Measuring value relevance in a (possibly) inefficient market

An interesting question in assessing value relevance of accounting variables is whether measures of value relevance are materially affected by market inefficiencies. We explore this question in two steps: First, we analytically examine the impact of market inefficiencies on the estimation of coefficients in value relevance regressions and derive a procedure that corrects potential biases caused by such inefficiencies. The procedure adjusts contemporaneous stock prices for future risk adjusted price changes, and yields value relevance coefficient estimates that capture both contemporaneous and delayed market reactions. Second, we apply this procedure to three types of studies that have attracted much attention in the accounting literature: 1) the value relevance of earnings and book values; 2) the value relevance of residual income value estimates; and 3) the value relevance of accruals and cash flows. We compare coefficient estimates obtained from conventional value relevance regressions with those from regressions employing our adjustment procedure, and find statistically significant differences in both level and return regression coefficient estimates. The magnitude of differences in coefficient estimates for return regressions is large enough to affect economic inferences. We find that coefficients of lagged price deflated residual income value estimates move significantly closer toward a predicted value of one implying a meaningful reduction of bias. Last, we find that cash flows now have significantly larger coefficient estimates than accruals consistent with their greater persistence.