Ex-ante determinants to delist or not delist targets after an M&A

The size effect in M&A data sample is well documented in the M&A literature with a strong skewness characteristic (Moeller et al. 2004). Betton et al. (2008) notice that the availability of a machine-readable database has led to develop empirical testing. Their global sample from the SDC database period 1980-2005 shows 21,476 available deals with an average deal size of USD 419 million, a standard deviation of USD 2,610 million and a median size of 36 million (see art. cit. Table 1). Looking at the public bidders to public targets subsample shows 7088 deals with an average size of USD 957 million, standard deviation of USD 4,337 million and median size of USD 116 million (Table 2). The size effect is important, as small sized deals may lead to poor or unavailable data. Transactions are strongly skewed regarding deal size, and small sized transactions may be overrepresented. Introducing a minimum size limit is explained because data availability and accuracy are better. Small-sized acquisitions covered by Zephyr (Bureau Van Dijk) and large ones covered by SDC (Thomson Reuters) are not similar. SDC has better accuracy (Bollaert and Delanghe 2015). A size limit is also justified because economic and financial comparisons may give different results according to the sample. The economic significance of large transactions compared to small sized transactions (i.e., USD 1 million) is higher. For instance, analyzing CARs by looking at unlisted and listed acquirer samples outlines a strong size effect, as CARs are different (Faccio et al. 2006). Moeller et al. (2004) highlight the size effect, as CARs at the announcement are opposite between small firms (gains) and large firms (losses). As larges deals provide more economically significant results, a high cutoff limit for deal size is typical when empirically testing M&As. Auguets-Pratsobrerocca et al. (2017) use a 500 million USD minimum size cutoff. Goergen and Renneboog (2004) use a USD 100 million USD cutoff. We test a USD 50 million limit that does not yield a much more numerous sample, as what constrains more the number of firms is the condition to obtain both listed acquirers and targets.

We need to rule out private equity funds as acquirers because they have specific reasons to maintain the target listed (Ljungqvist et al. 2016, Pour and Lasfer 2013). We thank an anonymous referee for mentioning this point.

Times to completion are very different. Sometimes, as several transactions are linked and carry on (takeover, second takeover to increase the targeted number of shares, squeeze out offer, etc.), the overall duration may be many months. Betton et al. (2008) show that the average time to completion is 71 days in public-to-public tender offers (median 49 days, highest quartile 98 days). Block trades and mergers may have a slightly longer time to completion (average 108 days, median 98 days) (art. cit. Table 3). Our limit of 270 days is comfortably above these figures and signals situations where the deal has been completed and the target firm survives after the deal completion. As this choice is arbitrary, we develop another set of empirical estimations with an alternate definition for “surviving” non delisted and “surviving” target firms using a 180-day (6-month) cutoff. The empirical results are strictly similar and not reported (available from the authors).

Another definition of a delisted and nonsurviving target comes from analyzing the activity on the stock market following the announcement of an acquisition. We filtered the daily sequence of prices and volume for each target’s stock on the market over the 10 to 220 business days after the start of the acquisition as per the Thomson Eikon database. We define a stock as inactive if it shows no quote, repeated identical prices, or zero transaction volume for five business days. This produces a list of 241 surviving firms (defined as traded for more than 180 business days, i.e., approximately 8–9 calendar months) and 550 delisted targets. The ex-post percentage of delisting is 77.8%. The paucity of the transaction data in Thomson Eikon made us prefer the first definition of delisting. The correlation of the dummy delisted variables according to the two measurement methods is high (+ 0.75).

Delisting may also happen after a noncompleted deal, for example, when the target’s managers decide on their own to go private.

Individuals with the same surnames were presumed to belong to the same family. VCs and private equity fund ownerships were cumulated if they were the most important shareholders. We assume that they share the same goal.

The market indices used to calculate abnormal returns are the Standard & Poor’s 500 Index for US and Canadian stocks and the Euro Stoxx 600 Index for European stocks. The CARs are calculated over the window from five business days before to five business days after.

Unreported tests using Logit models provide similar results.

See Table 1, Panel C.

Preexisting corporate control is defined as a minimum ownership stake of 20%.

Statistically different at the 1% level.

The estimation models of these variables are explained hereafter in Tables 6 and 7.

Looking at the coefficient magnitude, if 100% ownership is sought, this cancels out the very negative constant coefficient.

In Equation (5), we introduce year dummies. The results are identical to those of Table 4, Equation (3) and outline a significant positive effect with European acquirers announcing significantly more future delistings.

Table 6, Equation (1) has the highest number of forecasted observations.

Alternatively, we follow a Heckman procedure and use the lambda inverse Mills ratio calculated in the first step estimate to statistically condition the model for the successful completion of the acquisition process. Regressions similar to those of Panels B and C using the lambda Mills ratio instead of FIT_CONTROL_CPTD provide strictly similar results. They are not reported.