Discounting and Underpricing of REIT Seasoned Equity Offers

Discounting and Underpricing of REIT Seasoned Equity Offers Author Kimberly R. Goodwin Abstract For seasoned equity offerings, the discounting of the offer price from the closing price on the previous day is influenced by the level of asymmetric information surrounding the firm and the offering, as well as compensation to investors willing to purchase new shares. Discounting is important for the offering firm because it represents money left on the table, and this is the first paper to examine the degree of discounting and its determinants for real estate investment trusts (REITs). Studying the discounting of REIT seasoned equity offers from 1994 to 2006, this paper provides support for both the placement cost and value uncertainty hypotheses of discounting. When new REIT shares are more difficult to place with investors and there is less consensus about the valuation of REIT shares, investors will require greater discounting of the offer price. Discounting is a measure of the difference between the closing price of a particular stock on the day prior to a seasoned equity offer (SEO) and the offer price. Since we do not have a current market price available for initial public offering (IPO) firms, we are only able to calculate the degree of underpricing and not the degree of discounting. The amount of discounting in SEOs is important because it reflects the amount of potential equity capital the issuing firm does not receive. To date there are no existing studies looking at the phenomenon of real estate investment trust (REIT) discounting and no explanation offered for its possible determinants. REITs have several unique characteristics that are of particular interest when studying discounting and underpricing. First, REITs have to pay out a large portion of earning as dividends and must frequently access the capital markets to fund firm growth. Second, most REITs formed since the 1990s utilize the Umbrella Partnership REIT (UPREIT) organizational structure. Third, REITs are almost solely composed of real properties that are difficult for investors to value. Using data from 1994 to 2006, this is the first paper to study the discounting of REIT SEOs. Furthermore, this is among the first papers to test how sources of asymmetric information in REITs relate to the pricing of SEOs. One explanation for the phenomenon of SEO discounting is that it is a response to a downward-sloping demand curve. In this scenario, the market price must fall JRER Vol. 35 N o. 2 2 0 1 3

154 Goodwin to compensate for the increased supply of public shares. The problem with this explanation is that information about the supply of new shares in the market should already be incorporated in the share price. When the SEO announcement is made, the number of shares and filing range are also announced. Market efficiency suggests that investors will anticipate price pressure effects related to the shift in supply and adjust their demand accordingly. On average, the market should be able to account for the change in share price required to absorb the new shares. As a result, we would expect to see that on average, the offer price should equal the closing price on the previous trading day. Past evidence of price effects on the actual announcement date are mixed (Mikkelson and Partch, 1985; Asquith and Mullins, 1986; and Masulis and Korwar, 1986). Consistent with this view, however, Mikkelson and Partch (1985) find no evidence of a significant relation between issue size and price effects around secondary distributions. It is also possible that SEO discounting arises from a level of uncertainty about firm value resulting from asymmetric information problems. The degree of information asymmetry and its effect on equity offerings has not been studied in depth in the REIT literature. Since REITs are required to pay out such a large proportion of earnings as dividends, they are unable to finance expansion and acquisitions through retained earnings. As a result, REITs must access the capital markets frequently compared to industrial firms. To the extent that more frequent equity offerings result in more public information about the firm being generated, REITs should have a lower level of information asymmetry compared to industrial firms. If discounting is the result of uncertainty and asymmetric information, we would expect to see that REITs have lower levels of SEO discounting compared to industrial firms. In addition, holding other factors constant, REITs that access the equity markets more frequently should have lower levels of information asymmetry and lower discounting compared to REITs that seldom access the equity markets. It has been hypothesized that the UPREIT structure and investment in real property may make REITs more likely to suffer from asymmetric information problems compared to industrial firms. In an UPREIT there is a managing partner who may manage assets under several smaller REIT organizations. This structure may induce information asymmetry problems in that the manager s ability and performance are not easily observed, and he may be motivated to act in a way that is most beneficial to him rather than the REIT shareholders. In addition, REITs are composed of real properties that are difficult for shareholders to value. As a result, managers may have private information about the value of the REIT assets that the shareholders do not have. To the extent that REITs hold properties that are of a type or in a region with more valuation uncertainty, the asymmetric information problem is exacerbated. For the average REIT SEO, discounting represented nearly $2 million in lost proceeds and is a substantial cost to issuing firms. The amount of discounting is especially significant given the fact that secondary market pricing is easily observable. 1 This paper provides some indication of how lead banks price SEOs

