Asset Pricing and Portfolio AllocationNBER Reporter: Summer 2000
The stochastic discount factor (SDF) method provides an elegant and unified general framework for econometric analysis of linear and nonlinear assetpricing models, including derivative pricing models. But Jagannathan and Wang ask whether the generality of the SDF methodology comes at a cost in estimation efficiency. They show that for linear beta pricing models, they show that the SDF method provides estimates of factor risk premiums that are as precise as those obtained using classical regression methods. In the special case where the mean and variance of the factors are known, though, the common practice of ignoring the implied restrictions on the moments of the factors makes the SDF method substantially less precise than the regression method. However, proper incorporation of relevant restrictions can make the SDF method as asymptotically precise as the classical regression method.
Loewenstein and Willard describe a rational competitive equilibrium for an economy in which some investors' liquidity needs are uncertain in terms of timing. Apparent arbitrage opportunities, such as convergence trades, exist in this equilibrium. These arbitrages are sure to ultimately produce a profit, but their intermediate marked to market valuations are risky. Investors facing liquidity risk optimally do not exploit potential arbitrages, but those who can withstand temporary losses -- for example, leveraged hedge funds -- exploit arbitrages to a finite scale determined by the endogenous supply of credit. Loewenstein and Willard provide new results pertaining to the significant and sometimes counterintuitive effects of illiquidity on empirical tests.
Recent research in finance has shown the importance of time horizons in models of learning. The dynamics of how agents adjust to ultimately believe that the world around them is stationary may be just as crucial in the convergence to a rational expectations equilibrium as getting parameters and model specifications correct in the learning process. LeBaron explores this evolution in a simple agent-based financial market. His results indicate that, while the simple model structure he uses replicates the usual rational-expectations results with long-horizon agents, evolving a population of both long- and short-horizon agents to only long horizons may be difficult. Furthermore, populations with both long- and short-horizon agents increase the variability of returns and leave patterns in volatility and trading volume that are similar to actual financial markets.
Duffie, Pedersen, and Singleton construct a model for pricing sovereign debt which accounts for the risks of default and restructuring and for the compensation for illiquidity. Using a new and relatively efficient method, they estimate the model using data on Russian dollar-denominated bonds. The authors consider the determinants of the Russian yield spread; the yield differential across different Russian bonds; and the implications for market integration, relative liquidity, relative expected recovery rates, and implied expectations of different default scenarios.
Piazzesi develops an arbitrage-free time-series model of yields that incorporates central bank policy. The model introduces a class of linear-quadratic jump diffusions as state variables. She uses a special case of this setup to describe U.S. interest rates, the Federal Reserve's target rate, and key macroeconomic aggregates. The U.S. application captures: 1) target-rate moves on Federal Open Market Committee (FOMC) meeting days, 2) "exceptional" policy moves outside of FOMC meetings, and 3) releases of macroeconomic news that are likely to affect future Federal Reserve actions. To fit the model, she extends the simulated maximum-likelihood estimation to allow for jump diffusions. Introducing the target rate as a fourth, observable factor into a framework with three latent factors is a tractable way of improving the overall term-structure fit, especially at short maturities. A policy inertia factor influences the conditional probability of changes in the target. Fed policy is linked to the increased volatility of yields on FOMC meeting and release days and to the observed "snake-shaped" term structure of yield volatility.
Using pooled cross-sectional data from the 1962-3, 1983, 1989, 1992, and 1995 Surveys of Consumer Finance and panel data from TIAA-CREF, Ameriks and Zeldes examine the relationship between age and portfolio choice and especially between age and the fraction of wealth held in the stock market. They illustrate and discuss the importance of the well-known identification problem that prevents unrestricted estimation of age, time, and cohort effects in longitudinal data. Ameriks and Zeldes also document three important features of household portfolio behavior: significant nonstock ownership, wide-ranging heterogeneity in allocation choices, and the infrequency of active portfolio allocation changes. By including age and time effects (but excluding cohort effects), they show that equity ownership has a hump-shaped pattern with age, while equity shares conditional on ownership are nearly constant across age groups. Including age and cohort effects (but excluding time effects) shows that equity portfolio shares increase strongly with age. Following the same individuals over time, the authors find that almost half of the sample members made no active changes to their portfolio allocations over the nine-year sample period; the vast majority of those who did make changes increased their allocations to equity as they aged.
Kremer and Mehta examine the effect of reduced transactions costs in the international trading of assets on the ability of governments to issue debt. In their model, governments care about the welfare of their citizens and thus are more inclined to default if debt is held largely by foreigners. Reductions in transactions costs make it easier for domestic citizens to share risk by selling debt to foreigners. This may increase tendencies for governments to default and thus raise the cost of credit and reduce welfare. Kremer and Mehta find that even in the absence of transactions costs, there may be a home bias in the placement of government debt; in the presence of default risk, the return on government debt is correlated with the tax burden required to pay the debt. Asset inequality may reduce this home bias and, by increasing foreign ownership, increase incentives for default. Finally, if foreign creditors are less risk averse than domestic creditors, there may be one equilibrium in which domestic creditors hold the asset and default risk is low and another in which foreign creditors hold the asset and default risk is high.
Brav, Constantinides, and Geczy analyze household consumption of nondurables and services, as reconstructed from the Consumer Expenditure Survey database. They observe that the estimated co-efficient of relative risk aversion for the representative household decreases, and the estimated unexplained mean equity premium decreases, as inframarginal asset holders are eliminated from the sample. These results provide evidence of limited capital market participation. The estimated unexplained mean equity premium decreases when the assumption of complete consumption insurance is relaxed. The estimated correlation between the equity premium and the cross-sectional variance of the households' consumption growth is negative, as required, if the relaxation of market completeness is to contribute towards the explanation of the premium. The overall evidence from asset prices in favor of relaxing the assumption of complete consumption insurance is weak. An extensive Monte Carlo investigation highlights the relationship between the economic implications of limited participation and the resulting statistical properties of commonly used test statistics. The simulation results provide direct evidence relating observation error in consumption to the resulting small sample properties of the test statistics.