Print
E-MailRui Albuquerque is Professor of Finance at Boston College Carroll School of Management; Vyacheslav Fos is Associate Professor of Finance at Boston College Carroll School of Management; and Enrique Schroth is Professor of Finance at EDHEC Business School. This post is based on their recent paper. Related research from the Program on Corporate Governance includes The Long-Term Effects of Hedge Fund Activism by Lucian Bebchuk, Alon Brav, and Wei Jiang (discussed on the Forum here); and Dancing with Activists by Lucian Bebchuk, Alon Brav, Wei Jiang, and Thomas Keusch (discussed on the Forum here).
In our new paper, Value Creation in Shareholder Activism, we measure value creation by activist investors via structural estimation of a model of the choice between passive investment and activism. Our estimates imply that average returns following activist intent announcements consist of 74.8% expected value creation, or treatment, 13.4% stock picking, and 11.8% sample selection effects. Higher treatment values predict improvements in firm performance and lower proxy contest probabilities, whereas abnormal announcements returns do not, suggesting that our estimate identifies more effective activism campaigns.
Activist shareholders play an important role in modern corporate governance. A key question is how much value they create. The existing literature searches for answers in the abnormal stock returns observed around the announcements of new activist positions. The consensus is that these returns are significantly larger than those following the announcement of new passive positions. Indeed, in our data, the average return following the public announcement of activist intent via the filing of a Schedule 13D with the Securities and Exchange Commission (SEC) is 6.34%. Following announcements of passive investment, i.e., Schedule 13G filings, the return is only 0.59%.
Which aspects of activism could explain why the stock market rewards activist positions with a higher announcement return than passive positions? First, activist investors may indeed increase value by influencing the firm’s corporate policies, i.e., the treatment effect of activism. Second, activist investors could be better at identifying undervalued stocks. That is, if instead of filing a Schedule 13D the activist had filed a Schedule 13G, the announcement return would have been higher than the average return to a Schedule 13G announcement by other investors. Finally, because the investor strategically chooses to be an activist or to remain passive, the observed average announcement return for either type of filing includes a sample selection component.
There are several challenges to measure these three components of the Schedule 13D announcement return. The investor’s decision to be active or passive is endogenous and depends on unobservable quantities such as the expected returns to filing Schedule 13D versus filing Schedule 13G and the private costs of activism. In addition, the distributions of observed announcement returns for either type of filing schedule are each censored by the choice of the other type. It is hard to find good instruments for the unobservable components or get data on the non-censored distribution of announcement returns. In this paper, we propose to overcome these challenges and recover the three components of announcement returns with the structural estimation of an economic model of the investment strategy choice.
The central feature of our model is the investor’s decision to become an activist or to remain passive. The assumption that investors have that choice for every filing in the data is grounded on the observation that among all investors that filed at least one Schedule 13D any time in our sample, 78% also filed a Schedule 13G. To choose optimally between the two strategies, in the model the investor trades off the expected announcement return from filing a Schedule 13D, minus a private cost of activism, against the expected announcement return from filing a Schedule 13G. The implicit testable assumption is that the announcement return to a Schedule 13D filing is a noisy indicator of the expected profits of the activist campaign, gross of activism costs.
Our simple model produces the joint distribution of filing choices and announcement returns. The distribution of announcement returns to Schedule 13D filings is censored because announcement returns to Schedule 13D filings are not observed if the investor chooses to file Schedule 13G, giving rise to a sample selection component in the returns to both Schedule 13D and Schedule 13G filings. Using data on announcement returns and filing choices, we can estimate the unobservable quantities—expected returns and private cost of activism—but also each of the components—treatment, stock picking, and sample selection—in Schedule 13D announcement returns.
We find that the average treatment effect of activism, which is equal to the difference in expected returns from filing Schedule 13D instead of Schedule 13G, represents 74.8% of the average Schedule 13D announcement return, that is, 4.74% of 6.34%. The estimated average stock picking component, which equals the counterfactual expected return from filing Schedule 13G, amounts to 13.4% of the observed announcement return, or 0.85% of 6.34%. Finally, the sample selection component, which results from the left censoring of the distribution of expected returns to Schedule 13D filings and, therefore, increases average announcement returns above their unconditional mean, represents 11.8% of announcement returns. For Schedule 13G filings, the sample selection component is small and stock picking accounts for almost all of the observed announcement return.
Our estimates indicate that variation in expected returns is a first-order driving primitive of variation in the outcome variables. Indeed, even while allowing for variation in activism costs or other random sources of heterogeneity, our model best fits the data by loading most heavily on expected returns variation. Further, we find that all determinants of expected returns that are not related to the investor’s characteristics or to the existence of prior activist filings on the same target have equal estimated loadings on expected returns to Schedule 13D or Schedule 13G filings, even though the model places no such restrictions on their values. This finding implies that the market assigns an extra return to Schedule 13D vis-a-vis Schedule 13G filings based on the identity of the investor, not the target firm characteristics. In fact, the treatment component of announcement returns is determined almost entirely by the investors’ Schedule 13D filing experience.
We quantify the benefits of activism to investors and dispersed shareholders. We estimate that the mean Schedule 13D filer earns a net return of 1.74%. Dispersed shareholders save the costs and, therefore, enjoy higher returns from Schedule 13D filings than the activist does. However, our estimates also suggest that 62% of Schedule 13G filings would offer better counterfactual expected returns under Schedule 13D. According to our estimates, these Schedule 13G filings occur because the treatment effect is not enough to cover the private cost of activism.
The complete paper is available for download here.
