Although we believe the weight of the evidence to date suggests the use of the IR

factor as a proxy for information risk, research by Core et al. (2008) using a two-stage

cross-sectional regression approach reports that, in the second stage, the factor

loading on the IR factor from the first stage is not significantly associated with future

returns. However, Nichols (2006) also uses the two-stage cross-sectional regression

approach and finds that the factor loading on IR and the innate portion of IR are the

only risk factors for which there is an association with returns consistent with these

factors being priced.8 Furthermore, accepted risk factors in the literature, such as the

market, size, and book to market factors, are frequently insignificant in the second stage regression (including regressions reported by Core et al.) suggesting that the

interpretation of the results is not clear. A definitive finding that the IR factor is not a

priced risk factor would limit the inferences that we could draw about the association

between the cost of capital and restatements from our tests. Liu and Wysocki (2007) examine the association between several proxies for information quality (including the AQ metric) and proxies for the cost of capital (including an implied cost of equity capital measure). The AQ metric is generally significant when included without controls for underlying firm characteristics, including the five variables listed in Dechow and Dichev (2002) and used by Francis  et al. (2005) to partition the AQ metric into its innate and discretionary components.When these controls are included the AQ metric loses its significance consistent with the results in Francis et al. (2005) that innate AQ is more strongly associated with the cost of capital than the other, more discretionary component. Results in Francis et al. (2005), Liu and Wysocki (2007) and Nichols (2006) suggest that in broad samples, the innate component of AQ contains much of the  explanatory power of the overall AQ metric. In contrast, we exploit a setting and a sample where we expect investors’ concerns about the discretionary portion of IR to drive changes in the pricing of information risk.