Investment opportunities and share repurchases

Journal of Corporate Finance 23 (2013) 23–38

Contents lists available at ScienceDirect

Journal of Corporate Finance journal homepage: www.elsevier.com/locate/jcorpfin

Investment opportunities and share repurchases☆ Walter I. Boudry a,⁎, Jarl G. Kallberg b, Crocker H. Liu a a b

Cornell University, United States Washington State University, Arizona State University, United States

a r t i c l e

i n f o

Article history: Received 10 August 2012 Received in revised form 30 May 2013 Accepted 25 July 2013 Available online 3 August 2013 JEL classification: G35 Keywords: REITs Share repurchases Investment opportunities

a b s t r a c t We examine the over-investment motivation for share repurchases using a sample of 139 Real Estate Investment Trusts (REITs) between 1996 and 2010. By combining a REIT's property portfolio data with project ROAs from the underlying real estate market, we are able to create a unique measure of the firm's investment opportunity set. Controlling for other possible buyback rationales, we find that poor investment opportunities are related to higher levels of share repurchases. Conditioning on investment opportunities, we find that the level of cash is positively related to repurchases only for low investment opportunity set firms. We also find a negative relationship between share repurchase announcement returns and investment opportunities. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Share repurchase has been an active area of financial research for the last four decades. It has also been increasingly significant in global financial markets. For example, in 1999, for the first time in history, the dollar volume of share repurchase exceeded the total amount of dividends paid by U.S. firms.1 In addition, regulators in foreign markets have been relaxing restrictions on buybacks, leading to the rapid growth of repurchase outside the U.S.2 The majority of academic investigations into stock repurchase (these are briefly covered in the literature review in Section 2) examine the short- and long-term impact of the repurchase. These papers typically test theoretical buyback motives by employing an event study and/or a measurement of after-repurchase performance (either operating performance or abnormal stock returns). A number of more recent studies have used Logit and Tobit analyses to assess the determinants of the repurchase decision. One of the most cited motives for repurchase is the lack of attractive investment opportunities relative to the firm's cash position; this is usually referred to as the over-investment hypothesis. While the academic studies have shed considerable light on the empirical validity of potential motives for repurchase, they have presented only indirect evidence addressing this important interaction between the firm's repurchase decision and the return on its investment opportunities.3 The most recent studies have generally supported the over-investment hypothesis, but the evidence is not unequivocal. This is due to the obvious ☆ We would like to thank Craig Holden, Thies Lindenthal (our AREUEA discussant), Jeffry Netter, Qing Ma, Tobias Muhlhofer, Harold Mulherin, Milena Petrova, Cristian Ioan Tiu, Charles Trzcinka, Gregory Udell, David Weinbaum, Scott Weisbenner, and an anonymous referee for helpful comments and seminar participants at AREUEA 2011 (Denver), Cornell University, Michigan State University, the University of Illinois at Urbana Champaign, the University of Indiana, SUNY-Buffalo, and Syracuse University. We would also like to thank Sabri Sisman and Jim Stevens at SNL for help with the option data and the Center for Real Estate and Finance at Cornell University and the American Council of Life Insurers for providing data for our study. All errors are ours. ⁎ Corresponding author at: 465A Statler Hall, Ithaca, NY 14853, United States. Tel.: +1 607 255 9003. E-mail address: [email protected] (W.I. Boudry). 1 See Julio and Ikenberry (2004). 2 See Kim et al. (2005) for a description of the changing foreign restrictions. 3 In a rather different context, Dellavigna and Pollet (2013) show that equity issuance is related to long-term, future demand shifts due to demographic changes. 0929-1199/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jcorpfin.2013.07.006

24

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

difficulty in determining the firm's return on its possible projects. While these returns are available to firm insiders, they are not easily observable by researchers. Our study more directly attacks this essential aspect of the repurchase question by estimating the firm's expected project returns in order to determine the extent to which the firm's opportunity set influences the buyback decision. We achieve this objective by focusing on Real Estate Investment Trusts (REITs). While this is a significant restriction and potentially limits the generality of our results, it has a very powerful advantage: it allows us to use available data on capitalization rates (project ROAs) from observed market transactions on different property types and locations to proxy for the REIT's project opportunity set. REITs were established in 1960 to allow retail investors to invest in commercial and other forms of real estate. Because they are restricted to passive real estate investment and since they are required to pass on 90% of taxable income as dividends,4 REITs are essentially tax-free investment conduits.5 The reason for using REITs is that they are among the most transparent investment vehicles, and because of the severe restrictions on their allowable investments, we can use a diverse database of market transactions to determine cap rates. We then use the cap rates to estimate the REIT's feasible project returns at any point in time. It is reasonable to expect that cap rates are an excellent proxy for the REIT's anticipated IRRs. For example, an examination of our cap rates and expected IRRs obtained from surveys conducted by PWC Korpacz shows that these two measures have a correlation over 92%.6 So although we do not have actual IRRs, the measure we do have is very highly correlated with expected returns.7 Another advantage of using REITs in our study is that there are relatively few hostile REIT mergers.8 This virtually eliminates the use of repurchases as an anti-takeover device.9 Finally, restricting our analysis to one industry mitigates the confounding influence of varying levels of risk. While a more precise definition is given in Section 3, informally one can think of the capitalization rate as the property's Return on Assets. We obtain cap rates on 109,430 transactions with a total transaction value of $1.135 trillion over the period 1995 to 2010 from the CoStar and Real Capital Analytics databases. Our empirical analysis uses these transactions to create a time series of project returns for each of the 139 REITs in our sample. We then examine the relation between repurchase activity and project returns. The over-investment hypothesis asserts that REITs will be more likely to repurchase shares when expected project returns are relatively low, since this action would maximize shareholder value. Overall our data are strongly consistent with this hypothesis. After controlling for the impact of over-valuation, firm size, leverage and executive option characteristics, a one standard deviation decrease in capitalization rates makes the REITs in our sample 14.5% more likely to repurchase shares and leads to an increase in the level of repurchases of approximately 47% of the average level of repurchases observed in the sample. Conditioning on investment opportunities, we find that the level of cash is positively related to repurchase activity only for low investment opportunity firms. Since REITs tend to repurchase in smaller amounts and because they have essentially no internal capital available for financing projects, one would expect that these results would be even stronger for typical corporations, which have far less dependence on external capital markets. As a robustness check, we also examine announcement effects for firms announcing repurchase programs. Following the repurchase motivation methodology of Peyer and Vermaelen (2009), we find that firms with a stated repurchase motivation of facing poor investment opportunities have a 5.04% 3-day cumulative abnormal return around the repurchase announcement. Consistent with this, we also find that controlling for the size of the repurchase program, firms announcing repurchase programs in poor investment opportunity environments have 1.7% higher 3-day and 2.35% higher 5-day CARs than firms announcing in good investment opportunity environments. The organization of the remainder of this paper is as follows: Section 2 provides an overview of the academic literature. Section 3 describes the data and empirical methodology. Section 4 presents the empirical results and Section 5 concludes. 2. Literature review This section provides a brief review of the extensive academic literature on share repurchase, emphasizing those papers that are most relevant to the current study. A comprehensive treatment is presented in Allen and Michaely (2002). 2.1. Survey results Wansley et al. (1989) present survey data on the motives for share repurchase.10 They list six possible motives: (i) Repurchase can act as a substitute for dividend payments: firms may view both dividends and repurchase as alternative mechanisms for distributing cash to shareholders; the firm could prefer one mechanism over another because of tax, signaling implications etc.11 (ii) As a method to adjust leverage or control: repurchase will increase the firm's leverage and potentially give insiders more 4 This was 95% prior to 2001. See Boudry (2011) for an analysis of REIT dividend payout policy. This paper essentially shows that REIT managers set dividends so as to minimize the tax burden of their investors. See also Hardin and Hill (2008). 5 There is thus some difficulty in translating the typical motives for dividends versus repurchase as discussed in several studies of regular corporations; for example Grullon and Michaely (2002). 6 For Industrial properties the correlation is 86%, for Multifamily it is 97%, for Office it is 92% and for Retail it is 98%. 7 We do not use the Korpacz IRRs in this study due to the limited geographic dispersion available in these data. 8 There have been very few exceptions; see however Panovka (2007). 9 See Bagwell (1991). 10 See also Baker et al. (1981). 11 Guay and Harford (2000) also analyze the role that permanent versus transitory cash flow shocks play in this decision.

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

25

control of the firm. (iii) To provide shares for reissue (e.g., for option exercise). (iv) Because of a lack of investment opportunities or an excess of available cash. (v) To signal favorable information about the firm's prospects or to signal that insiders view the shares as undervalued (thus this is termed the under-valuation hypothesis).12 (vi) To transfer wealth to selling shareholders.13 The strongest agreement from survey respondents was for (v), (ii) and (iv). Point (iv) is the focus of this study. The idea is that as a firm matures or as market conditions evolve, its investment opportunities can diminish and/or its cash flows increase. Some firms, in order to maximize shareholder value, would then elect to buy back shares rather than invest in negative net present value projects.14 Brav et al. (2005) surveyed 348 financial executives with regard to their perspectives on dividends and share repurchase. Among their key findings were the following: repurchase decisions are made after investment decisions are undertaken; firms are concerned with the impact of repurchase on EPS15; firms opt to repurchase when they hold excess cash or have a lack of good investment opportunities. Again, the last of these three points is most relevant to this paper. A motivation not directly addressed in these early survey papers is the agency motive. A related area of particular academic interest is the interaction between repurchase and executive option compensation; these issues will be discussed further below.

