The Performance of Acquisitions in the Real Estate Investment Trust Industry

Transcription

1 The Performance of Acquisitions in the Real Estate Investment Trust Industry Author Olgun F. Sahin Abstract This study examines the performance of acquisitions in the Real Estate Investment Trust (REIT) industry around the acquisition announcement and in the long-run. The results suggest that the acquiring REITs experience statistically significant negative abnormal returns while the target REITs earn statistically significant positive abnormal returns during the three-day period around the announcement. The long-run performance of the acquiring REITs is analyzed using size benchmark portfolios with the buy-and-hold, cumulative average and mean calendar time abnormal returns, as well as the Fama French Three Factor Model. None of the other methods detect significant abnormal returns in the long-run with the exception of the buy-and-hold abnormal return. Further analysis shows that the long-run positive buy-and-hold abnormal return is consistent with an unexpected decline in cost of equity after acquisitions. Introduction The Real Estate Investment Trust (REIT) industry experienced relatively significant consolidation activity during 1990s. These acquisitions raise questions on the wealth effects in the short- and long-run for shareholders. Generally, firms make acquisitions for several reasons, such as market power, synergies in operations, removal of inefficient management, reduction of bankruptcy costs and tax loss benefits. The findings of previous studies suggest that the acquiring firm shareholders experience non-significant abnormal returns, while target shareholders earn significant positive abnormal returns around acquisition announcements. The long-run performance of the acquiring firms is not resolved. Early studies identify significantly negative abnormal returns to acquiring firms, although recent research points out problems associated with risk adjustment methods and biases in test statistics. This study examines the performance of REITs involved in acquisitions around the announcement and during a three-year period after the announcement. The market reaction to acquisition announcements reflects the market s assessment of acquisitions. The adjustment resulting from acquisition announcements in an JRER Vol. 27 No

2 322 Sahin efficient market takes place quickly, allowing no abnormal performance in the long-run or gradual adjustments over time. Therefore, the short-run analysis reflects the market s view about acquisitions while the long-run analysis has implications regarding market efficiency. The sample includes thirty-five acquisition transactions among REITs between 1994 and The market model and the market-adjusted return are used to estimate the announcement period abnormal returns. The results suggest that the acquiring REITs experience statistically significant negative abnormal returns while the target REITs earn statistically significant positive abnormal returns during the three-day period around the announcement. There are studies that examine the announcement period abnormal returns. However, the long-run performance of acquiring REITs has not been documented. In this study, the performance of an acquiring REIT is compared to a benchmark portfolio based on firm sizes measured by the market value of equity. The study utilizes four methods of abnormal return calculations after acquisitions: the Cumulative Average Abnormal Return (CAAR), the Buy-and-Hold Abnormal Return (BHAR), the Mean Calendar-Time Abnormal Return (MCTAR) and the Fama French Three Factor Model. No method detects significant abnormal returns in the long-run with the exception of the BHAR. The study also examines changes in cost of equity for acquiring REITs after the event. The findings suggest that the long-run positive buy-and-hold abnormal return is consistent with an unexpected decline in cost of equity after acquisitions. Literature Review and Hypotheses Allen and Sirmans (1987) examine the performance of thirty-eight successful acquisitions among REITs over the period. The sample acquisitions are collected from Moody s Bank and Finance Manual. Allen and Sirmans use the mean adjusted return method to determine the abnormal performance of acquiring REITs. The mean returns of sample firms are calculated using the data for days (120, 40) relative to the announcement day. The abnormal return, which is the event firm return minus its mean return, is estimated over days (40, 40). The results of suggest that acquiring REIT firms experience a statistically significant CAAR of 8.47% over days (10, 0) prior to the announcement. The CAAR for days (1, 40) is 0.71% and insignificant. The authors suggest that the pre-announcement gains might be due to improved managerial efficiency. To test for improved managerial efficiency, they identify the related and unrelated acquisitions. The assumption is that the related acquisitions lead to improved managerial and financial performance. The related acquisitions earn a statistically significant higher return (2.02%) than the unrelated acquisitions over days (1, 0). McIntosh, Officer and Born (1989) analyze the impact of acquisitions on specific target REITs. The sample includes twenty-three REITs that were subject to

