United States: Economics

In theory, a company’s current stock price reflects the present value of its expected future cash flows. In an informationally efficient market, the stock’s price reacts quickly to impound any news that affects the company’s future prospects. Importantly, however, a stock’s price may change even absent any company-specific news, due to changes in the broader market index or simply due to normal random fluctuations (noise). Therefore, economists typically use an analytical technique known as ‘event study’ to identify how much of a stock’s price change following some news release (or event) can reasonably be attributed to such news per se, rather than to a contemporaneous change in the broader market index or to noise.

For example, event studies can be helpful in gauging the extent to which a change in regulations will harm or help a firm, which may be hard to predict from economic theory alone. If news of the regulatory change is associated with appreciation in the firm’s stock price, then it suggests that investors expect the regulatory change to be favourable, while a decline in the stock price would suggest the opposite. Event studies have been used to estimate the effect of a variety of different types of events on firm value, including events related to firm earnings, stock splits, and changes in macroeconomic conditions.1

Event studies have a long history in antitrust analysis. They have been used to assess the effects on firm value of events such as mergers, acquisitions, antitrust litigation and regulatory enforcement actions.2 In antitrust, estimating the change in stock price is frequently a means to infer the competitive effects of the event of interest – for example, mergers have the potential to be procompetitive or anticompetitive. Measuring how security prices react to news about a merger may provide insight into the competitive implications of the merger that can complement the insights provided by the more traditional tools of antitrust analysis, such as the various measures of market concentration.3

In this chapter, we review the basic event study approach and outline some of the challenges inherent in trying to infer competitive effects using event studies.

Basics of an event study

To illustrate the basic event study approach, consider a hypothetical example. Oilco, an oil company whose stock trades publicly in an efficient market, has just announced the discovery of a new oil field. Its stock price rises 10 per cent following the announcement, from US$100 to US$110. Notably, the broad market index (the S&P 500) is also observed to have risen 5 per cent during the same interval. Therefore, at least some of Oilco’s 10 per cent price increase may be attributable to the change in the S&P 500 index, which would be expected even absent an announcement. To isolate the impact of the oil field announcement on Oilco’s stock price, one must therefore first calculate the stock’s expected return, given its relationship to the S&P 500. To do so, the historical relationship between Oilco’s returns and the S&P 500’s returns over an estimation period can be calculated using a regression analysis (eg, using daily return data for the year preceding the announcement). According to such a regression analysis, suppose the market ‘beta’ is estimated to be 1.2. This means that, on average, when the S&P 500 rises 1 per cent, Oilco’s stock price is expected to rise 1.2 per cent (and vice versa) even absent any company-specific news. So, given the S&P 500’s 5 per cent increase at the time of the announcement, our regression analysis suggests that Oilco’s stock price would be expected to increase by 6 per cent (equal to 1.2 x 5 per cent) even absent any company-specific news. Thus, Oilco’s market-adjusted (or residual) return following the announcement is 4 per cent (equal to 10 per cent – 6 per cent).

However, this residual return cannot be reasonably attributed to news of the oil field discovery until we consider that no regression model can perfectly predict a stock’s expected price change. There will always be a difference between observed movements of the stock price and the movements predicted by the model, even when there is no significant firm-specific event. Therefore, Oilco’s 4 per cent residual return following the announcement must be compared to the average variation observed over the estimation period (standard error). If, for instance, the regression model’s standard error is 2 per cent, then the Oilco’s residual return following its announcement is twice as large as it normally would be absent any news. The probability that such a large residual return is attributable to noise alone is less than 5 per cent, and therefore one may conclude that Oilco’s 4 per cent residual return following its announcement was ‘statistically significant’ at the 95 per cent confidence level. That is, US$4 of Oilco’s observed US$10 stock price increase could be attributed to news of the oil field discovery with a high degree of certitude (assuming everything has been correctly estimated). If Oilco had 1 billion shares outstanding at the time, the impact of the announcement on Oilco’s market value may be estimated to be US$4 billion. If the standard error of the regression model was larger, say 4 per cent, then Oilco’s residual return following the announcement would not be deemed statistically significant. In that case, despite the seemingly large 10 per cent change observed in Oilco’s stock price following the announcement, this change could not be attributed with a reasonable degree of certitude to the announcement per se, rather than to noise.