REIT Seasoned Equity Offers 155 in REITs and indicate the potential pricing impact of asymmetric information. Additionally, the paper illustrates that there are different forces at work determining the level of underpricing and the level of discounting. Issue-specific information significantly determines the level of discounting and indicates that the markets are efficiently incorporating information into the SEO price. Since those factors are no longer significant in the underpricing models, there is an indication that REIT equity markets are efficient and price changes after the distribution of new shares are only capturing new market and firm relevant information. The results provide a justification for making sure future studies in SEOs are specifying the proper variable of interest. This paper contributes to the literature in three main ways. First, this is the first paper to study the discounting of REIT SEOs. Second, utilizing REIT SEO data provides an interesting opportunity to examine the pricing effects of the unique UPREIT structure, as well as the effects of frequent capital offerings without introducing industry-specific variation. Third, the results have significant implications to the REIT industry. The current crisis in the credit and real estate markets has led to a restructuring of REIT balance sheets. Unable to issue debt at a reasonable cost, REITs have been turning to the equity markets far more than industrial firms. In 2009, REITs engaged in 87 SEOs totaling over $21.2 billion in raised capital. As this trend is expected to continue, a better understanding of these pricing dynamics is essential for both managers and investors. The following section contains a review of the literature and an explanation of various hypotheses on SEO discounting. The data and methodology are described next, followed by a presentation of the empirical results and a conclusion. Background and Influences Evidence from Industrial Firms While there is an extensive literature addressing the topic of IPO underpricing, far less is known about the topic of SEO underpricing or discounting. Smith (1977) was the first to report on significant SEO underpricing of industrial firms. Bhagat and Frost (1986) later found that regulated public utilities received a SEO premium of between 0.25% and 0.65%. Eckbo and Masulis (1992) further document the phenomenon of SEO underpricing in industrial firms and overpricing in regulated utilities. More recently in the finance literature, Altinkilic and Hansen (2003) and Corwin (2003) examine discounting in SEOs. Discounting reflects the uncertainty associated with firm value and is consistent with asymmetric information models of underpricing. There is also evidence that expected discounting covers placement costs and is a method of payment for positive information about issuers. REITs are excluded from these samples as a regulated financial entity, and there have been no papers specifically addressing the issue of expected discounting in REITs. JRER Vol. 35 N o. 2 2 0 1 3

156 Goodwin Evidence from REITs To date there have been a few papers examining SEOs by REITs. Howe and Shilling (1988) examine SEOs in the context of optimal capital structure theory for REITs. They find a significantly negative stock price reaction in response to the announcement of a SEO using data for 1970 1985. Ghosh, Nag, and Sirmans (1999) corroborate Howe and Shilling s findings using data for 1991 1995. Ghosh, Nag, and Sirmans also find a significantly negative return on the offer day that persists even after adjustment for the bid-ask bounce induced by institutional investors selling shares in the secondary market to take advantage of offer price discounts. They use firm size, underwriter rank, and trading volume to explain abnormal return on the SEO offering day. There is also a link between the SEO underpricing and IPO underpricing, as well as between the SEO and long-run performance. Ghosh, Nag, and Sirmans (2000a) also look at the signaling link between IPO-SEO pairs for REITs for 1992 1996 and find results that are consistent with signaling behavior. They find that the length of time from IPO to SEO decreases as IPO underpricing increases. They also find that firms with higher IPO underpricing likewise have higher SEO underpricing. Ghosh, Nag, and Sirmans (1997) look at the stock market performance of EREITs involved in SEOs for 1992 1995, specifically three months before and six months after a SEO. They find that REITs involved in a SEO outperform the Lehman REIT index but not necessarily the Lehman property type index. The authors propose that the result is due to the management success of the SEO REITs and indicates a window of opportunity for REIT managers. If markets have a negative reaction to SEOs and there is evidence that firms making SEOs underperform in the long run, managers should pay close attention to these factors when choosing to issue equity. A better understanding of the pricing of REIT SEOs is essential to managerial decision making. Ghosh, Nag, and Sirmans (2000b) examine SEO underpricing in REITs for 1991 1996 and find significant underpricing of 0.41% 0.89% that is dependent upon institutional ownership, issue size, and underwriter reputation. Ghosh, Nag, and Sirmans (2000b) actually use discounting as one measure of SEO underpricing but never utilize the variable in their reported regression analysis. They essentially end their analysis when they find no difference between pre- and post-1990 IPO REITs. Whether the IPO took place before or after 1990 could be a poor proxy for information asymmetry in the SEO market, and thus it is worthwhile to examine expected discounting and its determinants with a more complete methodology. Hypotheses Many theories exist to explain the underpricing of IPOs based on the degree of uncertainty and the information asymmetry between the parties involved in the transaction. According to Loderer, Sheehan, and Kadlec (1991), these theories can

