Conference on market microstructure.

The second formal meeting of the NBER's Market Microstructure Research Group, organized by NBER Research Associate Andrew Lo of MIT, took place in Cambridge on December 4. The discussion focused on the following papers:

Michael J. Fleming and Eli Remolona, Federal Reserve Bank of New York, "Price Formation and Liquidity in the U.S. Treasury Securities Market: The Response to Public Information"

Discussant: Torben Andersen, Northwestern University

Thomas Gehrig, University of Freiburg, and Matthew Jackson, Northwestern University, "Bid-Ask Spreads with Indirect Competition among Specialists"

Discussant: Eugene Kandel, Hebrew University

Martin Evans, Georgetown University, "The Microstructure of Foreign Exchange Dynamics"

Discussant: Richard K. Lyons, NBER and University of California, Berkeley 

Rajesh Chakrabarti and Richard Roll, University of California, Los Angeles, "Learning from Others, Reacting, and Market Quality"

Discussant: Blake LeBaron, NBER and University of Wisconsin

Eric Ghysels, Pennsylvania State University, and Mouna Cherkaoui, University Mohammed V, "Microstructure Reforms of an Emerging Market: Do They Really Improve Trading Costs?"

Discussant: Geert Rouwenhorst, Yale University

Robert Neal, Indiana University, and Simon Wheatley, University of New South Wales, "Adverse Selection and Bid-Ask Spreads: Evidence from Closed-End Funds"

Discussant: Lawrence Harris, University of Southern California 

Fleming and Remolona note that the arrival of public information in the U.S. Treasury securities market induces striking adjustment patterns for prices, trading volume, and bid-ask spreads. The release of a major macroeconomic announcement occasions a sharp and immediate price change with little trading volume, suggesting that price reactions to public information do not require trading. The bid-ask spread widens dramatically at announcement, and narrows shortly thereafter, apparently driven by inventory control rather than asymmetric information concerns. After the initial sharp price change, trading volume surges and persists with high price volatility and a slightly higher bid-ask spread, suggesting a sluggish process of price formation as the market reconciles investors differential private views.

Gehrig and Jackson examine the prices quoted by specialists (or dealers) who have monopoly power to set prices (bids and asks) for a given asset, but who face indirect competition from other specialists who trade in related assets. They compare the equilibrium spreads as the number (and factor structure) of the assets that each specialist controls varies. For some constellations of initial portfolio holdings and asset covariance, it is socially preferable to have competing specialists, while for others it is socially preferable to have actions coordinated (or to have one specialist control several assets). In some situations it is beneficial to have specialist power concentrated within industries; in other situations, across industries; and in other situations, not to be concentrated at all.

Evans studies the high frequency behavior of the interbank foreign exchange market with a newly created dataset that provides a comprehensive picture of activity across the market. His analysis indicates that trade activity within the interbank market is distinct from the posting of indicative quotes. Trading and quote-making decisions are linked, but the links are complicated and poorly understood. He also documents the existence of a strong relationship between exchange rate movements and a measure of excess dollar demand. Empirically, this effect appears important in the determination of exchange rates at both high frequencies and over the longer time spans that are relevant in international macroeconomics.

Traders in security markets take into account the actions of their peers in making their own trade decisions. In this paper, Chakrabarti and Roll compare a market in which traders "learn" from one another with a market in which they ignore each other's actions. The authors measure volatility, volume, and the accuracy of market price as a forecast of value in the "learning" and the "no learning" cases. While bubbles and cascades do arise some times, on average "learning" reduces price and return volatility and volume, and improves the accuracy of the market price as an indicator of value. The authors examine the marginal effects of different parameters on these market characteristics, and find that the "learning" process has a complex and nonlinear impact.

To enhance their understanding of emerging markets, Ghysels and Cherkaoui study an unusual dataset containing all the transaction records of a market over a long span. The market, which was included in 1996 in the International Finance Corporation database, operated under a dual trading system, consisting of an upstairs market for large block trades and a trading floor exchange. Transactions were recorded separately for both segments of the market. The authors: assess the quality of the market through the various stages of reform; examine the effect of microstructure reforms on the emergence of the market; and consider the price impact of large block trades. They also test whether the costs of trading have changed significantly since the stock market reforms. Their results show that the effective spreads and costs of trading have, if anything, increased except for the most actively traded stock.

Neal and Wheatley examine the performance of two commonly used empirical models for estimating the adverse selection component of a firm's bid-ask spread. They use the models to estimate the adverse selection components of a sample of closed-end funds and a matched sample of common stocks. In contrast to the stocks, closed-end funds report their net asset values weekly, all but eliminating uncertainty about their current liquidation values. Thus the authors expect the adverse selection component to be much smaller for the funds than for the stocks. Estimates of the component, however, are large and significant for both samples. This suggests that either adverse selection arises primarily from factors other than current liquidation value or that the empirical models are misspecified.