An important determinant of the validity of the event study is the event window, the period of time over which the change in stock price is measured. For example, one window to consider could be from the day prior to Oilco’s announcement to the end of the day on which the announcement was made. The choice of window is often motivated by the timing of the information disclosure and the speed with which it is expected to reach investors. If rumors of the discovery circulate before the official announcement, an appropriate event window may begin before the official announcement. Similarly, if a series of announcements about the discovery were made over several days, a longer event window may be appropriate. In some situations, intraday event windows (eg, 3:00pm to 3:15pm) may be most relevant because information may circulate and be impounded into prices very quickly.4

There is often a trade-off in selecting the event window length. Longer event windows may be more likely to capture the effect of the information at issue, but also increase the likelihood of capturing the effect of extraneous information. In modern securities markets, which generally react quickly to new information, shorter windows may be preferred. For example, an analysis of price responses to earnings and dividend announcements found that, while significant returns could be detected one day after earnings announcements that exceeded analyst forecasts, ‘by far the largest portion of the price response occurs in the first five to fifteen minutes after the disclosure’.5

When appropriately specified, event studies can yield useful insights into the effect of an event on stock prices and firm market value. Large differences between the expected returns predicted by the event study model and the actual return observed in the stock market price may reflect the effect of the announcement on the stock price and, therefore, the value of the company.

Event studies in antitrust

Applying this event study framework in an antitrust context requires added caution, as event studies evaluate the effects of information on security prices while analysis of competition generally relates to conditions in the markets for products or services. Reaching definitive conclusions about product quality or product prices based on the results of an event study can therefore be difficult. For this reason, the literature addressing the use of event studies in antitrust advises caution in the development of hypotheses and the interpretation of results.6 Below, we outline three specific challenges identified in the literature when trying to infer competitive effects using event studies.

Alternative interpretations of results

Changes in a firm’s stock price may not provide an unambiguous prediction of expected economic effects.7 For example, a statistically significant share price increase following an announced merger might result from higher expected future product prices, suggesting that the merger may have anticompetitive effects. But the very same stock price increase might alternatively imply anticipated procompetitive efficiency gains. This stark ambiguity with respect to the underlying source of the observed share price increase implies that the event study findings may well be interpreted inappropriately without further analysis of the likely economic consequences of the transaction.

Changes in the valuation of rival firms have been suggested as another means of identifying potentially anticompetitive mergers, based on the prediction of standard economic models that anticompetitive mergers will raise product prices to the benefit of competitors, while efficiency-enhancing mergers will harm competitors and cause their share prices to decrease.8 But here, too, there may be other explanations for these share price movements. If investors interpret the announced merger as a signal that other firms in the same market may merge, then competitors’ share prices may increase even with no expectation that the announced merger will lead to significant price increases.

Multi-product firms

Another point of caution when applying the event study approach in an antitrust context arises when a firm is involved in a wide range of business activities. Even if the event under examination has a large effect on sales of one of the firm’s products, it may not significantly change its market value if that product accounts for only a small portion of overall profits. Consistent with this logic, the stock market’s response to an antitrust action taken by the United States Department of Justice or the Federal Trade Commission has been found to be directly related to the proportion of the targeted firm’s revenue affected by the breadth of the lawsuit.9

A similar result has been found in the context of alleged price fixing for firms with substantial market values that were typically involved in several lines of business. In this context, the ‘unavoidable inclusion of irrelevant business lines into their analyses introduced statistical noise which subsequently increased their chance of incorrectly failing to test the significance of an abnormal return.’10 In addition, the same basic finding has been documented with respect to the use of event studies to evaluate the effects of mergers and cartels in multi-product firms.11

