Merger and Acquisition Performance Event Study
- Introduced by
- Gregor Andrade, Mark Mitchell & Erik Stafford; David King, Dan Dalton, Catherine Daily & Jeffrey Covin
- Year
- 2001
- Type
- Event-study pipeline for the wealth effects of merger and acquisition announcements
- Family
- Strategic management / corporate strategy and finance
What does it do?
A merger and acquisition event study measures the stock-market reaction to a deal announcement to infer how much value the deal is expected to create or destroy for acquirers and targets.
The logic is that in an efficient market the share-price jump around the announcement capitalizes investors' revised expectations of future cash flows attributable to the deal. Andrade, Mitchell and Stafford's 2001 survey distilled the empirical regularities: targets earn large positive abnormal returns, combined acquirer-plus-target returns are modestly positive, while acquirers themselves often earn around zero or slightly negative returns. King, Dalton, Daily and Covin's 2004 meta-analysis confirmed that acquirers, on average, do not gain and pointed to unidentified moderators, motivating cross-sectional models that link abnormal returns to deal and firm characteristics.
The intuition behind the method
When a company announces it is buying another, investors immediately reprice both firms based on what they think the deal will be worth. If the market believes the acquirer overpaid or that the strategic logic is weak, the acquirer's stock barely moves or drops; if the target is being bought at a premium, its stock jumps. The genius of the event-study design is that it isolates the deal's effect by first learning each stock's normal relationship to the overall market, then asking how far the stock departs from that normal behavior in the narrow window around the announcement. Anything beyond normal is the 'abnormal return,' and summed over the event window it becomes a clean, forward-looking estimate of value creation — measured the moment the news hits, before operating results could possibly show up.
History
Event-study methodology was established for finance by Fama, Fisher, Jensen and Roll (1969) and Brown and Warner's methodological work, then became the dominant tool for evaluating corporate events, M&A foremost among them. Andrade, Mitchell and Stafford's 2001 Journal of Economic Perspectives survey synthesized decades of evidence, establishing the now-standard stylized facts that targets earn large positive announcement returns, combined returns are modestly positive, and acquirers earn approximately zero. In strategic management, King, Dalton, Daily and Covin's 2004 Strategic Management Journal meta-analysis aggregated event-study and accounting results to conclude that acquisitions do not, on average, improve acquirer performance and that the field's standard moderators leave most variance unexplained. Together these works frame the modern M&A event study as a search for the contingencies under which acquisitions create value.
When to use it
Use an M&A event study when you want a forward-looking, market-based estimate of the value implications of a deal and you have a clean announcement date plus liquid, frequently traded stocks with available daily returns. It is ideal for assessing acquirer and target wealth effects, for comparing value creation across deal types, and for testing which deal or firm characteristics drive abnormal returns. It is poorly suited to private firms or thin-trading stocks (no reliable market prices), to settings where the announcement date is ambiguous or leaked well in advance, or to events confounded by other simultaneous news. When the question is realized operating improvement rather than expected value, accounting-based long-run performance studies are the appropriate complement, as both Andrade and colleagues and King and colleagues emphasize.
Limitations and considerations
- Validity hinges on market efficiency and on a clean, uncontaminated event date; leakage or confounding news biases the estimate.
- It measures expectations, not realized outcomes; expected synergies need not materialize, so CARs can mislead about long-run performance.
- It requires publicly traded firms with adequate liquidity, excluding private targets and thin markets.
- King and colleagues found that standard moderators explain little of post-acquisition performance variance, signaling unmodeled heterogeneity.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Provides a forward-looking, market-based measure of expected value creation available the instant the deal is announced.
- The short event window cleanly isolates the deal's effect from slow-moving operating results and most other news.
- Standardized and replicable, enabling comparison across deals, periods, and studies and supporting meta-analysis as in King and colleagues.
- Cross-sectional extensions link abnormal returns to deal and firm moderators, turning the design into a theory-testing tool.
Applications
- Estimating acquirer and target shareholder wealth effects around merger and acquisition announcements.
- Comparing value creation across cash versus stock deals, related versus unrelated acquisitions, and friendly versus hostile bids.
- Testing whether acquirer governance, CEO characteristics, or prior acquisition experience moderate announcement returns.
- Assessing market reactions to other corporate events such as divestitures, alliances, and CEO successions using the same design.
- Supplying effect sizes for meta-analyses of acquisition performance across many studies.
How it works
How it works
The analyst first dates the event precisely — the day the deal is first publicly announced — and defines two windows: an estimation window well before the announcement, used to learn the stock's normal behavior, and a short event window around day zero (often a few days such as minus one to plus one) over which the deal's effect is measured. Andrade, Mitchell and Stafford rely on tight windows because they sharpen the inference: the shorter the window, the less other news contaminates the measured reaction, while the estimation window must be clean of the event so the benchmark is untainted.
