On estimating market efficiency

Abstract

In this paper, we discuss commentaries by Maynes and Hjorth-Andersen on our earlier paper (1990), and then go on to discuss the merits of various measures of market efficiency which have appeared in the literature. While Maynes criticized the model in our 1990 paper for a lack of realism, we argue that our limited objective of demonstrating that price-quality correlations are not necessarily related to market efficiency did not require a model which was realistic in all details. We also demonstrate that our basic conclusion that the price-quality correlation need not measure market efficiency does not depend on our theoretical model. Hjorth-Andersen advanced a number of alternative reasons why price-quality correlations may not measure efficiency, and we view his arguments largely as complementary to ours. The basic conclusion is that there are severe problems with interpreting measured price-quality correlations as measures of efficiency. We go on to discuss alternative measures. While no alternatives are completely satisfactory, we argue that measures based on deviations from an efficient frontier have some attractive properties, and are currently the most desirable alternative.