Research

Working Papers

Intermediary Inventory Risk and the Pricing Kernel

Job Market Paper

When market makers are short index options in an intermediary-based pricing model, the pricing kernel is U shaped in market returns. The U shape is driven more by the inability of market makers to perfectly hedge returns that are large in absolute value and less by unhedgeable stochastic volatility risk. The model provides a joint explanation for historically observed U-shaped kernels, large jump risk premia for both upward and downward jumps in the market, and expensive out-of-the-money calls and puts. Changes in end-user demand since the financial crisis coincide with a disappearance of the U shape in the pricing kernel and a decrease in index option prices, as predicted by the model.

Is 24/7 Trading Better?

with Patrick Blonien

In a dynamic model of large traders who manage inventory risk, we show that a daily market closure coordinates liquidity. Some length of closure is welfare-improving relative to 24/7 trade, as the coordination of liquidity improves allocative efficiency, fully offsetting the costs of the closure. A long closure is optimal for traders in small markets, while traders in large markets would benefit from extending trading hours to near 24/7. A calibration of our model to several large equity exchanges that have proposed extending trading hours suggests that implementing such proposals would benefit traders.

Policy Comments and Media: SEC comment letter, Columbia Blue Sky Blog, Markets Media

Cyclical Labor Market Search Effort and Portfolio Risk

A model of portfolio choice and labor market search implies that individuals who own stocks should exhibit countercyclical search intensity. Empirically, the data indicate that countercyclical search by this group of individuals contributes significantly to the countercyclical aggregate labor market search documented in empirical studies, consistent with the model. Moreover, endogenous search intensity during unemployment hedges stock market risk, leading to higher optimal equity allocations. The ability to search on the job also induces higher optimal equity allocations, although this effect is weaker for high-wage workers who engage in less search. 

The Empirical Virtue of Complexity in Simple Economic Models

with Kerry Back and Seth Pruitt

We show that the empirical virtue of complexity arises in simple economic models. We show that machine-learning estimates, in data simulated from well-known economic frameworks, are superior to those from simpler empirical models. Machine-learning empirical models more accurately estimate the true SDF than sort-based or regression-based methods. Moreover, machine-learning methods more accurately estimate firms’ SDF weights, thus providing a superior depiction of the economically important firm heterogeneity that the economic models were designed to explain.