REIT Seasoned Equity Offers 157 be extended to a SEO framework. One model (Rock, 1986) explains underpricing as necessary compensation for uninformed investors participation in new issues. If informed investors only participate in good issues, uninformed investors will be left with a high proportion of bad issues and will eventually refuse to participate in the new issues market. Underpricing (or discounting of SEOs) is compensation to induce future participation by uninformed investors. Beatty and Ritter (1986) show that the winner s curse problem results in a positive relationship between value uncertainty and underpricing. Chan, Wang, and Yang (2009) present a novel approach to the IPO underpricing puzzle that may be particularly relevant to REIT equity offers. Their model includes parameters for the deadweight cost if an issuance fails, the additional information search costs for an investor if alternative investments are investigated, and the likelihood that investors can get consensus opinion on firm value. The current paper extends the theories of underpricing to the discounting of REIT SEOs. In particular, it examines the evidence for three hypotheses linking uncertainty and information asymmetry. Placement Cost Hypothesis. In the placement cost hypothesis, the amount of discounting is related to the ease with which the underwriters can place the new shares. When marketing a SEO, underwriters solicit potential investors, who provide feedback in the form of more or less explicit, but always non-binding, indications of interest. On the basis of these indications of interest and state of the market, the investment bank proposes an offer price to the company. Once priced, investors are asked to confirm their indications of interest, and shares are allocated. Under the placement cost hypothesis, SEOs with larger proceeds and a higher relative amount of new shares would be more difficult to place. The larger the number of shares being offered, the more likely they will end up in the hands of uninformed investors with higher informational costs. All else equal, the larger the offering (measured by proceeds and relative amount of new shares), the higher the expected amount of discounting. Value Uncertainty Hypothesis. In the value uncertainty hypothesis, the amount of discounting is related to the level of uncertainty about the true value of the firm. In contrast to the placement cost hypothesis, larger issues (measured by total proceeds) would have lower asymmetric information and lower discounting since they are likely related to larger firms with more information available. As the volatility of the holding period return increases, there will be a higher level of value uncertainty and higher discounting. To the extent that stocks with lower prices are less likely to have the interest of institutional investors and analysts, they will be associated with higher levels of asymmetric information. So, an increase in 1/stock price is associated with higher levels of uncertainty and greater discounting. In addition, to the extent that the UPREIT structure does introduce the possibility for agency problems and increased asymmetric information between the manager and investors, being organized as an UPREIT should be associated with more value uncertainty and higher levels of discounting. It is also possible that specific property types may have differing levels of risk and uncertainty associated with their valuation. JRER Vol. 35 N o. 2 2 0 1 3