3 The Performance of Acquisitions 323 takeovers between July 1962 and December The authors use the market model to determine the performance of the REITs. In successful acquisitions, the REITs experience a significant abnormal return of 3.34% during the day before the announcement of the acquisition in The Wall Street Journal. The abnormal returns are small and insignificant over days (1, 30). Campbell, Ghosh and Sirmans (1998) calculate the three- and five-day abnormal returns for acquiring the target REITs. The sample contains twenty-five transactions that took place between January 1989 and January Because the sample period ends at the beginning of January, a significant number of acquisitions that took place in 1998 are not included. The study finds that acquiring REITs lose 1.1% while target shareholders gain 5.2% over the fiveday period (2, 2) relative to the announcement. The returns are adjusted based on the Willshire Daily REIT Index. Unfortunately, the statistical significance of these results is not reported. Campbell, Ghosh and Sirmans (2001) analyze the information content of the payment method in REIT acquisitions involving publicly traded acquiring and target REITs and privately held target REITs. The sample contains eighty-five transactions (of which forty involve publicly traded REITs) between 1994 and The three-day (1, 1) abnormal return for acquiring and target REITs are 0.6% and 3.0%, respectively. The study attributes the negative market reaction to the information signaling that the stock is overvalued. This study examines the market performance of REITs involved in acquisitions around the announcement and during a three-year period after the announcement. The market reaction to acquisition announcements provides insights on how the market views the future prospects of acquisitions. To examine the possible wealth effects around the announcement, this study tests the hypotheses that there are no abnormal returns to shareholders of acquiring and target firms. In efficient markets, event firm prices quickly reflect the impact of acquisitions allowing no abnormal performance in the long-run or gradual adjustments over time. The study also tests the hypothesis that the long-run performance of acquiring firms is not significantly different from zero. The announcement period results reflect the market s view about acquisitions while the long-run analysis has implications regarding market efficiency. Data and Methodology Data In order to identify the acquisitions, the CRSP database is searched for REITs delisted between 1990 and 2000 due to acquisitions (delisting codes between 200 and 203). The 2001 edition of the Directory of Obsolete Securities, published by Financial Information Incorporated, is used to identify acquiring REITs for each JRER Vol. 27 No

4 324 Sahin delisted REIT. The Wall Street Journal Index is searched to identify the announcement dates of these acquisitions. The resulting sample contains thirtyfive transactions among REITs. Due to data availability, the number of acquiring REITs included in the analyses varies. Panels A and B of Exhibit 1 show the distribution of acquisition announcements over time and the market value data as of two days prior to the announcements, respectively. REIT returns are obtained from the CRSP. Methodology The standard market model and the market-adjusted return are used to measure the abnormal returns during the announcement period, days (1, 1). The parameters of the market model and market-adjusted return are estimated using the data over days (200, 21). The significance of the market model and marketadjusted return is conducted following Bosch and Hirschey (1989) and Beneish and Gardner (1995), respectively. The market in both models is represented by the CRSP value-weighted market index and a value-weighted REIT index constructed to include only non-event REITs. The abnormal dollar return method of Malatesta (1983) is also applied to assess the wealth effects of acquisitions in absolute terms. Malatesta defines the dollar abnormal return as the price of an acquiring firm times the number of shares outstanding (both taken two days before the announcement date) times the abnormal return over days (1, 1). Moeller, Schlingemann and Stulz (2003) estimate the percentage net present value (NPV) of acquisitions. The net present value of a transaction is the same as the dollar abnormal return of Malatesta. Moeller, Schlingemann and Stulz divide the acquiring firm NPV by the total transaction value to determine the percentage NPV. The long-run performance of acquiring REITs is measured for a three-year period. The benchmark portfolios are constructed based on the market values of non-event REITs, following the evidence suggested by McIntosh, Liang and Tompkins (1991) and Chen, Hsieh, Vines and Chiou (1998). The benchmark portfolios are formed as follows: At the end of each month, all REITs are ranked based on market values (price per share times the number of shares outstanding) and placed into quintile portfolios. The number of REITs used to form these benchmark portfolios are provided in Panel C of Exhibit 1. The event firm returns are calculated starting with the first trading day of the next calendar month to allow for benchmark portfolio formation. The expected returns of acquiring REITs are represented by the returns on the corresponding size portfolios. This study utilizes three techniques for calculating abnormal returns: the CAAR, the BHAR and the MCTAR. Early research on long-run performance after corporate events estimates abnormal performance using the CAAR. The CAAR is estimated as follows:

5 The Performance of Acquisitions 325 Exhibit 1 Sample Characteristics Panel A: Announcements during the sample period Year Announcements Total 35 Panel B: Market value statistics of acquiring and target firms (in thousands except number of firms) Acquirer Target Average 1,240, ,311 Std Dev. 1,278, ,686 Min. 13,262 11,673 Max. 5,088,552 2,028,109 Med. 756, ,646 Firms Panel C: Number of REITs used to form size benchmark portfolios (annual average of monthly number of REITs) Year REITs Notes: In order to identify the acquisitions, the CRSP database is searched for REITs delisted between 1990 and 2000 due to acquisitions identified by the delisting codes between 200 and 203. The Wall Street Journal Index is searched to identify the announcement dates of these acquisitions. The resulting sample contains 35 transactions announced and completed between January 1994 and December Panel B of the table is based on market values as of two days prior to the acquisition announcements for firms with available data. JRER Vol. 27 No

