Determining security speed of adjustment coefficients

Abstract

Speeds of adjustment of asset prices towards their intrinsic values will provide direct measures of the degrees of over and underreactions in financial markets. The Amihud and Mendelson (J. Finance 42 (1987) 533) partial adjustment with noise model provides the framework for the analysis presented in this paper. Speed of adjustment coefficient estimators are developed which have advantages over existing estimators in that they can be adjusted for thin trading and have associated sampling distributions. The empirical properties of these estimators are found to be superior to extant estimators. Stock prices are found to be characterised by speeds of adjustment less than complete at short differencing intervals, while evidence of overreaction at longer differencing intervals is found. Large capitalisation stock speeds of adjustment coefficients are found to be higher in most cases than for small capitalisation stocks, even after adjusting for thin trading.