2.2. Repurchases: Theoretical background This section reviews some of the most relevant theoretical papers in order to frame our econometric analysis. Since repurchase and dividends can be viewed as substitutes,16 many of the theoretical arguments that address dividend policy can be applied to share repurchase as well. These include the early signaling models of Bhattacharya (1979), Vermaelen (1984), Miller and Rock (1985) and Ofer and Thakor (1987). Broadly stated, these asymmetric information models imply that dividends and share repurchase are credible signals of the firm's future prospects. Thus, the signaling hypothesis will imply that share repurchase is a signal of future over performance. A closely related theory, which has considerable support from the above survey results, is the under-valuation hypothesis, another version of the asymmetric information story. Managers have superior knowledge about their firm's true value and repurchase their shares when the market undervalues the firm. The parallel between dividend theories and repurchase also applies to the diverse agency cost arguments. From the earliest, Jensen (1986), these papers have focused on dividends and repurchase as a solution to the over-investment problem. These payouts reduce divertible cash flow and mitigate managements' incentives to invest in negative net present value projects. As noted above, we refer to this as the over-investment hypothesis.

2.3. Repurchases: Empirical results The original empirical work focused on the market reaction to the announcement of open-market and tender offer repurchases. The earliest papers, Dann (1981), Vermaelen (1981) and Bartov (1991) supported the signaling story. They showed that the announcement effects were strongly positive, and that long-term returns were also positive, partially because of EPS growth. This point was refined in Nohel and Tarhan (1998), which showed that the EPS gain came from high book-to-market firms, consistent with the over-investment hypothesis.17 Dittmar (2000) and others have provided significant support for the under-valuation hypothesis. The more recent literature favors the over-investment hypothesis over the signaling story. Ikenberry et al. (1995), Guay and Harford (2000) and Jagannathan and Stephens (2003) show a decline in earnings after the repurchase. However, Lie (2005) correcting for pre-repurchase performance, attributes much of this decrease to mean reversion. These results are generally robust when applied to non-U.S. data. See, e.g., Ikenberry et al. (2000) and Rau and Vermaelen (2002).18 A further aspect of the agency cost problem is the debt overhang problem as articulated in Myers (1977): simply put, high degrees of leverage can induce under-investment. This theory has received considerable empirical support; see, for example, Grullon and Michaely (2004), Hennessy (2004) and Cai and Zhang (2011). Another twist on the agency implications of buybacks is developed in Kahle (2002), which studies how the structure of employee stock options influences repurchase behavior, suggesting that the recent growth in repurchase activity could be a result of the great increases in executive option compensation. She further finds that the presence of executive options leads to greater buyback activity.19 Weisbenner (2000) shows that while 12

Some studies, e.g., Kahle (2002), refer to the latter as the free cash flow theory. A number of studies also address the tax issues of repurchase versus dividends as a means of returning capital to investors; these papers date back to Bierman and West (1966) and Elton and Gruber (1967). See also Bagwell and Shoven (1989). This issue is more complex for REITs since their dividends do not qualify for the usual favorable dividend tax treatment. 14 The recent $15 billion repurchase plans announced by IBM illustrate this point. According to one source (www.v3.ro.uk/vnunet/news/2210896/ibm-setsaside-15bn-stock-repurchase. Feb.29. 2008.): “On the surface the decision may seem to be a positive one for shareholders, but on the flipside this essentially means that IBM has nothing better to spend the money on than trying to raise the share price by financial machinations rather than by selling more product.” 15 Bens et al. (2003) study the relation between repurchase and EPS. They find that managers tend to increase buybacks in order to maintain a target rate of EPS growth. 16 This exchange of repurchase for dividends is in fact dubbed the substitution hypothesis; see Grullon and Michaely (2002). 17 Lie (2005), using data from 1981–2000, estimated announcement effects for open-market repurchase averaging 3.0%, a figure that is comparable to the initial studies. 18 The latter paper suggests that the U.K.'s stricter regulatory environment dampens the adverse information effect, making short- and long-term abnormal returns lower than the figures found in comparable U.S. studies. 19 See also Fenn and Liang (2001). 13

26

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

overall option grants can predict repurchase behavior, there is no link between the option compensation of the firm's top executives and repurchase activity. De Cesari et al. (2012) analyze the impact of liquidity and ownership structure on the repurchase decision. They find a non-linear relation between buyback profits and insider ownership. They also show a positive relation between buyback profits and firm liquidity. An important issue in open-market repurchases is the extent to which firms actually buy back the stated number of shares. Recent evidence, e.g., Lie (2005), has shown that actual repurchase, rather than merely the announcement of an open market repurchase program, leads to superior operating performance after the repurchase. Finally, there are a number of papers (e.g., Kahle, 2002) that deal with the darker motives for repurchase. It is possible that insiders can use their valuable private information to time repurchase decisions, for example, to maximize the value of their performance-based compensation or to engage in profitable insider trading. Related work on the market timing ability evidenced by Hong Kong stock buybacks is found in Brockman and Chung (2001). 2.4. REIT repurchases Brau and Holmes (2006) analyze the various motives for repurchase in the case of REITs. They analyze the cumulative abnormal returns around repurchase announcements in order to determine the extent to which the signaling hypothesis can be tested in light of the competing theories. They argue that the unique structure of REITs (in particular the restriction on paying out 90% of taxable income as dividends) makes the signaling and over-valuation hypotheses dominate the alternatives. Ghosh et al. (2008) address REIT repurchases from a different perspective. Using repurchase announcements from 1997 to 1999, they partition REITs into the group that did buybacks and the group with no buybacks. They then use a Logit framework to show, consistent with Kahle (2002) and others, that option compensation increases the possibility of a buyback. Neither of these papers directly address the over-investment issue.20 3. Data and empirical approach Our basic empirical approach follows Fenn and Liang (2001) and others by employing Logit and Tobit models to determine the influence of various possible repurchase motives on the decision to buy back shares. The unique aspect of our analysis is that, in addition to the typical controls used in related buyback studies, we use the cap rate as a possible explanatory variable to proxy for the dynamics of the firm's investment opportunity set. 3.1. Repurchases Although a common variable of interest, the actual value of shares repurchased is not a straightforward variable to measure. The most accurate measure is the actual shares repurchased multiplied by the average price of repurchased shares as reported in the firm's financial statements. Unfortunately, detailed disclosure of repurchase activity has only been a requirement since 2004. In a hand collected sample of firms, Banyi et al. (2008) horse race various measures of share repurchases against the measure of share repurchases made in the firms' financial statements. They show that the line item “shares repurchased” from the cash flow statement is the best proxy for this measure once it has been adjusted for preferred stock repurchases. In order to expand our sample to before 2004, we employ a hybrid measure of share repurchases. Where available, from SNL Financial we obtain the number of shares repurchased and the average price of repurchased shares from the firm's 10-K.21 When these variables are available, the dollar value of share repurchases is the product of these two variables. When these variables are not available, we use the line item common shares repurchased from the cash flow statement.22 Where no data are reported in SNL, we examine the firm's 10-K and either verify that there were no share repurchases made during the year or reconstruct the amount of share repurchases from the information in the 10-K.23 Table 1 reports descriptive statistics of stock repurchases made by 139 REITs between 1996 and 2010. Repurchases are measured as the dollar value of share repurchases made during the year, divided by the market capitalization of the firm at the start of the year and are reported in percentages. Of the 139 firms in our sample, 101 repurchased stock during the sample period so it does not appear that repurchases are driven by a few firms. Overall, the mean value of repurchases is 0.77%, which is lower

20 It is not straightforward to compare their results with those in this study because of their different timeframe, and because they use announcements rather than actual buybacks. 21 SNL Financial is the standard industry source for financial information on REITs. 22 REITs typically report both common and preferred stock repurchased in the cash flow statement. SNL collects these line items separately, so no adjustment is necessary for preferred stock repurchases. 23 A blank line item in SNL is caused by the absence of the line item in the firm's statement of cash flows. This can either be due to the firm not conducting any repurchase activity during the year or SNL not recording the line item because the firm aggregated common and preferred share repurchases. By examining the discussion in the 10-K we were able to distinguish the amounts due to common and preferred share repurchases.

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

27

Table 1 Repurchases. Table reports descriptive statistics for share repurchases by the 139 REITs in the sample between 1996 and 2010. Repurchases are measured as the dollar value of common share repurchases during the year, divided by market capitalization at the start of the year. Where available, dollar value of common share repurchases is measured as common shares repurchased multiplied by average price of common shares repurchased reported in the company's financial statement. If these variables are not available, the dollar value of repurchases is measured as ‘Common Stock Repurchased’ reported in the firm's statement of cash flows. Repurchases data were obtained from SNL. Shares outstanding is from CRSP. Repurchases are reported in percentages. Year

Firms

Firms repurchasing

Mean

Std. dev

Min

Max

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Average

76 78 96 97 93 90 87 86 90 85 86 83 81 78 78 85

8 6 26 48 48 41 33 26 13 20 27 43 36 31 22 29

0.102 0.054 0.555 1.815 2.194 1.104 0.750 0.490 0.175 0.593 0.738 1.439 0.733 0.115 0.204 0.767

0.434 0.250 1.790 3.390 3.922 2.384 1.864 1.570 0.748 1.804 2.037 3.391 2.215 0.338 1.148 2.250