6 326 Sahin ar r E(r ), (1) i,t i,t i,t where t is the month relative to the event month, ar i,t is the abnormal return on asset i for month t, r i,t is the return on asset i for month t and E(r i,t ) is the expected rate of return on asset i for month t. The Average Abnormal Return (AAR t )ona portfolio of N stocks for event month t is: N 1 N i1 AAR ar. (2) t i,t The CAAR from event month q to event month s is: s tq CAAR AAR. (3) q,s t Ritter (1991) calculates the t-statistic for AAR t for each month as: N t t-statistic AAR t, (4) sd t where sd t is the cross-sectional standard deviation of the adjusted return for month t. The t-statistic is used to form the hypothesis that the ARR of sample firms during event month t is equal to zero. The t-statistic for CAAR 1,t is computed as follows: N t t-statistic CAAR 1,t, (5) csd t and csd (t)(var) 2(t 1)cov, (6) t

7 The Performance of Acquisitions 327 where t is the event month, var is the average cross-sectional variance over the assumed time horizon and cov is the first-order auto-covariance of the AAR t series. The t-statistic tests the hypothesis that the CAAR over months (q, s) relative to the event month is equal to zero. There are conceptual and statistical problems associated with the use of the CAAR to measure the abnormal performance in the long-run. The CAAR does not measure the actual investor experience and ignores the compounding of returns that provides biased results according to Barber and Lyon (1997). Conrad and Kaul (1993) show that the CAAR method cumulates not only the actual abnormal returns, but also the upward or downward biases in periodic returns. This might be due to the bid-ask effect over long periods leading to substantially high or low CAARs. However, Fama (1998) suggests that despite problems associated with the CAAR methodology, it still experiences fewer statistical inference problems than its alternative, the BHAR method. Since Ritter (1991), the BHAR has become one of the most commonly used measures of long-run performance. Following Mitchell and Stafford (2000), the BHAR on firm i over T period can be calculated as: T T i t1 i,t t1 control,t BHAR (1 r ) (1 t ). (7) The average BHAR over N firms is: N 1 N i1 BHAR BHAR. (8) i The t-statistic for the BHAR is: N t-statistic BHAR, (9) BHAR where BHAR is the cross-sectional standard deviation of individual firm BHARs. The t-statistic tests the hypothesis that the mean buy-and-hold abnormal return of sample firms over T period after the event is equal to zero. Although the BHAR takes into account the investor experience, it suffers from skewness and cross-sectional dependence problems. Barber and Lyon (1997) JRER Vol. 27 No

8 328 Sahin suggest that long-run BHARs are positively skewed because it is possible to observe an event firm return greater than 100%, although not on a control portfolio. The positive skewness of the BHARs leads to an inflated estimate of the cross-sectional standard deviation and, therefore, downwardly biased test statistics. The effect of the cross-sectional dependence is the opposite. The crosssectional correlation in the BHARs leads to a tight distribution, which results in upwardly biased test statistics. Pseudo-portfolios are used to compute the empirical p-value under the null hypothesis of zero abnormal performance to solve the skewness associated with the BHAR methodology. Pseudo portfolios are used by Brock, Lakonishok and LeBaron (1992), Ikenberry, Lakonishok and Vermaelen (1995) and Lee (1997) to generate the empirical p- values. The method is based on generating the empirical distribution under the null hypothesis of zero abnormal performance. For each event firm, a representative firm with similar risk characteristics is chosen to be included in the pseudo portfolio. This procedure is continued until all the event firms are represented by non-event firms. The BHAR for firms in the pseudo portfolio is calculated using the same method. The grand average of individual firm BHARs yields an observation of abnormal return. The formation of the representative portfolio is repeated 10,000 times, leading to 10,000 observations of abnormal returns. Lee (1997) calculates the empirical p-value for 10,000 trials as: empirical p-value Number of trials with BHAR less than or equal to sample BHAR. 10,000 (10) Although the empirical p-value method eliminates the skewing bias, the crosssectional dependence is still problematic because the procedure assumes that the BHARs are independent. The MCTAR eliminates cross-sectional dependence. This method is based on forming portfolios of firms that complete an event during the prior three-year period. These portfolios are formed every month in calendar-time by adding new firms and dropping firms that reach the end of the holding period. The crosssectional dependence is not problematic because portfolios are formed in calendartime. Following Lyon, Barber and Tsai (1999), the MCTAR is calculated as follows: AR R R, (11) i,t i,t p,t

