Corporate Exit Strategies: Evidence from Real Estate Investment Trusts

Transcription

1 Corporate Exit Strategies: Evidence from Real Estate Investment Trusts Lisa A. C. Frank, * Robert D. Campbell, ** and Zhilan Feng *** * Central Connecticut State University ** Hofstra University *** Union Graduate College Abstract Of the 200 public equity REITs that exist in 1996, more than half exit from the population by This study systematically examines the nature of these exits. The most likely candidates for exit are smaller, less levered REITs with lower profit margins and the traditional structure. Public-public merger is the most common exit vehicle. Specialized REITs are more likely to exit by public-public merger, UPREITs are more likely to exit by public-private merger, and larger REITs are less likely to exit by public-public merger or liquidation. Returns are positive for exit announcements and greatest when firms merge private or liquidate.

2 Corporate Exit Strategies: Evidence from Real Estate Investment Trusts 1. Introduction There are 200 equity REITs on the NAREIT member list of On the NAREIT list for 2006, 135 REITs appear, reflecting a decline in the total REIT population of 65 firms over the elevenyear period. Yet these numbers obscure the true degree of change in the overall REIT population. Of the 200 REITs appearing on the 1996 NAREIT list, only 73 reappear on the 2006 list, and thus the remaining 62 are new REITs. This means that the total number of firms that exit the REIT cohort during this interval is 127, over one-half of the total 1996 equity REIT population. These numbers imply that the process of exiting from the REIT population has important implications for REIT management and for investors. Yet this process has been largely neglected in the literature to date. The primary purpose of this paper is to systematically examine the nature of these exits. The study identifies the mechanisms for exit, finding that absorption in public-public merger is the most common exit strategy, but that REITs also often choose to merge private, to liquidate, or simply eliminate election of REIT status. Using standard event study methodology, the study measures shareholder returns around the announcement of the exit. Abnormal returns are significantly positive for all forms of exit. Consistent with previous literature, we find that abnormal returns in announcements of exit by merging with other public firms are positive, but are smaller than target returns in takeover announcements for conventional companies. To date, the literature has been silent with regard to wealth effects when public REITs merge private. The research finds that shareholder returns are highest in announcements of exit by merging private, at approximately 14%. Logistic regression is used to test conditional likelihoods for the population of REITs that choose to exit, examining the relationship between the likelihood that a particular exit strategy is chosen, and certain structural characteristics of the REIT. We further test the likelihood that a REIT exits at all, using logistic regression to compare exiting REITs with both the surviving REITs that were in existence in 1996, and a portfolio of firms matched by size in the year prior to the exit event. There is significant evidence that REITs investing in diversified property portfolios are far less likely to exit by public-public merger than are REITs with specialized property portfolios. Further, firms with larger market values are significantly less likely to exit by public-public merger or liquidation than smaller firms. The next section reviews the literature related to real estate investment trusts. The following section describes the data and presents descriptive statistics for exit firms, surviving firms and match firms. Sections 4 and 5 discuss abnormal returns to shareholders and conducts logistic 2

3 regression analyses of probability of exit. Section 6 presents the results of the regression analyses, and section 7 concludes. 2. Real Estate Investment Trusts United States REITs are corporations or trusts investing in real estate that meet requirements specified in Title 26 of the United States Internal Revenue Code which effectively provide for exemption from federal income taxes. 1 There are a number of requirements a firm must meet to qualify for the REIT pass-through structure, but the more stringent requirements are related to ownership structure, asset composition, and dividend payout. Firms must elect to take on REIT status by filing such election with its tax return. Once REIT status is elected, it must be revoked or terminated in order for the firm to forgo REIT status in subsequent years. Once a firm s REIT status is terminated, it may not elect REIT status for five years after the year in which the termination became effective. REITs may be classified according to their investment approach as mortgage, equity, or hybrid. Mortgage REITs are in the business of lending to real estate owners and operators. They may also invest in loans or mortgage-backed securities. Equity REITs are those that own incomeproducing real estate. Hybrid REITs combine the characteristics of mortgage and equity REITs by both owning properties and lending to owners and operators. The focus of this study is equity REITs, which comprise the majority of REITs in existence in 1996 and today. Equity REITs are further distinguished by the nature of their assets. Some equity REITs are specialized with greater than 75% of assets concentrated in one type of property investment, such as office buildings, apartment complexes, or retail shopping centers, while others maintain a diversified portfolio of real estate assets. Both specialized and diverse equity REITs are included in this study, and potential differences in exit strategy between these classifications are analyzed. A final distinguishing characteristic of REITs considered in this study is the REIT corporate structure. Traditionally REITs owned, and in most cases managed, investment properties directly. More recently, however, the Umbrella Partnership REIT (UPREIT) structure has gained popularity, and is the structure of choice for the majority of REITs in existence today. Transactions that entail the exchange of real property for REIT stock are taxable events. The UPREIT structure allows for deferment of capital gains taxes that the real estate owner would otherwise realize on the taxable event. This structure is created as the REIT forms a partnership to which the property owner contributes real estate assets in exchange for partnership units. Once the partnership units are exchanged for cash or REIT stock, a taxable event occurs. The real 1 For the complete definition of REITs see Section 856 of the Internal Revenue Code: Title 26, Subtitle A, Chapter 1, Subchapter M, Part II. 3

