Abstract
Even after controlling for other observable factors, reciprocal deposits are associated with higher bank risk as measured by the probability of failure and the Z-score. These results are consistent with the moral hazard hypothesis and reject the risk substitution hypothesis.
Notes
1 This requirement is stipulated in Section 1506 of the Dodd-Frank Act; see Adler (Citation2011).
2 The most common such product is CDARS, offered by Promontory Interfinancial Network, LLC. A similar product has been patented by Intrasweep LLC.
3 This Z-score, which represents a distance to default, is not the same as Altman’s Z-score, which is used in corporate finance and fixed income studies.
4 Some studies use a hazard model rather than a logit model, but both types of models have found similar financial ratios to be significant predictors of failure.
5 This window of 21 months is within the 12- to 24-month failure window used by most early warning studies.
6 Prior studies have documented abnormal performance for younger banks (De Young and Hasan, 1998; Shaffer, Citation1998)
7 Banks were also excluded if they reported equity growth exceeding 100% per quarter from the prior December.
8 The highest ratio in the sample was 44.4%.
9 Recall that banks in the sample exhibited reciprocal deposit ratios as high as 44%.
10 The unconditional mean probability of failure within the sample window is 1 665 827 = 0.02849.
11 An alternate specification using annualized September and year-to-date June figures, not reported here, gave qualitatively similar results when banks with excessively large Z-scores were excluded due to the possibility of strategic earnings smoothing by banks. Note that quarterly Z-scores will typically exceed annual Z-scores due to smaller quarterly standard deviations of ROA; the sample average for the first three quarters of 2009 was 322.68 with a SD of 576.73.