158 Goodwin Underwriter Certification Hypothesis. Underwriters act as an intermediary between the issuer and the investor and attempt to mitigate some of the uncertainty and information asymmetry involved in issuing new equity. Appropriate pricing is also important from the underwriter s perspective. Nanda and Yun (1997) find that overpricing in IPO markets leads to a decrease in the lead underwriter s own stock market value, whereas moderate levels of underpricing are associated with an increase in stock market value. Dunbar (2000) finds that banks subsequently lose IPO market share if they either underprice or overprice too much. To the extent that higher ranked underwriters perform this role better, we would expect to see that higher ranked underwriters would be associated with lower levels of discounting. There is some element of selection bias in the matching of underwriter and issuer. Not only do issuers choose the underwriter, but the underwriter also chooses the issuer. Higher ranked underwriters will likely choose only to work with issuers who are larger, of higher quality, and have the potential to bring in more underwriting business in the future. Thus, the underwriter certification hypothesis asserts that higher ranked underwriters should be associated with less asymmetric information and lower discounting. Data Data on REIT SEOs (SIC code 6798) was collected from the SDC New Issues Database from 1994 to 2006. The initial sample of REIT SEOs contained 919 observations. Exhibit 1 lists the number of SEOs per year for REITs for 1990 2006. Removing rights offers, warrants, and unit offerings left a sample of 913 observations. Daily stock returns and shares outstanding come from CRSP. Information of REIT property type and UPREIT structure comes from the SNL REIT database and NAREIT. The SEOs collected from SDC form the initial data set for the study. After removing 188 observations for which CRSP did not contain daily stock return data and one observation where there was a stock split on the same day as the SEO, the sample contains 725 observations. We removed an additional 28 observations from the sample whose REIT classification was that of a hybrid REIT (containing both equity and mortgage components) or could not be classified through NAREIT. The final sample contains 697 SEOs by equity and mortgage REITs. Lease, Masulis, and Page (1991), Eckbo and Masulis (1992), and Safieddine and Wilhelm (1996) documented that the stated offer dates are incorrect and often inappropriate for analysis since some offers take place after the close of trading. Safieddine and Wilhelm suggest a volume-based offer date correction, which is used to correct the offer date listed in SDC. Specifically, if trading volume on the day following the SDC listed offer date is (1) more than twice the trading volume on the SDC offer date and (2) more than twice the average daily trading volume over the previous 250 trading days, then the day following the SDC offer date is designated as the actual offer date. This volume-based correction results in an offer date change for 52% of the sample.

REIT Seasoned Equity Offers 159 Exhibit 1 Seasoned Equity Offerings from all REITs listed in SDC Year REIT SEOs 1990 6 1991 8 1992 16 1993 34 1994 33 1995 57 1996 88 1997 154 1998 156 1999 10 2000 8 2001 66 2002 57 2003 80 2004 74 2005 55 2006 87 Notes: Exhibit 1 lists the number of seasoned equity offerings from all REITs listed in SDC for 1990 2006. Rights offers, unit offerings, and offer with warrants are excluded from the sample. Methodology Following the identity defined by Altinkilic and Hansen (2003), discounting is defined as the ratio of the closing price on the day before the offer to the offer price. Underpricing is defined as the ratio of the closing price on the offer day to the offer price. Offer day rate of return is defined as the ratio of the offer day closing price to the previous day s closing price. Discounting and underpricing are both included as independent variables. While Ghosh, Nag, and Sirmans (200b) reported some factors impacting the level of underpricing, this study examines both variables to establish whether they may be capturing distinct events influenced by separate factors. Discounting is then regressed on a number of explanatory variables, including property type, amount of the SEO (offer proceeds), relative number of shares offered, number of seasoned equity issues by the firm in the previous one year, 1/stock price five days prior to the SEO date, volatility, and underwriter reputation. The relative number of shares is defined as the total number of shares JRER Vol. 35 N o. 2 2 0 1 3

