Scott Tolton and Michael Drake, School of Accountancy

We behave differently when we are being watched. A large body of research finds that when observed, individuals are more likely to act in accordance with the perceived expectations or desires of the observer as a means to achieve a specific goal (e.g., positive appraisal, promotion, penalty avoidance, etc.).¹ In the corporate space, prior studies provide considerable evidence on how attention or monitoring from various parties such as auditors, analysts, business press reporters, institutional investors, and debt holders affects the financial reporting choices of managers. The current literature finds almost unilateral evidence that increased attention from these outsiders is associated with higher quality financial reports (Becker et al., 1998; Sweeney, 1994; Chung et al., 2002; Dyck et al., 2008; Yu, 2008). We extend this literature by investigating how attention paid to the firm by a previously unexamined group of outsiders—its customers—impacts the aggressiveness of managers’ financial reporting choices.

To test the relation between customer attention and financial reporting aggressiveness, we introduce a new measure of customer attention based on the number of individuals who visit the firm’s homepage during the year.⁴ We validate this measure as a proxy for customer attention in two ways. First, under the assumption that customers will pay attention to the announcement of a new product, we examine changes in our customer attention measure around product releases. Consistent with the measure capturing customer attention, we find that the level of firm website traffic increases significantly (it is approximately six times greater) on product release dates than on the average day, and that it is significantly greater on product release dates than on earnings announcement dates. Second, we show that our homepage visitor measure is highly, positively associated with current and future sales, which are relations we would expect to observe if the measure captures customer attention.

We measure aggressive financial reporting using two sets of proxies common in the literature. The first set includes financial misstatements (as revealed through subsequent restatements) and revenue misstatements. The second set consists of two variables related to the likelihood of meeting or just beating analyst earnings forecasts by one cent or less (Brown and Caylor, 2005), and the likelihood of meeting or just beating analysts’ revenue forecasts (Ertimur et al., 2003; Rees and Sivaramakrishnan, 2007).

Our empirical tests are based on a sample of 7,804 firm-year observations from 2007 to 2013. Consistent with our prediction, we document a significantly positive relation between customer attention and aggressive financial reporting behavior. We obtain these results after controlling for a broad set of firm and market characteristics, as well as the activities of other important monitors of the firm. These findings are consistent with our prediction that increased customer attention is associated with managers being more aggressive in their financial reporting behavior.

Because the number of homepage visitors is not randomly determined, our use of nonexperimental data may produce results that are inconsistent or biased. We address this possibility in two ways. First, we estimate a two-stage Heckman selection model under the assumption that the level of homepage visitors is a function of an underlying, unobservable selection process. To satisfy the exclusion restriction, we use an indicator for whether customers can access the homepage in order to transact business directly with the firm online. We find that all our inferences hold after controlling for a selection bias correction term in our main models.

Second, we recognize that our multiple regression analyses could suffer from functional form misspecification, which we address using a propensity-score matched sample (Armstrong et al., 2010; Shipman et al., 2016). Specifically, we construct a sample that matches firms that are similar along a host of relevant observable dimensions (i.e., firm and market characteristics, ability to transact business directly with the firm via the homepage, attention from other monitors of the firm, industry membership) that may be associated with a firm’s level of homepage visitors, but that are different in the amount of homepage visitors they actually receive.

Next, we consider alternative measures of aggressive financial reporting including exante fraud risk (AGR), absolute discretionary accruals, and discretionary revenue. We find that each of these measures is also positively associated with our measure of customer attention, thus providing corroborating evidence for our main findings.