4 estate owner may therefore delay capital gains taxes until partnership units are converted to REIT shares or cash. In a study of REIT merger activity, Campbell, Ghosh and Sirmans (2001) find that the UPREIT status of the acquiring firm is positively associated with acquirer returns. Vogel (1997) discusses factors behind REIT growth and consolidation in the 1990 s and predicts a continuing trend of mergers and consolidations, but argues these consolidations are not designed to take advantage of economies of scale. Campbell, Ghosh and Sirmans (1998) challenge the conventional thought that the advantages of size typical for conventional companies are applicable for REITs. They argue that systematic structural problems such as negative returns for most acquirers and the absence of effective hostile takeovers impair merger activity in the REIT market. Panovka (2000) suggests that not only smaller, but also larger cap REITs consider the option of going private as an attractive alternative to the possibility of selling out to a large competitor. In particular, the author argues that this is a result of the gap between valuations in the private real estate markets and Wall Street valuation of REITs. Consistent with this view, Weir and Wright (2006) study 96 public-to-private acquisitions of non-reit firms and find evidence that some transactions may result from situations in which the market s perspective of the company s prospects differs from that of management. Empirical studies focusing on non-reit companies generally find positive excess returns for shareholders of acquired firms in public-public mergers. Huang and Walkling (1987) find a high excess return of 26%, Hayn (1989) finds excess returns of 17.5%, while Davidson and Cheng (1997) find somewhat smaller excess returns of 12.4%. Campbell, Ghosh and Sirmans (1998) illustrate that acquired equity REITs underperform compared to their non-reit counterparts. They record five-day cumulative abnormal returns of 5.2% to the target firms in these transactions. Research on REIT liquidations has been relatively thin with only a handful of relevant papers appearing in real estate and economics journals. Ghosh, Owers, and Rogers (1991) identify the organizational and financial characteristics of firms that choose to voluntarily liquidate, indicating that voluntary liquidation is a less frequent, but more extreme form of restructuring. They assert that management perceives liquidation as an optimal response to a choice between hostile takeover and costly bankruptcy. 3. Data This is a cohort study of the entire population of 200 publicly traded equity REITs listed as NAREIT members in the 1996 REIT Handbook. 2 At the end of 2006 there are only 135 REITs REIT Handbook, National Association of Real Estate Investment Trusts (NAREIT). 4

5 listed as NAREIT members, indicating a reduction in the total population by 65 REITs. 3 A direct comparison of the REITs in the 1996 population to those existing at the end of 2006 reveals that only 73 of the REITs are the same. This means that 127 REITs, or 63.5% of the 1996 cohort, exit the REIT population during the period of this study. 4 MVE ($ billions) Debt Ratio (%) Profit Margin (%) Property Sector Table I Descriptive Statistics as of Year-End 1995: 1996 NAREIT Member Public REITs Total Cohort (n=200) Survivor Firms (n=73) Exit Firms (n=127) Difference in means (p-value) Mean < St. Dev Max Min Mean St. Dev Max Min Mean St. Dev Max Min Diversified Office & Industrial Retail Health Care Apartment Other Traditional Structure UPREIT Total (n=200) is the entire cohort of 1996 NAREIT member public equity REITs. Survivors (n=73) are REITs in the 1996 cohort that remain NAREIT members as of December 31, Exits (n=127) are REITs in the 1996 cohort that are no longer NAREIT members as of December 31, Descriptive statistics are measured as of year-end MVE is market value of equity, determined by multiplying the company s total common shares outstanding by its market price as of year-end 1995, as obtained from the CRSP database. Debt Ratio is measured as total debt divided by total assets. Profit Margin is measured as net income divided by total revenue. Property Sector and Structure are collected from the REIT Handbook and the SNL database. Other is defined by NAREIT to include REITs that specialize in lodging, resorts, specialty, or self-storage. Traditional includes REITs that do not utilize the UPREIT or DownREIT structure. P-values for difference in means between the survivor firms and exit firms are reported in the last column. 3 REITWatch, January 2007, National Association of Real Estate Investment Trusts, 4 REITs that changed name but did not exit are excluded from the exit population. 5

