John D. Lyon

Detecting Long-Run Abnormal Stock Returns: The Empirical Power and Specification of Test Statistics

We analyze the empirical power and specification of test statistics in event studies designed to detect long-run (one- to five-year) abnormal stock returns. We document that test statistics based on abnormal returns calculated using a reference portfolio, such as a market index, are misspecified (empirical rejection rates exceed theoretical rejection rates) and identify three reasons for this misspecification. We correct for the three identified sources of misspecification by matching sample firms to control firms of similar sizes and book-to-market ratios. This control firm approach yields well-specified test statistics in virtually all sampling situations considered.

Improved Methods for Tests of Long-Run Abnormal Stock Returns

We analyze tests for long-run abnormal returns and document that two approaches yield well-specified test statistics in random samples. The first uses a traditional event study framework and buy-and-hold abnormal returns calculated using carefully constructed reference portfolios. Inference is based on either a skewnessadjusted t-statistic or the empirically generated distribution of long-run abnormal returns. The second approach is based on calculation of mean monthly abnormal returns using calendar-time portfolios and a time-series t-statistic. Though both approaches perform well in random samples, misspecification in nonrandom samples is pervasive. Thus, analysis of long-run abnormal returns is treacherous. COMMONLY USED METHODS TO TEST for long-run abnormal stock returns yield misspecified test statistics, as documented by Barber and Lyon ~1997a! and Kothari and Warner ~1997!. 1 Simulations reveal that empirical rejection levels routinely exceed theoretical rejection levels in these tests. In combination, these papers highlight three causes for this misspecification. First, the new listing or survivor bias arises because in event studies of long-run abnormal returns, sampled firms are tracked for a long post-event period, but firms that constitute the index ~or reference portfolio! typically include firms that begin trading subsequent to the event month. Second, the rebalancing bias arises because the compound returns of a reference portfolio, such as an equally weighted market index, are typically calculated assuming periodic ~generally monthly! rebalancing, whereas the returns of sample firms are compounded without rebalancing. Third, the skewness bias arises because the distribution of long-run abnormal stock returns is positively skewed, * Graduate School of Management, University of California, Davis. This paper was previously entitled “Holding Size while Improving Power in Tests of Long-Run Abnormal Stock Re

Detecting Abnormal Operating Performance: The Empirical Power and Specification of Test Statistics

This research evaluates methods used in event studies that employ accounting-based measures of operating performance. We examine the choice of an accounting-based performance measure, a statistical test, and a model of expected operating performance. We document the impact of these choices on the test statistics designed to detect abnormal operating performance. We find that commonly used research designs yield test statistics that are misspecified in cases where sample firms have performed either unusually well or poorly. In this sampling situation, the test statistics are only well specified when sample firms are matched to control firms of similar pre-event performance.

The Importance of Business Risk in Setting Audit Fees: Evidence from Cases of Client Misconduct

Previous research provides evidence that, for the clients of a large audit firm, audit clients with higher perceived business risk bear the expected costs of this risk with higher audit fees. We extend the literature, which focuses on the relation between litigation risk and audit fees, by examining alleged client misconduct that is not illegal but possibly increases business risk. In particular, we examine the relation between audit fees and business risk for audit clients doing business in developing countries where bribery of top government officials has been an accepted business practice. We hypothesize that bribery-paying clients are riskier because of both client business risk and audit business risk. Using data collected from Securities and Exchange Commission filings and audit fee data in the 1970s, before the passage of the Foreign Corrupt Practices Act, we provide evidence that audit fees were higher for clients that disclosed paying bribes. This evidence is consistent with an audit market where auditors assess business risk at the client level, then pass their expected costs to the client in the form of higher audit fees.

