On January 10, the Federal Reserve Bank of Minneapolis (FRBM) released its final plan to essentially double capital requirements for large banks. The final FRBM plan was based on an earlier proposal published in November 2016. The day after that earlier proposal was released, TCH published a blog post that reported that, using a combination of numbers directly from the FRBM proposal and estimates from the Basel Committee on Bank Supervision (BCBS), the costs of the plan in terms of lost economic output would substantially exceed the benefits of the plan resulting from a reduced probability of a financial crisis.

In its final proposal, the FRBM stated that there were two errors in our blog post demonstrating the substantial cost of the FRBM’s proposal, which we address here.

First, the FRBM found a simple miscalculation which resulted in the TCH post slightly overstating the cost. We quickly acknowledged the mistake, updated and corrected the blog post, and credited the FRBM. Our original blog post stated the FRBM proposal would cost (inclusive of benefits) every man, woman, and child $16,000, but after correcting the calculation error caught by the FRBM, the cost turns out to be $13,000 for every man, woman, and child. We sincerely regret our error.

However, and more importantly, the FRBM also alleged a second error in our blog post, on which they’re wrong. The FRBM incorrectly asserts that the Basel Committee’s estimated cost of a financial crisis would have been much higher if it had been calculated in a manner consistent with the assumptions in the TCH blog post. The remainder of this post explains why this second FRBM comment on our analysis is incorrect. In a nutshell, the FRBM misinterpreted the BCBS report.

In its own calculation of the benefits and costs of raising banks’ capital requirements, the FRBM adopted the approach developed by the Basel Committee when it assessed the impact of Basel III in 2010.^{1} Using the Basel Committee approach, the FRBM calculated that the cost of their proposal would be a 0.6 reduction in the level of GDP each year resulting from a contraction in bank credit supply. They also calculate that the higher capital would reduce the annual probability of a financial crisis by 0.6 percentage points.

To convert the latter reduction in the annual probability of a financial crisis to a per-year benefit that can be then compared to the proposal’s per-year cost, it is necessary to multiply it by an estimate of the total cost of a financial crisis if one were to occur. We analyzed this aspect of the proposal in our blog post, using an estimate of the cost of a crisis of 63 percent of GDP, the same number that the BCBS used in its baseline estimates in the 2010 study.

Using that estimate, the annual benefit of the FRBM proposal is 0.375 percent of GDP (the 0.6 percentage point reduction in the probability of a crisis multiplied by the total cost of a crisis equal to 63 percent of GDP) and the net benefit of the crisis is *negative*0.225 percent of GDP *each year*.

There is no disagreement between TCH and the FRBM about this conclusion. Using the baseline estimate of the cost of a crisis of 63 percent of GDP from the Basel Committee’s 2010 report, the estimated annual net benefit of the FRBM proposal is negative.

Where we diverge is the implication of how TCH converts those annual negative benefits to a present value for the BCBS’s estimate of the cost of a crisis.

TCH calculates the present value using a 5 percent discount factor and an assumption that nominal GDP will grow at 4 percent a year. The FRBM asserts that the 63 percent figure used by the BCBS is calculated using “…a 5 percent interest rate (the same as TCH), but assumes no growth” (p.10). The FRBM goes on to assert that if the crisis cost were instead calculated assuming that GDP grows at 4 percent—as we did in our blog post analysis—the crisis cost estimate would rise to 303 percent of GDP, resulting in “…per capita net benefits of $72,000.”

However, the 63 percent figure is not calculated assuming no growth and thus, the additional adjustment that the FRBM says one must make to reflect GDP growth (and produces the higher net benefit number) is neither necessary nor appropriate. As indicated in Table A1.2 (p.38) of the BCBS analysis, the crisis cost estimate of 63 percent of GDP is taken directly from Boyd et al (2005).^{2}^{ } The BCBS chose the figure because it is “the median cost of crises across all comparable approaches reported in the literature” (p. 13). Boyd et al derive their estimate by applying a discount factor to estimates of the immediate and future crisis costs that grow at the same rate as trend GDP (see Boyd et al, Figure 1, p. 981, and the explanation on p.984).

It is understandable how the FRBM got tripped up. In order to facilitate a comparison across estimates in the literature, the BCBS report converts the 63 percent figure into an estimate of per-year perpetual cost of 3 percent of GDP using a 5 percent discount factor and an assumption of no growth in GDP (p. 13). If you then take that 3 percent figure and mistakenly convert it back to a present value assuming that GDP is growing at 4 percent, you conclude that the present value of the crisis is 303 percent, the same figure cited by FRBM. But doing so would be incorrect. Thus, we stand by our conclusion that, using the Basel Committee’s baseline estimate of the costs of financial crisis, the FRBM plan would impose significant net costs. ($13,000 per American in net costs, to be exact.)

Addendum

We should also note here, as we did in our blog post, that although the FRBM employs the Basel Committee’s approach, they do not agree with using the Basel Committee’s baseline 63 percent of GDP estimate for the cost of a financial crisis. Rather, they prefer to use 158 percent – which is the highest of a range of three estimate used in the relevant Basel Committee report. As reported in Table 2 of that study, the BCBS uses three estimates that vary depending on the assumption made about whether a crisis permanently reduces GDP: 19 percent of GDP (no permanent effect), 63 percent (a long-lasting or small permanent effect), and 158 percent (a large permanent effect). It is worth quoting in full why the BCBS cautions against using the 158 percent figure:

The high-side estimates are based on studies that extrapolate a significant portion of the observed post-crisis shortfall in output into the indefinite future. However, the longer lasting the reduction in output, the greater the chance that it could reflect other factors, such as a persistent slowdown in trend productivity growth that occurred independently of the financial crisis; in fact, such factors may be an underlying cause of the financial crisis itself. Given this risk, it seems prudent to take a conservative approach and focus on the two lower sets of estimates in this analysis. (p. 13)

*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, and are not intended to be, and should not be construed as, legal advice of any kind.*