), who examine the corporate demography of credit unions in New York, find that older credit unions fail at a faster rate than their younger counterparts. The authors attribute this pattern to either liability of obsolescence, or liability of senescence, or both.

  25. We find a higher liquidity ratio leads to disappearance (the numerator of our liquidity ratio is defined as cash on hand, cash on deposit and cash equivalents) and a low loan to assets ratio also leads to disappearance. An explanation for these findings is that a credit union that hordes cash, or does not create a loans portfolio of a size commensurate with its deposits, may be either an attractive target for an acquirer that believes itself capable of earning a higher return by expanding the loans portfolio, or at risk of failure due to an inability to generate an adequate return.

  26. For convenience, we refer below to the three sub-periods as 1995–2000, 2000–08 and 2009–12.

  27. The differences between these estimated rates of mean reversion are statistically significant in all cases. A z-test of the hypothesis α 11  = α 21 in (2) (equivalent to ∑λ 1 m  = ∑λ 2 m ), interpreted as the condition that the rates of mean reversion are the same for well capitalized credit unions and credit unions adequately capitalized and below, 1995–2000, yields a z-statistic of 10.7 (p-value = .000). The test of α 31  = α 41 in (2), as above for 2000–08, yields z = 8.21 (p = .000); and the test of α 51  = α 61 in (2), as above for 2009–12, yields z = 2.22 (p = .026). An F-test of α 11  = α 31  = α 51 in (2), interpreted as the condition that the rates of mean reversion for well capitalized credit unions are the same for all three sub-periods, yields F = 33.7 (p = .000). Despite these three coefficients being qualitatively similar the test is highly significant, presumably owing to the large number of observations for well capitalized credit unions. The test of α 21  = α 41  = α 61 in (2), as above for credit unions adequately capitalized and below, yields F = 24.3 (p = .000).

  28. The estimations reported by Shrieves and Dahl (1992, p452, Table III) imply a corresponding rate of mean reversion measure of 0.942 for US banks with capitalization above the mid-1980s regulatory threshold (7 % equity capital-to-assets ratio), and 0.729 for banks below this threshold. Accordingly, the rate of mean reversion is slower for the former category, and faster for the latter. The corresponding estimates for the Tier 1 leverage ratio reported by Aggarwal and Jacques (2001, p1151, Table III) are: 0.857, 0.888, 0.895 for well capitalized banks in 1991, 1992 and 1993, respectively; 0.834, 0.771, 0.773 for adequately capitalized banks; and 0.637, 0.220, 0.565 for undercapitalized banks (same 3 years).

  29. For j = 1, the estimated rates of mean reversion are 0.8702 (estimation without the sample-selection correction) and 0.8735 (estimation with the correction). For j = 2, the corresponding estimated rates of mean reversion are 0.6860 (without) and 0.7041 (with); for j = 3, 0.8627 (without) and 0.8643 (with); for j = 4, 0.6995 (without) and 0.7045 (with); for j = 5, 0.8708 (without) and 0.8717 (with); and for j = 6, 0.9403 (without) and 0.9462 (with). Although the exclusion of the sample-selection correction does not seem to carry any serious qualitative implications for the estimation of the capital-adjustment equation, these results suggest that exclusion introduces a modest downward bias into the estimated coefficients on the lagged dependent variable in the capitalization regression. The sample-selection correction attributes some of the variation in capitalization to a survivorship effect. Exclusion of this correction therefore increases the unexplained proportion of the variation in capitalization, making the latter appear more random, and making capitalization appear less highly autocorrelated.

  30. For credit unions adequately capitalized or below the pattern is less consistent, presumably owing to the smaller numbers of observations on which the coefficient estimates are based. For j = 2 (1995–2000), a coefficient on Δk i,t–1 in Table 6 of −0.1689 is replaced by very similar estimates of −0.1680 for Δk i,t–1 ≥ 0 and −0.1691 for Δk i,t–1 < 0. For j = 4 (2000–2008), a coefficient on Δk i,t–1 in Table 6 of 0.0413 is replaced by coefficients of −0.0026 for Δk i,t–1 ≥ 0 and 0.0423 for Δk i,t–1 < 0. For j = 6 (2009–2012), a coefficient on Δk i,t–1 in Table 6 of 0.0451 is replaced by coefficients of 0.2491 for Δk i,t–1 ≥ 0 and -0.0087 for Δk i,t–1 < 0.