6 Descriptive data for market value of equity (MVE), debt ratio (DR), profit margin (PM), property specialization, and corporate structure for the 1996 and 2006 equity REIT populations are presented in Table I. Data are obtained from the CRSP database, NAREIT, and SEC filings, and are measured as of year-end Market value of equity is measured in billions of dollars as the total number of common shares outstanding multiplied by the market price and is included as a proxy for firm size. Debt ratio is used as a proxy for financial stability and is measured as total debt divided by total assets. As a proxy for operational efficiency profit margin is measured as the ratio of net income to total revenue. T-tests for difference in means between the survivor and exit groups for each variable are conducted and p-values are reported in the last column. All statistical tests use a 95% confidence interval. In the full 200 firm cohort of 1996 public equity REITs, 45 specialize in retail properties, 43 specialize in apartment and residential, 44 specialize in other property types, 30 in office and industrial, 13 specialize in health care, and 25 are diversified. 5 In 1996, 129 REITs are traditionally structured and 71 formed the UPREIT structure. 68.2% of the traditionally structured REITs that were members of NAREIT in 1996 exit by the end of 2006, while 54.9% of firms with the UPREIT structure exit in the same timeframe. The Office and Industrial and Other property sectors exhibit the greatest loss of REITs in the 11-year period with 70% and 73% exiting, respectively. Comparing the exiting sample with the non-exiting sample, there are significant differences in the market value of equity and profit margin between the two groups; both measures are smaller for the exiting firms. Table II presents descriptive statistics for the 127 firms that exit from the equity REIT population during the period of study, classified by type of exit. The most frequent exit strategy is publicpublic merger (N=83), but public-private merger (N=23) and liquidation (N=16) are also popular exit strategies. Firms that simply de-reit remain ongoing concerns but elect to give up REIT status, or no longer meet the regulatory requirements to qualify for REIT status. This is the least popular of all forms of exit, with only 5 out of the 127 exiting firms choosing or being forced to de-reit. The first column of Table II presents the market value of equity, debt ratio, profit margin, and breakdown by property type and property sector as of year-end 1995 for the full sample of exiting REITs. The remaining columns present these statistics for REITs exiting via public-topublic merger, public-to-private merger, liquidation, and de-reit methods, respectively. T-tests are conducted for difference in means between the particular exit group and all other exits and p- values are reported in parentheses under the respective means. The size (MVE) of REITs exiting by public-private merger and liquidation are significantly different from the size of other exiting REITs, with mergers being larger and liquidations smaller. There seems to be no significant difference in debt ratio and profit margin between the exit groups. 5 The Other property sector defined by NAREIT includes lodging, resorts, specialty and self-storage. 6

7 MVE ($ billions) Debt Ratio (%) Profit Margin (%) Property Sector Structure Table II Descriptive Statistics as of Year-End 1995: Exiting REITs by Method of Exit Total Public-Public Public-Private Exits (n=127) Merger (n=83) Merger (n=23) Liquidation (n=16) Mean (P-value) (.504) (.099) 54.3 (.025) de-reit (n=5) 20.7 (.151) St. Dev Max Min Mean (P-value) (.301) 52.6 (.534) 59.4 (.141) 34.5 (.245) St. Dev Max Min Mean (P-value) (.208) 19.2 (.875) 8.7 (.140) 7.5 (.361) St. Dev Max Min Diversified Office & Industrial Retail Health Care Apartment Other Traditional UPREIT Exiting REITs are 1996 NAREIT member public equity REITS that are no longer members as of December 31, Descriptive statistics are measured as of year-end Firms that de-reit remain ongoing concerns but elect to give up REIT status or no longer meet the requirements to qualify for REIT status. MVE is market value of equity determined by multiplying the company s common shares outstanding by its market price as of year-end 1995, obtained from CRSP. Debt Ratio is measured as total debt divided by total assets. Profit Margin is measured as net income divided by total revenue. Other is defined by NAREIT to include REITs that specialize in lodging, resorts, specialty, or self-storage. Traditional includes REITs that do not utilize the UPREIT or DownREIT structure. P-values for difference in means between the particular exit group and all other exits are reported in parentheses below the respective means. Table II reveals that REITs of all different property sectors are represented in the exiting population. It appears from these descriptive data that diversified REITs are less inclined to be absorbed in public-public mergers than are REITs investing in specialized property types. Of the diversified REITs that exit, 35% exit via public-public merger, as compared to 70% of specialized REITs. This result is consistent with recent literature regarding the effects of focus 7

