The accuracy of dynamic stress-test capital models remains undocumented. Three methodologies: a CLASS-style approach, Bayesian model averaging, and a Lasso specification are used to forecast the performance of 14 large US banks during the financial crisis. Individual bank models are calibrated using bank historical data while regulatory models are calibrated using representative bank data. Representative bank model forecasts differ dramatically from the forecasts from bank-specific models and from actual outcomes. The Lasso methodology is most accurate, but its superiority may be sample-specific and is only apparent ex post. The results highlight the policy uncertainty inherent in regulatory stress tests.

Multi-year forecasts of bank performance under stressful economic conditions determine large institution regulatory capital requirements and yet the accuracy of these forecasts is undocumented. I compare the accuracies of alternative stress test model forecasts using the financial crisis as the stress scenario. Models include specifications that mimic the Federal Reserve CLASS model and alternatives that use Lasso, the AIC and an abridged set of explanatory variables. A simple single-equation Lasso model has, by far, the best forecast accuracy. Large differences in model forecast accuracy are undetectable from estimation sample statistics. These findings highlight the need for new methods for validating bank stress test models.

The efficacy of the Financial Stability Board's proposed requirement for minimum “total loss absorbing capacity” (TLAC) at global systemically important banks (G-SIBs) is assessed using a stylized model of a bank holding company and an equilibrium asset pricing model to value financial claims. I identify a number of G-SIB strategies that satisfy minimum TLAC requirements but fail to reduce implicit safety net subsidies that accrue to G-SIB shareholders or increase the resources available to recapitalize a failing G-SIB subsidiary. To meet the FSB's stated goals, TLAC requirements must impose minimum TLAC at all subsidiaries and restrict how TLAC funds can be invested. An equivalent, but much simpler solution is to significantly increase regulatory capital requirements on systemically important bank subsidiaries.

The Single Point of Entry (SPOE)—the FDIC strategy to implement its new Dodd–Frank Orderly Liquidation Authority (OLA)—promises to reduce the financial market turmoil caused by the failure of a large complex financial institution by using parent holding company resources to recapitalize its large failing subsidiary banks. We identify legal and financial impediments that may prevent the use of a SPOE strategy for this purpose. Dodd–Frank does not authorize bank recapitalizations through SPOE or otherwise, and the OLA cannot be invoked unless the failure of a subsidiary puts the parent in danger of default. The imprecise legislative language that authorizes OLA creates a new source of systemic risk. Regulations required to operationalize SPOE will require bank holding companies to issue a substantial volume of new subordinated debt, increasing these institutions’ leverage and financial fragility. Unless the Dodd–Frank Act is amended, OLA could well magnify and not reduce market instability in the next financial crisis.

A simple overlapping generations model is used to characterize the effects of initial margin requirements in the volatility of risky asset prices. Investors are assumed to exhibit heterogenous preferences for risk-bearing, the distribution of which evolves stochastically across generations. This framework is used to show that imposing a binding initial marginal requirement may either increase or decrease stock price volatility, depending upon the microeconomic structure behind fluctuations in economywide average risk-bearing propensity. The ambiguous effect on volatility similarly arises when the source of heterogeneity is noise trader beliefs.

We measure the effect of bank failures on economic growth using data from 1900 to 1930, a period without active government stabilization policies and several severe banking crises. VAR model estimates suggest bank failures have long-lasting negative effects on economic growth. A bank failure shock involving one percent of system liabilities leads to a 6.5% reduction in GNP growth within three quarters and a measurable reduction for 10 quarters. Panel VAR model estimates for the 48 states show bank failures aggravate commercial non-bank failures. Institutional and regulatory features affect the intensity of the bank failure effect. We find that bank failures have a larger impact in states with deposit insurance, in states more heavily concentrated in agriculture, and in states with fewer large firms. However, because a number of states exhibit all three characteristics, we are not able to clearly identify the true marginal effects of these factors independently.

We estimate the impact of a poor bank examination rating on the growth rates of individual bank loan portfolios. We use a novel approach to control for loan demand variation and estimate a fixed-effect model using an unbalanced panel with over 381,000 bank-quarter observations from the period 1994–2011. Our estimates show that a poor examination rating has a large negative impact on bank loan growth, even after controlling for the impact of monetary policy, bank capital and liquidity conditions, and any voluntary reduction in lending triggered by weak legacy loan portfolio performance or other bank losses. This previously unidentified effect is consistent with the hypothesis that the bank supervision process successfully constrains the lending activities of banks operating in an unsafe and unsound manner.

A simple overlapping generations model is used to characterize the effects of initial margin requirements on the volatility of risky asset prices. Investors are assumed to exhibit heterogeneous preferences for risk‐bearing, the distribution of which evolves stochastically across generations. This framework is used to show that imposing a binding initial margin requirement may either increase or decrease stock price volatility, depending upon the microeconomic structure behind fluctuations in economy‐wide average risk‐bearing propensity. The ambiguous effect on volatility similarly arises when the source of heterogeneity is noise trader beliefs.

A simple overlapping generations model is used to characterize the effects of initial margin requirements on the volatility of risky asset prices. Investors are assumed to exhibit heterogeneous preferences for risk-bearing, the distribution of which evolves stochastically across generations. This framework is used to show that imposing a binding initial margin requirement may either increase or decrease stock price volatility, depending upon the microeconomic structure behind fluctuations in economy-wide average risk-bearing propensity. The ambiguous effect on volatility similarly arises when the source of heterogeneity is noise trader beliefs. BEYOND THEIR PRUDENTIAL FUNCTIONS, the effects of initial margin requirements on securities markets are not well understood. Many have argued, based upon empirical findings, that the level of margin requirements has no discernable impact on stock price behavior.' Nevertheless, the view that high margin requirements dampen speculative excesses and reduce excess price volatility remains popular. The simultaneous existence of such views highlights the uncertainty surrounding this issue. In fact, there exists no rigorously articulated theory of the interaction between initial margin requirements and stock prices and their returns' characteristics. In this paper we construct a simple dynamic model useful for characterizing the effects of initial margin requirements on security prices and returns. To analyze the impact of margins, we introduce investor heterogeneity into an otherwise simplified overlapping generations (OLG) economy.2 The OLG models are constructed so that each generation displays heterogeneous preferences or beliefs, the distribution of which evolves stochastically across generations. In each of two heterogeneous taste models we consider there are two types of agents, one of which is more risk tolerant than the other. In one model, the