9 The Performance of Acquisitions 329 where t is the calendar month t, AR i,t is the abnormal return on firm i in calendar month t, R i,t is the actual return on firm i in calendar month t, and R p,t is the return on a control portfolio in calendar month t. The MCTAR in calendar month t, is: n t t i,t i,t i1 MCTAR x AR, (12) where n t is the number of firms in calendar month t, and x i,t is the weight for firm i in calendar month t. The grand mean of monthly abnormal returns is: T 1 T t1 MCTAR MCTAR, (13) t where T is the total number of months. A t-statistic is calculated using the timeseries standard deviation of the mean monthly abnormal returns as follows: T t-statistic MCTAR. (14) MCTAR The t-statistic tests the hypothesis that the mean monthly abnormal return of the event firm portfolio is equal to zero during the study period. Jaffe (1974) standardizes the monthly abnormal returns using within month variance and Mitchell and Stafford (2000) require at least ten firms in any calendar month to form a portfolio to adjust for heteroscedasticity, due to changing portfolio composition over time. This study also employs the Fama French Three Factor Model based on Fama and French (1992, 1993). Buttimer, Hyland and Sanders (2001) apply the Factor Model to analyze the long-run performance of REIT Initial Public Offerings (IPOs) between 1980 and These authors find no significant abnormal performance except the sub-period when IPO portfolio returns are equally weighted. The following regression is estimated to examine the long-run performance of acquiring REITs using the Fama French Three Factor Model: R R RMRF ssmb hhml e, (15) p,t ƒ,t p p t p t p t p,t JRER Vol. 27 No

10 330 Sahin where R p,t is the equal or value weighted monthly return on the event firm portfolio in calendar month t, R ƒ,t is the risk-free rate of return, RMRF t is the excess return on the value-weighted market portfolio, SMB t is the return difference between the portfolios of small and large company stocks, HML t is the return difference between the portfolios of high and low book-to-market stocks and p is the average monthly abnormal return on the event portfolio of firms. 1 The regression is run for a 36-month period before and after the event, excluding the event month. The test hypothesis is that p is equal to zero. Results Announcement Period Abnormal Returns Exhibit 2 shows the market model and market-adjusted abnormal returns for acquiring and target REITs over days (1, 1) relative to the day of the announcement. Abnormal returns could not be estimated for five acquiring and two target REITs because the market model and market-adjusted return require historical return data. Both models identify statistically significant abnormal returns to target firms. The three-day abnormal return for target REITs varies between 4.31% and 4.45%, depending on the assumed market portfolio and estimation procedure for abnormal returns. The abnormal returns for target REITs are statistically significant. Although small in magnitude, the three-day abnormal returns for acquiring REITs are 1.21% and 1.16% using the market model and the market-adjusted return with the CRSP value-weighted market index, respectively. The abnormal return figures are statistically significant for the acquiring firms as well. When the value-weighted REIT index is used, similar abnormal returns for acquiring REITs are identified. These findings are consistent with Campbell, Ghosh and Sirmans (1998, 2001) that acquiring firms experience small negative returns while target shareholders gain around the time of the announcement. On the other hand, the findings contradict Allen and Sirmans (1987) that acquiring REIT shareholders benefit from acquisitions. 2 Moreover, the results indicate lower gains to target REIT shareholders during the announcement period, as opposed to the findings of McIntosh, Officer and Born (1989). 3 Differences in sample periods may account for the contradicting results since both groups use samples of acquisitions prior to Abnormal dollar returns around acquisition announcements are reported on Exhibit 3. Calculations in Panel B of are based on Malatesta (1983). On average, an acquiring REIT loses about $38.8 million while a target REIT gains about $19.7 million over days (1, 1). In aggregate, acquiring REIT losses amount to about $1.16 billion (for thirty acquiring REITs) while target gains are around $ million (for thirty-three target REITs). The aggregate dollar return for

11 The Performance of Acquisitions 331 Exhibit 2 Abnormal Returns to Acquirer and Target REITs Around Acquisition Announcements Panel A: Acquiring REIT performance around the announcement CRSP Value-Weighted Market Index Market Model Market-Adjusted Abnormal return (1, 1) t-statistics 2.61*** 2.26*** N Value-Weighted REIT Index Market Model Market-Adjusted Abnormal return (1, 1) t-statistics 2.54*** 2.33*** N Panel B: Target REIT performance around the announcement CRSP Value-Weighted Market Index Market Model Market-Adjusted Abnormal return (1, 1) t-statistics 9.15*** 8.40*** N Value-Weighted REIT Index Market Model Market-Adjusted Abnormal return (1, 1) t-statistics 9.45*** 9.07*** N Notes: This table shows the three-day (1, 1) cumulative abnormal returns around acquisition announcements. The cumulative abnormal returns are estimated using the market model and the market-adjusted return. Market model parameters are estimated using the data over days (200, 21) relative to the announcement. In Panel A, results are shown for acquirers. Panel B presents the results for target REITs. The market portfolio is represented by the CRSP value-weighted market index or value-weighted REIT index. ***Significant at the 1% level. JRER Vol. 27 No