  31. One-tail z-tests of null hypotheses of zero restrictions on the sums of the coefficients on the lagged differences of c, n, e and g produce the following results: test β11 + β12 = 0 (sum of coefficients on Δc i,t-1 and Δc i,t-2 = 0), z = −7.56, p = 0.000; test β21 + β22 = 0 (sum of coefficients on Δn ,t-1 and Δn i,t-2 = 0), z = −14.9, p = 0.000; test β31 + β32 = 0 (sum of coefficients on Δe i,t-1 and Δe i,t-2 = 0), z = −8.57, p = 0.000; test β41 + β42 = 0 (sum of coefficients on Δg i,t-1 and Δg i,t-2 = 0), z = 7.56, p = 0.000. These results indicate that permanent changes in the rates of growth of loans, non-performing loans ratio, or ratio of non-interest expenses to assets would produce a reduction in the equilbrium capital-to-assets ratio; and a permanent change in the rate of growth of gross state product would produce an increase in the equilbrium capital-to-assets ratio.

  32. These results suggest that the link between survivorship and capitalization is stronger for large than for small credit unions. Our conjecture is that the survival or disappearance of large credit unions is likely to be driven primarily by their capitalization. For smaller credit unions, by contrast, other factors may be paramount: for example, a small credit union might be acquired and absorbed into another credit union if its common bond becomes obsolete owing to the closure of the employer of its members, or because its present management wishes to retire.

References

Download references

Acknowledgments

The authors would like to thank an anonymous referee for extensive comments and suggestions on a previous draft of this paper. We also thank Christine Brown, Santiago Carbo Valverde, Rebel Cole, Bob DeYoung, Kevin Davis, Scott Frame, Michael Goldstein, Jens Hagendorff, Mohammad Kabir Hassan, Michael King, Phil Molyneux, Rob Nijskens, Steven Ongena, Matthew Osborne and Barry Quinn for useful comments and suggestions on a previous draft of this paper. The usual disclaimer applies.

Author information

Authors and Affiliations

  1. Bangor Business School, Bangor University, Bangor, UK

    John Goddard

  2. Queen’s University Management School, Queen’s University Belfast, Belfast, UK

    Donal McKillop

  3. Centre for Responsible Banking and Finance, School of Management, University of St Andrews, The Gateway, North Haugh, St Andrews, Fife, KY16 9SS, UK

    John O. S. Wilson

Authors
  1. John Goddard
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Donal McKillop
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. John O. S. Wilson
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to John O. S. Wilson.

Appendix

Appendix

Definitions of variables

This appendix provides definitions of the variables used in the empirical analysis. The Call Report data item numbers are as follows:

Total Assets, 010; Net Worth, 997; Liquid Assets, 730A + 730B + 730C; Non-performing Loans, 041B; Loans, 025B; and Non-interest Expense, 671

s i,t :

natural logarithm of Total Assets [ASSET] for i = 1,…,N, t = 1995.2 to 2012.2

k i,t :

Capital-to-Assets ratio = Net Worth [NW]/Total Assets

q i,t :

Liquid Assets [LIQ]/ Total Assets

n i,t :

Non-performing Loans [NP] / Total Assets

x i,t :

Loans [LOAN]/ Total Assets

e i,t :

Non-interest Expense [NIE] / Total Assets

a i,t :

natural logarithm of Age

c i,t :

natural logarithm of Loans

g i,t :

natural logarithm of Gross State Product

d s,i,t :

(for s = 1995.2,…,2012.2) are 0–1 dummy variables for each semi-annual observation

I 1 i,t  = 1:

if credit union i has k i,t–1 ≥ 0.07, for t = 1995.2 to 2000.1 inclusive, 0 otherwise

I 2 i,t  = 1:

if credit union i has k i,t–1 < 0.07, for t = 1995.2 to 2000.1 inclusive, 0 otherwise

I 3 i,t  = 1:

if credit union i has k i,t–1 ≥ 0.07, for t = 2000.2 to 2008.2 inclusive, 0 otherwise

I 4 i,t  = 1:

if credit union i has k i,t–1 < 0.07, for t = 2000.2 to 2008.2 inclusive, 0 otherwise

I 5 i,t  = 1:

if credit union i has k i,t–1 ≥ 0.07, for t = 2009.1 to 2012.2 inclusive, 0 otherwise

I 6 i,t  = 1:

if credit union i has k i,t–1 < 0.07, for t = 2009.1 to 2012.2 inclusive, 0 otherwise

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Goddard, J., McKillop, D. & Wilson, J.O.S. Regulatory Change and Capital Adjustment of US Credit Unions. J Financ Serv Res 50, 29–55 (2016). https://doi.org/10.1007/s10693-015-0217-y

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10693-015-0217-y

Keywords

JEL Classification

Search

Navigation