The research question

The classic M&A question is whether a merger creates value and how that value splits between the two sides. Three decades of evidence find that targets capture most of the announcement gains while acquirer returns are close to zero or modestly negative (Andrade, Mitchell & Stafford, 2001, Journal of Economic Perspectives). An event study quantifies this for any sample of deals.

Sample and event-date definition

The event date is the first public announcement of the deal, not the completion or shareholder-vote date. Getting this right is the single most important step: a date one day late misses the reaction, and using the completion date measures the wrong event. For each deal, record the acquirer ticker, the target ticker, and the announcement date, and split the sample into an acquirer panel and a target panel.

Choosing windows and the return model

Use a short event window because M&A news is incorporated quickly: $[-1,+1]$ is the workhorse, with $[-5,+5]$ to capture pre-announcement run-up. Estimate the market model over $[-250,-11]$ in event time. The market model is the right benchmark here, and a $[-1,+1]$ window keeps confounding-news risk low.

Computing CAR for the target vs the acquirer

For each firm, $CAR_i=\sum_{t\in[-1,+1]}AR_{i,t}$ with $AR_{i,t}=R_{i,t}-(\hat\alpha_i+\hat\beta_i R_{m,t})$. The illustrative numbers below are representative of the M&A literature for a $[-1,+1]$ cash-deal sample:

Side	N	Mean CAR [-1,+1]	BMP t	Reading
Target	200	+23.4%	18.7	Large positive premium, highly significant
Acquirer	200	-1.1%	-2.3	Small negative, marginally significant
Combined (value-weighted)	200	+1.8%	3.4	Net value creation, significant

The combined entity earns a small positive abnormal return, consistent with mergers creating modest aggregate value, while the split is heavily toward the target.

Cross-sectional CAAR and significance testing

Average CAR across firms within each panel to get CAAR, then test $H_0: E(CAAR)=0$. Report the BMP test as the headline because announcement events raise volatility, which would inflate a plain Patell Z:

$$CAAR=\frac{1}{N}\sum_{i=1}^{N}CAR_i, \qquad t_{BMP}=\sqrt N\,\frac{\overline{SCAR}}{S_{\overline{SCAR}}}, \quad SCAR_i=\frac{CAR_i}{S_{CAR_i}}.$$

For the target panel the BMP t of 18.7 leaves no doubt; for the acquirer panel the small negative CAAR is only marginally significant, which is itself the standard finding: acquirer returns are close to zero and noisy.

Interpreting the asymmetry against the literature

The large target premium and near-zero acquirer return mean the deal transfers value to target shareholders, who must be paid above the pre-bid price to tender. A positive combined CAAR says the market expects synergies net of the premium; a negative combined CAAR (common in stock-financed or empire-building deals) says the market doubts them. Always interpret the acquirer number in light of deal financing: cash deals tend to show better acquirer returns than stock deals, where the issuance signals overvaluation.

Robustness: confounding, leakage, subsamples

• Leakage. A pre-announcement run-up in the target inflates the $[-5,+5]$ CAR relative to $[-1,+1]$. Report both; a gap signals information leakage or rumour trading.
• Confounding events. Exclude deals with same-window earnings releases or other material news.
• Subsamples. Split by cash vs stock financing and by relative size; the acquirer result in particular varies sharply across these cuts.
• Clustering. Merger waves cluster deals in time; use the BMP or Kolari-Pynnonen adjusted test (see Significance tests).

Reproduce this analysis free in ARC

Build a request file with one row per firm (acquirer and target panels separately), upload the matching price data, choose the market model on Expected-return models, and the ARC calculator returns the CAR, CAAR and test statistics in the table format above.

FAQ