160 Goodwin Exhibit 2 Summary Statistics for Key Variables Mean Median Std. Dev. Min. Max. Proceeds ($ million) 100.8943 64.0993 115.6874 3.7791 1183.9163 Relative amount 0.2729 0.1071 2.1449 0.0018 56.2207 Underwriter rank 8.0293 9.1000 1.6168 0.0000 9.1000 Volatility 0.0133 0.0119 0.0058 0.0062 0.0543 Issues in past 1 year 0.8752 0.0000 1.2527 0.0000 8.0000 Offer price 26.7947 25.2500 13.8843 5.5000 134.5000 Discount (%) 1.7703 1.1929 2.2868 7.5349 19.0104 Underpricing (%) 1.2066 0.4454 2.4858 7.2455 18.7516 Return (%) 0.5637 0.4684 2.0411 9.0684 7.4627 Notes: Summary statistics for key variables from 1994 to 2006. Proceeds are expressed in millions of dollars. Relative proceeds are number of shares being offering relative to the total shares outstanding before the new issue. Volatility is the standard deviation of holding period return during 30 previous trading days ending 11 days before the offer date. Underwriter ranking is the Carter-Manaster ranking obtained from Jay Ritter s website. offered relative to the total number of shares outstanding prior to the SEO. The number of issues in the previous one year is a count variable defining the number of SEOs the firm has made in the past 12 months. Volatility is defined as the standard deviation of return for 30 trading days ending 11 trading days prior to the offering date. Underwriter reputation is defined using the data updated Carter- Manaster rankings obtained from Jay Ritter s website. Where there was more than one lead underwriter listed in SDC for a particular offering, the highest ranked underwriter is chosen. Where the underwriter rank is not listed in Jay Ritter s comprehensive list, a ranking of 0 is assigned. The general condition of the equity markets and investor sentiment are proxied by the CRSP market return in the previous month. Firm size is controlled by using the total assets in the previous year as reported in Compustat. This may be a noisy indication of firm size due to the fact that real estate assets are recorded at book value rather than current market value. Not all variables are available for each of the 697 offerings, but every observation with complete data is included in the regression analysis. Summary statistics of important variables are listed in Exhibit 2. Exhibit 3 illustrates the frequency level of different rates of underpricing and discounting for REIT SEOs from 1994 to 2006. The figures indicate that both discounting and underpricing may be censored near zero. Exhibit 4 shows that during the 1994 1997 period, discounting was highly concentrated at 0%. This practice, however, started to change in the late 1990s as the use of positive discounting increases. An increase in the rate of discounting is also reported in

Exhibit 3 Frequency Distributions of Rates of Discounting and Underpricing Panel A: Rate of Discounting Frequency Distribution of Discounting: 1994-2006 225 JRER Vol. 35 N o. 2 2 0 1 3 200 175 150 125 100 75 50 25 0-3 -2.5-2 -1.5-1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 More Discounting (%) REIT Seasoned Equity Offers 161

Panel B: Rate of Underpricing Exhibit 3 (continued) Frequency Distributions of Rates of Discounting and Underpricing 162 Goodwin Frequency Distribution of Underpricing: 1994-2006 180 160 140 120 100 80 60 40 20 0-3 -2.5-2 -1.5-1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 More Underpricing (%)

REIT Seasoned Equity Offers 163 Exhibit 4 Summary Statistics for the Level of Discounting by Year 1994 2006 1994 1995 1996 1997 1998 1999 Mean 1.77 0.40 0.56 0.83 1.22 3.48 2.77 Std. Error 0.09 0.19 0.18 0.20 0.19 0.24 1.07 Median 1.20 0 0 0 0.17 4.66 2.77 Std. Dev. 2.29 0.83 1.11 1.50 2.06 2.64 2.61 Relative Kurtosis 5.60 4.77 8.32 2.58 10.71 1.41 2.98 Skewness 1.31 2.25 2.75 1.70 2.60 0.39 0.04 Range 26.55 2.99 5.10 6.90 15.33 9.66 5.78 Min. 7.53 0 0 1.54 1.73 1.74 0 Max. 19.01 2.99 5.10 5.36 13.61 7.91 5.77 Count 698 19 38 58 114 121 6 # of unique issuers 18 34 45 71 52 6 # of multiple issuers 1 4 9 28 27 0 t-test 20.42 2.10 3.11 4.21 6.33 14.50 2.60 2000 2001 2002 2003 2004 2005 2006 Mean 2.59 1.24 1.99 2.4 1.68 1.3 1.44 Std. Error 3.14 0.36 0.34 0.23 0.16 0.18 0.19 Median 0 0.23 1.43 2.22 1.98 1.26 1.51 Std. Dev. 8.30 2.57 2.3 1.82 1.25 1.25 1.57 Relative Kurtosis 2.82 1.73 1.22 2.52 0.20 0.80 4.97 Skewness 1.34 0.51 0.87 1.33 0.39 0.33 0.47 Range 26.55 14.37 12.35 8.34 5.37 5.13 11.33 Min. 7.53 5.87 2.73 0 0.22 0.90 5.16 Max. 19.01 8.50 9.61 8.34 5.15 4.23 6.18 Count 7 50 46 62 63 49 65 # of unique issuers 6 33 37 48 45 42 46 # of multiple issuers 1 11 7 13 15 6 13 t-test 0.83 3.41 5.87 10.38 10.67 7.28 7.39 Notes: Exhibit 4 reports summary statistics for the level of discounting by year. Relative kurtosis defines the distribution with respect to the normal distribution and is negative when relatively flat and positive when relatively peaked. Number of unique issuers is the number of individual REITs that made a SEO in the calendar year. Number of multiple issuers is the number of REITs that made more than one seasoned equity offer in the calendar year. The t-test is the result from the test that the sample mean is significantly different from zero. JRER Vol. 35 N o. 2 2 0 1 3

