EVALUATING MERGER PERFORMANCE ON A LoNGITUDINAL BASIS: AN EMPmICAL INvEsTIGATION Ram Subramanian Grand Valley State University Allendale, Michigan Essam Mahmoud American Graduate School of International Management Glendale, Arizona Mary S. Thibodeaux University of North Texas Denton,Texas Bahman Ebrahimi University of North Texas Denton, Texas Introduction In the past, researchers from a myriad of disciplines have attempted a number of studies on merger performance and related issues. Even using a wide range of performance measures (e.g., stock returns, return on investment, risk, market share), the essence of their results was that mergers benefit the acquired firm and its stockholders, but do not result in any significant benefit to the acquiring firm or its stockholders ([14], [16]). However, researchers from the field of strategic management have recently found evidence that contradicts the earlier findings of scholars who were predominantly from the finance discipline. The strategic management researchers ([21], [32]), using the same methodology as the finance scholars but with different time frames and slightly different assumptions, have concluded that mergers yield significant benefits to stockholders of both firms. Thus, a question currently exists as to whether mergers benefit only one (the acquired) or both firms involved. A priori, from the recent merger wave [4] it would appear as though both firms benefit. The clarion call for researchers in the area of mergers and acquisitions research is for more micro studies (focusing on individual industries) rather than macro approaches (pooling firms from different industries into one single sample) that examine merger outcomes from a longitudinal perspective ([9], [29]). This is seen as a way of resolving the current controversy that exists regarding merger performance. The purpose of this study was to measure merger performance on a longitudinal basis using a micro perspective. Specifically, this study looked at the performance of 88 Journal olBusiness Strategies Vol. 9, No.2 a sample of mergers drawn from the food and kindred products industry, Standard Industrial Classification Code 20, for a period of five years before and five years after the mergers using two performance measures. The performance measures, namely market returns to stockholders and return on investment, have been used extensively in the literature ([17], [18], [21], [24], [36]) to study the performance of mergers, albeit on macro samples. Because previous research provides conclusive support for increased returns following a merger to stockholders of acquired firms ([14], [16], [21]), this study uses the two measures mentioned earlier to determine returns only to the acquiring firms' stockholders. Also, because the effect of a merger can best be felt only in the long run [29], performance was measured for a period of five years after the merger and compared to performance five years prior to the merger. Such a longitudinal scope ensured that the anticipated synergies ([6], [7]) have had a chance to be effective. Theoretical Framework While Table 1 provides detailed descriptions of an illustrative list of studies on merger performance, this section discusses those studies that have a direct impact on the current work and help in the development of the research hypotheses. Table 1: mustrative Studies on Merger Performance Author(s) Measure: Returns Halpern (1983) Jensen and Ruback (1983) Lubatkin (1983) Methodology Conceptual- Review of literature. Conceptual- Review of literature. Conceptual- Review of literature. Findings While empirical evidence points out that mergers bring significant returns to stockholders of acquired firms, not enough evidence exists to suggest the same for the acquiring firms' stockholders. Stockholders of acquired firms benefit, but not stockholders of acquiring firms. Empirical studies point out that all significant benefits go to the acquired firm. Fall 1992 Subramanian, et aL: Evaluating Merger Performance Table 1: D1ustrative Studies (Continued) 89 Lubatkin (1987) Singh and Montgomery (1987) Measure: Risk Langeteig, Haugen, and Wichern (1980) Lubatkin and O'Neill (1987) Empirical- Archival data. Sample size 439 acquiring fIrms and 340 acquired fIrms. Event study approach using market-based performance measure. Empirical- Archival data. 105 fIrms from the period 1975-1980. Used a market-based performance measure. Empirical- Archival data. Sample of 149 firms from the period 1929- 1969. Used a self- developed measure of risk. Empirical- Archival data. Sample of 297 firms from the period 1954- 1973. Used Beta as a measure of risk. However, the strategic management literature suggests that the acquiring fIrms get tremendous benefIts. These claims have not been supported by empirical evidence. Mergers do lead to perman- nent gains in stockholder value for both acquiring and acquired rums, but differences not significant across merger types. Acquired rums in related acquisitions performed better than acquired fIrms in unrelated acquisitions. Mergers increase the risk for the merged fIrm. Part of the risk is due to increased leverage, other parts not explained by study results. Lowered risk is a valid rationale for mergers. All mergers increased the unsystematic risk for the merged fum, while related acquisitions lowered the systematic and total risks. 