Identification of competitors

As noted above, antitrust-related event studies are often concerned with estimating changes in the market value of competitors. But identifying the most appropriate set of competitors can be another challenge in this context. In evaluating the trade-off between broad and narrow definitions of the set of rival firms, it has been found that ‘overly broad measures introduce excessive noise because the “rivals” are only loosely related to the merging firms, and overly narrow measures introduce excessive noise because results rely too much on the idiosyncratic shocks of the small number of firms considered to be rivals’.12

Conclusion

Event studies can, at times, be informative in assessing whether a particular company action had anticompetitive implications. But it is important to appropriately specify the analysis and control for all relevant factors unrelated to the competitive action at issue. These steps are crucial in ensuring that the implications of the results are valid. Of course, careful specification and interpretation of results are not unique to the use of event studies in antitrust analysis. However, because event studies examine the effects of information on security prices while antitrust analysis concerns itself with production and prices in the markets for products or services, the application of event studies in antitrust requires particular care and comes with specific pitfalls that researchers must work carefully to avoid.

Notes

1. See, generally, A Craig MacKinlay, ‘Event Studies in Economics and Finance,’ Journal of Economic Literature, Vol. 35, No. 1, March 1997, pp. 13-39. See also Eugene F Fama, Lawrence Fisher, Michael C Jensen and Richard Roll, ‘The Adjustment of Stock Prices to New Information,’ International Economic Review, Vol. 10, No. 1, February 1969, pp. 1-21.
2. See, generally, Michael Cichello and Douglas J Lamdin, ‘Event Studies and the Analysis of Antitrust,’ International Journal of the Economics of Business, Vol. 13, No. 2, 2006.
3. See, for example, B E Eckbo, ‘Horizontal mergers, collusion, and stockholder wealth,’ Journal of Financial Economics, Vol. 11, 1983, pp. 241–273; R Stillman, ‘Examining antitrust policy towards horizontal mergers,’ Journal of Financial Economics, Vol. 11, 1983, pp. 225–240.
4. See Belinda Mucklow, ‘Market Microstructure: An Examination of the Effects on Intraday Event Studies,’ Contemporary Accounting Research, Vol. 10, No. 2, 1994, pp. 355-382.
5. James M Patell and Mark A Wolfson, ‘The intra-day speed of adjustment of stock prices to earnings and dividend announcements,’ Journal of Financial Economics, Vol. 13, 1984, pp. 223-252.
6. For a detailed overview, see Cichello (2006), McAfee and Williams (1988) and Beigi and Budzinski (2013).
7. S Davies and P Ormosi, ‘A Comparative Assessment of Methodologies Used to Evaluate Competition Policy,’ Journal of Competition Law & Economics, Vol. 8, No. 4, December 2012, pp. 769-803.
8. S Davies and P Ormosi, ‘A Comparative Assessment of Methodologies Used to Evaluate Competition Policy,’ Journal of Competition Law & Economics, Vol. 8, No. 4, December 2012, pp. 769-803.
9. K D Garbade, et al, ‘Market reaction to the filing of antitrust suits: an aggregate and crosssectional analysis,’ Review of Economics and Statistics, Vol. 64, Issue 4, 1982, pages 686-691.
10. Bosch, J C and Eckard, E W, ‘The profitability of price fixing: evidence from stock market reaction to federal indictments,’ Review of Economics and Statistics, Vol. 73, Issue 2, 1991, pp. 309-317.
11. S Davies and P Ormosi, ‘A Comparative Assessment of Methodologies Used to Evaluate Competition Policy,’ Journal of Competition Law & Economics, Vol. 8, No. 4, December 2012, pp. 769-803; R P McAfee and M A Williams, ‘Can event studies detect anticompetitive mergers?’ Economics Letters, Vol. 28, Issue 2, 1988, pp. 199–203.
12. D Filson, S Olfati and F Radoniqi, ‘Evaluating Mergers in the Presence of Dynamic Competition Using Impacts on Rivals,’ Journal of Law & Economics, Vol. 58, No. 4, 2015, pp. 915-934.