12 332 Sahin Exhibit 3 Announcement Abnormal Returns, Dollar Abnormal Returns and Percentage NPV of Acquisitions Acquirers Targets Panel A: Abnormal return over days (1, 1) (in percent except the number of firms) Mean 1.21*** 4.31*** Median 0.41* 5.07*** Negative percent N Panel B: Mean and total abnormal dollar returns over days (1, 1) Mean dollar return 35,206,329 17,819,971 Mean dollar return (2001) 38,775,809 19,692,987 Total dollar return 1,056,189, ,059,044 Total dollar return (2001) 1,163,274, ,868,562 N Panel C: Percentage NPV of acquisitions over days (1, 1) Mean percent NPV Median percent NPV 5.31** 1.02* Negative percent N 29 Notes: Panel A reports the market model abnormal returns over days (1, 1). Calculations in Panel B of the table are based on Malatesta (1986). The author defines the dollar abnormal return as the price of an acquiring firm times the number of shares outstanding (both taken two days before the announcement date) times the abnormal return over days (1, 1). Mean and total abnormal dollar returns are reported in current and 2001 dollars. Panel C shows percentage NPV of acquisitions. The NPV of a transaction is the same as the dollar abnormal return of Malatesta (1986). Moeller, Schlingemann and Stulz (2003) divide the acquiring firm NPV by the total transaction value to determine the percentage NPV. Reported percent NPVs are based on acquiring NPV divided by the market value of the target two days before the announcement. The number of percentage NPV observations is one fewer because one of the target firm s market value could not be calculated due to lack of data availability. The test of significance for the median is based on the sign test. *Significant at the 10% level. **Significant at the 5% level. ***Significant at the 1% level.

13 The Performance of Acquisitions 333 acquiring REITs is likely to be underestimated because of the disproportionate number of REITs. Panel C of shows the percentage NPV of acquisitions. The percentage NPV of a transaction is the dollar abnormal return divided by the total transaction value. Reported percentage NPVs are based on the acquiring NPV divided by the market value of the target REIT two days before the announcement. The total number of NPV observations is one less because a target firm s market value could not be calculated due to lack of data availability. According to the results, the mean percentage NPV is 5.31% and significant. Given the findings, it appears that target REIT shareholders benefit from acquisitions while acquiring REIT shareholders face losses at the announcement. Acquiring REIT Performance in the Long-Run Exhibit 4 presents the results of the AAR and the CAAR for acquiring REITs. The CAAR reaches the highest level during the eighth month and falls below zero during the twelfth month. The CAAR is 1.41% over the 36-month period. Neither the AARs nor the CAARs are significant in any month after the announcement. Panel A of Exhibit 5 reports the BHAR. The average BHAR for acquiring REITs is 3.56% during the three-year period after the announcement and insignificant using the conventional t-statistic. The cross-sectional standard deviation of acquiring REIT BHARs is 37.41%. The median BHAR is 11.62% and is significant at the 10% level. Approximately 26% of the acquiring REITs experience negative BHARs. Pseudo-portfolios are formed to calculate the empirical p-value to eliminate the potential skewness problem. Panel B of Exhibit 5 suggests that the empirical p-value of 3.56% means the BHAR is 0.72, indicating that 72% of the time non-event firm portfolio BHARs are below the sample BHAR. The median BHAR of 11.62% has an empirical p- value of.953, which indicates significance at the 5% level. The mean of the simulation procedure is 0.13% and the standard error is 0.10% after 5,000 iterations. Fifty-two percent of the time pseudo-portfolios have negative abnormal returns. Exhibit 6 shows the distribution of the pseudo-portfolio BHARs. The sample REIT portfolio BHAR clearly does not fall outside of the critical values of the distribution generated under the null hypothesis of zero abnormal performance. The result of the MCTAR method is reported on Exhibit 7. The portfolios are formed in calendar-time to account for cross-sectional dependence. Each event firm is held in the portfolio for thirty-six months. Since the calendar-time portfolios are formed each month, portfolio composition changes every month. If there are fewer than ten event REITs in a month, that particular month is not included in the statistical testing. The portfolio return is calculated using equaland value-weighted individual REIT returns each month. The average abnormal returns are 0.01% and 0.19% per month or 0.07% and 2.32% per year using JRER Vol. 27 No

14 334 Sahin Exhibit 4 Average Abnormal Return (AAR) and Cumulative Average Abnormal Return (CAAR) to Acquirers Month Number of Firms AAR t-statistics CAAR t-statistics Notes: The abnormal returns on each acquiring firm are generated using size matching portfolios. Each month represents twenty-one trading days, starting two days after the announcement. The test statistics are generated by following Ritter (1991).