90 Journal ofBusiness Strategies Table 1: llIustrative Studies (Continued) Vol. 9, No.2 Measure: Market Share Mueller (1985) Empirical- Archival data. Sample of 332 firms from the period 1950-1972. Mergers result in a loss of market share to the acquired firm. Hopkins (1987) Other Measures Montgomery and Wilson (1986) Neely and Rochester (1987) Empirical- Archival data. Sample of 64 firms from Fortune 1000 for 1965. Empirical- Archival data. Samples of 434 firms that were aquired during 1967-1969. Empirical- Archival data from 37 S&Ls that merged matched with 37 that did not. Market share decreased for the acquired firm after the merger except in the case of marketing-related mergers where it went up. Used resale value to measure performance. Not enough evidence to suggest that unrelated acquisitions are bad. Merged savings and loans firms showed significant increases in profits and return on net worth. In their extensive review of the literature on merger performance, Subramanian, Ebrahimi, and lbibodeaux [37] emphasized the existing controversy regarding merger benefits. Halpern [14] reviewed finance literature on merger performance using an event study approach. He concluded that there exists strong evidence to suggest that higher than normal returns accrue to stockholders of acquired firms following a merger. The findings of the reviewed literature by Halpern [14] were, however, inconclusive regarding the returns to the stockholders of acquiring firms. While Langeteig [18] and Michel, Shaked, and Yobaccio [26] found empirical evidence to suggest that the returns to the stockholders of acquiring firms actually declined after a merger, Mandelkar [24] found that the returns were similar to other investments of comparable risk. Lubatkin [21] found evidence to suggest gains accruing to stockholders of both the acquired and the acquiring firms. Pettway and Yamada [32] found similar evidence in their study of mergers in Japan. Their conclusion is also supported by the empirical Fall 1992 Subramanian, et aL: Evaluating Merger Performance 91 research of Burgman [5], Chatterjee [7], Shelton [35], and Singh and Montgomery [36]. Thus, while the finance and strategic management disciplines agree on returns to stock- holders of acquired firms, they are divided as to the returns to stockholders of acquir- ing firms. Table 2 summarizes the conclusions of prior researeh on merger performance. Table 2: Summary of Conclusions From Prior Research on Merger Performance Using Market Return as Measure (A) Inconclusive evidence regarding merger benefits Researcher(s) *Mandelker (1974) Langeteig (1978) Michel, Shaked and Yobaccio (1983) Halpern (1983) Jensen and Ruback (1983) (B) Conclusive evidence regarding merger benefits Researcher(s) Lubatkin (1983) Burgman (1984) Pettway and Yamada (1986) Chatterjee (1986) Singh and Montgomery (1987) Lubatkin (1987) Shelton (1988) Methodology Empirical - event study Empirical - event study Empirical - event study Review of literature Review of literature Methodology Review of literature Empirical - event study Empirical - event study Empirical - event study Empirical - event study Empirical - event study Empirical - event study *actually concluded that mergers provided returns comparable to investments of similar risk Several explanations, such as different time frames (daily stock data versus monthly stock data) and the use of "clean" data (discarding firms engaged in multiple mergers from sample) by the finance researchers, have been offered to explain these conflicting results [23]. Additionally, recent researchers have emphasized the need to control for industry effects [10] and also to examine merger performance using a longitudinal instead of a cross-sectional time frame in order to fully capture its effect. 92 Journal ofBusiness Strategies Methodology Vol. 9, No. 2 The study method used for this research was statistical analysis of historical data obtained from published sources. While prior studies ([21], [36]) have used samples of fmns drawn from a variety of industries, the analysis group for the current study was composed entirely of fmns in the food and kindred products industry (SIC code 20). This study considered all publicly reported mergers and acquisitions (for which data were available) for the period from 1968 through 1984. The United States Fed- eral Trade Commission's (FTC) Large Merger Series contained in its Statistical Re- port on Mergers and Acquisitions [11] as well as "Merger Rosters" of various issues of Mergers and Acquisitions were used to identify target firms in the food and kin- dred products industry. There are