A recent working paper released by economists at the Federal Reserve (Firestone et al., 2017; hereafter “Fed analysis”), concluded that “optimal [tier 1] bank capital levels in the United States range from just over 13 percent to over 26 percent [relative to risk-weighted assets] (p. 1),” based on an assessment of costs and benefits of bank capital in the U.S. In contrast, a working paper released by economists at the Bank of England (Brooke et al., 2015, hereafter “BoE analysis”) – which followed a very similar methodology – estimates the optimal level of bank capital to be between 10 and 14 percent Tier 1 capital relative to risk-weighted assets. As we explain below, the difference in the level of capital requirements primarily reflects that the BoE analysis takes into account the reduction in expected costs of future financial crisis and changes to the probability of a financial crisis resulting from post-crisis financial reforms – in particular, a credible resolution regime and the imposition of stringent minimum total loss-absorbing capacity (TLAC) requirements for global systemically important banks – while the Fed analysis largely does not.
The framework used in both the BoE and Fed analyses to compare the macroeconomic costs and benefits of higher capital requirements entails an examination of the expected net benefits of an increase in capital requirements. Specifically, the optimal level of capital is determined when the costs of an additional percentage point increase in capital requirements (measured in terms of its impact on GDP) exceeds the economic benefits. The net benefit of higher capital requirements is defined as:
The net benefit of higher capital requirements depends on estimates for the following three terms: (i) the probability of a financial crisis as a function of the level of capital in the system; (ii) the net present cost of a financial crisis; and (iii) reduction in GDP due to higher costs of borrowing (i.e. lending spreads). Each component is estimated using models and data on past financial crisis and relying on a large number of assumptions. Although an in-depth review of each assumption used in each paper is outside the scope of this post, it is worthwhile noting that the probability of a financial crisis and the costs of borrowing are both negatively correlated with the level of capital in the banking system, while the cost of a crisis is assumed to be unaffected by the amount of capital in the banking sector.
|Table 1: Average Benefits and Costs of Increasing Tier 1 Capital Requirements
(from 12.5% to 13.5% relative to risk-weighted assets)
|Fed analysis (2017)||11.5 bps||17 bps||5.5 bps|
|BoE analysis (2015)||3 bps||6 bps||3 bps|
|Note: The benefits column measures the average increase in the level of GDP of a one percentage point increase in capital requirements relative to the case where capital requirements are kept unchanged. The costs column measures the average decrease in the level of GDP of a one percentage point increase in capital requirements relative to the case where capital requirements are kept unchanged. The average corresponds to the midpoint of the range presented in the Fed and BoE analyses.|
To understand the key differences in the two papers listed above, we compare average estimates of each of the components included in equation (1) above. Table 1 presents average estimates of the net benefits of increasing tier 1 capital requirements from 12.5% to 13.5%. As shown in the first column of the table, the net benefits of increasing capital requirements from 12.5% to 13.5% are higher in the Fed analysis relative to the BoE analysis. As a result, the level of capital requirements at which the costs of an increase in capital requirements exceeds the benefits is higher in the Fed analysis. As indicated in the following two columns, the higher net benefits of the Fed analysis are in most part driven by the higher benefits of an additional increase in Tier 1 capital requirements. In contrast, the average estimate of the cost of higher capital requirements is only slightly lower in the BoE analysis. In the remainder of the note, we focus on the difference in the sensitivity of the economic benefits to an increase in capital requirements across the two studies.
|Table 2: Impact on Average Benefits of Increasing Tier 1 Capital Requirements
(from 12.5% to 13.5% relative to risk-weighted assets)
|Reduction in Probability of Financial Crisis||Cost of Financial Crisis (% of GDP)|
|Fed analysis (2017)||24 bps||0.7%|
|BoE analysis (2015)||15 bps||0.4%|
|Note: The average corresponds to the midpoint of the range presented in the Fed and BoE analyses.|
Table 2 further analyzes the average benefits of additional capital into two subcomponents, namely the reduction in the probability of a financial crisis and the cost of a financial crisis. About two-thirds of the difference in average benefits estimated across the two studies is explained by the higher estimate of the average cost of a financial crisis found in the Fed analysis. The remaining one-third, is explained by the larger impact of capital in reducing the probability of a crisis reported in the Fed analysis. As explained below, both differences are driven by how each study incorporates the impact of the post-crisis reforms.
Both studies rely on recent work by Romer and Romer (2015) to estimate the costs of financial crisis using data from crisis experiences from advanced economies. However, the BoE analysis (2015) makes important adjustments to those estimates to take into account the regulatory reforms introduced since the global financial crisis, namely the introduction of minimum TLAC requirements and the creation of a credible resolution regime. In particular, the BoE analysis reduces the cost of a financial crisis using the estimates provided by FSB (2015). It also assumes that a credible resolution regime and minimum TLAC requirements would allow an economy to recover more quickly post-crisis, lower the impact on GDP by imposing losses on bondholders instead of taxpayers, and reduce the rise in yields of sovereign debt and thus the cost of private sector borrowing. Taken together, these three channels lower the cost of a financial crisis by more than 60 percent relative to the case where minimum TLAC requirements and a credible resolution regime are absent. Similarly, the sensitivity of the odds of a financial crisis to the level of capital is lower in the BoE analysis in part because of the new resolution framework. Specifically, holders of bank debt and equity holders are expected to induce bank management to make less risky investment choices as they no longer anticipate to be bailed out in the case a bank fails (e.g., removal of government support assumptions). Lastly, it is worthwhile noting that most of the BoE adjustments use the same method as the FSB’s TLAC Impact Assessment, which relies on the results of several academic research papers.
In summary, the Fed analysis estimate a significantly higher range for the optimal Tier 1 capital ratio because it fails to comprehensively take into account the benefits to financial stability of the extensive reforms to bank regulation and supervision that have taken place since the aftermath of the global financial crisis, such as the creation of a credible resolution regime as well as the introduction of stringent minimum TLAC requirements. Incorporating the impact of post-crisis reforms is the key challenge for the research comparing the macroeconomic costs and benefits of higher bank capital requirements and it appears the Fed analysis (2015) has not taken into account the effects of recent regulatory reforms in addressing the question of optimal capital requirements.
Disclaimer: The views expressed in this post are those of the author(s) and do not necessarily reflect the position of The Clearing House or its membership.