164 Goodwin Exhibit 5 Discounting, Underpricing, and Offer-day Return Panel A: REIT sample Underpricing Discounting Offer-day Return Mean Median Mean Median Mean Median REITs 1994 2006 1.21 0.45 1.77 1.19 0.56 0.47 Panel B: REITs Ghosh, Nag, and Sirmans (1998) 1.06 0.94 Ghosh, Nag, and Sirmans (2000) 0.74 1.14 a 0.44 Panel C: Industrial firms Altinkilic and Hansen NYSE/AMEX offers 1.78 0.80 1.47 0.53 0.23 0.00 NASDAQ/NYSE/AMEX offers 2.58 1.28 2.47 1.52 0.03 0.00 Safieddine and Wilhelm 0.45 0.54 0.69 Yeoman 0.27 0.45 0.29 Notes: Exhibit 5 compares discounting, underpricing, and offer-day return for a select number of studies during similar time periods as this study. Ghosh, Nag, and Sirmans (1998) look at EREIT SEOs during 1997. Ghosh, Nag, and Sirmans (2000) use REIT data from 1991 1996. Altinkilic and Hansen (2003) look at industrial firms from 1990 1997. Safieddine and Wilhelm (1996) look at industrial firms from 1988 1991. Yeoman (2001) looks at industrial firms, utilities, and financial firms from 1998 1993. a This variable would be positive according to the definition of discounting used in this paper but is negative based on the definition by Ghosh, Nag, and Sirmans (2000b). industrial SEOs. The time-varying nature of the dependent variable and the possible left censoring require the use of year dummy variables, as well as both OLS and Tobit regression models. Correcting for any clustering effect of equity offerings did not change the significance of the results and are thus not included. Exhibit 4 also reports the number of unique REITs making SEOs in a calendar year, as well as the number of REITs who are multiple seasoned equity issuers in a year. There seems to be more instances of multiple offerings in 1997 and 1998, when there was a hot market for overall equity issuances. Results In Exhibit 5, the rate of discounting is compared to prior findings for REITs and industrial firms. The reported values for discounting and underpricing are higher than the values reported by Ghosh, Nag, and Sirmans (2000a). Underpricing is

