W. Bruce Johnson

Valuation Model Use and the Price Target Performance of Sell-Side Equity Analysts

This study investigates the influence of inferred valuation model use on the investment performance of sell-side equity analysts’ published price target opinions. There is limited and inconclusive evidence on how analysts’ price targets are determined and on their value for investment decisions. Using a broad sample of 45,693 price targets provided to First Call by sell-side analysts during 1997 through 2003, we first show that price targets have investment value because they predict future stock returns. Next, we develop and implement an innovative large-sample procedure for inferring valuation model use from the observed correlation between analysts’ price targets and two researcher-constructed stock valuation estimates that differ in simplicity and rigor. Reliance on a less rigorous valuation model may diminish the investment advantage associated with an analyst’s more accurate earnings forecasts but it may also mitigate the disadvantage of less accurate forecasts. We test whether the apparent use of a more rigorous valuation technique yields higher quality price targets as measured by realized investment returns over a 12-month horizon, controlling for possible differences in earnings forecast accuracy. The central message from our data is that price targets exhibit superior investment performance when analysts appear to be using a fundamental residual income (RIM) stock valuation technique rather than a simple price-earnings-growth (PEG) valuation heuristic. This investment advantage is reduced when analysts’ earnings forecasts are inaccurate. Our results underscore the importance of valuation model choice to analyst’s stock investment evaluation process.

Contrarian Share Price Reactions to Earnings Surprises

A persistent (but overlooked) feature of the cross-sectional distribution of quarterly earnings announcement returns is that the measured earnings surprise and share price response to that surprise are often in the opposite direction. Extending the study by Kinney, Burgstahler, and Martin, this study provides evidence on the prevalence, determinants, and consequences of contrarian stock returns at the earnings announcement date. Using the most recent Institutional Brokers’ Estimate System (I/B/E/S) consensus earnings per share forecast as the earnings benchmark, the authors find that contrarian returns occur for roughly 40% of the more than 230,000 quarterly earnings announcements that comprise their sample. Contrarian returns are only slightly less prevalent in extreme earnings surprise deciles and are evident each quarter during 1985-2005. The incidence of contrarian returns is statistically related to “noise” in the measured earnings surprise (stale I/B/E/S consensus forecasts, preannouncement stock returns, and the presence of Generally Accepted Accounting Principles [GAAP] exclusions) and “noise” in the share price response to announced earnings (discordant revenue changes, discordant earnings forecast revisions, return volatility, bid-ask spread, and discordant prior quarter earnings surprises). Finally, contrarian stocks exhibit little post-earnings-announcement drift.

The Materiality of Accounting Errors: Evidence from SEC Comment Letters: The Materiality of Accounting Errors: Evidence from SEC Comment Letters

To gain unique insights into the materiality judgments of financial statement preparers, we examine firms’ responses to SEC comment letters that inquire about accounting error materiality. We document that preparers typically use multiple quantitative benchmarks in their materiality analyses and frequently acknowledge departures from the traditional “5 percent of earnings” rule of thumb when deeming errors immaterial. These departures are explained by asserting that the benchmark hurdle is abnormally low and/or that the error does not consistently exceed the hurdle across affected periods. We also document substantial variation in the extent to which qualitative factors are mentioned as considerations. Leveraging these comment letter insights, we propose and test a parsimonious model of materiality judgments that uses data commonly available to researchers. Our tests affirm that a materiality benchmark based on the three-year average of earnings predicts actual materiality judgments better than quantitative proxies used in prior research. Further, we find that including both a control for financial statement misclassification errors and the existence of multiple errors dramatically improves model explanatory power. Our findings shed new light on firms’ complex and nuanced materiality evaluations and help to improve the materiality proxies used in empirical archival research.