Disciplinary Role of Subordinated Debt and CoCos

The idea that subordinated debt can impose market discipline on banks is longstanding. Sironi (2003) notes that there have been numerous proposals for mandatory subordinated debt since the late 1980s.

Using data for US banks in 1980-2004, Ashcraft (2008) examines the influence of an increase in the ratio of subordinated debt to regulatory capital (the sum of subordinated debt and equity capital). He begins with recognition that the standard (Merton model) influence of greater debt leverage is higher risk and hence that the disciplining role must be great enough to more than offset the higher leverage risk for the net influence of subordinated debt to be favorable. His empirical tests deal with endogeneity by using variations in state corporate tax rates to instrument the subordi­nated debt ratio.

Ashcraft finds that for stand-alone banks, an increase in the (instru­mented) ratio of subordinated debt to regulatory capital reduced the prob­ability of bank distress before the 1990 implementation of the original Basel accord but increased it thereafter. He attributes the difference to the Basel requirement that to be included in tier 2 capital, subordinated debt could not include restrictive covenants, thereby reducing investor influ­ence. In contrast, for banks affiliated with bank holding companies, the influence of subordinated debt in reducing distress probability persisted through the full period. He attributes this difference to the fact that bank holding companies provided a direct channel of surveillance of the affili­ated banks.

Evanoff, Jagtiani, and Nakata (2011) use data on the publicly traded subordinated bonds of 19 large US banks and 39 bank holding companies in 1990-99 to examine the relationship between spreads and risk indica­tors. Their sample of about 600 representative bonds is evenly split between an “issuance” subsample, for which banks had recently issued new debt, and a “nonissuance” subsample for others. The authors find that spreads are significantly related to two measures of risk: nonperforming loans as a percent of assets (positive coefficient) and return on assets (negative coef­ficient). A third risk variable, the market leverage ratio (the ratio of liabili­ties to the market value of common stock plus the book value of preferred stock), has the correct (positive) sign but is not statistically significant. Macroeconomic variables have significant influences in the expected direc­tions (with positive coefficients on the Treasury bill rate and unemploy­ment and a negative coefficient for periods of economic expansion), and long-term debt has a higher spread. The authors emphasize that although both subsamples tend to obtain these same results, the significance and degree of explanation are considerably higher for the “issuance” subsample (R² of about 0.65 versus about 0.50). They argue that if subordinated debt were made a mandatory part of the regulatory structure, markets would be much deeper and issuance more frequent. Citing the “issuance” results as a closer representation of those conditions, they argue in support of mandatory subordinated debt requirements, and maintain that this ap­proach should not be rejected based on the argument that in the past the signals in market data for subordinated debt spreads have been too noisy to warrant such an approach.

Calomiris and Herring (2013) argue that there should be a major CoCo requirement to complement the common equity requirement. Its purpose would be to provide strong incentives to bank management to carry out prompt recapitalization to deal with losses rather than waiting until it is too late and the bank has lost access to the equity market. They propose that CoCos equivalent to 10 percent of total assets be issued. Conversion would be triggered when the quasi-market value of equity ratio—the market capitalization value of equity divided by that value plus total debt (as a com­prehensive measure of firm value)—falls below 4 percent, based on a 90-day moving average. The conversion rate would be set such that CoCo holders lost no value. The dilution consequences would be so severe for existing shareholders that bank management would make every effort to avoid conversion, providing discipline on risk taking ex ante and pressure to act promptly if recapitalization did become necessary.

Without providing a cost-benefit analysis, they also suggest that the common equity requirement should be set at 10 percent of total assets, but they explicitly recognize that contrary to those who invoke the Modigliani- Miller offset, “equity is costlier to raise than debt for fundamental reasons associated with both information and managerial agency problems” (p. 42). The authors maintain that interest on CoCo bonds would “continue to exploit the tax shield provided by the asymmetry of treatment between interest and dividends in the tax codes of most countries” (p. 44). However, the lack of tax deductibility in the United States has so far limited their use primarily to European banks.^[155]

Their emphasis on CoCos triggered by market equity prices stems in part from their skepticism regarding supervisors' capacity to keep up with banks' ability to exploit accounting and regulatory loopholes, as well as po­litical pressures on regulators to grant excessive forbearance. To illustrate, they cite lurid episodes from the 2008 financial crisis, including Lehman Brothers' use of disguised repos that in effect understated assets, Northern Rock's shift to internal model risk weighting, and “re-remic” resecuritiza­tion of subprime collateral debt obligations to arbitrage risk weightings of tranches. Their CoCo proposal is meant to reinforce official supervision with market discipline.