many gaps in the publicly reported mergers and acquisitions activity data. Therefore, some firms were eliminated from the population of firms that were engaged in merger and acquisition activity during the period from 1968 to 1984. Although this reduced the sample size, the number of mergers and acquisitions that occurred during this period remained large enough for statistical analysis. Based on the guidelines suggested by the FTC [11], the sample of fIrms was classifIed into vertical, horizontal, product extension, market extension, and pure conglomerate merger types. A vertical merger occurs when a firm merges with or acquires a firm. that supplies it with inputs or are customers for its outputs. A merger or acquisition involving competitors is called a horizontal merger. Merging with or acquiring a firm. that makes related products involves a product extension merger. When a firm merges with or acquires another fum for the purpose of increasing its market coverage, it is involved in a market extension merger. Finally, merging with or acquiring a firm that is in a totally unrelated industry is an example of a pure conglomerate merger. This classification system has been used by other researchers [e.g., 21] to study merger pedormance. The dependent variable examined in this research study was pedormance. For the variable pedormance, two measures were used: accounting return on assets and market return to stockholders. The accounting return on assets (ROA) measure was calculated based on after-tax earnings (including extraordinary items) on year-end book value of total assets. The use of accounting-based pedormance measures have been criticized in the literature ([14], [23], [27]). Their use in this study is, however, justifIed on the ground that they measure an important aspect of pedormance, namely, the earnings stream that is at the disposal of the acquiring firm. as a percentage of the assets employed to earn the return. As an ex-post facto measure, return on assets complements an ex-ante measure such as market return to give a complete picture of pedormance. While there are several methods to measure market pedormance [38], this study used the relationship suggested by Kusewitt [17] in his work on factors associated with acquisitions pedormance. The justifIcation for using this formula was that it is simple Fall1992 Subramanian, et al.: Evaluating Merger Performance 93 to understand, easy to use, and is based on sound theoretical grounds. Moreover, it has been used to measure merger and acquisition performance in the past [e.g., 17]. Using Kusewitt's [17} formula, the individual year's market return was computed for each acquiring firm as follows: R = (Pt + Dt ) -------------- -1 where, Pt-l R = return on acquirer's stock for the year, Pt = arithmetic mean of high and low market price per share of stock in calendar year t, Pt-l = arithmetic mean of high and low market price per share of stock for the previous year, and Dt = dividend per share in year t. The market performance of a flIID for a period of five years before and five years after a merger was obtained by taking the geometric mean of the individual years' returns. The geometric mean was used because it is a more conservative average than the arithmetic mean and is better suited to account for outliers [8}. Merger type was the independent variable used in the study. Based on the categorization scheme suggested by the FTC [11}, the sample of firms was grouped into the five merger types mentioned previously. This independent variable assessed the impact of merger type on performance. Using Moody's Industrial Manual and Value Line Investment Surv(!y. data on each acquiring company's net income, year-end book value of assets, high and low stock prices, and dividends per share were obtained. Because both Moody's Industrial Manual and Value Line Investment Survey report data only on selected publicly held companies, several flIIDs had to be eliminated from the list. Incomplete data also resulted in the elimination of another set of firms. Finally, 80 of a possible 138 firms for which data were complete were assembled for the study. Since meeting the data requirements was the criterion employed in choosing firms, the sample was not a random sample in the probabilistic sense and any resulting biases could not be avoided. Results The descriptive statistics for the analysis group are shown in Table 3. As indi- cated by the means and standard deviations, the range of performance in the analysis group is remarkably large, particularly when one considers that this is over a ten-year period. Some acquiring firms did extremely well with their acquisition program while others did very poorly. This is indicated very clearly, especially by the jump in the average market return from 9.54 percent before merger to 30.44 percent after merger. This performance variability is consistent with other research fmdings [e.g., 17]. The descriptive statistics broken down by the type of merger are shown in Table 4. 