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

2.49 1.59 13.39 16.93 22.96 12.38 13.75 8.55 4.88 8.65 11.50 24.04 14.18 1.89 9.63 24.04

than the value reported for studies focusing on non-REITs.24 Due to their dividend payout requirement,25 REITs have far less internal cash flow than regular corporations, which helps explain the lower overall level of repurchases.26 Even though the average repurchase is lower, Table 1 shows that there is considerable variation in the amount repurchased by both time and firm. In fact, the maximum repurchase in any given year is quite large, up to 23%, suggesting that REITs do at times undertake very significant repurchase activity. For example, in 2000, 48 of the 93 REITs in the sample repurchased shares, leading to a reduction in 2.2% of shares outstanding. In contrast, in 1997 only 6 of the 78 REITs in the sample repurchased shares, leading to a very small reduction, .05%, in shares outstanding. Given that these firms are constrained by their dividend payout requirement, any results we do find are likely to be more pronounced in a sample of regular corporations, which do not face this constraint. 3.2. Investment opportunities A poor investment opportunity set is a common explanation for buyback decisions. When faced with a lack of attractive investment opportunities, managers should return capital to investors rather than invest in negative net present value projects. The difficulty the prior literature has faced is that, in general, the researcher cannot observe the firm's investment opportunity set. As such, various proxies have been used to indirectly measure the firm's investment opportunity set.27 We mitigate this problem by examining a set of firms whose assets trade in public and private markets. A REIT's main asset is a portfolio of properties. These properties trade directly in private markets as well as through public markets via trading in the REITs themselves. By obtaining pricing information from the underlying real estate markets and relating these market transactions to the REIT's portfolio, we are able to construct the investment opportunity set that the REIT faces at a given point in time. The main metric we use to measure the investment opportunity in the underlying real estate market is the Capitalization Rate, or cap rate. The cap rate – a property's Net Operating Income (NOI) divided by its transaction price – is a standard valuation metric in real estate.28 Since NOI is the property equivalent of Earnings Before Interest, Taxes, Depreciation, and Amortization (EBITDA), by definition the cap rate is equivalent to the pre-tax Return on Assets.29 While the cap rate is not the property's IRR, we believe that it is a very good metric for measuring investment opportunities for two reasons. First, from a practical perspective we know that real estate investors' cap rate and IRR expectations are closely linked. As noted earlier, the correlation between our cap rates and expected IRRs obtained from surveys conducted by PWC Korpacz is over 92%. Second, using a version of Campbell and Shiller's (1988) dynamic Gordon model, Plazzi et al. (2010) show that cap rates are related to future return expectations. This result once again suggests that the cap rate is capturing relevant information with which to gauge the firm's investment opportunity set. 24

See, for example, Fenn and Liang (2001). See Boudry (2011) for a discussion of REIT dividend payout requirements. NAREIT reports that the average dividend yield for equity REITs over the sample period was 5.86%. Aggregating this with our repurchases data gives an average total payout of 6.63% which is considerably higher than the total payout reported by Fenn and Liang (2001) of 2.5% or Jagannathan et al. (2000) of 3%. 27 See Gaver and Gaver (1993) for a discussion. 28 See Brueggeman and Fisher (2008) for a textbook treatment. 29 Note however that for the firms in our sample the cap rate's correlation with the EV/EBITDA multiple is −0.2. 25 26

28

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

Table 2 Capitalization rates. Table reports descriptive statistics of capitalization rates for 109,430 property transactions in the CoStar and Real Capital Analytics databases between 1995 and 2010. Panel A reports statistics for Industrial and Multifamily properties while Panel B reports statistics for Office and Retail properties. Properties were excluded if the transaction cap rate was less than 1%, the price was below $1 million or the transaction was a distressed sale. Panel A: Industrial and Multifamily Industrial

Multifamily

Year

Obs

Mean

Std

Min

Max

Obs

Mean

Std

Min

Max

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

349 508 527 561 740 721 769 713 937 1046 1281 1382 1520 671 337 403

10.30 10.32 10.05 9.57 9.66 9.68 9.59 9.23 8.87 8.21 7.76 7.22 7.08 7.34 8.40 8.57

2.25 1.93 1.69 1.68 1.82 1.81 1.59 1.56 1.59 1.63 1.71 1.47 1.43 1.53 1.79 1.75

4.01 5.62 4.52 4.01 4.22 1.22 3.03 1.29 1.94 3.16 2.00 2.00 2.60 2.00 3.79 1.50

24.39 23.66 23.06 15.34 20.87 19.47 18.03 15.04 18.00 15.00 15.49 14.50 15.00 12.27 17.00 22.28

719 1003 1642 2002 2513 2816 3242 4087 4580 5689 6029 4961 4562 2050 1591 1805

9.21 9.45 9.24 8.83 8.80 8.60 8.54 7.92 7.20 6.53 5.98 5.99 6.04 6.18 6.87 6.81

2.01 2.14 1.98 1.93 1.92 1.98 1.86 1.77 1.78 1.80 1.82 1.74 1.64 1.67 1.67 1.68

3.05 2.00 2.00 2.83 1.41 1.12 2.04 3.17 1.50 2.08 1.18 1.20 1.62 1.59 1.65 2.00

23.15 19.44 18.84 17.78 17.68 17.82 17.48 15.81 18.30 16.96 26.00 23.97 22.37 15.00 17.41 16.37

10.34 10.33 9.65 9.42 9.77 9.67 9.65 9.29 8.82 8.10 7.42 7.00 6.70 7.01 8.13 7.88

2.31 2.30 1.82 1.80 1.78 1.65 1.72 1.67 1.56 1.65 1.60 1.45 1.44 1.48 1.66 1.74

4.18 3.86 3.73 3.20 3.49 2.48 2.31 1.84 2.62 2.20 1.30 2.00 2.00 1.60 3.12 2.60

19.12 26.01 16.78 16.48 23.37 15.71 20.89 25.11 14.82 16.14 14.98 14.00 13.20 13.60 14.93 13.30

10.26 10.44 10.14 9.65 9.67 9.47 9.39 9.05 8.42 7.71 7.10 6.79 6.79 6.95 7.97 8.10

2.04 2.20 1.77 1.70 1.75 1.70 1.61 1.49 1.53 1.60 1.47 1.28 1.27 1.26 1.56 1.46

4.85 3.62 3.34 4.06 3.00 3.51 2.05 3.57 2.68 2.45 1.87 1.98 1.20 1.58 2.20 3.46

23.31 19.58 18.74 18.06 17.69 18.00 17.00 14.88 16.00 15.00 20.80 14.52 24.53 13.96 15.33 20.00

Panel B: Office and Retail Office 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

273 445 699 819 1074 1166 1168 1310 1443 1829 1998 2011 2174 905 470 585

Retail 302 465 762 876 1007 1120 1367 1912 2407 2771 3196 3677 4361 1944 1442 1696

We obtain transactional cap rates from CoStar and Real Capital Analytics, the two leading industry sources for such data, for the period 1995 to 2010. In order to make the properties comparable to the underlying REIT portfolios, we place four filters on the transactions data. First, we limit the sample to core property types (Office, Industrial, Multifamily and Retail). The reason for including only these four property types is simply one of practicality. Other property types, such as Hotel and Health Care, do not have enough transactions to create an accurate time series of cap rates for these property types across different locations. Also, because REITs tend to focus solely on core property types or solely on specialty property types, aside from Hotel and Health Care REITs, which we exclude from the sample, we are able to obtain very good matches to the REITs' portfolios with just these four property types. Second, the reported cap rate has to be greater than 1%.30 Third, the price of the property has to be greater than $1 million.31 Finally, the transaction cannot be a distressed sale. Table 2 reports the descriptive statistics for the cap rate data by year and property type. Overall there were 109,430 property transactions in the sample. 12,465 were Industrial; 49,291 were Multifamily; 18,369 were Office; 29,305 were Retail. Industrial properties had the highest average cap rate at 8.53%, then Office at 8.32%, then Retail at 7.93% and finally Multifamily at 7.19%. The minimum and maximum cap rates show that there is a large degree of dispersion in property cap rates both by time and location. This dispersion in cap rates suggests that any REIT investing in different property types and locations will face differing investment opportunity sets. Since most real estate investing is done on a levered basis, the interaction between cap rates and borrowing rates is likely to be an important determinant of the investment decision. To capture this, we calculate the spread between the property's cap rate and the 30

Fewer than 1% of the cap rates in the sample are less than 3%. Our results are not sensitive to alternative cutoffs. The average property in the sample has a price of $10.37 million with office having the largest average price of $23.56 million; Retail has an average of $7.93 million; Multifamily has an average of $7.72 million; Industrial has an average of $7.18 million. The smallest properties in the sample are $1 million in value while the largest is $5.4 billion. 31

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

29

Table 3 Spread between cap rates and mortgage rates. Table reports descriptive statistics of the spread between cap rates and mortgage rates for 109,430 property transactions in the CoStar and Real Capital Analytics databases between 1995 and 2010. Mortgage rates were obtained from the ACLI. Panel A reports statistics for Industrial and Multifamily properties while Panel B reports statistics for Office and Retail properties. Properties were excluded if the transaction cap rate was less than 1%, the price was below $1 million or the transaction was a distressed sale. Panel A: Industrial and Multifamily Year

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Industrial

Multifamily

Obs

Mean

Std

Min

Max

Obs

Mean

Std

Min

Max

349 508 527 561 740 721 769 713 937 1046 1281 1382 1520 671 337 403

2.27 2.38 2.35 2.61 1.51 2.04 2.63 3.53 3.62 2.87 2.16 1.30 1.06 1.09 1.14 3.10

2.37 1.97 1.69 1.68 1.82 1.85 1.60 1.57 1.58 1.64 1.76 1.48 1.43 1.53 1.82 1.87

−5.17 −2.53 −3.44 −2.93 −4.17 −6.19 −3.90 −4.49 −3.31 −2.24 −3.95 −4.20 −3.49 −4.28 −3.63 −3.76

16.88 15.49 15.44 8.29 12.83 12.29 11.23 9.26 12.71 9.81 9.57 8.30 8.96 5.99 10.00 17.61

719 1003 1642 2002 2513 2816 3242 4087 4580 5689 6029 4961 4562 2050 1591 1805

1.21 1.62 1.63 1.90 0.80 1.09 1.65 2.24 2.08 1.19 0.38 0.13 0.12 −0.31 0.43 1.73

2.10 2.16 1.99 1.93 1.94 1.99 1.87 1.77 1.79 1.79 1.86 1.75 1.66 1.71 1.69 1.74

−5.02 −6.01 −5.53 −4.07 −7.06 −6.61 −5.09 −2.89 −3.42 −3.18 −4.70 −4.55 −4.37 −5.24 −5.07 −3.73