8 and diversification in real estate companies. Cronqvist, Hogfeldt, and Nilsson (2001), and Capozza and Seguin (1999) find a positive relationship between property specialization and real estate firm value. Campbell, Ghosh and Sirmans (2001) find evidence that diversifying REIT capital transactions have negative implications for value. The negative value implications of diversification add to the costs of acquiring diversified firms, and make it much more difficult for acquiring specialized firms to capture merger synergies, thus placing the diversified seller at a competitive disadvantage in the merger market. The specialized buyer would dispose of most of the acquired properties in any event in order to maximize value, thus incurring substantial transaction and holding period costs that erode value. This being the case, the diversified seller will be more inclined to sell properties individually or in portfolios, to a set of specialized buyers. Table III presents descriptive statistics in the same manner as Table II, but the data are collected in the year immediately prior to the exit event. Again, liquidating firms and firms that exit via public-private merger are significantly different in size (MVE) than firms that exit by other means. Liquidating firms are significantly smaller than firms exiting by other means. REITs that merge private are significantly larger than other exiting REITs in both Table II and Table III. Privately merging firms are characterized by a significantly higher debt ratio in the year prior to exit, and firms merging public have the highest profit margin. It is also apparent from this data that the profit margins are significantly different for all types of exit. Table IV compares the characteristics of the exit firms with those of match firms. Match firms are selected by first identifying the REITs that did not exit during the entire study period. The non-exiting REITs are sorted by market value of equity and the REIT with the market value of equity closest to the event firm in the year prior to the exit is selected as the match firm. 6 All descriptive statistics are measured for the exit firm and its match firm in the year prior to exit. The evidence indicates the exiting REITs have a significantly lower profit margin and debt ratio than the match firms. That the market value of equity is insignificantly different in these two groups is expected as the match firms were selected based on their resemblance to the size of the exit firm. 6 Due to the limited number of REITs we are unable to find a match for 14 of the exiting REITs. 8

9 Table III Descriptive Statistics as of the Year-End Prior to Exit: Exiting REITs by Method of Exit Total Exits (n=127) Public-Public Merger (n=83) Public-Private Merger (n=23) Liquidation (n=16) de-reit (n=5) MVE ($ billions) Debt Ratio (%) Mean (P-Value) (.843) (.054) (.099) (.264) St. Dev Max Min Mean (P-Value) (.004) (.002) (.893) (.399) St. Dev Max Min Profit Margin (%) Mean (P-Value) (.0001) (.0651) (.0356) 0.8 (.0027) St. Dev Max Min Exiting REITs are 1996 NAREIT member public REITs that are no longer NAREIT members as of December 31, Descriptive statistics are measured as of the year-end prior to the exit event. MVE is market value of equity, determined by multiplying the company s total common shares outstanding by its market price as of year-end prior to the exit event, as obtained from the CRSP database. Debt Ratio is measured as total debt divided by total assets. Profit Margin is measured as net income divided by total revenue. Firms that de-reit remain ongoing concerns but elect to give up REIT status or no longer meet the requirements to qualify for REIT status. P-values for difference in means between the particular exit group and all other exits are reported in parentheses below the respective means. 9

10 MVE ($ billions) Debt Ratio (%) Profit Margin (%) Table IV Descriptive Statistics as of the Year-End Prior to Exit: Exiting REITs and Matched REITs. Total Exit Firms Year Prior to Exit Match Firms Difference in Means (p-value) Mean St. Dev Max Min Mean St. Dev Max Min Mean St. Dev Max Min Exit Firms (n=113) are 1996 NAREIT member public REITs that are no longer NAREIT members as of December 31, Match Firms are firms that did not exit during the sample period and are closest in MVE to the exiting firm as of the year-end prior to the event date. Descriptive statistics are measured as of the year-end prior to the exit event. MVE is market value of equity, determined by multiplying the company s total common shares outstanding by its market price as of year-end prior to the exit event, as obtained from the CRSP database. Debt ratio is measured as total debt divided by total assets. Profit margin is measured as net income divided by total revenue. Difference in means reports the p-value for the hypothesis that the difference in means between the exiting firms and match firms is zero. 4. Abnormal Shareholder Returns Table V reports abnormal shareholder returns for the one-day, two-day, and three-day windows around the announcement of exit. Abnormal returns to shareholders over a three-day event period window (-1,+1) are computed for the total portfolio of eliminated firms and for each exit strategy using standard event study methodology following Mikkelson and Partch (1988). 7 The S&P 500 index is used as a proxy for the market portfolio in abnormal return calculations. Market model parameters are estimated using daily stock returns with a 255 trading-day 7 The estimation procedure is conducted using Eventus software. We employ a single-factor model consistent with the REIT merger literature (for example, see Campbell, Ghosh, and Sirmans, 2005). 10

