This study provides evidence that hyperlinking a firm's audited financial statements to unaudited information in a web-based environment leads investors to blend the unaudited information with the audited statements. I obtain evidence of this blending effect using an experiment where investors assessed a firm's earnings potential by evaluating the firm's audited financial statements and a subsequent optimistic unaudited letter to shareholders from the firm's management. Investors who viewed hyperlinked materials on the Web misclassified more unaudited information as audited and assessed the credibility of the unaudited information as higher than did investors who viewed hardcopy materials. Those investors who assessed the unaudited information as more credible also judged the firm's earnings potential to be higher. Notifying users with an “AUDITED/NOT AUDITED” label attenuated these effects. This evidence suggests that firms can influence financial report users' perceptions by hyperlinking unaudited information to information in their audited financial statements, and that a simple disclosure rule reduces this influence.

This study shows experimentally that when individuals use information on intangibles expenditures to predict future profits, expensing (vs. capitalizing) the expenditures significantly reduces the accuracy, consistency, consensus, and self-insight of individuals' subjective profit predictions. The experimental design allows us to eliminate several competing explanations for this apparent fixation on accounting. Subjects do not base their judgments on a nai¨ve prior belief that expensing precludes effects on future profits; a preexperiment question shows that subjects expect intangibles expenditures will affect future profits even when expensed. Moreover, subjects do not lack, or fail to use, data that would allow them to learn the exact expenditure-profit relation. They receive data on intangibles expenditures and profits as a basis for learning, and in some respects the learning is quite successful even when intangibles are expensed; subjects' profit predictions accurately reflect the mean and standard deviation of actual profits. Nevertheless, consistent with psychological theories of learning, subjects do not learn the exact magnitude of the effect of intangibles on future profits as well when the intangibles are expensed. Although the mean of their predictions is accurate, they do not discriminate well between cases with high and low actual profits. In consequence, their prediction accuracy, consistency, consensus, and self-insight are lower when intangibles are expensed. Thus, in this case, learning does not mitigate fixation on accounting, because accounting affects the learning process itself.

We consider how auditors assess the risk of fraudulent financial reporting and plan their audit where a possibly fraudulent auditee anticipates the assessment and planning process. The auditor uses the auditee's (possibly fraudulent) earnings report to revise his beliefs about the likelihood of fraud when formulating an audit plan. We find that as underlying earnings increase, a fraudulent auditee increases reported earnings. In turn, as the auditee's reported earnings increase, the auditor increases audit effort. We also find that the auditee (who knows the auditor will use the report for audit planning) selects reports that increase his own expected payoff, relative to reports he would select if the auditor did not observe the report before finalizing the audit plan. By contrast, the auditor is no better off using the auditee's report for audit planning. Inherent risk, detection risk, and overall audit risk can increase when the auditor uses the auditee's report. Thus, because of the dynamic interaction between the auditor and auditee, procedures that aid in assessing audit risk may not reduce that risk or result in more efficient audits.