15.64 11.43 11.31 10.93 9.21 9.93 10.64 10.36 13.01 11.72 20.78 17.92 16.23 8.17 11.34 11.50

2.27 2.43 2.02 2.48 1.60 2.01 2.62 3.73 3.46 2.71 1.75 1.01 0.74 0.76 0.88 2.51

2.40 2.30 1.83 1.80 1.79 1.69 1.71 1.66 1.56 1.65 1.66 1.46 1.44 1.49 1.69 1.79

−4.86 −4.40 −4.01 −3.59 −4.62 −4.81 −4.55 −4.13 −2.59 −3.14 −4.38 −4.07 −4.12 −4.36 −4.26 −2.82

11.22 17.75 8.93 9.46 15.26 8.42 13.83 19.04 9.47 10.80 9.30 8.14 7.43 7.28 7.55 7.56

2.27 2.59 2.51 2.58 1.51 1.66 2.35 3.30 3.00 2.25 1.36 0.82 0.78 0.75 0.50 2.46

2.12 2.23 1.79 1.70 1.76 1.75 1.61 1.50 1.55 1.60 1.51 1.29 1.29 1.28 1.62 1.55

−3.68 −4.19 −4.22 −2.82 −5.54 −4.69 −5.17 −2.20 −2.76 −2.97 −4.21 −4.25 −4.87 −4.58 −5.42 −2.39

15.12 12.15 11.46 10.98 9.64 9.77 10.01 9.52 10.78 9.58 15.02 8.75 18.39 7.86 7.86 14.39

Panel B: Office and Retail Office 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

273 445 699 819 1074 1166 1168 1310 1443 1829 1998 2011 2174 905 470 585

Retail 302 465 762 876 1007 1120 1367 1912 2407 2771 3196 3677 4361 1944 1442 1696

borrowing rates that an investor could obtain on that property type at that point in time. As the borrowing rate we use the rates reported by the American Council of Life Insurers (ACLI).32 Table 3 reports the spread between cap rates and mortgage rates by year and property type. Once again, Panel A reports results for Industrial and Multifamily, while Panel B reports results from Office and Retail. Consistent with Table 2, we observe both time series and cross sectional variation in the spread. Industrial has the highest average spread of 2.14, then Office with 2.00, then Retail with 1.69 and finally Multifamily with 1.04. In order to create an accurate investment opportunity set we exploit the portfolio characteristics of the REIT industry. If REITs held properties of all different property types in all locations, then they would all face the same investment opportunity set. Fortunately for our study, REITs tend to be focused by property type, location or both.33 This makes sense given the potential informational advantages of specializing by property type, location, or both. The focused nature of REITs also provides a logical way of characterizing a REIT's investment opportunity set. When considering a new marginal project, a REIT is likely to consider a property in a market and property type in which it already operates. In this sense the current geographic and property type profile of the REIT provides a plausible filter with which to select projects that the REIT might undertake. Thus by examining a REIT's portfolio at a point in time and matching this to transactional cap rates observed in those markets, we can measure the likely investment opportunities that the REIT faces at a given point in time. Calculating a REIT's investment opportunity set is a three step process. First, for each Core Based Statistical Area (CBSA), State, National Council of Real Estate Investment Fiduciaries (NCREIF) region and then Nationally, we calculate the average transactional cap 32 The ACLI is the trade association for the life insurance industry. Life lenders are a standard source of mortgage capital for the real estate market. The ACLI collects loan originations from its members and reports average loan terms for each quarter. 33 For example, Washington REIT invests in multiple property types, but all within driving distance of Washington, DC. Simon Property Group invests only in retail properties, but invests across the US. Nearly all of SL Green's portfolio consists of office properties in Manhattan.

30

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

Table 4 Investment opportunity set. Table reports descriptive statistics of REIT portfolio cap rates for the 139 REITs in our sample between 1996 and 2010. For each year, the properties in the REIT's portfolio are matched by property type and location (CBSA, State, NCREIF region and Nationally) to transaction based cap rates from CoStar and Real Capital Analytics. Average Cap Rate is the average cap rate across the REIT's properties in a given year. Cap Rate Spread is the average spread between cap rates and mortgage rates. Mortgage rates were obtained from the ACLI. Data are reported in percentages. Year

Firms

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Average

76 78 96 97 93 90 87 86 90 85 86 83 81 78 78 85

Cap Rate

Cap Rate Spread

Mean

Std

Min

Max

Range

Mean

Std

Min

Max

Range

10.42 10.39 9.99 9.89 9.90 9.77 9.61 9.21 8.64 7.98 7.44 6.96 6.74 7.01 8.03 8.80

0.58 0.57 0.45 0.56 0.39 0.51 0.47 0.52 0.56 0.58 0.61 0.59 0.60 0.64 0.66 0.55

8.73 8.57 8.14 8.17 8.04 7.77 7.88 7.26 6.50 5.81 5.36 5.16 5.12 5.31 5.81 6.91

11.83 11.56 10.91 12.10 10.73 10.61 10.35 10.19 9.67 9.12 8.49 8.17 7.77 8.48 8.91 9.93

3.10 2.99 2.77 3.93 2.69 2.83 2.48 2.93 3.17 3.32 3.13 3.01 2.65 3.17 3.10 3.02

2.51 2.54 2.38 2.91 1.73 2.12 2.62 3.55 3.34 2.58 1.76 1.01 0.76 0.74 0.84 2.09

0.64 0.60 0.45 0.54 0.37 0.52 0.46 0.53 0.52 0.58 0.61 0.59 0.58 0.68 0.58 0.55

0.65 0.73 0.66 1.24 0.05 −0.02 0.97 1.58 1.37 0.47 −0.24 −0.69 −0.79 −1.14 −0.69 0.28

4.14 3.81 3.30 5.04 2.58 2.99 3.31 4.54 4.31 3.77 2.90 2.29 1.74 2.27 2.13 3.27

3.49 3.08 2.64 3.80 2.52 3.01 2.35 2.96 2.95 3.31 3.14 2.98 2.54 3.41 2.82 3.00

rate and spread for the four different property types Office, Industrial, Multifamily and Retail.34 For each year, we then have an estimate of the cap rate and spread an investor would face in each property type in each given geographic region. Second, for each REIT we build a geographic profile of the REIT's portfolio for each year. We obtain the REIT's property portfolio information from SNL, which provides details of property type and location. Finally, for each year we then match the REIT's property portfolio by property type with the transactional cap rate and cap rate spread from Step 1, first at the CBSA level, then the State level, then at the NCREIF region level, and finally at the National level until all properties are matched. Of the 346,319 property/years in the REITs' portfolios, 273,788 (79.1%) were matched at the CBSA level, 44,733 (12.9%) were matched at the State level, 25,453 (7.3%) were matched at the NCREIF region level and 2345 (0.7%) were matched at the National level.35 Having matched each REIT property to a transactional cap rate and spread in each year, we calculate Average Cap Rate and Cap Rate Spread, which are the averages of these cap rates and spreads across the REIT's portfolio each year. These two metrics provide a measure of the average Return on Assets that a REIT would face on a typical project in its portfolio and the spread between asset returns and the cost of debt. Since repurchase decisions are likely to be made throughout the year, in our empirical specification for repurchases made in year t, we use the average cap rate and cap rate spread from year t− 1. Table 4 reports descriptive statistics for Average Cap Rate and Cap Rate Spread for the 139 REITs in our sample between 1996 and 2010. The most noticeable result from Table 4 is that there is both cross-sectional and time series variation in the investment opportunity set that REITs face. This indicates that the variation observed in transactional cap rates in Table 2 and from the cap rate spread in Table 3 has not been averaged out by the composition of the REITs' portfolios. It also shows that REITs do not all have the same investment opportunity set, which is critical for our study. On average there is a 302 basis point difference between the cap rate faced by the REITs with the best and worst investment opportunity sets. Overall the pattern of investment opportunities reported in Table 4 matches the anecdotal evidence from the market. Between 2003 and 2008 there was a large inflow of capital into commercial real estate, which drove down cap rates. This is now considered by many to be a bubble period in the commercial real estate market with properties becoming increasingly over-valued. Consistent with this, both our investment opportunity measures show that the investment opportunity set was worsening during this period. The effects of the global financial crisis are also apparent in the data. After 2008 we observe an improvement in the investment opportunity set. This coincides with a price correction in the commercial real estate market. During this period, aggregate commercial real estate prices as measured by the MIT\TBI commercial real estate price index were down 40% from their peak.

34 NCREIF divides the US into 8 regions: North-East (ME, VT, NH, NY, CT, RI, MA, PA, NJ), Mid-East (MD, WV, VA, KY, NC, SC, DC, DE), South-East (TN, GA, FL, AL, MS), East North Central (MI, IL, OH, IN, WI), West North Central (MN, IA, MO, KS, SD, ND, NE), South West (TX, OK AR, LA), Mountain (ID, WY, UT, CO, NM, AZ) and Pacific (WA, OR, CA, AK, HI). 35 The data from CoStar and Real Capital Analytics are heavily biased towards larger metro areas, so it is unlikely that they are representative of cap rates in all U.S. locations. Fortunately REIT portfolios are biased in the same manner, with REITs mainly investing in larger metro areas, which is why we are able to match the majority of the sample at the CBSA level.

