Why the Right Capital Increase Can Come in a Boom

The Federal Reserve paper's three shock experiments test a deceptively simple question: if capital is supposed to prevent banks from crossing into risky lending, should the requirement rise in a downturn and fall in an expansion? This chapter compares the model's response to a productivity contraction, an investment-led boom, and a rise in risky-return volatility. The comparison gives the reader a practical way to judge a capital signal: follow the shock to the safe loan return and the bank's relative incentive, rather than inferring the policy response from GDP or credit alone.

The paper's no-failure rule raises capital whenever doing so is necessary to keep safe lending more attractive. That can happen while the economy is weak, while investment is temporarily strong, or while the real economy barely notices the disturbance. The repeated direction of the policy response is not a contradiction. It reflects one safety objective operating through different channels.

A downturn and a boom can weaken the safe alternative

A productivity contraction

Start with a contractionary total-factor-productivity shock. Productivity determines how much output the economy can generate from its labor and capital. In the model, the shock is persistent, so households and firms expect the lower productivity to affect the return on capital in the next period as well as today. The expected marginal product of safe capital falls, and with it the expected return on a safe loan.

That decline changes the bank's comparison. The risky project has not become socially better; its idiosyncratic shock still has a negative average effect. But the safe alternative has become less attractive to the bank's equity holders. The protected upside of risky lending now has greater value relative to a safe loan whose expected return has been compressed. If capital is held at the lowest steady-state level that is safe in normal conditions, the economy can cross into the risky regime.

Once the switch occurs, the contraction is amplified by the financial response. Risky firms produce less on average. Some risky banks fail, destroying equity and requiring deposit-insurance payouts. Output and income fall, and households reduce consumption. The credit-to-GDP ratio initially rises and later falls in the paper's simulation: loans decline more quickly than GDP at one stage, then both the financial and real adjustments evolve as the shock dissipates and the capital stock falls.

The no-failure rule responds in the opposite direction from what a mechanical “support lending in a downturn” rule might suggest. It raises capital just enough to keep safe loans attractive, then lets the requirement fall back as the productivity shock fades. The increase is very small in the model—only a fraction of a basis point in the illustrated response—because the rule is correcting the distance to an incentive boundary, not trying to offset the entire fall in productivity.

That distinction matters. The capital increase does not repair the productivity shock. Households still experience lower net worth and reduce consumption, as in a standard real-business-cycle response. What the policy prevents is a second loss caused by banks changing the composition of their lending when safe returns weaken. A prudential rule can therefore be useful in a recession without pretending that it can make the underlying recession disappear.

The comparison also reveals why the credit-to-GDP ratio is an imperfect guide even in this case.