Transaction costs and security return behavior: The effect on systematic risk estimation and firm size

This study examines the effect of total transaction costs on security returns and develops an alternative return generating process that incorporates the effect of transaction costs on security returns. The demonstrable influence of transaction costs is the incidence of zero returns. The market model, in neglecting the zero returns, assumes a continuous relation between security returns and the market index. However, transaction costs inhibit this continuous relationship because of the zero returns. These zero returns suggest that the risk-return relationship is non-linear. The non-linearity between risk and return results in an expected return relation that prices transaction costs and the asset's own variance in addition to systematic risk. The market model, by pricing only systematic risk, has omitted variables. The omitted variables induce downward biased systematic risk estimates which, in turn, lead to a size effect wherein firm size replaces systematic risk as the primary means of explaining portfolio returns. When transaction costs are specifically modeled, the resulting systematic risk estimates supplant firm size as the primary means of explaining portfolio returns. The results are robust with respect to either the NYSE/AMEX or NASDAQ exchange.;Finally, an application of both the market model and transaction costs pricing relation is examined for firms that had experienced a 'subject to' audit qualification. Neither the market model nor the transaction costs pricing relation are able to detect abnormal performance on the public announcement date of an audit qualification contained in the 10-K disclosure. However, there is a significantly negative market reaction to the reporting of a delay in the scheduled 10-K release, or an NT 10-K. This is obtained whether or not an NT 10-K was filed in the previous year. In effect, non-financial disclosures of information are not significant, but delays in reporting the financial information are significant. The delay in filing a 10-K is viewed by market participants as bad news.;The transaction costs pricing relation is also compared to the market model in an event study simulation for both the NYSE/AMEX and NASDAQ exchanges. Although the market model produces downward biased systematic risk estimates, the market model is well specified in an event study context. The transaction costs pricing model does not improve the performance in detecting abnormal performance.