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

31

Table 5 Descriptive statistics. Table reports descriptive statistics for the independent variables in the analysis. Data are for the 139 REITs in the sample between 1996 and 2010. Average Cap Rate is the REIT's average portfolio cap rate (in percentages) and Cap Rate Spread is the average spread between the portfolio cap rates and mortgage rates (in percentages). Exercisable Options are unexercised but exercisable options divided by shares outstanding (in percentages) and Unexercisable Options are unexercisable options divided by shares outstanding (in percentages). Free Cash Flow is funds from operations (FFO) divided by total assets (in percentages) and Free Cash is cash and marketable securities divided by total assets (in percentages) winsorized at 99%. Leverage is the REIT's total debt plus preferred stock to total capital (in percentages). Market to Book is book assets minus book equity plus market value of equity all divided by total assets and Total Assets is the log of total book assets. Average Cap Rate and Cap Rate Spread are calculated using data from CoStar and Real Capital Analytics and mortgage rates from the ACLI, while all other data are from SNL except the options data which is from EXECUCOMP or the firm's proxy statements. Variable

Average Cap Rate Cap Rate Spread Exercisable Options Unexercisable Options Free Cash Flow Free Cash Leverage Market to Book Total Assets

No repurchase

Repurchase

t test

Obs

Mean

Std

Min

Max

Obs

Mean

Std

Min

Max

856 856 856 856 856 856 856 856 856

8.94 2.23 1.40 0.87 0.06 0.02 40.88 1.31 13.88

1.34 1.02 2.13 1.44 0.03 0.05 15.58 0.31 1.38

5.35 −0.51 0.00 0.00 −0.29 0.00 0.00 0.52 9.36

12.10 5.04 17.90 21.75 0.28 0.40 97.80 2.98 17.20

428 428 428 428 428 428 428 428 428

8.63 1.91 1.45 1.07 0.05 0.02 43.45 1.23 14.32

1.44 1.05 1.68 2.11 0.03 0.04 16.38 0.23 1.25

5.12 −1.14 0.00 0.00 −0.16 0.00 0.00 0.60 9.33

11.44 4.30 8.27 28.65 0.13 0.40 89.19 2.24 17.07

−3.84 −5.22 0.45 2.00 −2.24 −2.79 2.74 −4.74 5.61

3.3. Other characteristics In our empirical design we control for other potential explanations of repurchase behavior posited in the prior literature. In particular we control for stock options, agency cost explanations of repurchases, leverage, under valuation, and external financing costs.

3.3.1. Stock options Executive compensation is a factor potentially affecting repurchase behavior.36 Managers' incentives can be aligned with shareholders' interests by awarding managers stock options. Fenn and Liang (2001) argue that because of aligned incentives, managers would be more likely to pay out free cash flow to investors rather than invest in bad projects. Thus we would expect a positive relationship between stock options and repurchases. Another channel through which stock options could affect repurchases is by making repurchases preferred to dividends as a mechanism to return capital to investors. Lambert et al. (1989) argue that since executive stock options (like all call options) are negatively impacted by dividends, if managers wish to maintain the same overall payout ratio they would prefer repurchases to dividends. Once again we would expect a positive relationship between the level of option compensation and repurchase activity. Options data are obtained from two sources. First, where available we obtain both unexercisable and unexercised but exercisable options from EXECUCOMP. EXECUCOMP's coverage of the REIT universe is quite limited, so where data are missing we collect the options data from the firm's proxy statements. Following Kahle (2002), we create two variables: Unexercisable Options, which is outstanding unexercisable unexercised options divided by total shares outstanding, and Exercisable Options, which is unexercised exercisable options divided by total shares outstanding.

3.3.2. Agency costs Under agency cost explanations of repurchases, we would expect to observe firms with high levels of free cash flow repurchasing stock. By repurchasing stock the firm reduces Jensen's (1986) free cash flow agency problems. We employ two variables to proxy for free cash flow. First, Free Cash Flow is the firm's funds from operations divided by total assets. We use funds from operations because it is the standard measure of cash flow in the REIT industry.37 The second variable, also employed by Dittmar (2000), is Free Cash, which is the REIT's cash and marketable securities divided by total assets.38 In both cases we would expect to observe a positive relationship with repurchases.

36 37 38

See Kahle (2002) for a detailed discussion. FFO is essentially equal to a REIT's net income, excluding gains or losses from sales of property, and adding back real estate depreciation. We winsorize Free Cash at the 99% level. Our results are similar if we use unwinsorized data.

32

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

Table 6 Correlations. Table reports correlations between the independent variables in the analysis. Data are for the 139 REITs in the sample between 1996 and 2010. Average Cap Rate is the REIT's average portfolio cap rate (in percentages) and Cap Rate Spread is the average spread between the portfolio cap rates and mortgage rates (in percentages). Exercisable Options is unexercised but exercisable options divided by shares outstanding (in percentages) and Unexercisable Options are unexercisable options divided by shares outstanding (in percentages). Free Cash Flow is funds from operations (FFO) divided by total assets (in percentages) and Free Cash is cash and marketable securities divided by total assets (in percentages) winsorized at 99%. Leverage is the REIT's total debt plus preferred stock to total capital (in percentages). Market to Book is book assets minus book equity plus market value of equity all divided by total assets and Total Assets are the log of total book assets. Average Cap Rate and Cap Rate Spread are calculated using data from CoStar and Real Capital Analytics and mortgage rates from the ACLI, while all other data are from SNL except the option data which is from EXECUCOMP or the firm's proxy statements.

Average Cap Rate Cap Rate Spread Exercisable Options Unexercisable Options Free Cash Flow Free Cash Leverage Market to Book Total Assets

Average Cap Rate

Cap Rate Spread

Exercisable Options

Unexercisable Options

Free Cash Flow

Free Cash

1.00 0.71 0.16 0.26 0.24 −0.11 −0.02 −0.26 −0.36

1.00 0.11 0.15 0.16 −0.09 −0.02 −0.13 −0.21

1.00 0.29 −0.08 0.07 0.14 −0.04 −0.02

1.00 −0.01 −0.06 0.06 −0.05 −0.02

1.00 −0.07 −0.57 0.37 0.00

1.00 −0.21 0.01 −0.20

Leverage

1.00 −0.49 0.03

Market to Book

Total Assets

1.00 0.11

1.00

3.3.3. Leverage Firm leverage may also influence buyback decisions. Since repurchasing stock will increase leverage, firms with high leverage are less likely to repurchase stock since higher leverage will potentially increase expected bankruptcy costs. Leverage can also affect repurchase decisions because it reduces free cash flows.39 Leverage could also increase the volatility of the firm's free cash flow making it less likely to make repurchases because of a precautionary savings motive. In each case, we expected a negative relationship between share repurchases and leverage. We calculate Leverage as total debt plus preferred stock divided by total capital (market value of equity plus the book value of debt and the book value of preferred stock). 3.3.4. Under valuation Managers may choose to repurchase stock when the firm is undervalued. This is in fact a common explanation of the market's reaction to share repurchase announcements. To capture any potential under valuation, we include the market to book ratio in our empirical specification. Following Dittmar (2000), Market to Book is calculated as the firm's book value of assets minus book value of equity plus market value of equity, all divided by book value of assets.40 3.3.5. External funding costs Firms that face lower external funding costs are likely to be less constrained in their ability to repurchase stock since funding shortfalls can be corrected at low cost in the future. We proxy external funding costs using Total Assets, the log of the firm's total assets. As argued by Smith and Watts (1992) and Opler and Titman (1993), larger firms are likely to be more transparent and to have more stable cash flows. The latter is also likely to be true for REITs, because as Boudry et al. (2011) show, firm size is a key determinant of analyst coverage. 3.4. Descriptive statistics Table 5 reports descriptive statistics for the independent variables in the analysis. The sample is bifurcated into repurchasing and non-repurchasing firms and a t test of the difference in means between repurchasing and non-repurchasing firms is reported in the right most column. Consistent with the notion that poor investment opportunities lead to higher repurchases, we observe that the average cap rate and the spread between the average cap rate and the borrowing rate are higher for non-repurchasing than repurchasing firms. So, at least in a univariate analysis, there appears to be some support for the over-investment hypothesis. The average value of Exercisable Options and Unexercisable Options is larger for repurchasing than non-repurchasing firms, with the difference being statistically significant for unexercisable options. The average values of 1.40% and 0.87% for non-repurchasing and 1.45% and 1.07% for repurchasing firms are similar to the averages reported in Ghosh et al. (2008). A relationship that runs counter to expectations is the variables related to agency costs. The average values of Free Cash Flow and Free Cash are higher for non-repurchasing firms especially for Free Cash.41 If firms with higher free cash flow pay out this cash 39

See Jensen (1986). An alternate proxy for overvaluation in the REIT market is the price to NAV ratio reported by analysts (see, for example, Boudry et al., 2010). Using this variable is quite problematic because not all analysts report NAVs and it isn't commonly recorded in databases such as IBES. For a select subsample of firms where we were able to obtain NAVs, we find that the price to NAV ratio has a correlation of 45% with the market to book ratio. Given that using the price to NAV ratio would impose considerable censoring on the sample, in the light of the high correlation we use the market to book ratio as our measure of overvaluation. 41 The level of cash held by REITs appears quite low compared to COMPUSTAT firms. However, on a cash to sales basis REITs appear quite similar to the COMPUSTAT sample reported by Harford et al. (2008) with a mean cash to sales of 17.6% compared to 18% for regular C-corporations. Furthermore, if unused credit lines are included in cash, the average cash plus unused credit lines to total assets ratio is 10.7%. 40

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

33

Table 7 Decision to repurchase. Table reports estimation results for a random-effect panel Logit model. The dependent variable is equal to one if the firm repurchased shares during the year and zero otherwise. Marginal effects are reported for a one standard deviation change in the independent variable and are reported in percentages. Data are for the 139 REITs in the sample between 1996 and 2010. Average Cap Rate is the REIT's average portfolio cap rate (in percentages) and Cap Rate Spread is the average spread between the portfolio cap rates and mortgage rates (in percentages). Exercisable Options are unexercised but exercisable options divided by shares outstanding (in percentages) and Unexercisable Options are unexercisable options divided by shares outstanding (in percentages). Free Cash Flow is funds from operations (FFO) divided by total assets (in percentages) and Free Cash is cash and marketable securities divided by total assets (in percentages) winsorized at 99%. Leverage is the REIT's total debt plus preferred stock to total capital (in percentages). Market to Book is book assets minus book equity plus market value of equity all divided by total assets and Total Assets is the log of total book assets. Average Cap Rate and Cap Rate Spread are calculated using data from CoStar and Real Capital Analytics and mortgage rates from the ACLI, while all other data are from SNL except the option data which is from EXECUCOMP or the firm's proxy statements. Estimation includes unreported year effects.

