The Case for Simplicity Under Uncertainty
The Federal Reserve paper's title poses a policy question, not a slogan: why can a static capital buffer beat a rule that appears smarter because it follows the credit cycle? The answer emerges when the model's bank incentive, regime switch, shock experiments, and rule comparisons are treated as one system. This final synthesis gives the reader a practical test for capital policy—identify the risk mechanism, price the liquidity cost, measure the information burden, and ask whether the evidence supports that degree of precision—without turning a calibrated model result into a universal regulatory prescription.
The surprising result is not that variation in capital requirements is never useful. It is that variation is valuable only when it is connected to the thing the regulator is trying to control. In the paper, that thing is the bank's changing relative incentive to finance risky rather than safe projects. A credit measure, a business-cycle label, or a sophisticated-looking formula can be a poor substitute for that information.
One mechanism, three constraints
The entire argument can be reduced to a chain of incentives and consequences. Limited liability protects a bank's equity holders from losses below zero. Deposit insurance protects depositors and prevents them from charging the bank for the default risk created by its portfolio. Because deposits are valued for liquidity by households, they are also a cheap source of bank funding. Together, these features make it possible for the bank to capture the upside of a risky loan while not bearing its full downside.
That is the risk wedge. The risky project in the model has a lower expected return than the safe project because its additional idiosyncratic shock has a negative mean. It is not socially attractive merely because it is volatile. It becomes privately attractive when the payoff to the bank's residual claim is protected on the downside and open on the upside. A change in the economy can then make risky lending look better without making the project better for society.
Capital requirements push against this wedge in two related ways. They force more equity into the bank's balance sheet, so the bank's own loss-absorbing stake is larger. They also reduce the amount of cheap deposit funding available for a given loan portfolio. More equity means more skin in the game and a smaller value to walking away from bad outcomes. But it also means fewer deposits, and therefore fewer liquidity services for households. Capital is not a free safety device in this model; it is a purchase made partly with a valued funding instrument.
The third link is the regime change. Because the bank's expected dividend is convex in its risky exposure, the model does not produce a smooth interior portfolio choice. Banks choose the minimum or maximum permitted risky share, with a mixed outcome possible only when safe and risky equity returns are exactly equal. A modest change in a safe loan return, risky-return volatility, the penalty attached to risky lending, or the capital requirement can therefore change which extreme is preferred.
The consequences of crossing the boundary are discontinuous. In the safe regime, the bank does not fail. In the risky regime, bad idiosyncratic outcomes can make net worth negative, destroy equity, and activate deposit insurance. Risky firms produce less on average, so output and income fall; the government covers insured deposits through taxes; households experience the event through lower production, lost equity, and changes in deposits and consumption. A one-percentage-point policy movement can therefore have very different effects depending on whether it remains within the safe regime or crosses the boundary.
This chain creates three constraints on policy. The first is safety: capital must be high enough to prevent incentive-driven risk-taking when the relevant shock arrives. The second is liquidity: capital above that threshold displaces deposits that households value. The third is information: locating the threshold requires knowing which shock has changed the relative payoff and how much capital is needed to restore the safe choice.
The no-failure rule solves the first two constraints perfectly within the model because it is allowed to know the third. It sets capital at the lowest level that prevents the risky regime for the current state.