Scott Condie

Under the Fundamental Review of the Trading Book, most trading banks compute market risk capital using the standardized approach. This sensitivities-based method requires the entire charge to be computed three times—under a high, a medium, and a low prescribed correlation scenario—with the largest of the three binding. Taking the worst of a set of correlation models echoes the maxmin criterion of Gilboa and Schmeidler (1989). This paper studies what that design does to the banks that face it. The first result is that the rule penalizes hedging: a hedged book is always charged at the regulator’s lowest correlation, so hedging is credited only up to that correlation and charged beyond it, and a fully matched hedge carries strictly more capital than an open position whenever the prescribed correlation is below two thirds—a threshold that covers the equity buckets, the base-metals bucket, and most cross-bucket aggregations. The second result is that a bank optimizing against the constraint has an inaction region, an interval of news over which its capital holdings against risk do not adjust to information and, where it is the marginal investor, its quotes do not move. The same freezing occurs when the traded factor’s volatility changes and its correlation does not, because the charge is computed from the regulator’s risk weights rather than the asset’s true current volatility. The region’s width is the shadow cost of regulatory capital times the regulator’s scenario spread, so it widens in stress, and because every bank faces the same three correlation matrices these regions lie at the same portfolio configurations for all banks. What is synchronized across the system is not how much risk banks perceive but where they stop responding to news. Calibrated at the prescribed Basel correlations, the scenario spread reaches 40 percentage points and the mechanism is shown to be quantitatively meaningful at current policy correlations. Policy suggestions are given to help reduce the negative consequences of this design without undermining its prudential intent.

This paper considers the effect on asset prices of investors contemplating the possible occurrence of unexpected and unprecedented events that they have no basis to evaluate. We build a Capital Asset Pricing Model (CAPM) where, in addition to regular risk, investors are aware that they are potentially unaware of some events. We show that when investors feel that there exist states about which they are unaware, asset prices contain an unawareness premium. A driving force is that the “risk free” asset is no longer considered to be truly risk free. We develop a methodology that enables us to estimate the systematic portion of the unawareness premium, and we estimate it using daily data from 1980 to 2021. This unawareness premium implies a theoretical motivation behind the correlation between estimated asset alphas and betas in the cross section. We find evidence in support of the hypothesis that unawareness, in addition to risk, is a determinant of expected equity returns. This additional factor adds insights into asset market behavior around market run ups like those during the dot com boom and the pre-financial crisis market outperformance.

This paper studies equilibrium order book formation in a limit-order market by building a search-theoretic model where the shape of the order book and its spread are determined jointly in equilibrium. The model characterizes liquidity as a function of differences in valuation between sellers and buyers, beliefs about the probability distribution of the arrival of buyers and sellers, exchange fees, and preference parameters like patience and beliefs about the expected lifespan of information. The efficiency of the market is characterized as a function of these parameters. We demonstrate how to extract the model’s parameters from order book data and, using a sample of data from Coinbase’s Bitcoin/U.S. Dollar exchange, characterize market liquidity as a function of these factors. We derive the optimal timing of frequent batch auctions in our model and show how this timing can be calculated for all limit-order markets.

This paper studies the welfare and asset pricing consequences of market panics driven by ambiguity aversion. When information is perceived to be highly ambiguous, ambiguity-averse traders exit the market entirely, raising the risk premium and creating opposing effects for expected-utility traders who remain: a wealth loss from falling prices and an expanded trading opportunity from reduced risk sharing. This paper characterizes the size of these effects and their impact on welfare. In a two-period extension the premium associated with the possibility of a future panic is calculated. Even when all fundamental uncertainty is normal, the distribution of interim prices is left-skewed and fat-tailed. Price variance decomposes into a fundamental component and an ambiguity-driven component, providing structural microfoundations for well-documented non-normalities in equity returns.