Abstract: We study aversion to model ambiguity and misspecification in dynamic portfolio choice. Investors with relative risk aversion gamma above 1 fear return persistence, while risk-tolerant investors (gamma between 0 and 1) fear return mean reversion, to confront model misspecification concerns when facing a model with IID returns. The intuition is that risk-averse (risk-tolerant) investors who are keen to hedge (speculate) intertemporally worry about an endogenous worst-case misspecification where returns persist (mean-revert) so that hedging (speculation) is impossible. A log investor is myopic and unaffected by model misspecification, therefore only worrying about model ambiguity among IID models. Rather than the multiplier approach of Hansen and Sargent (2001) we utilize a constraint approach, preserving homotheticity and tractability. Our model can explain evidence for the experience hypothesis, for nonparticipation in equity markets, as well as for extrapolative return expectations. In equilibrium, we show that model misspecification, unlike model ambiguity, can generate excess volatility, even with IID fundamentals.
Discussant: Philipp Illeditsch, Texas A&M University
Abstract: We use new monthly security-level data on portfolio holdings, flows, and returns of U.S. households to understand asset demand across multiple asset classes. Our data cover a wide range of households across the wealth distribution – including ultra-high-net-worth (UHNW) households – and holdings in many asset classes, including public and private assets. We first develop a descriptive model to summarize households’ rebalancing behavior. We find that less wealthy households rebalance from liquid risky assets to cash during market downturns, while UHNW households tend to purchase risky assets during those periods and thus stabilize market fluctuations. This pattern is particularly pronounced for U.S. equities. Across risky asset classes, three factors explain most of the variation in portfolio rebalancing and those factors target the long-term equity premium, the credit premium, and the premium on municipal bonds. Next, we develop a new framework to estimate demand curves across asset classes. While nesting traditional models as a special case, our framework allows for a muted response of asset demand to fluctuations in asset prices and easily extends to account for inertia. Our new estimator of asset demand curves exploits variation in second moments of returns and portfolio rebalancing, and can even be used when only a fraction of all holdings in a market can be observed. Our preliminary results indicate that asset demand elasticities are smaller than those implied by standard theories, vary significantly across the wealth distribution, and are negative for various groups, pointing to positive feedback trading. In sum, we think that our framework and data paint a coherent picture of U.S. households that captures, quite uniquely, their rebalancing behavior across the wealth distribution and across broad asset classes.
Discussant: Valentin Haddad, University of California-Los Angeles
Abstract: We investigate the impact of financial windfalls on household portfolio choices and risk exposure. Exploiting the randomized assignment of lottery prizes in three Swedish lotteries, we find a windfall gain of $100K leads to a 5-percentage-point decrease in the risky share of household portfolios. We show theoretically that negative wealth effects are consistent with both constant and decreasing relative risk aversion and analyze how our empirical estimates help distinguish between competing models of portfolio choice. We further show our results are quantitatively aligned with the predictions of a calibrated dynamic portfolio choice model with nontradable human capital and consumption habits.
Discussant: Sylvain Catherine, University of Pennsylvania
Abstract: We propose a dynamic theory of wealth concentration in which investors at the top of the wealth distribution choose to hold concentrated stakes in illiquid assets. In this framework, there are trading rents because wealth concentration leads to imperfect competition in financial markets. To capture these rents, wealthy investors consume too little and choose to remain under-diversified. As a result, fortunes are constantly created and destroyed. We provide conditions under which wealth concentration leads to inflated asset values, unequal returns to wealth, and low liquidity, and we discuss the feedback between real and financial market concentration and implications for measurement.