Value creation and the probability of success in merger and acquisition transactions

Abstract

This paper presents and tests a hypothesis that expands existing explanations of value creation in merger and acquisition (M&A) transactions. The main premise is that value creation is determined by not only the target’s pre-acquisition value, as indicated by numerous studies in extant literature, but also by the acquirer’s competency (among other factors) demonstrated by their pre-merger financial ratios. The paper shows that M&A transactions create value in the longer-run and the gain is commensurate with the acquirer’s historical performance and the target’s pre-acquisition value. Further, the paper employs statistical procedures and model-building techniques in order to develop and validate parsimonious Altman-style predictive models. The models reasonably identify successful M&A deals and are statistically significant and consistent with a few existing theories. Specifically, the evidence on liquidity supports internal capital markets hypothesis but does not support the theories of agency problems, while the evidence on financial leverage supports the view that lower debt enhances corporate control.

Appendices

Appendix 1: Tobin’s Q, excess value, and abnormal return

We use the following notation:

N :

Number of diversifiers (parent company) in the sample of this paper

i = 1, 2, …, N :

Subscript that identifies diversifier i in the sample

M :

Number of segments in the parent (diversifier) company

j = 1, 2, …, M :

Subscript that identifies segment j in the parent company

K :

Number of competitors in the same market (firms with the same SIC code as the parent company)

k = 1, 2, …, K :

Subscript that identifies competitor k in the market

Q :

Number of matching pure-plays of a segment (firms with the same last two digits of SIC code as the segment)

q = 1, 2, …, Q :

Subscript that identifies matching pure play q

1.1 Tobin’s Q (TQ)

Tobin’s Q is the ratio of market value of assets to the book value of assets. Tobin’s Q is a classical measure of performance used in diversification literature (see Lang and Stulz 1994; Servaes 1996; Steiner 1996; Dong et al. 2006) we calculate Tobin’s Q as,

$$TQ_{i} = \frac{{MVE_{i} + BVL_{i} }}{{BVE_{i} + BVL_{i} }}$$

(3)

$MVE_{i}$, Market Value of Equity; $BVL_{i}$, Book Value of Liabilities; $BVE_{i}$, Book Value of Equity.

1.2 Excess value (EX)

We follow Berger and Ofek’s (1995) procedure in estimating the imputed value of each segment. First we calculate imputed value of each segment j in a diversifier i as median ratio of total capital to total assets (or sales²⁵) of matching pure-play firms multiplied by total assets (or sales) of segment j. Matching pure play firms are in the same SIC code of segment j, have at least $20 million in sales, and have sufficient data on Compustat. Mathematically,²⁶

$$\begin{aligned} IV_{j}^{i} & = \left[ {Median \left( {\frac{{TC_{1} }}{{TA_{1} }},\frac{{TC_{2} }}{{TA_{2} }}, \ldots ,\frac{{TC_{q} }}{{TA_{q} }}, \ldots ,\frac{{TC_{Q} }}{{TA_{Q} }}} \right)} \right]TA_{j}^{i} \\ TC_{q} & = MVE_{q} + BVL_{q} \\ \end{aligned}$$

(4)

$IV_{j}^{i}$, Imputed value of segment j in firm i; $TA_{j}^{i}$, Total assets of segment j in firm i (identifiable total assets in Compustat segments database); $TC_{q}$, Total capital of a pure-play firm q in the same industry of segment j; $MVE_{q}$, Market Value of Equity of a pure-play firm q in the same industry of segment j; $BVL_{q}$, Book Value of Liabilities of a pure-play firm q in the same industry of segment j; $TA_{q}$, Total Asset of a pure-play firm q in the same industry of segment j.

The sum of the imputed values of the M segments in firm i becomes the imputed value of the firm i (its value if all of its M segments were operated as stand-alone businesses):

$$IV_{i} = \mathop \sum \limits_{j = 1}^{M} IV_{j}^{i}$$

(5)

Next, we measure the percentage difference between diversifier i’s total value and the sum of imputed values for its M segments as stand-alone entities. If market value (total capital) of firm i is $TC_{i}$ then we calculate percentage difference (excess value) as:

$$\begin{aligned} EV_{i} & = \ln \left( {\frac{{TC_{i} }}{{IV_{i} }}} \right) \\ TC_{i} & = MVE_{i} + BVL_{i} \\ \end{aligned}$$

(6)

$TC_{i}$, Total capital of a firm i; $MVE_{i}$, Market Value of Equity of firm i; $BVL_{i}$, Book Value of Liabilities of a firm i.