11 estimation window ending 46 days prior to the event date. The event date is measured as the day the exit announcement appeared in the Wall Street Journal, PR Newswire, or Dow Jones Newswire, if announced prior to 3:30 PM, and the next trading day for announcements appearing after 3:30 PM. 8 Table V Abnormal Shareholder Returns for 1996 NAREIT Member Public REITs Exiting from REIT Status During , by Method of Exit. Elimination Method AR (0) t-stat % Pos AR (0,1) t-stat % Pos AR (-1,1) t-stat % Pos Public-Public Merger Public-Private Merger Liquidation de-reit Total Eliminated Abnormal Returns around the announcement of exit from the public REIT population for equity REITs exiting by December 31, ARs are percent abnormal returns to REIT shareholders around the date of the first public announcement of merger, liquidation or elimination of REIT status (de-reit). Announcements must appear in the Wall Street Journal, the Dow Jones Newswire, or the PR Newswire prior to 3:30 PM on a trading day. For announcements made after 3:30 PM, the first trading day following the announcement is used. Standard event study methodology is employed. The market proxy is the CRSP Value- Weighted Index of the S&P 500. Statistical significance is based on a two-tailed t-test. % Pos is the percentage of observations with positive abnormal returns. Significantly positive shareholder returns are observed for public-public merger, public-private merger, and liquidation exit strategy announcements, with the average return at approximately 7% and significant for all announcements. The finding of positive shareholder returns in the 5% range for targets in public-public mergers is consistent with previous literature. McIntosh, Officer and Born (1989) and Campbell, Ghosh and Sirmans (2001) find positive returns for REIT targets in the 5% range and 3% range, respectively. These returns are much smaller than usual 10% to 15% target returns reported for conventional firms (Jensen and Ruback (1983); Huang and Walkling (1987); Hayn (1989); Datta, Pinches and Narayanan (1992); Davidson and Cheng (1997)). 8 Announcements are collected from Factiva or LexisNexis. 11

12 The literature is silent on the question of shareholder wealth effects when REITs merge private. This study finds that shareholder returns are highest, in the 14% range, when this exit strategy is used. This finding is evidence that the limitations of the REIT structure can be negative for value, at least in some situations. Campbell and Sirmans (2002) note that the creation of taxadvantaged real estate firms is not unambiguously positive for value. On the one hand, exemption from taxes tends to add value, at least for those firms that qualify for it. On the other hand, the tax exemption entails constraints on managerial choices designed to limit unfair competition with taxable firms, reducing managerial flexibility to respond to changes in the business opportunity set. By merging with private non-reits, REIT owners lose the advantage of access to public capital markets, but gain the advantage of greater operational and financial flexibility. Abnormal returns for REITs announcing liquidation are significantly positive at approximately 7% on the event date, and in the 11% range over the two-day and three-day windows. This result is consistent with Hite, Owers, and Rogers (1987) who report average abnormal returns of 12.24% upon announcement of proposals to liquidate. Although in the same direction, the magnitude of our result is different from that reported by Skantz and Marchesini (1987), who record strong positive returns of 21% in the month of liquidation announcement for conventional firms. 5. Logistic Analysis Using logistic regression analysis, this study tests the relationship between REIT structural characteristics and the potential for exit. It also tests the relationship between structural characteristics and the conditional probability that, given the REIT exits at all, it will exit by a particular strategy. Results for the logistic regressions are reported in Table VI. The population tested in Panel A is the 1996 cohort of 200 firms. The dependent variable is a binary variable equal to one if a REIT existing in 1996 exits from the NAREIT population by the end of 2006, and zero otherwise. The cause variables are measured as of December 31, 1995 and include market value of equity (MVE), debt ratio (DR), and profit margin (PM). Two indicator variables are included to control for the structure of the REIT. RT is a variable that takes the value of one if the REIT maintains a traditional structure traditionally and a value of zero if the firm is structured as an UPREIT. DIV takes the value of one if the REIT maintains a diversified portfolio of assets and a value of zero if the portfolio is specialized. The coefficients are reported in log odds units. 12

13 Table VI Logistic Models of Probability of Exit During Panel A: Exiting REITs and Surviving REITs as of Year-End 1995 Variable Log Odds Chi-Sq Constant 1.82*** 7.65 MVE *** 9.85 DR PM ** 6.22 RT 0.587* 2.74 DIV Likelihood ratio Prob > chi Panel B: Exiting REITs and Matched REITs as of the Year-End Prior to Exit Variable Log Odds Chi-Sq Constant 4.001*** MVE ** 4.2 DR ** 6.33 PM ** 5.56 RT DIV Likelihood ratio Prob > chi Likelihood ratio test, nested model H 0 : MVE coefficient=0 Prob > chi The dependent variable is a binary variable equal to one if a REIT existing in 1996 exits from the REIT population by December 31, 2006, and zero otherwise. Independent variables are measured as of the year-end 1995 (Panel A) or prior to the exit event (Panel B). Exiting firms are those that are no longer NAREIT members as of December 31, Match firms are firms that did not exit during the sample period and are closest in MVE to the exiting firm as of the year-end prior to the event date. MVE is market value of equity, determined by multiplying the company s total common shares outstanding by its market price as of yearend 1995 (Panel A) or prior to the exit event (Panel B), as obtained from the CRSP database. Debt Ratio (DR) is measured as total debt divided by total assets, expressed as a percentage. Profit Margin (PM) is measured as net income divided by total revenue, expressed as a percentage. RT is an indicator variable equal to one of the REIT has a traditional structure and zero otherwise. DIV is an indicator variable equal to one if the REIT manages a diverse portfolio of properties and zero otherwise. * Indicates statistical significance at 10% level of certainty ** Indicates statistical significance at 5% level of certainty *** Indicates statistical significance at <1% level of certainty REITs that are smaller, with a lower profit margin and traditional structure are significantly more likely to be in the exit population. This result is consistent with Vogel s (1997) prediction that the UPREIT structure will be a significant influence in REIT growth and consolidation. The likelihood ratio indicates the overall model is significant. 13