Average Cap Rate Cap Rate Spread Exercisable Options Unexercisable Options Free Cash Flow Free Cash Leverage Market to Book Total Assets Obs ρ χ2(ρ = 0)

Marginal effect

Z-stat

−14.50

−2.71

1.13 2.66 −0.34 −0.74 −1.55 −10.66 7.17 1284 0.41 153.8

0.58 1.62 −0.15 −0.37 −0.56 −3.52 2.44

Marginal effect

Z-stat

−9.06 1.08 2.69 −0.43 −0.77 −1.59 −10.36 7.33 1284 0.42 157.1

−2.3 0.55 1.63 −0.19 −0.38 −0.57 −3.42 2.48

flow to investors in the form of repurchases in order to avoid agency costs, we would expect to see a positive relationship with repurchases. The average market to book ratio for non-repurchasing firms is 1.31 and the average for repurchasing firms is 1.23. This difference is highly statistically significant and if one assumes that firms with lower market to book ratios are more likely to be undervalued, then the relationship is consistent with the under-valuation hypothesis. Finally, repurchasing firms are larger than non-repurchasing firms. This is consistent with larger firms facing lower external financing costs and as such being less constrained in their ability to repurchase stock. Table 6 reports a correlation matrix of the independent variables. As expected, Free Cash Flow has a strong negative correlation with Leverage of − 0.57. This is consistent with leverage reducing free cash flow. Interestingly, Average Cap Rate and Market to Table 8 Amount of repurchases. Table reports estimation results for a random-effect panel Tobit model. The dependent variable Repurchase, is equal to the dollar amount of common shares repurchased during the year divided by market capitalization at the start of the year multiplied by 100. To aid interpretation, marginal effects are reported for a one standard deviation change in the independent variable and also as a percentage of the average level of Repurchases. Data are for the 139 REITs in the sample between 1996 and 2010. Average Cap Rate is the REIT's average portfolio cap rate (in percentages) and Cap Rate Spread is the average spread between the portfolio cap rates and mortgage rates (in percentages). Exercisable Options are unexercised but exercisable options divided by shares outstanding (in percentages) and Unexercisable Options are unexercisable options divided by shares outstanding (in percentages). Free Cash Flow is funds from operations (FFO) divided by total assets (in percentages) and Free Cash is cash and marketable securities divided by total assets (in percentages) winsorized at 99%. Leverage is the REIT's total debt plus preferred stock to total capital (in percentages). Market to Book is book assets minus book equity plus market value of equity all divided by total assets and Total Assets is the log of total book assets. Average Cap Rate and Cap Rate Spread are calculated using data from CoStar and Real Capital Analytics and mortgage rates from the ACLI, while all other data are from SNL except the option data which is from EXECUCOMP or the firm's proxy statements. Estimation includes unreported year effects.

Average Cap Rate Cap Rate Spread Exercisable Options Unexercisable Options Free Cash Flow Free Cash Leverage Market to Book Total Assets Obs Left Censored ρ χ2(ρ = 0)

Marginal effect

Marginal effect (% of mean)

Z-stat

−36.63

−47.76

−2.52

−3.19 7.62 −0.30 10.71 −6.45 −38.25 15.01 1284 856 0.37 119.9

−4.16 9.93 −0.39 13.97 −8.41 −49.87 19.57

−0.59 1.95 −0.05 2.24 −0.86 −4.55 1.93

Marginal effect

Marginal effect (% of mean)

Z-stat

−21.49 −3.38 7.72 −0.61 10.82 −6.74 −37.55 15.52 1284 856 0.38 122.5

−28.01 −4.40 10.07 −0.79 14.11 −8.78 −48.96 20.23

−2.01 −0.62 1.97 −0.09 2.25 −0.89 −4.45 1.97

34

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

Book have a negative correlation of − 0.26. This can be explained by the fact that when cap rates are low, property prices are high, which should be reflected in the market value of the firm. It also suggests that they are capturing different effects.

4. Estimation results 4.1. Probability of repurchase We examine the relation between investment opportunities and share repurchases in two ways. First, we treat share repurchases as a binary outcome and estimate the relation between repurchasing shares and the independent variables in a Logit framework. Second, we treat share repurchases as a censored variable and estimate a Tobit model of share repurchases on the independent variables. Table 7 reports estimation results for the random effects Logit model. The dependent variable takes the value of 1 if Repurchases are greater than zero and 0 otherwise. To aid interpretation, we report the change in the probability of repurchasing shares for a one standard deviation increase in the independent variables. Marginal effects are reported in percentages. The overall model fit is good; the χ2 values are 153.8 and 157.1 in the two estimations. Each of the following estimations uses year dummies, which are unreported. The marginal effects in Table 7 consistently support the hypothesis that poor investment opportunities lead to more repurchases. Measuring the investment opportunity set using Average Cap Rate, we observe that a firm is 14.5% less likely to repurchase stock if the average cap rate faced by the firm increases by one standard deviation, or 138 basis points. Turning to Cap Rate Spread we draw similar conclusions. A one standard deviation increase in the spread between the cap rates and mortgage rates makes a firm 9.06% less likely to repurchase stock. The strong statistical significance of these marginal effects demonstrates that a firm's investment opportunity set does indeed play a role in the firm's choice to repurchase shares. Consistent with the prior literature, we find that market to book ratio has a strong negative relationship with share repurchases.42 A one standard deviation increase in the market to book ratio makes the firm 10.66% and 10.36% less likely to repurchase stock, respectively in the two estimates. Interpreting high market to book ratios as a signal of possible over valuation, this result is consistent with the under-valuation hypothesis. External funding costs also appear to play a role in the firm's decision to repurchase stock. Consistent with larger firms being less constrained in their ability to raise future funding, larger firms are significantly more likely to repurchase stock. In both estimations, a one standard deviation increase in total assets makes the firm approximately 7% more likely to repurchase stock. The magnitude of this marginal effect is not surprising given that REITs are capital constrained. As such it is natural to expect that the ability to raise future capital is a key determinant of the decision to repurchase stock today.43 Note that in this table, our measures of option compensation are never significant, which differs from the findings of Ghosh et al. (2008).44 These estimates are consistent with the notion that REITs do not use repurchase to acquire treasury stock for reissue as executive compensation. Similarly, leverage is insignificant, which suggests that REITs do not use repurchase to manipulate capital structure.

4.2. Amount of repurchase Table 8 reports estimation results for a random effects Tobit model of Repurchases on the independent variables. Once again, to aid interpretation we report the marginal effects as the change in the dependent variable for a one standard deviation increase in the independent variable. As in the Logit estimation in Table 7, there is a statistically significant marginal effect for Average Cap Rate and Cap Rate Spread in Table 8. A one standard deviation increase in the average cap rate faced by the REIT leads to a 36.63 basis point decrease in repurchases. In terms of the spread between cap rates and mortgage rates, a one standard deviation decrease in the spread leads to a 21.49 basis point increase in repurchases. Given that the average level of repurchases in the sample is 77 basis points, this is a large effect. In terms of the average level of repurchases, these coefficients represent a decrease of 47.76% and 28.01% respectively. Once again we also find that under valuation and external financing costs have a significant impact on share repurchases. A one standard deviation increase in the market to book ratio leads to a 38.25 and 37.55 basis point decrease in repurchases respectively. A one standard deviation increase in total assets leads to a 15.01 and 15.52 basis point increase in repurchases respectively. Interestingly, Free Cash Flow is not significant in the estimation. This may be due to high payout requirements effectively diminishing the magnitude of this repurchase channel relative to regular corporations. However, Free Cash is positive and significant, implying that cash levels have a greater influence on the amount repurchased than on the probability of repurchase. Leverage is never significant. Again, as in Table 7, we find that the two option compensation variables are not significant.

42

Results are similar if we use alternative measures of undervaluation, such as the U-Index of Peyer and Vermaelen (2009) or prior year stock returns. Since market size is a proxy for liquidity, this result is also consistent with previous findings on buybacks and liquidity. 44 Notice that the comparison to Ghosh et al. (2008) is not a direct one, since they examine repurchase program announcements over a very limited 2-year window. Our result does not change if we omit our cap rate variables from the estimation. 43

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

35

Table 9 Low investment opportunities: Average cap rate. Table reports estimation results for a random-effect panel Tobit model. The dependent variable Repurchase, is equal to the dollar amount of common shares repurchased during the year divided by market capitalization at the start of the year multiplied by 100. To aid interpretation, marginal effects are reported for a one standard deviation change in the independent variable and also as a percentage of the average level of Repurchases. Data are for the 139 REITs in the sample between 1996 and 2010. Low refers to the bottom quartile of observations based on Average Cap Rate, while Not Low is the remainder of the sample. Average Cap Rate is the REIT's average portfolio cap rate (in percentages). Exercisable Options are unexercised but exercisable options divided by shares outstanding (in percentages) and Unexercisable Options are unexercisable options divided by shares outstanding (in percentages). Free Cash Flow is funds from operations (FFO) divided by total assets (in percentages) and Free Cash is cash and marketable securities divided by total assets (in percentages) winsorized at 99%. Leverage is the REIT's total debt plus preferred stock to total capital (in percentages). Market to Book is book assets minus book equity plus market value of equity all divided by total assets and Total Assets are the log of total book assets. Average Cap Rate is calculated using data from CoStar and Real Capital Analytics, while all other data are from SNL except the option data which is from EXECUCOMP or the firm's proxy statements. Estimation includes unreported year effects. Low

