Serhiy Kozak

Summary

Exploited staggered removals of interstate banking restrictions to identify the causal effect of credit access on households' stock market participation and asset allocation. Documented that loosened credit constraints lead households to enter the stock market and increase allocation to risky assets, establishing a direct link between credit access and household investment decisions.

Summary

Proposed a method using the 'kernel trick' from machine learning to construct a stochastic discount factor (SDF) when the set of characteristics is extended to a potentially infinitely-dimensional set of non-linear functions. Found that allowing for interactions and non-linearities leads to substantially more efficient SDFs, doubling out-of-sample Sharpe ratios for implied MVE portfolios.

Summary

Developed a Bayesian tensor model to impute missing or infrequently observed financial or economic data, incorporating multiple imputation, joint modeling of dependencies, and low-dimensional tensor representation. Documented that firms with missing or unusual characteristic values exhibit abnormal expected returns, enhancing the understanding of the risk-return tradeoff.

Published by

Journal of Financial Economics, Revise & Resubmit

Summary

Investigated conditions under which the stochastic discount factor (SDF) that prices individual stocks can be represented as a factor model. Demonstrated that factors constructed heuristically do not span the SDF unless the covariance matrix of stock returns has a specific structure, and showed how dimension reduction is possible under certain conditions without inverting a large covariance matrix.

Published by

Journal of Finance

Summary

Utilized a large cross-section of equity returns to estimate a rich affine model of equity prices and dividends, successfully pricing dividend strips without using strip data. The model extends equity term-structure data over 45 years, generating term structures for various equity portfolios and providing novel empirical moments for asset pricing models.

Published by

Journal of Monetary Economics

Summary

Developed a production-based equilibrium model that jointly prices bond and stock returns, producing time-varying correlation between stock and real treasury returns. The model incorporates time-varying risk aversion and two physical technologies, demonstrating how bonds hedge risk-aversion shocks and cash-flow shocks drive co-movement, predicting relative bond-stock and long-short equity returns.

Published by

Journal of Financial Economics

Summary

Developed a new empirical methodology to investigate whether increases in equity risk premia represent good or bad news to rational investors, finding a negative price of discount rate risk. This suggests high discount rates during periods of high marginal utility of wealth, with covariances driving observed patterns in stock and bond expected returns.

Published by

Review of Financial Studies

Summary

Proposed and implemented a method to characterize the optimal factor timing portfolio, finding that market-neutral equity factors are strongly and robustly predictable. Exploiting this predictability leads to substantial improvement in portfolio performance relative to static factor investing, posing new challenges for theories matching cross-section of stock returns.

Published by

Journal of Financial Economics

Summary

Constructed a robust Stochastic Discount Factor (SDF) that effectively summarizes the joint explanatory power of numerous cross-sectional stock return predictors. Achieved robust out-of-sample performance by imposing an economically motivated prior on SDF coefficients, demonstrating that while sparse SDFs from a few factors are insufficient, an SDF derived from a small number of principal components performs well.

Published by

Journal of Finance

Summary

Argued that tests of reduced-form factor models and 'horse races' cannot effectively distinguish between alternative investor belief models. Demonstrated that, due to substantial commonality in asset returns, the Stochastic Discount Factor (SDF) can be represented by a few dominant sources of return variation, a conclusion applicable even when sentiment drives cross-sectional variation in expected returns, provided arbitrageurs are present.