1.3 Abnormal returns

Abnormal return is the difference between observed stock return and a hypothetical return for the scenario where the event did not happen. The hypothetical return is computed using the market model approach (three-factor model of Fama and French). The parameters of the model are estimated over the window (−720, −90) trading days preceding the deal announcement date. The exclusion of 90-day pre-event window minimizes the effect of possible leakage of information (Scholes and Williams 1977).

For each firm in the sample, we compute cumulative abnormal return (CAR), which reflects total valuation effect of the merger. CARs are calculated beginning 30 days after the event through 3 years after the event, that is, within an event window spanning (+30, +1080) trading days after the deal announcement date. The exclusion of the 30-day post-event windows eliminates the effect of possible adjustment to the share price following the announcement (Scholes and Williams 1977). The relatively long horizon surely raises several concerns (discussed below), but it allows the M&A effect to be fully impounded in stock prices. The work of Hund et al. (2010) and Alhenawi and Krishnaswami (2015) indicate that it takes time for the M&A advantages to materialize and, hence, short-run analysis does not capture the full effect of M&A events on value.²⁷

In computing cumulative abnormal return, we were conscious of the pitfalls of long-term event studies. For instance, consecutive acquisitions made by the same acquirers are treated as separate events and the data is clean of observation with confounding merger events near the studied window (McWilliams and Siegel 1997). Second, this study uses a sample of 454 observations; this mitigates the outlier and robustness-of-significance problems (McWilliams and Siegel 1997). Nevertheless, several other empirical concerns remain present in the analysis. For instance, we cannot ensure that other events are not driving the long-term abnormal returns. In addition, this test does not account for potential interaction between financial ratios and does not capture the relative strengths of pre-event conditions in predicting post-merger performance. These pitfalls are accounted for in regression analysis explained in subsequent sessions. Finally, long-term event studies suffer sizable type I and type II errors. We believe this issue is mitigated by the fact we use three other long-term measures of performance: Tobin’s Q, asset-based excess value, and sales-based excess value. Further, the predictive models use the dichotomous variables Z3 and Z4; not raw CAR. We believe this mitigates, at least partially, CAR problems (see Sect. 4.3 for more details).

Appendix 2: Post-acquisition performance measures

2.1 Using Tobin’s Q and excess value

Hund et al. (2010) show that the synergies of M&A deals materialize over time. Specifically, they show that an average M&A may initially erode value but exhibits sustained recovery afterward. In a similar vein, Barraclough et al. (2013) show that short-term price reaction to a M&A announcement tends to underestimate true synergies of the deal. Accordingly, we gauge post-acquisition performance over three post-acquisition years. For each transaction, we compute time-series averages of TQ, excess value based on total assets (EVTA), and excess value based on sales (EVSL) (one at a time) of the combined firm using data from years 0, 1, 2, and 3. That is,

$$V_{m, Avg} = \left( {\mathop \sum \limits_{t = 0}^{t = 3} V_{m, t} } \right){/}4$$

(7)

where $m = 1, 2, \ldots , 454$ indexes diversifiers and t indexes time distance (in years) from event. Thus, $V_{m, t}$ is the measure of performance of diversifier m computed in the tth year and $V_{m,Avg}$ is the time-series average of $V_{m, t} .$ ²⁸ For each of the three measures of post-acquisition performance (TQ, EVTA, and EVSL), we compute the difference $\Delta Y = Y_{t = 3} - Y_{t = - 1}$ where $t = - 1$ indicates the year ending immediately prior to the M&A completion and $t = 3$ indicates 3 years after completion. For each measure, a positive $\Delta Y$ indicates success and a negative $\Delta Y$ indicates failure.

2.2 Dichotomous measures of M&A success

We create the first dichotomous variable Z1, where $Z1 = 1$ when the M&A generates three positive differences (i.e., $\Delta TQ > 0$ and $\Delta EVTA > 0$ and $\Delta EVSL > 0$) and $Z1 = 0$ with three negative differences (mixed combinations are eliminated from the sample). Then, we create the second dichotomous variable Z2, where $Z2 = 1$ when the M&A generates three differences above the corresponding medians of differences (i.e., $\Delta TQ > Median\,\Delta TQ$ and $\Delta EVTA > Median\,\Delta EVTA$ and $\Delta EVSL > Median\,\Delta EVSL$) and $Z2 = 0$ with three below-median differences (mixed combinations are eliminated from the sample). Variables Z1 and Z2 supply a conservative measure of M&A event success.