Exercisable Options Unexercisable Options Free Cash Flow Free Cash Leverage Market to Book Total Assets Obs Left Censored ρ χ2(ρ = 0)

Not Low

Marginal effect

Marginal effect (% of mean)

Z-stat

Marginal effect

Marginal effect (% of mean)

Z-stat

−13.33 4.10 0.38 17.56 −11.59 −22.71 −0.78 328 193 0.59 48.22

−17.31 5.33 0.49 22.80 −15.05 −29.49 −1.01

−1.64 0.69 0.06 2.94 −1.4 −2.67 −0.12

−0.51 5.01 −3.53 −4.08 −6.00 −29.57 20.82 956 293 0.31 64.44

−0.66 6.51 −4.58 −5.29 −7.79 −38.40 27.04

−0.1 1.29 −0.54 −0.67 −0.82 −3.24 3.09

4.3. Low investment opportunity firms and repurchase To examine further the potential relation between agency costs and investment opportunities, we sort our sample into low and not low investment opportunity firms. We identify observations in the bottom quartile based as those that face low investment opportunities. Specifically, we compute the average cap rate and spread over all observations in the sample. We then classify firms as low investment opportunity firms if their average cap rate or spread lies in the corresponding lowest quartile. We

Table 10 Low investment opportunities: Cap rate spread. Table reports estimation results for a random-effect panel Tobit model. The dependent variable Repurchase, is equal to the dollar amount of common shares repurchased during the year divided by market capitalization at the start of the year multiplied by 100. To aid interpretation, marginal effects are reported for a one standard deviation change in the independent variable and also as a percentage of the average level of Repurchases. Data are for the 139 REITs in the sample between 1996 and 2010. Low refers to the bottom quartile of observations based on Cap Rate Spread, while Not Low is the remainder of the sample. Cap Rate Spread is the average spread between the portfolio cap rates and mortgage rates (in percentages). Exercisable Options are unexercised but exercisable options divided by shares outstanding (in percentages) and Unexercisable Options are unexercisable options divided by shares outstanding (in percentages). Free Cash Flow is funds from operations (FFO) divided by total assets (in percentages) and Free Cash is cash and marketable securities divided by total assets (in percentages) winsorized at 99%. Leverage is the REIT's total debt plus preferred stock to total capital (in percentages). Market to Book is book assets minus book equity plus market value of equity all divided by total assets and Total Assets is the log of total book assets. Cap Rate Spread is calculated using data from CoStar and Real Capital Analytics and mortgage rates from ACLI, while all other data are from SNL except the option data which is from EXECUCOMP or the firm's proxy statements. Estimation includes unreported year effects. Low

Exercisable Options Unexercisable Options Free Cash Flow Free Cash Leverage Market to Book Total Assets Obs Left Censored ρ χ2(ρ = 0)

Not Low

Marginal effect

Marginal effect (% of mean)

Z-stat

Marginal effect

Marginal effect (% of mean)

Z-stat

−5.21 0.00 4.90 19.26 −5.83 −22.37 −5.65 329 197 0.58 40.93

−6.77 0.01 6.37 25.01 −7.58 −29.05 −7.34

−0.55 0 0.54 2.52 −0.53 −2.08 −0.63

−1.25 5.95 −8.98 −7.08 −12.61 −43.35 24.76 955 296 0.29 56.85

−1.62 7.72 −11.67 −9.19 −16.38 −56.29 32.15

−0.23 1.37 −1.19 −1.06 −1.5 −3.32 2.85

36

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

then repeat the estimation from Table 8 in Tables 9 and 10. If the agency cost argument is correct, we would expect to see Free Cash Flow or Free Cash to have a positive relationship with Repurchases for the low investment opportunity firms.45 Tables 9 and 10 report estimation results for the two types of investment opportunity set firms based on Average Cap Rate and Cap Rate Spread respectively. We also report results for the remainder of the sample to test if any differential effect exists between those firms facing good and bad investment opportunity sets. The results of both Tables 9 and 10 are consistent with the agency cost argument. In both sets of estimations, we observe a significant positive marginal effect on Free Cash in the low investment opportunity set sample and an insignificant effect in the not low sample. Using Average Cap Rate, Table 9 shows that a one standard deviation increase in Free Cash will increase the repurchase amount by 17.56%, which means that the average repurchase amount increases by 22.8%. Using Cap Rate Spread, Table 10 shows that a one standard deviation increase in Free Cash will increase the repurchase amount by 19.26%, which means that the average repurchase amount increases by 25.01%. In both of these tables neither of the cash variables is significant for the not low investment opportunity REITs. Free Cash Flow is not significant in any of these estimations. 4.4. Market reaction To provide some evidence on the market reaction to repurchases for the firms in our sample, we collect announcement dates from Securities Data Corporation's repurchase database and also obtain any repurchase motivations the firm may have by reading press releases related to the announcement on LexisNexis.46 For our sample of firms we are able to identify 136 repurchase program announcements and of these 76 provide a motivation for the repurchase program. The most common motivation is undervaluation, with 49 announcements referring to the firm's stock being undervalued. This is consistent with both our empirical findings that firm undervaluation is an important determinant of actual repurchase behavior, and also with Peyer and Vermaelen (2009) who show that undervaluation is the most common motivation for share repurchases in a broader sample of firms. So it appears in this context that our sample of REITs looks quite similar to regular firms. The second most common motivation was poor investment opportunities with 43 references. This was usually made concurrently with the firm being undervalued, but there were three announcements where the firm's sole motivation was poor investment opportunities with no reference to being undervalued. This is also consistent with our empirical finding that firms repurchase stock when facing poor investment opportunities. A potential explanation for the weak relationship we find between executive stock options and repurchase behavior is that REITs are rarely motivated to repurchase stock because of stock options. Only 7 announcements refer to executive stock options programs as a motivation to repurchase stock.47 We estimate abnormal returns in the 3-day (−1 to 1) event window surrounding the firm's repurchase program announcement. The average abnormal return is 1.17% and statistically different from zero. It is however smaller than the announcement effect reported for the larger sample of firms reported in Peyer and Vermaelen (2009). To provide some evidence on the market's reaction to firms with low investment opportunities, we observe that the announcements where only poor investment opportunities are provided as a motivation, have an abnormal return of 5.04%. While the sample is quite limited, it does suggest that the market reacts positively to firms with poor investment opportunities repurchasing stock. To further examine the relationship between investment opportunities and announcement effects, we estimate cross-sectional regressions of 3-day and 5-day CARs for firms announcing repurchase programs during the “low” and “not low” investment opportunity environments described in Table 9. Controlling for the size of the announced repurchase program, firms that announce repurchase programs in low investment opportunity environments have 1.7% higher 3-day CARs and 2.35% higher 5-day CARs than firms making repurchase announcements when investment opportunities are not low. Both of these differences are statistically significant. This once again supports the notion that markets react positively to repurchase announcements when firms face low investment opportunities. It is possible that the market reacts not only to the announcement of a share repurchase program, but also to the behavior of the firm with regard to its actual share repurchases. To examine this we calculate 12-month abnormal returns in the year subsequent to repurchase activity. In this sense our sample is the same as that used in the previous empirical work (Tables 7–10), but instead of explaining repurchase activity, we examine the subsequent performance of firms that repurchase stock in different investment opportunity environments. Firms that repurchase stock in our sample have a 12-month abnormal return of −3.72% versus −4.73% for firms that do not repurchase stock. Although in both cases the return is not statistically significant. Using our definition of low investment opportunities employed in Table 9, firms that face low investment opportunities experience 12-month abnormal returns of −16.06% which is highly statistically significant. This compares to 0.5% for the remainder of the sample. Although it is an in-sample result, it suggests that our proxy of the firm's investment opportunity set is measuring future investment opportunities. We further split our sample between firms that repurchase and don't repurchase stock in low and not low investment environments. We find that the key driver of subsequent returns is the investment opportunity set. Repurchasing firms have −15.76% abnormal returns while non-repurchasing firms have −16.28% abnormal returns in low investment opportunity environments. In this 45

Barth and Kasznik (1999) show that repurchase amounts are positively related to cash levels. This is the same approach used by Peyer and Vermaelen (2009). 47 14 announcements in the sample have some “other” motivation. The most common “other” reason was simply that the old repurchase program had expired, but no motivation was given for starting a new one. 46