15 The Performance of Acquisitions 335 Exhibit 5 Three Year Buy-and-Hold Abnormal Return (BHAR) to Acquirers Panel A: Conventional t-statistic Mean BHAR 3.56 Cross-sectional standard deviation t-statistic 0.53 Median 11.62* Negative percent N 31 Panel B: Pseudo portfolios Mean BHAR of the sample 3.56 Empirical p-value 0.72 Notes: BHARs are calculated using size-matching portfolios. The matching portfolios are formed at the end of every month by ranking all REITs based on market values. Panel A reports the conventional t-statistic and sign test for the median. Panel B shows the results of the simulation procedure to generate BHARs under the null hypothesis of zero abnormal performance. The simulation procedure employed requires random selection of a non-event REIT for each actual event REIT matched based on size portfolio assignment. Abnormal returns for the non-event REITs are calculated the same way as actual event REITs. After forming a pseudo portfolio of non-event REITs, the mean BHAR of the portfolio is computed. This formation is repeated 5,000 times to generate the empirical p-value. The empirical p-value for the sample is the number of trials with BHAR less than or equal to sample BHAR divided by 5,000. *Significant at the 10% level. equal and value weights, respectively. Although these abnormal returns are not statistically significant, they suggest that large REITs perform better after the acquisitions than smaller REITs. The results of the Fama French Three Factor Model are reported on Exhibit 8. The Fama French Three Factor Model does not adequately capture the acquiring REIT returns, especially during the pre-event period. This indicates that there might be other significant factors in explaining REIT returns. The acquiring REITs experience insignificant positive abnormal returns during the 36-month pre-event period. Intercepts of the post event regressions are negative with equal and value weighting; however, they are not statistically significant. Additionally, the abnormal return of value-weighted portfolios is half the abnormal return of equalweighted portfolios, which indicates that large REITs perform better than small REITs. The analysis thus far shows that there is some evidence of positive abnormal performance after acquisitions based on the BHAR method. The next section of this study addresses potential explanations of positive abnormal performance. JRER Vol. 27 No

16 700 Exhibit 6 Frequency Distribution under the Null Hypothesis 336 Sahin Frequency % % % % -8.00% -4.00% 0.00% 4.00% 8.00% 12.00% 16.00% 20.00% 24.00% 28.00% This graph shows the frequency distribution of pseudo portfolio BHARs after 5,000 simulations. Each bar represents the number of times a particular value of BHAR is observed for size matching portfolio of non-event REITs. Sample mean and median BHARs are 3.56% and 11.62%, respectively.

17 The Performance of Acquisitions 337 Exhibit 7 Mean Monthly Calendar-Time Abnormal Return (MCTAR) Equal-Weighted Value-Weighted Mean calendar-time abnormal return t-statistics Number of months Notes: Results are based on forming portfolios of event firms in calendar time. Abnormal return for each firm in each calendar month is calculated by using the size-matching portfolios. A calendar-time portfolio is formed only if there are ten REITs that acquire other REITs during the last 36-month period. An event firm is kept in the portfolio for 36 months after the announcement of the acquisition. The t-statistics are computed on abnormal returns that are standardized by withinmonth variance, which changes the sign of the statistics for equal-weighted calendar-time portfolios. Boehme and Sorescu (2002) suggest that in an Efficient Markets framework, post event positive abnormal returns can have two explanations that are not mutually exclusive and are jointly complete. First, a decrease in the Factor Model coefficients indicates an unexpected reduction in the cost of equity that is not related to acquisitions. The post event lower cost of equity is not anticipated at the announcement. As investors realize the lower cost of equity, returns improve. Second, according to Fama and French (1997), decreases in the Factor Model coefficients are related to unexpectedly improved cash flows. If unexpectedly high performing firms dominate a sample, the positive abnormal performance and lower Factor Model coefficients are likely to be observed. Boehme and Sorescu do not offer a test for the second potential explanation. Risk Changes and the Long-Run Performance Boehme and Sorescu (2002) test the first explanation by estimating the following firm level Fama French Three Factor Model regression: R R D RMRF ssmb hhml i,t ƒ,t i t i i t i t i t DRMRF s DSMB h D HML e, i t t i t t i t t i,t (16) where R i,t is the return of an acquiring or target REIT, R ƒ,t is the one-month Treasury bill rate, RMRF t is the excess return on the value-weighted market portfolio, SMB t is the return difference between the portfolios of small and large JRER Vol. 27 No

18 338 Sahin Exhibit 8 The Fama French Three Factor Model Acquirers Targets Panel A: Equal-weighted portfolios Pre-event Post event Pre-event Intercept (0.66) (1.29) (0.44) RMRF 0.55*** 0.65*** 0.47*** (3.24) (6.50) (3.25) SMB 0.55*** 0.32*** 0.51*** (2.88) (3.59) (3.12) HML 0.71*** 0.75*** 0.51** (2.62) (6.22) (2.23) Adj. R Panel B: Value-weighted portfolios Intercept (0.68) (0.53) (0.47) RMRF 0.64*** 0.68*** 0.56*** (3.44) (5.31) (3.51) SMB 0.53** 0.34*** 0.49*** (2.55) (3.02) (2.76) HML 0.68** 0.82*** 0.60** (2.33) (5.34) (2.38) Adj. R Notes: The Fama French Three Factor Model is based on the following regression: R R RMRF ssmb hhml, p,t f,t p p t p t p t p,t where R p,t is the value or equal-weighted returns on the acquiring or target REIT portfolio, R f,t is the one-month Treasury bill rate; RMRF t is the excess return on the value-weighted market portfolio; SMB t is the return difference between a portfolio of small and a portfolio of large firms; and HML t is the return difference between a portfolio of high book-to-market stocks and a portfolio of low book-to-market stocks. The regression is run for 36-month periods before and after the event, excluding the event month. The Model for target REITs is estimated for the preevent period. The regression provides coefficients of p, s p and h p. The intercept term, p, measures the mean monthly abnormal return. **Significant at the 5% level. ***Significant at the 1% level.

