Noise In the Price Discovery Process: A Comparison of Periodicand Continuous Auctions

We examine market volatility across an automated periodic auction mechanism and a continuous automated auction, using data on five futures contracts traded on the GLOBEX trading system. The analysis is supplemented by a comparison of the periodic market with floor trading. Our data permit virtually no lag between the close of continuous trading and the periodic auction just preceding further continuous trading, avoiding ambiguities in results due to periods of non-trading. Theoretical results are summarized, which suggest greater volatility in the continuous market, but which also imply equality of price volatility across periodic and continuous mechanisms for some market conditions. The empirical analysis consists both of direct comparisons via variance ratios and variance ratio tests conditional on a model of price and value adjustment previously suggested by Amihud and Mendelson (1987). We show how to directly estimate the structural parameters of the model, and derive a test of market efficiency conditional on the model's structure. Our results indicate that the periodic auction yields the same volatility as the automated continuous market for a stock index contract, but that the periodic market is associated with relatively lower volatility for currency futures. Tests fail to reject market efficiency for the index contract and three out of four currency futures.