94 Journal ofBusiness Strategies Table 3: Descriptive Statistics of Analysis Group (0 =SO) Vol. 9, No.2 Return on Assets Mean Standard Deviation Market Return Mean Standard Deviation • significant at the GO.05 level Before 6.66 2.92 9.54 25.82 F 4.41* 9.46* After 7.92 3.58 30.44 24.47 Table 4: Descriptive Statistics of Analysis Group By Merger Types (n = SO) Merger Typea 1 2 3 4 5 F Mean ROA Before 5.49 7.31 6.80 6.78 6.72 SD Before 1.32 4.01 3.31 2.63 2.55 1.05b Mean ROA After 6.91 7.65 7.34 12.33 8.06 SD After 2.16 2.62 3.02 6.02 4.01 Mean Mkt. Ret. Before SD Before Mean Mkt. Ret. After SD After 7.20 40.18 39.87 48.91 20.23 33.35 34.83 26.79 12.57 18.75 32.19 17.61 6.80 11.79 27.25 11.97 4.68 29.45 1.7~ 23.34 14.82 n 16 12 22 10 20 a 1 = horizontal, 2 = vertical, 3 = product extension, 4 = market extension, and 5 = pure conglomerate b not significant at the GO.05 level Yijk = Y = T· =I Fall 1992 Subramanian, et aL: Evaluating Merger Performance 95 Yertical mergers reported the highest ROA before merger, and market extension mergers showed the highest ROA after merger. While vertical mergers showed the highest market return before merger, the mean market return after merger was the highest for horizontal mergers. The standard deviations for market return was much higher than those for ROA Two hypotheses were generated from a review of the literature. The fIrst dealt with merger performance in general while the second sought to identify differences in performance among merger types. The prior literature is not conclusive as to the benefIts accruing to the stockholders of the acquiring firm, nor to differences in merger performance across merger types [21]. Therefore, Ho1: In the food and kindred products industry, mergers do not result in any change in performance by acquiring fIrms following the merger. Ho2: In the food and kindred products industry, there is no difference in performance among merger types. The following analysis of variance (ANOYA) model was constructed to test the hypotheses: I.l. +Ti +Bj +Ck(i) +l:iik where, performance measure, I.e., market return or ROA, merger type (1 = horizontal, 2 = vertical, 3 =market extension, 4 = product extension, and 5 = conglomerate), B· = time period where, 1 = before merger 2 = after merger, Ck(i) = company k within merger type i, and l:ijk = error term. The Imt hypothesis was rejected at the ao.05 level (F = 9.46, PR>F = 0.0034 for market return; F = 4.41, PR>F = 0.0653 for ROA) providing support for the contention that mergers did result in improved performance on both measures. However, neither performance measure yielded results to indicate that the difference among merger types was statistically signiftcant. This is consistent with prior research findings by Lubatkin [21]. Discussion This study attempted to analyze merger performance within an industry-specific domain. Using two performance measures, it offered significant statistical support for the hypothesis that mergers benefIt the acquiring fIrm and its stockholders. This evidence, while being consistent with Lubatkin's [21] findings, contradicts earlier results from the finance discipline reported by Halpern [14] and Jensen and Ruback [16]. This could be attributed to two possible reasons. 96 Journal ofBusiness Strategies Vol. 9, No.2 First, researchers from the finance discipline used short time frames to measure merger performance--the most common time period used was 180 days before and 180 days after the merger [25]. Presumably, this short time period was used to reduce bias associated with extraneous events. Because market models have been shown to factor in the effects of time [13], it is argued by strategic management researchers [e.g., 23] that a five-year time period captures the strategic impact of a merger better than a shorter time frame. Thus, Lubatkin's [21] study, using a five-year time frame both before and after the merger, reported results that contradicted earlier findings of researchers from the finance discipline. Second, none of the earlier researchers studying merger performance controlled for industry effects. In other words, their samples included firms from a myriad of industries with no countervailing checks-and-balances to account for this diversity that may have contaminated the results. Lubatkin [21] argues that market models adjust for industry variation. This, of course, assumes a perfectly efficient market. But, other researchers ([3], [10]) report that inefficiencies in the market create "noise," rendering it less than perfect. Limiting the sample to firms from only one industry is one of the ways of controlling for industry effect [10]. In this study, the average ROA for the focus group increased from 6.66 percent to 7.92 percent after the merger-an increase of 18.9 percent compounded for the five- year period. A look at the standard deviation (2.92 before and 3.58 after the merger) indicates that the range of performance in the analysis group was remarkably large. Some a