14 Panel B of Table VI presents the results of a logistic regression in which the exiting firms are compared to the match firms in testing the cause variables for exit. The dependent variable is equal to one if the firm is in the exiting population and zero if the firm is a non-exiting, match firm. Smaller firms with a lower debt ratio and profit margin are more likely to exit the REIT population. These results maintain the same sign as those reported in the comparison of exiting firms with their non-exiting 1996 cohorts, yet the debt ratio gains statistical significance. In terms of probabilities, an increase of 1% in debt ratio decreases the expected probability of exit by approximately 50%, holding all other variables constant. Again, the overall model is statistically significant as measured by the likelihood ratio. The likelihood ratio test chi-squared statistic for the null hypothesis that the coefficient for MVE is zero is reported at the bottom of Table VI, Panel B. This statistic is significant, indicating the full model is superior to the nested model in explaining potential for exit. 9. Panel A of Table VII reports the results of a test of the full cohort of 200 firms for probability of exit by a particular strategy. Variables are defined in the same manner as in Table VI. The likelihood ratio tests indicate the overall models for public-public merger and liquidations are significant. There is significant evidence that diversified REITs are less likely to merge public than are specialized REITs, and larger REITs are less likely to exit by liquidation or publicpublic merger than public-private merger. These results are consistent with the conclusions drawn from the data presented in Table II, and with recent literature that finds a positive relationship between property specialization and real estate firm value (Cronqvist, Hogfeldt, and Nilsson (2001); Capozza and Seguin (1999); Campbell, Ghosh and Sirmans (2001)). The same test is conducted with the exiting firms and match firms and results are reported in Panel B of Table VII. The results for this test are substantially the same for exits by public-public merger, except that we gain significance on the debt ratio variable. In terms of probabilities, a 1% increase in debt ratio decreases the expected probability of exit via public-public merger by approximately 45%. 9 Match firms were selected by market value of equity, therefore we analyze a nested model in which we eliminate market value of equity from the regression. Results are essentially the same as for the full model and are available upon request. 14

15 Table VII Logistic Models of Probability of Exit by a Particular Strategy During Panel A: Exiting REITs and Surviving REITs as of Year-End 1995 Public-Public Merger Public-Private Merger Liquidation Variable Log Odds Chi-Sq Log Odds Chi-Sq Log Odds Chi-Sq Constant ** ** 4.33 MVE ** ** 3.88 DR PM RT DIV -1.53*** Likelihood ratio Prob > chi Panel B: Exiting REITs and Matched REITs as of the Year-End Prior to Exit Public-Public Merger Public-Private Merger Liquidation Variable Log Odds Chi-Sq Log Odds Chi-Sq Log Odds Chi-Sq Constant ** MVE * DR ** * PM RT DIV * Likelihood ratio Prob > chi Nested Model: H 0 : MVE coefficient= Prob > chi The dependent variable is a binary variable equal to one if a REIT existing in 1996 exits from the REIT population, and zero otherwise. Independent variables are measured as of the year-end 1995 (Panel A), or prior to the exit event (Panel B). Exiting firms are those that are no longer NAREIT members as of December 31, Match firms are firms that did not exit during the eleven-year sample period and are closest in MVE to the exiting firm as of the year-end prior to the event date. MVE is market value of equity, determined by multiplying the company s total common shares outstanding by its market price as of year-end 1995 (Panel A) or prior to the exit event (Panel B), as obtained from the CRSP database. Debt Ratio (DR) is measured as total debt divided by total assets, expressed as a percentage. Profit Margin (PM) is measured as net income divided by total revenue, expressed as a percentage. RT is an indicator variable equal to one of the REIT has a traditional structure and zero otherwise. DIV is an indicator variable equal to one if the REIT manages a diverse portfolio of properties and zero otherwise. * Indicates statistical significance at 10% level of certainty ** Indicates statistical significance at 5% level of certainty *** Indicates statistical significance at <1% level of certainty 15