A comparison of tests for heteroscedasticity

We study the relationship, the size, the power, the sensitivity to high leverage, outliers and the normal error distribution of eight likelihood ratio and score tests for heteroscedasticity. Four versions of likelihood ratio tests are investigated which include the ordinary likelihood ratio test, the conditional likelihood ratio test obtained from Hondas conditional likelihood function, the residual likelihood ratio test derived from Verbylas residual likelihood function and the modified likelihood ratio test adapted from Simonoff and Tsai. Furthermore, four score tests are examined which contain Breusch and Pagans score test and its Studentized version (Koenkers score test), Verbylas score test and the White score test which has been adopted by the SAS Institute. For normal errors, Monte Carlo results show that Verbylas residual likelihood ratio test holds its null size well and is more powerful than other tests. For long-tailed or contaminated error distributions, Koenkers score test holds its null size better than other tests. In addition, Breusch and Pagans and Verbylas score tests are robust against high leverages. Finally, Monte Carlo studies show that Koenkers test outperforms Whites test and should be used routinely in place of Whites test. An example is presented to illustrate the test statistics.

Firm growth and the valuation relevance of earnings levels, earnings innovations and dividends

The objective of this paper is to empirically assess the valuation relevance of earnings levels, earnings innovations, and past dividends for firms with abnormal and normal growth profiles. Abnormal growth firms are firms for which investors expect the firm’s future cash flows to deviate substantially from accounting measurements of its past transactions (i.e., book value of equity differs substantially from the firm’s market value of equity). However, investors’ expectations for normal growth firms closely mirror the firm’s past transactions such that the firm’s market value deviates only modestly from its book value. The intuition behind our modeling strategy follows from the notion that book value (owner’s equity) is a long-run anchor for firm value (Ohlson [1991b]). Book value is a long-run anchor in the sense that total dividends (over the life of the firm) are equal to capital contributions plus total earnings given clean-surplus accounting. Thus, for normal growth firms (firms for which market and book values are close) book value may be (nearly) value sufficient; on the other hand, for abnormal growth firms (firms for which market and book value deviate substantially) current earnings and dividends may be more important indicators of future dividends than current book value. In turn, this implies that changes in firm value are reflected in earnings levels in the former case and changes in firm value are reflected in changes in earnings and dividends in the latter case.’

Improved tests for the first-order autoregressive model with heteroscedasticity

In the first-order autoregressive model with nonconstant variance, likelihood ratio tests, score tests and modified likelihood ratio tests are derived to test each of the following hypotheses: (i) independence and homoscedasticity, (ii) independence, and (iii) homoscedasticity. Monte Carlo studies show that likelihood ratio tests can be very anticonservative, score tests are conservative and modified likelihood ratio tests are reliable and powerful tests. In addition, modified likelihood ratio tests for testing hypotheses (ii) and (iii) are robust in the presence of heteroscedasticity and autocorrelation, respectively.

Earnings Announcements, Aggregate Earnings, and Individual-Firm Stock Returns: A Signal Extraction Perspective

There is an extensive stream of research that documents a positive association between earnings surprises and stock returns at the individual firm level. We posit that individual firms’ earnings surprises have systematic and firm-specific components that differ in their persistence, implying that the market reaction to individual firms’ earnings surprises should depend upon the relative magnitudes of the underlying systematic and firm-specific earnings surprise components. We further posit that investors behave as if they solve a signal extraction problem that allows them to estimate from aggregate (e.g., market) earnings the systematic earnings surprise component. Our signal extraction framework implies that the market reaction to individual firms’ earnings surprises is increasing in the cross-sectional mean earnings surprise and that the magnitude of the mean effect is inversely related to the cross-sectional dispersion of the earnings surprises. Our results are consistent with these predictions. Also consistent with signal extraction, we find that the effect of the crosssectional mean earnings surprise is significantly larger for firms that announce their earnings early in the quarter. We also find that signal extraction occurs for firms with a large percentage of individual investors, but not a large percentage of institutional investors, consistent with institutional investors having private information that allows them to partition a firm’s earnings surprise into its systematic and firm-specific components. Overall, our results suggest that to understand the market reaction to individual firms’ earnings announcements, one must consider aggregate earnings.