Asset pricing program.

The NBER's Asset Pricing Program met in Cambridge on April 25. Stanley E. Zin, NBER and CarnegieMellon University, organized the session and chose the following papers for discussion:

David Backus, NBER and New York University; Silverio Foresi, Abort Mozumdar, and Liuren Wu, New York University, "Predictable Changes in Yields and Forward Rates"

Discussant: Erzo Luttmer, Northwestern University

William T. Roberds, Federal Reserve Bank of Atlanta, and Charles H. Whiteman, University of Iowa, "Endogenous Term Premia and Anomalies in the Term Structure of Interest Rates: Explaining the Predictability Smile"

Discussant: Robert J. Hodrick, NBER and Columbia University

Qiang Dai, Stanford University, and Kenneth J. Singleton, NBER and Stanford University, "Specification Analysis of Affine Term Structure Markets"

Discussant: David Marshall, Federal Reserve Bank of Chicago 

Wayne E. Ferson, NBER and University of Washington, and Andrew F. Siegel, University of Washington, "The Efficient Use of Conditioning Information in Portfolios"

Discussant: John C. Heaton, NBER and Northwestern University

William Fung, Paradigm LDC, and David A. Hsieh, Duke University, "Empirical Characteristics of Dynamic Trading Strategies: the Case of Hedge Funds"

Discussant: Andrew W. Lo, NBER and MIT 

W. Brian Arthur, John H. Holland, Richard Palmer, and Paul Tayler, Santa Fe Institute; and Blake D. LeBaron, NBER and MIT, "Asset Pricing Under Endogenous Expectations in an Artificial Stock Market"

Discussam: Bryan Routledge, Carnegie-Mellon University

The first three papers study the behavior of long-term bond yields in relation to short-term interest rates that is, the term structure of interest rates. Recent studies have explored the ability of spreads between long and short yields to forecast subsequent movements in interest rates, and have found that spreads do forecast interest rate movements over short horizons up to about three months, and over long horizons beyond about two years, but do not forecast interest rate movements at intermediate horizons. This pattern is sometimes called a "predictability smile," because a graph of interest rate predictability has the shape of a smile. One explanation for this pattern is that long yields are influenced not only by interest rate movements, but also by changes in risk premiums on long-term bonds.

The papers by Backus et al. and by Roberds and Whiteman ask whether simple models with endogenous time-varying risk premiums can explain the predictability smile. Roberds and Whiteman explore a single-factor model proposed in a well-known paper by Cox, Ingersoll, and Ross; in this model the volatility and level of the short-term interest rate move together and drive the movements of the whole term structure. Roberds and Whiteman argue that this model has the potential to explain the predictability smile; Backus et al. respond that it can fit the smile or the average levels of yield spreads, but cannot explain both the smile and average spreads at the same time.

Backus et al. and Dai and Singleton explore more complicated multifactor models in which the level, long-run mean, and volatility of the short rate can move independently. For tractability, they consider "affine" models in which bond yields are related to each other linearly. Dai and Singleton provide a general framework in which to analyze such models, and they fit several models to the joint distribution of short- and long-term interest rates. Backus et al. find that these models have the potential to explain both the predictability smile and other properties of interest rates, although there are some remaining empirical difficulties at long maturities.

Dynamic trading strategies are increasingly popular in financial markets, and the next two papers on the program study the returns on such strategies. Ferson and Siegel ask what portfolio allocation rules produce the highest possible unconditional expected returns given their unconditional variances of returns, when the portfolio weights can be adjusted dynamically in response to conditioning information. They provide closed-form solutions for these unconditionally mean-variance efficient portfolios.

Fung and Hsieh present some new results on an unexplored dataset on hedge fund performance. Hedge funds are unregulated, private investment pools. Their trading strategies are dramatically different from mutual funds, and their returns have low correlation to the major asset markets. Yet most hedge funds use the same asset markets as mutual funds to generate returns, which supports the claim that hedge fund strategies are highly dynamic. The authors find five dominant investment styles in hedge funds, which can account for only 43 percent of their cross-sectional variance. This means that hedge funds use many different dynamic trading strategies to generate returns.

Arthur et al. propose a theory of asset pricing based on heterogeneous agents who continually adapt their expectations to the market that these expectations aggregatively create. They explore the implications of this theory computationally using a simulation of an artificial stock market. They show that within a regime where investors explore alternative expectational models at a low rate, the market settles into the rationalexpectations equilibrium of the efficient-market literature. Within a regime where the rate of exploration of alternative expectations is higher, the market self-organizes into a complex pattern. It acquires a rich psychology, technical trading emerges, temporary bubbles and crashes occur, and asset prices and trading volume show statistical features characteristic of actual market data.