16 Table VIII Logistic Models of Probability of Exit by a Particular Strategy Given That Exit Occurs During Panel A: Exiting REITs by Type of Exit as of Year-End Public-Public Merger Public-Private Merger Liquidation Variable Log Odds Chi-Sq Log Odds Chi-Sq Log Odds Chi-Sq Constant MVE DR PM RT DIV Likelihood ratio Prob > chi Panel B: Exiting REITs by Type of Exit as of the Year-End Prior to the Exit Event. Public-Public Merger Public-Private Merger Liquidation Variable Log Odds Chi-Sq Log Odds Chi-Sq Log Odds Chi-Sq Constant ** MVE DR ** ** PM * RT DIV * Likelihood ratio Prob > chi The dependent variable is a binary variable equal to one if a REIT existing in 1996 exits from the REIT population by a particular exit strategy, and zero otherwise. Independent variables are measured as of the year-end 1995 (Panel A), or prior to the exit event (Panel B). Exiting firms are those that are no longer NAREIT members as of December 31, MVE is market value of equity, determined by multiplying the company s total common shares outstanding by its market price as of year-end 1995 (Panel A) or prior to the exit event (Panel B), as obtained from the CRSP database. Debt Ratio (DR) is measured as total debt divided by total assets, expressed as a percentage. Profit Margin (PM) is measured as net income divided by total revenue, expressed as a percentage. RT is an indicator variable equal to one of the REIT has a traditional structure and zero otherwise. DIV is an indicator variable equal to one if the REIT manages a diverse portfolio of properties and zero otherwise. * Indicates statistical significance at 10% level of certainty ** Indicates statistical significance at 5% level of certainty *** Indicates statistical significance at <1% level of certainty Table VIII Panel A reports results for the conditional log odds that a particular exit strategy will be chosen, given that exit occurs at all. The population in this model consists of the 127 firms existing in 1996 that exit during the subsequent period of study. The firm characteristics are measured as of year-end Again we find significant evidence that REITs with diversified property portfolios are less likely to exit by public-public merger than to exit by other methods, and smaller REITs are more likely to liquidate. The negative result for the diversified nature of 16

17 REITs property holdings is consistent with existing financial theory as previously noted. A consistent body of literature argues that diversification is negative for the value of real estate firms (Capozza and Sequin (1999); Cronqvist, Hogfeldt, and Nilsson (2001)). The acquisition of diversified REITs entails either decreases in firm value resulting from greater property diversification, or increased transactions costs required to dispose of properties. These circumstances make diversified firms less attractive as takeover targets. Further, findings show evidence that REITs using the UPREIT organizational structure are more likely to choose the public-private merger form of exit than are traditionally-structured REITs. The reason for this finding may relate to taxation and to the relationship between the UPREIT structure and the costs of restructuring. UPREITs usually hold all of their real estate in the form of Operating Partnerships or Limited Liability Corporations (LLCs), entities that are similar to the structural forms of the private firms with which they are merging. As a result the IRS will treat the merger as a tax-exempt combination, moving existing property cost bases forward to the combined entity, at least to the extent that the financial assets acquired take the form of OP or LLC subsidiary shares. Thus, many of the special tax advantages available to UPREITs when they acquire properties from private sellers also apply in reverse when the firm is sold to private buyers. Because, as we have shown here, merging private is positive for value, this is another argument in favor of the advantages of the UPREIT structure. Panel B of Table VIII shows the same analysis with the characteristics measured as of the year prior to exit. The results for diversified REITs are the same as in Panel A indicating a lower likelihood of exiting via public-public merger. The results for liquidating firms are no longer significant, however there is significant evidence that REITs with greater debt ratios are less likely to merge public and more likely to merge private, while REITs with greater profit margins are more likely to exit via public-public merger. 6. Regression Analysis A regression analysis is conducted to measure the relationship between abnormal returns to shareholders over the three-day window around the exit announcement and market value of equity, debt ratio, profit margin, REIT structure, and asset diversification. A second model tests these relationships after controlling for the form of exit. The results are reported in Table IX, Panel A for firm characteristics measured as of 1995, and in Panel B for firm characteristics measured as of the year prior to exit. The results are similar in both panels. The profit margin variable is negatively related to abnormal returns, even after controlling for type of exit. Furthermore the traditional structure (RT) indicator variable has a significantly negative impact on abnormal returns. This result implies the UPREIT structure garners a premium over the traditional structure upon announcement of an exit strategy. However, this effect is eliminated once the analysis controls 17

18 for exit type. The public-private merger and liquidation forms of exit are significantly positively related to abnormal returns around the event date, as compared to the abnormal returns for public-public merger announcements. Table IX Regression Analysis: Abnormal Returns around Exit Announcements Panel A: REIT Characteristics as of Year End 1995 Variable Coefficient P> t Coefficient P> t Constant 10.78*** *** MVE DR PM ** ** RT -3.17* DIV PRI 10.54*** LIQ 5.73** Adj. R-squared Variable Coef P> t Coef P> t Constant 10.18*** *** MVE DR PM ** ** RT ** DIV PRI 10.37*** LIQ 5.47** Adj. R-squared The dependent variable is the abnormal return over the 3-day event window around exit announcement. Independent variables are measured as of the year-end 1995 (Panel A), or prior to the exit event (Panel B). Market value of equity (MVE) is measured as total common shares outstanding multiplied by market price as of year-end 1995 (Panel A) or year-end prior to exit (Panel B). Debt ratio (DR) is measured as total debt divided by total assets, expressed as a percentage. Profit margin (PM) is measured as net income divided by total revenue, expressed as a percentage. RT is an indicator variable equal to one of the REIT has a traditional structure and zero otherwise. DIV is an indicator variable equal to one if the REIT manages a diverse portfolio of properties and zero otherwise. PRI is an indicator variable equal to one if the REIT exits via public-private merger and zero otherwise. LIQ is an indicator variable equal to one if the REIT exits via liquidation and zero otherwise. * Indicates statistical significance at 10% level of certainty ** Indicates statistical significance at 5% level of certainty *** Indicates statistical significance at <1% level of certainty 7. Conclusion This study examined the changes taking place in the cohort of 200 public NAREIT member equity REITs listed in It finds that 127 of these firms, more than half of the cohort, exit from the public equity REIT population by the end of Examining the method of exit shows that public-public merger is the most popular exit strategy, but that public-private merger and liquidation are also frequently used. Standard event study methodology is used to measure abnormal returns to shareholders around the exit announcement. Consistent with existing literature, results show that shareholder returns around the announcement of exit by public- 18