19 The Performance of Acquisitions 339 Exhibit 9 The Fama French Three Factor Model and Post Event Risk Changes Acquirers Targets Panel A: Average pre-event coefficients i *** ** (7.61) (2.23) s i *** *** (8.34) (4.18) h i *** ** (7.28) (2.38) Panel B: Average change in coefficients i (0.04) s i h i *** (3.02) (0.12) Panel C: Average monthly change in the required return due to each Fama French Factor loading (percent per month) RMRF (0.06) SMB HML 0.014*** (2.92) (0.11) Notes: The Fama French Three Factor Model is based on the following event firm regressions: R R D RMRF ssmb hhml i,t f,t i i i 1 t i t i t DRMRF s DSMB h DHML e, i t t i t t i t t i,t where R i,t is the return of an acquiring or target REIT, R f,t is the one-month Treasury bill rate; RMRF t is the excess return on the value-weighted market portfolio; SMB t is the return difference between a portfolio of small and a portfolio of large firms; and HML t is the return difference between a portfolio of high book-to-market stocks and a portfolio of low book-to-market stocks. The firm specific regressions for acquirers are run over the months (36, 36) excluding the event month. D t is one if a month is in the post event period and zero otherwise. The regression provides coefficients of i, s i and h i for the pre-event period and i, s i and h i as the post event changes in the Fama French Three Factor Model loadings. The regressions for target REITs are run for the 36-month pre-event period. Percentage changes are estimated by multiplying the average factor loading changes by the average of the Fama French Factors over the period from August 1992 to November **Significant at the 5% level. ***Significant at the 1% level. JRER Vol. 27 No

20 340 Sahin company stocks and HML t is the return difference between the portfolios of high and low book-to-market stocks. The firm-specific regressions for acquiring firms are run over the months (36, 36) excluding the event month. D t is set to one if the month is in the post event period and zero otherwise. The regression provides coefficients of i, s i and h i for the pre-event period and i, s i and h i as the post event changes in the Fama French Three Factor Model loadings. The regressions for target REITs are run for the 36-month pre-event period only. The results of the regressions are reported in Exhibit 9. Panel A reports the preevent average factor loadings for acquiring and target REITs. Average changes in factor loadings are shown in Panel B. Acquiring REITs display a loading increase in the market factor with decreases in loadings of size and book-to-market factors. The average change in the size factor loading is statistically significant. These results indicate that REITs became more sensitive to the market factor and less sensitive to size and book-to-market factors after acquisitions. The resulting monthly percentage change in the required return is reported in Panel C. Percentage changes are estimated by multiplying the average factor loading changes by the average of the Fama French factors over the period from August 1992 to November The time period reflects the window of months where individual firm returns are used to estimate Equation (16). Panel C shows that the average reduction in the required return on equity due to size factor is 0.014% per month, which is significant. Overall, the results support the notion that an unexpected post event decrease in the cost of equity might lead to improved stock returns as investors realize the lower cost of equity. Conclusion This study examines the performance of acquiring REITs around the acquisition announcement and during the three-year period after the announcement. The announcement period abnormal returns are estimated using the market model and market-adjusted return where the market is represented by the CRSP equalweighted market index and a non-event REIT index. The results suggest that target REITs experience statistically significant positive abnormal returns of 4.31% during the three-day announcement window. The abnormal returns to acquiring REITs are small and negative (1.21%), and statistically significant. The results of this and other studies suggest an important difference in the market response to acquisitions in the REIT industry before and after Acquisitions prior to 1990 resulted in positive abnormal returns to acquirers [Allen and Sirmans (1987)] while abnormal returns to acquirers were negative [Campbell, Ghosh and Sirmans (1998, 2001) and the current study)] during the 1990s around the announcement. To estimate the abnormal performance in the long-run, the firm size is used to establish benchmark portfolios of REITs at the end of each month during the study period. Abnormal returns are calculated and tested based on the CAAR, the BHAR, the MCTAR and the Fama French Three Factor Model. Excluding the BHAR results, none of the methods detect significant abnormal returns in the