Formula
t=0:\text{announcement};\ \text{estimation }[t_0,t_1],\ \text{event }[\tau_1,\tau_2]\ (e.g.\ [-1,+1])Over the estimation window, the analyst models each stock's normal return, most commonly with the market model that regresses the firm's daily return on the market index return to recover its alpha and beta. This calibrates how the stock typically moves with the market in the absence of firm-specific news. The fitted relationship becomes the counterfactual: it predicts what the stock would have returned during the event window had the deal not been announced, which is the benchmark every subsequent step is measured against.
Formula
R_{it}=\alpha_i+\beta_i R_{mt}+\epsilon_{it}\ \text{ over }[t_0,t_1]During the event window, the abnormal return on each day is the actual return minus the normal return predicted by the estimated market model. Summing these daily abnormal returns across the event window gives the cumulative abnormal return (CAR), the headline quantity: the total deal-attributable price movement for that firm. Andrade, Mitchell and Stafford report CARs separately for acquirers and targets, since the wealth effects differ starkly — targets capture large premiums while acquirers typically see little — and the combined CAR speaks to whether the deal creates value overall.
Formula
AR_{it}=R_{it}-(\hat\alpha_i+\hat\beta_i R_{mt}),\quad CAR_i=\sum_{t=\tau_1}^{\tau_2} AR_{it}Because any single deal's CAR is noisy, the analyst aggregates across many deals, averaging CARs and testing whether the mean differs significantly from zero with parametric t-tests and often nonparametric or robust alternatives that handle event clustering and cross-correlation. This aggregation is what yields Andrade, Mitchell and Stafford's stylized facts: significantly positive target returns, modestly positive combined returns, and acquirer returns statistically indistinguishable from zero. Proper test statistics — accounting for variance inflation in the event window and for overlapping events — are essential to avoid spurious significance.
Formula
\overline{CAR}=\tfrac{1}{N}\sum_i CAR_i,\quad t=\dfrac{\overline{CAR}}{\widehat{\sigma}(\overline{CAR})}Mean effects mask wide variation, so the final step regresses individual deal CARs on deal and firm characteristics — method of payment, relatedness, relative size, acquirer governance, prior experience — to explain which deals create or destroy value. King, Dalton, Daily and Covin's meta-analysis showed that common moderators studied this way explain little of the variance in post-acquisition performance and inferred that important moderators remain unidentified, a finding that frames cross-sectional event-study models as a search for the conditions under which acquisitions pay off.
Formula
CAR_i=\gamma_0+\boldsymbol{\gamma}'\mathbf{z}_i+\xi_i,\quad \mathbf{z}=\text{deal/firm moderators}Common pitfalls
- Using an event window so wide that unrelated news contaminates the abnormal returns, or so narrow that pre-announcement leakage is missed.
- Letting the estimation window overlap the event, which contaminates the normal-return benchmark.
- Ignoring event clustering and cross-sectional correlation, which deflate standard errors and inflate apparent significance.
- Equating positive announcement CARs with realized success, despite evidence that acquirer gains are weak and long-run results often disappoint.
Frequently asked questions
Why do acquirers often earn near-zero returns while targets gain so much?
Targets are typically bought at a substantial premium over their pre-bid price, so their shareholders capture most of the anticipated gains, producing large positive target CARs. Acquirers, by contrast, often pay away much of the expected synergy in that premium and may face investor skepticism about the deal's logic or the risk of overpayment, leaving their announcement returns around zero or slightly negative. Andrade, Mitchell and Stafford document exactly this pattern: combined acquirer-plus-target value rises modestly, but the acquirer's own shareholders rarely gain on average.
How long should the event window be?
There is a tension. A short window (such as minus one to plus one trading days) keeps the measured reaction tightly focused on the announcement and minimizes contamination from unrelated news, which is why Andrade, Mitchell and Stafford favor short windows for clean inference. But if information leaks before the formal announcement, too narrow a window misses the pre-announcement run-up. The standard compromise is a short symmetric window around day zero, sometimes widened slightly to capture leakage, with the estimation window kept well clear of the event so the normal-return benchmark stays uncontaminated.
Does a positive announcement return mean the acquisition succeeded?
Not necessarily. An event study measures the market's expectation of value creation at announcement, not the realized outcome. Expected synergies may fail to materialize, integration may falter, and long-run accounting performance can diverge from the initial reaction. King, Dalton, Daily and Covin's meta-analysis found that acquirers do not, on average, improve their post-acquisition performance and that standard moderators explain little of the variation, underscoring that announcement CARs are a forward-looking forecast that must be complemented by long-run performance evidence.
Related methods
Sources
Andrade, G., Mitchell, M., & Stafford, E. (2001). New evidence and perspectives on mergers. Journal of Economic Perspectives, 15(2), 103-120.
Open DOI referenceCopied.King, D. R., Dalton, D. R., Daily, C. M., & Covin, J. G. (2004). Meta-analyses of post-acquisition performance: Indications of unidentified moderators. Strategic Management Journal, 25(2), 187-200.
Open DOI referenceCopied.