19 public merger are positive, but smaller than those observed for takeover targets in conventional firms. The study adds to the literature by finding that abnormal returns from announcements of merging private are very high at about 14%, and that returns from announcements of liquidation are positive in the 11% range. The result related to merging private is evidence that the restrictions on operational flexibility that is part of the REIT institutional environment can outweigh the advantages of access to public capital, at least in some cases. Using logistic regression the study tests the relationship between the choice of particular exit strategies and certain structural characteristics of the REIT. REITs with diversified property portfolios, as compared to specialized REITs, are less likely to be absorbed in a public-public merger, traditional REITs are less likely to be absorbed in public-private merger, and larger REITs are less likely to liquidate. The result for diversified REITs is confirmed by firm characteristics measured as of the year prior to the exit event. REITs with higher debt ratios are less likely to exit the REIT population via public-public merger and more likely to be absorbed in a public-private merger. REITs with higher profit margins are more likely to exit the REIT population via public-public merger than by other types of exit. 19

20 References Campbell, Robert D., Chinmoy Ghosh, and C.F. Sirmans, 2001, The information content of method of payment in mergers: Evidence from Real Estate Investment Trusts (REITs), Real Estate Economics 29, Campbell, Robert D., Chinmoy Ghosh, and C.F. Sirmans, 1998, The great REIT consolidation: Fact or fancy? Real Estate Finance 15, Campbell, Robert D., and C.F. Sirmans, 2002, Policy implications of structural options in the development of Real Estate Investment Trusts in Europe, Journal of Property Investment and Finance 20, Capozza, Dennis R., and Paul J. Seguin, 1999, Focus, transparency and value: The REIT evidence, Real Estate Economics 27, Cronqvist, Henrik, Peter Hogfeldt and Mattias Nilsson, 2001, Why agency costs explain diversification discount, Real Estate Economics 29, Datta, Deepak K., George E. Pinches, and V. K. Narayanan, 1992, Factors influencing wealth creation from mergers and acquisitions: A meta-analysis, Strategic Management Journal 13, Davidson III, Wallace N., and Louis T. W. Cheng, 1997, Target firm returns: Does the form of payment affect abnormal returns?, Journal of Business Finance & Accounting 24, Ghosh, Chinmoy, James E. Owers, and Ronald C. Rogers, 1991, The financial characteristics associated with voluntary liquidations, Journal of Business Finance and Accounting 18, Hayn, Carla, 1989, Tax attributes as determinants of shareholder gains in corporate acquisitions, Journal of Financial Economics 23, Hite, Gailen L., James E. Owers, and Ronald C. Rogers, 1987, The market for interfirm asset sales: Partial sell-offs and total liquidations. Journal of Financial Economics 18, Huang, Yen-Sheng, and Ralph A. Walkling, 1987, Target abnormal returns associated with acquisition announcements: Payment, acquisition form, and managerial resistance. Journal of Financial Economics 19, Jensen, Michael C., and Richard S. Ruback, 1983, The market for corporate control: The scientific evidence, Journal of Financial Economics 11, McIntosh, Willard, Dennis T. Officer, and Jeffrey A. Born, 1989, The wealth effects of merger activities: Further evidence from Real Estate Investment Trusts, Journal of Real Estate Research 4,

21 Mikkelson, Wayne H., and M. Megan Partch, 1988, Withdrawn security offerings, Journal of Financial and Quantitative Analysis 23, Panovka, Robin, 2000, Focus on REITs, Real Estate Issues 25, 58. Skantz, Terrence R., and Roberto Marchesini, 1987, The effect of voluntary corporate liquidation on shareholders wealth, Journal of Financial Research 10, Vogel, John, 1997, Why the conventional wisdom about REITs is wrong, Real Estate Finance 14, Weir, Charlie, and Mike Wright, 2006, Governance and takeovers: Are public-to-private transactions different from traditional acquisitions of listed corporations?, Accounting and Business Research 36,