Asset Pricing

The NBER's Program on Asset Pricing met on March 24 on the campus of the University of California, Los Angeles. Program Director John H. Cochrane of the University of California, Los Angeles, and Monika Piazzesi, NBER and University of California, Los Angeles, organized the meeting and chose the following papers for discussion:

Duffie, Garleanu, and Pedersen study the impact on asset prices of illiquidity associated with search and bargaining in an economy in which agents can interact only when they find each other. Even when market makers are present, investors' abilities to meet directly are important. Prices are higher and bid-ask spreads lower if investors can find each other more easily. Prices approach the Walrasian price if investors' search intensity increases or if market makers, who do not have all the bargaining power, search more intensely. Endogenizing search intensities yields natural implications. Finally, the authors show that information can fail to be revealed through prices when searching is difficult.

Veronesi reinterprets standard axioms in choice theory to introduce the concepts of "belief dependent" utility functions and aversion to "state uncertainty." Within a standard pure-exchange economy in which investors ignore the long-run drift of consumption growth ("the state"), such preferences help explain the various stylized facts of stock returns, including a high-equity risk premium, a low risk-free rate, high return volatility, stock return predictability, and volatility clustering. Since the long-run drift of consumption determines the (average) path of future consumption, aversion to state uncertainty suggests aversion to the dispersion of long-run consumption paths. This differs from the standard notion of (local) risk aversion in its temporal dimension. When the model is calibrated to real consumption, it generates unconditional moments for asset returns that confirm the empirical observations. In addition, when it is estimated using consumption data, the fitted model produces posterior distributions on the drift rate of consumption that are relatively dispersed, further motivating the notion of aversion to long-run risk. In theory, an investor can make infinite profits by taking unlimited positions in an arbitrage. In reality, investors must satisfy margin requirements that completely change the economics of arbitrage.

Liu and Longstaff derive the optimal investment policy for a risk-averse investor in a market in which arbitrage opportunities exist. They show that it is often optimal to underinvest in the arbitrage by taking a smaller position than margin constraints allow. In some cases, it is actually optimal to walk away from a pure arbitrage opportunity. Even when the optimal policy is followed, the arbitrage strategy may underperform the riskless asset or have an unimpressive Sharpe ratio. Furthermore, the arbitrage portfolio typically experiences losses at some point before the final convergence date. These results have important implications for the role of arbitrageurs in financial markets.

Recent equity carve-outs in U.S. technology stocks appear to violate a basic premise of financial theory: identical assets have identical prices. Lamont and Thaler explore a sample of company shareholders who expect to receive shares of another company. A prominent example involves 3Com and Palm. The authors find that arbitrage does not eliminate blatant mispricings resulting from short sale constraints, so that one stock is overpriced but expensive or impossible to sell short. Evidence from options prices shows that shorting costs are extremely high, eliminating exploitable arbitrage opportunities.

Mitchell, Stafford, and Pulvino examine the impediments to arbitrage in 82 situations between 1985 and 2000, where the market value of a company is less than the sum of its publicly traded parts. These situations, often referred to as "negative stub values," suggest clear arbitrage opportunities and provide an ideal setting in which to study the risks and market frictions that prevent arbitrageurs from forcing prices to fundamental values. The authors find that 30 percent of negative stub deals terminate without converging. In addition, they estimate that the returns to a specialized arbitrageur would be 50 percent larger if the path to convergence was smooth rather than volatile, but the marginal investor in negative stub values is not likely to be so specialized. Short-selling frictions do not appear to be a major impediment to arbitrage in their sample. Information transaction costs appear to be the most serious limit to arbitrage.

Wachter proposes a model that captures the ability of the yield spread to predict excess returns on bonds as documented in empirical studies. The model, a generalization of Campbell and Cochrane (1999), also captures the predictability of stock returns by the price-dividend ratio, a high equity premium, excess volatility, positive excess returns on bonds, and an upward-sloping average yield curve. The model implies a joint process for interest rates and consumption. Controlling for contemporaneous consumption growth, long lags of consumption predict the interest rate. Thus the success of the model is based on a more realistic process for consumption and the interest rate, rather than on additional degrees of freedom in the utility function.

Bossaerts, Plott, and Zame develop structural tests of asset pricing theory to apply to data from experimental financial markets. The tests differ from those used with field data because they verify the consistency between prices and allocations, rather than merely testing whether only prices satisfy equilibrium restrictions. The authors test two large-scale financial market experiments in order to resolve an apparent price-allocation paradox in the experiments: namely, why asset prices can satisfy theoretical restrictions (CAPM) when allocations deviate substantially. They find that allocations seem not to be any closer to CAPM predictions than when prices grossly violate CAPM restrictions. Their theory explains the paradox in both sets of experiments; that is, when end-of-period prices are such that the market portfolio is close to mean-variance efficient (the CAPM pricing prediction), the allocations and their theory explain why. When end-of-period prices make the market portfolio far from mean-variance optimal, their tests reject.