We create another two variables, Z3 and Z4, using abnormal returns. $Z3 = 1$ if the M&A generates positive abnormal return; otherwise, $Z3 = 0$. $Z4 = 1$ if the M&A generates above-median abnormal return. If neither of these conditions is met, $Z4 = 0$. While Z3 and Z4 are less conservative than Z1 and Z2, they do not require reduction in sample size.

Appendix 3: Measures of pre-acquisition performance and other control variables

3.1 Financial ratios

A list of 20 pre-acquisition financial ratios is computed for each acquirer. Financial ratios are sensitive to transitional activities. For instance, a non-ordinary transaction (e.g., sale of an asset) may artificially boost liquidity. In order to smooth out transitional effects, we compute two time-series averages (of each ratio) as follows,²⁹

$$\begin{aligned} R_{m, 4} & = \left( {\mathop \sum \limits_{Q = - 4}^{Q = - 1} R_{m, Q} } \right) /4 \\ R_{m, 20} & = \left( {\mathop \sum \limits_{Q = - 20}^{Q = - 1} R_{m, Q} } \right) /20 \end{aligned}$$

(8)

where $m = 1, 2, \ldots , 454$ indexes diversifiers and Q indexes time distance (in quarters) from event. Thus, $R_{m, Q}$ is the financial ratio R of diversifier m in the Qth quarter and $R_{m, 4}$ is the time-series average of $R_{m, Q}$ that captures the acquirer’s performance immediately before the completion of the deal. Since $R_{m, 4}$ might be subject to potential window dressing behavior (i.e., misreporting in the face of event-related pressure³⁰), we also use $R_{m, 20} ,$ which is computed with 20 pre-acquisition quarters.

3.2 Relatedness and target’s pre-acquisition value

Several authors have shown that a target’s characteristics play a significant role in post-acquisition performance of the combined firm. For instance, Graham et al. (2002) and Campa and Kedia (2002) suggest that diversification discount can be explained by the diversifiers’ tendency to acquire already discounted business units. Following Graham et al. (2002), we compute the targets’ excess value immediately prior to the acquisition, that is, in their last year of operating as stand-alone firms (year −1). The findings are reported in Table 1 above. Pre-acquisition excess value of target firms (TGTEX) is included in subsequent analysis to control for the potential effect of target pre-acquisition performance.

We also control for relatedness (REL), which is almost a standard in M&A literature because related mergers tend to create increased post-acquisition synergies relative to unrelated mergers (see Bettis and Mahajan 1985; Wernerfelt and Montgomery 1988; Berger and Ofek 1995). A survey study by Mukherjee et al. (2004) confirms existence of operational and strategic advantages of diversification and indicates that managers anticpate higher synergies in related acquisitions.³¹ Rumelt (1974) and Lewellen (1971) noted that operational advantages of a merger are more likely to materialize in intra-industry mergers (i.e., related events). Following Berger and Ofek (1995), an M&A transaction is considered “related” if the first two digits of the SIC code of the acquirer match the first two digits of the target. Thus, the dichotomous variable REL equals 1 for related acquisitions and 0 otherwise.

3.3 Other control variables

We control for size (SIZE) and access to capital market (ATC). We also control for transaction-specific characteristics, such as target’s public status (TGT), method of payment (PMT), and deal’s size relative to the acquirer’s market value (DEAL). Size is an important factor commonly used in M&A literature. For instance, Moeller et al. (2004) finds that M&A announcement returns are lower for acquisitions by larger firms. Lewellen (1971) pointed out that diversification strategy could be motivated by the desire to enhance access to capital markets. In effect, the measures of M&A performance might be upward biased because it reflects improvement in borrowing capacity. Thus, we include a dummy variable (ATC) to capture access to capital markets. ATC equals 1 if the firm pays a dividend and 0 otherwise.³² Prior research has shown negative market reaction to M&A announcement when the target is a public company (e.g., Chang 1998; Fuller et al. 2002), when stock is used as a method of payment (Moeller et al. 2004), and when the ratio of target equity to the acquirer’s equity is smaller (Asquith et al. 1983).³³