Differences in the Value Relevance of Earnings in Knowledge-Based and Traditional Industries

Abstract

This paper evaluates the value relevance of earnings in knowledge-based and traditional industries. We focus on two factors - earnings' lack of timeliness and noise - that contribute to the weak contemporaneous association between earnings and stock returns in knowledge-based industries. Earnings' lack of timeliness represents the non-contemporaneous association between earnings and stock returns, whereas noise - arising from uncertainty in estimating the expected benefits of investments in intangibles - is unassociated with either contemporaneous, past, or future returns. To correct earnings' lack of timeliness, we derive a new model that includes future earnings changes, an earnings-to-price ratio, and future stock returns; we abate noise by aggregating yearly data within the same industry. We find adjusting for lack of timeliness is more effective in traditional industries, and diversifying away noise in knowledge-based industries increases the association of earnings and stock returns substantially. The value relevance of earnings in both types of industries is indistinguishable when we jointly control noise and lack of timeliness. We conclude that noise is the main factor contributing to the weak association between earnings and stock returns for firms in knowledge-based industries.