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

37

sense there appears little difference between repurchasing and not when investment opportunities are low. We observe a similar result in the not low investment opportunity sample. Repurchasing firms have a 12-month abnormal return of 1.54%, while non-repurchasing firms have an abnormal return of −1.36%. In both cases neither is statistically significant. 5. Conclusion Previous studies on buyback behavior have demonstrated three general points. (i) Valuation plays a significant role in the repurchase decision: firms tend to buy back shares when they believe that the shares are undervalued. (ii) Increases in option compensation tend to lead to more repurchase activity. (iii) Using market-to-book to proxy for investment opportunities, there is indirect evidence that the firm's opportunity set can influence the share repurchase decision. This study, while offering evidence on each of these three issues, is novel in its much more direct investigation of point (iii). Using data on cap rates (a REIT's projected Return on Assets for a given property) we are able to construct a series of potential project returns for the 139 REITs in our sample, i.e., an approximation to the REIT's opportunity set at a given point in time. Thereby we address more precisely the over-investment hypothesis, firms repurchasing shares in lieu of investing in negative NPV projects, by linking these investment returns (as well as these returns in excess of a REIT's cost of borrowing) with buyback activity over time. With regard to finding (ii), we find no support for the hypothesis that option compensation is a significant influence on REIT buyback activity. This result may be due to the fact that the average level of repurchases in our sample is lower than corresponding levels for industrial firms. We also find no support for the notion that leverage and free cash flow play a significant role in the repurchase decision for the firms in our sample. We find significant evidence for the over-investment hypothesis: REITs increase their repurchase activity significantly when cap rates are low and decrease it significantly when cap rates are high. Table 8 shows that a one standard deviation increase in the average cap rate leads to the repurchase amount falling by almost one half. Conditioning on investment opportunities, we find that the amount of free cash available has a positive and significant influence on repurchase only for firms with poor investment opportunities. These results are noteworthy since our sample of REITs (because of institutional constraints on dividend payout) has far less divertible cash flow than industrial firms. Thus we would expect these results to be even stronger for non-REITs. We are also able to distinguish this effect from the over-valuation effect. We provide support for this hypothesis by using market-to-book in our Logit and Tobit analyses and we find the expected negative impact on buyback activity: high market to book REITs are less likely to repurchase shares. Finally, consistent with our other findings, we find that the market reacts positively to repurchase program announcements made when a firm faces low investment opportunities. This results holds for the sample where the firms themselves have stated that a lack of good investment opportunities is the reason for starting the repurchase program and also for the sample where we identify firms with poor investment opportunities using our cap rate based measure of investment opportunities. References Allen, Franklin, Michaely, Roni, 2002. Market timing and capital structure. J. Finance 62, 1–32. Bagwell, Laurie, 1991. Share repurchase and takeover deterrence. RAND J. Econ. 22, 72–88. Bagwell, Laurie, Shoven, John B., 1989. Cash distributions to shareholders. J. Econ. Perspect. 3, 129–140. Baker, H., Gallagher, P., Morgan, K., 1981. Managements' view of share repurchase programs. J. Financ. Res. 4, 233–247. Banyi, Monica, Dyl, Edward, Kahle, Kathleen, 2008. Errors in estimating share repurchases. J. Corp. Finance 14, 460–474. Barth, Mary, Kasznik, Ron, 1999. Share repurchases and intangible assets. J. Account. Econ. 28, 211–241. Bartov, Eli, 1991. Open-market stock repurchase as signals for earnings and risk change. J. Account. Econ. 14, 275–294. Bens, Daniel, Nagar, Venky, Skinner, Douglas, Franco Wong, M.H., 2003. Employee stock options, EPS dilution, and stock repurchases. J. Account. Econ. 36, 51–90. Bhattacharya, Sudipto, 1979. Imperfect information, dividend policy, and ‘the bird in the hand’ fallacy. Bell J. Econ. 10, 259–270. Bierman, Harold, West, Richard, 1966. The acquisition of common stock by the corporate issuer. J. Finance 4, 687–696. Boudry, Walter I., 2011. An examination of REIT dividend payout policy. Real Estate Econ. 39, 601–634. Boudry, Walter I., Kallberg, Jarl G., Liu, Crocker H., 2010. An analysis of REIT security issuance decisions. Real Estate Econ. 38, 91–120. Boudry, Walter I., Kallberg, Jarl G., Liu, Crocker H., 2011. Analyst behavior and underwriter choice. J. Real Estate Finance Econ. 43, 5–38. Brau, James C., Holmes, Andrew, 2006. Why do REITs repurchase stock? Extricating the effect of managerial signaling in open market share repurchase announcements. J. Real Estate Res. 28, 1–23. Brav, Alon, Graham, John R., Harvey, Campbell R., Michaely, Roni, 2005. Payout policy in the 21st century. J. Financ. Econ. 77, 483–527. Brockman, Paul, Chung, Dennis Y., 2001. Managerial timing and corporate liquidity: evidence from actual share repurchases. J. Financ. Econ. 61, 417–448. Brueggeman, William B., Fisher, Jeffrey D., 2008. Real estate finance and investments. New York, McGraw-Hill. Cai, Jie, Zhang, Zhe, 2011. Leverage changes, debt overhang and stock prices. J. Corp. Finance 17, 391–402. Campbell, John Y., Shiller, Robert J., 1988. The dividend–price ratio and expectations of future dividends and discount factors. Rev. Financ. Stud. 1, 195–228. Dann, Larry, 1981. Common stock repurchase: analysis of returns to bond-holders and shareholders. J. Financ. Econ. 9, 113–138. De Cesari, A., Espenlaub, S., Khurshed, A., Simkovic, M., 2012. The effects of ownership and stock liquidity on the timing of repurchase transactions. J. Corp. Finance 18 (5), 1023–1050. Dellavigna, Stefano, Pollet, Joshua, 2013. Capital budgeting versus market timing: an evaluation using demographics. J. Finance 68, 237–270. Dittmar, Amy, 2000. Why do firms repurchase stock? J. Bus. 73, 331–355. Elton, Edwin, Gruber, Martin, 1967. The effect of share repurchase on the value of the firm. J. Finance 23, 135–149. Fenn, George, Liang, Nellie, 2001. Corporate payout policy and managerial stock incentives. J. Financ. Econ. 60, 45–72. Gaver, Jennifer J., Gaver, Kenneth M., 1993. Additional evidence on the association between the investment opportunity set and corporate financing, dividend and compensation policies. J. Account. Econ. 16, 125–160. Ghosh, Chimnoy, Harding, John, Sezer, Ozcan, Sirmans, C.F., 2008. The role of executive stock options in REIT repurchases. J. Real Estate Res. 30, 27–44. Grullon, Gustavo, Michaely, Roni, 2002. Dividends, share repurchases and the substitution hypothesis. J. Finance 57, 1649–1684. Grullon, Gustavo, Michaely, Roni, 2004. The informational content of share repurchase programs. J. Finance 59, 656–680.

38

W.I. Boudry et al. / Journal of Corporate Finance 23 (2013) 23–38

Guay, Wayne, Harford, Jarrad, 2000. The cash-flow permanence and information content of dividend increases versus repurchases. J. Financ. Econ. 57, 385–415. Hardin, William, Hill, Matthew, 2008. REIT determinants: excess dividends and capital markets. Real Estate Econ. 36, 349–369. Harford, Jarrad, Mansi, Sattar A., Maxwell, William F., 2008. Corporate governance and firm cash holdings in the US. J. Financ. Econ. 87, 535–555. Hennessy, Christopher, 2004. Tobin's Q, debt overhang and investment. J. Finance 59, 1717–1742. Ikenberry, David, Lakonishok, Josef, Vermaelen, Theo, 1995. Market underreaction to open market share repurchases. J. Financ. Econ. 39, 181–208. Ikenberry, David, Lakonishok, Josef, Vermaelen, Theo, 2000. Share repurchases in Canada: performance and strategic trading. J. Finance 55, 2373–2397. Jagannathan, Murali, Stephens, Cliford, 2003. Motives for multiple open-market repurchase programs. Financ. Manag. 71–91. Jagannathan, Murali, Stephens, Cliford, Weisbach, Michael, 2000. Financial flexibility and the choice between dividends and stock repurchases. J. Financ. Econ. 57, 355–384. Jensen, Michael, 1986. Agency costs of free cash flow, corporate finance, and takeovers. Am. Econ. Rev. 76, 323–329. Julio, Brandon, Ikenberry, David, 2004. Reappearing dividends. J. Appl. Corp. Finance 16, 89–100. Kahle, Kathleen M., 2002. When a buyback isn't a buyback: open market repurchases and employee options. J. Financ. Econ. 63, 235–261. Kim, Jaemin, Schremper, Ralf, Varaiya, Nikhil, 2005. Survey on open market repurchase regulations: cross-country examination of the ten largest stock markets. Working Paper. San Diego State University. Lambert, Richard A., Lanen, William N., Larker, David F., 1989. Executive stock option plans and corporate dividend policy. J. Financ. Quant. Anal. 24, 409–425. Lie, Erik, 2005. Operating performance following open market share repurchase announcements. J. Account. Econ. 39, 411–436. Miller, Merton, Rock, Kevin, 1985. Dividend policy under asymmetric information. J. Finance 40, 1031–1051. Myers, Stewart, 1977. Determinants of corporate borrowing. J. Financ. Econ. 5, 147–175. Nohel, Tom, Tarhan, Vefa, 1998. Share repurchases and firm performance: new evidence on the agency cost of free cash flow. J. Financ. Econ. 49, 187–222. Ofer, Aharon R., Thakor, Anjan V., 1987. A theory of stock price responses to alternative corporate cash disbursement methods: stock repurchases and dividends. J. Finance 42, 365–394. Opler, Tim, Titman, Sheridan, 1993. The determinants of leveraged buyout activity: free cash flow versus financial distress costs. J. Finance 48, 1985–1999. Panovka, Robin, 2007. REITs face a new reality. knowledge.wprarey.asu.edu/artirle.rfm?artirleid=1560. Peyer, Urs, Vermaelen, Theo, 2009. The nature and persistence of buyback anomalies. Rev. Financ. Stud. 22, 1693–1745. Plazzi, Alberto, Torous, Walter, Valkanov, Rossen, 2010. Expected returns and expected growth in rents of commercial real estate. Rev. Financ. Stud. 23, 3469–3519. Rau, P. Raghavendra, Vermaelen, Theo, 2002. Regulation, taxes and share repurchase in the U.K. J. Bus. 75, 245–282. Smith, Clifford, Watts, Ross L., 1992. The investment opportunity set and corporate financing, dividends and compensation policies. J. Financ. Econ. 32, 263–292. Vermaelen, Theo, 1981. Common stock repurchases and market signaling: an empirical study. J. Financ. Econ. 9, 138–183. Vermaelen, Theo, 1984. Repurchase tender offers, signaling and managerial incentives. J. Financ. Quant. Anal. 19, 163–181. Wansley, James, Lane, William, Sarkar, Salil, 1989. Managements' view on share repurchase and tender offer premiums. Financ. Manag. 18, 97–110. Weisbenner, Scott J., 2000. Corporate share repurchases in the 1990s: what role do stock options play? Federal Reserve Board working paper. www.v3.ro.uk/vnunet/news/2210896/ibm-sets-aside-15bn-stock-repurchase (Feb.29. 2008).