21 The Performance of Acquisitions 341 long-run after acquisitions. The long-run performance is consistent with market efficiency and any significant positive long-run performance appears to be related to an unexpected decline in equity risk after acquisitions. The change in equity risk indicates that REITs became less sensitive to the size factor of the Fama French Three Factor Model after acquisitions. Endnotes 1 Thanks to Kenneth R. French for kindly providing RMRF, SMB and HML factor returns at his website: mba.tuck.dartmouth.edu/ pages/ faculty/ ken.french/ data library. html. 2 Allen and Sirmans (1987) report a significant abnormal return of 5.78% over event days (10, 0) and an insignificant return of 1.66% over event days (1, 10). For the current study, the abnormal returns are calculated to be 1.49% for the first period and 0.10% for the second period. These abnormal returns are not statistically significant. 3 The cumulative abnormal return to target REITs in successful acquisitions over days (1, 1) is calculated to be 6.88% based on abnormal returns. References Allen, P. R. and C. F. Sirmans, An Analysis of Gains to Acquiring Firm s Shareholders: The Special Case of REITs, Journal of Financial Economics, 1987, 18, Barber, B. and J. Lyon, Detecting Long-run Abnormal Stock Returns: The Empirical Power and Specification of Test Statistics, Journal of Financial Economics, 1997, 43, Beneish, M. D. and J. C. Gardner, Information Costs and Liquidity Effects from Changes in the Dow Jones Industrial Average List, Journal of Financial and Quantitative Analysis, 1995, 30, Boehme, R. D. and S. M. Sorescu, The Long-run Performance following Dividend Initiations and Resumptions: Underreaction or Product of Chance?, Journal of Finance, 2002, 57, Bosch, J-C. and M. Hirschey, The Valuation Effects of Corporate Name Changes, Financial Management, 1989, Winter, Brock, W., J. Lakonishok and B. LeBaron, Simple Technical Trading Rules and the Stochastic Properties of Stock Returns, Journal of Finance, 1992, 47, Buttimer, R. J., D. C. Hyland and A. B. Sanders, The Long-run Performance of REIT IPOs, Working paper, Campbell, Robert D., C. Ghosh and C. F. Sirmans, 1998, The great REIT consolidation: Fact or fancy? Real Estate Finance, Summer, , The Information Content of Method of Payment in Mergers: Evidence from Real Estate Investment Trusts, Real Estate Economics, 2001, 29, Chen, S-J., C. Hsieh, T. W. Vines and S-N. Chiou, Macroeconomic Variables, Firm-Specific Variables and Returns to REITs, Journal of Real Estate Research, 1998, 16, Conrad, J. and G. Kaul, Long-term Market Overreaction or Biases in Computed Return?, Journal of Finance, 1993, 48, JRER Vol. 27 No

22 342 Sahin Fama, E. F., Market Efficiency, Long-term Returns, and Behavioral Finance, Journal of Financial Economics, 1998, 49, Fama, E. F. and K. R. French, The Cross-section of Expected Stock Returns, Journal of Finance, 1992, 46, , Common Risk Factors in the Returns of Stocks and Bonds, Journal of Financial Economics, 1993, 33, , Industry Cost of Equity, Journal of Financial Economics, 1997, 43, Ikenberry, D., J. Lakonishok and T. Vermaelen, Market Underreaction to Open Market Share Repurchases, Journal of Financial Economics, 1995, 39, Jaffe, J., Special Information and Insider Trading, Journal of Business, 1974, 47, Lee, I., Do Firms Knowingly Sell Overvalued Equity?, Journal of Finance, 1997, 52, Lyon, J., B. Barber and C-L Tsai, Improved Methods For Test For Long Run Abnormal Stock Returns, Journal of Finance, 1999, 54, Malatesta, P. H., The Wealth Effect of Merger Activity and the Objective Functions of Merging Firms, Journal of Financial Economics, 1983, 11, McIntosh, W., D. T. Officer and J. A. Born, The Wealth Effects of Merger Activities: Further Evidence from Real Estate Investment Trusts, Journal of Real Estate Research, 1989, 4, McIntosh, W., Y. Liang and D. L. Tompkins, An Examination of the Small-Firm Effect Within The REIT Industry, Journal of Real Estate Research, 1991, 6, Mitchell, M. L. and E. Stafford, Managerial Decisions and Long-Term Stock Price Performance, Journal of Business, 2000, 73, Moeller, S. B., F. P. Schlingemann and R. M. Stulz, 2003, Do Shareholders of Acquiring Firms Gain from Acquisitions?, Charles A. Dice Center working paper, Ohio State University, Ritter, J., The Long-run Performance of Initial Public Offerings, Journal of Finance, 1991, 46, The author wishes to thank two anonymous referees for their valuable comments. Olgun F. Sahin, Minnesota State University Moorhead, Moorhead, MN or sahin@mnstate.edu.