REIT Seasoned Equity Offers 165 lower and discounting is in the range that Altinkilic and Hansen (2003) found for industrial firms within this same time period. Similarly, Dolvin and Pyles (2009) find that REITs engaging in IPOs between 1986 and 2004 also have lower underpricing and less price revision than industrial firms. 2 The results indicate that whatever factors are driving the level of underpricing and discounting are similar but may not be as strong in REITs as in industrial firms. The regressions explaining the rate of discounting are reported in Exhibit 6. 3 There is a concern that REITs investing in some property types may be more likely to choose the UPREIT structure than others. As a result, the UPREIT structure and property type are controlled for in separate regressions. The years 1994 1997 are significant and may be reflecting the higher proportion of offerings that were discounted exactly 0% during that period. There is evidence for the value uncertainty hypothesis and placement cost hypothesis in that the amount of proceeds is negative and significant. REITs engaging in larger equity issues tend to be larger and have less uncertainty and asymmetric information problems. They also tend to be easier to place in the market. This result remains significant even after controlling for firm size. Firm size can be related to some of the other variables in the model and thus models with and without firm size are included. To the extent that the UPREIT structure leads to more information asymmetries, they are more difficult to correctly value. As a result, the UPREIT structure reflects more value uncertainty and the equity offerings will experience a higher rate of discounting. As expected, the results show that the greater the number of times a firm has issued seasoned equity in the past year, the lower the rate of discounting. Firms who are frequent issuers have more information about them available in the market and have lower value uncertainty. Consistent with the placement cost hypothesis, increasing the relative amount of shares offered increases the level of discounting. The placement cost theory asserts that a higher level of discounting would be needed to induce investors to take the newly issued shares. This effect, however, is no longer significant when 1/stock price 2 is added to the model. This term accounts for the fact that the demand curve for stocks of different prices is considered to be u-shaped. Thus, it may be possible that the demand for stocks further away from the average price of $25 negates any pricing effect related to the number of new shares. In addition, discounting is higher when the market return is higher and indicates a high level of investor sentiment and the probability of a larger number of uninformed investors in the market. Discounting may be higher during these time periods as a placement cost to compensate informed investors for active participation in the issue. The regressions explaining the rate of underpricing are reported in Exhibit 7. 4 Consistent with the value uncertainty hypothesis, the UPREIT structure results in a higher rate of underpricing. There is some evidence that a higher number of seasoned equity issues in the past year are associated with less value uncertainty and results in a lower rate of underpricing. To the extent that more reputable underwriters are more skilled at valuing firms, higher ranked underwriters are JRER Vol. 35 N o. 2 2 0 1 3

166 Goodwin Exhibit 6 OLS Regression Results Explaining the Level of Discounting 1 2 3 4 5 6 Intercept 3.283** 3.138** 3.334** 1.784 1.861 2.154 2.38 2.27 2.49 0.61 1.32 1.56 Amount of proceeds 0.002* 0.002* 0.002* 0.002* 0.002* 0.002* 1.78 1.75 1.95 1.85 1.84 1.95 Relative amount 0.103** 0.095** 0.113 0.128** 0.125** 0.040 3.45 3.15 0.85 4.76 4.63 0.30 1/stock price 0.403 0.544 18.704 2.184 1.864 17.458 0.08 0.11 1.59 0.52 0.44 1.50 1/stock price 2 145.89* 116.964 1.69 1.34 Underwriter reputation 0.077 0.073 0.061 0.023 0.025 0.031 0.44 0.73 0.61 0.39 0.42 0.53 Volatility 9.520 12.684 9.015 3.764 2.486 6.737 0.33 0.43 0.31 0.19 0.13 0.33 Number of issues 0.151 0.153 0.174* 0.155* 0.158* 0.173** in past year 1.62 1.61 1.82 1.84 1.85 2.00 Market return 5.026* 4.679 4.976 6.281** 5.970** 6.098** 1.68 1.53 1.63 2.40 2.24 2.29 Firm size 0.000* 0.000** 0.000** 0.000** 1.95 2.32 2.38 2.69 UPREIT 0.480** 0.491** 0.527** 2.55 2.60 2.78 Mortgage 0.724 0.720 0.754 1.20 1.19 1.25 Diversified 0.642 0.605 0.553 1.06 0.99 0.91 Health 0.055 0.056 0.057 0.09 0.09 0.09 Hotel 1.000* 1.073* 1.069* 1.69 1.81 1.81 Industrial 0.921 0.862 0.813 1.49 1.40 1.33 Office 0.840 0.819 0.746 1.43 1.40 1.27 Residential 0.768 0.749 0.658 1.28 1.25 1.10 Retail 0.472 0.480 0.419 Self-Storage 1.202 1.185 1.133 1.42 1.40 1.33

REIT Seasoned Equity Offers 167 Exhibit 6 (continued) OLS Regression Results Explaining the Level of Discounting 1 2 3 4 5 6 Specialty 1.281* 1.295 1.206 1.65 1.61 1.53 1994 2.630** 2.643** 2.264** 2.066 2.215** 1.910* 2.87 2.95 2.71 0.73 2.02 1.80 1995 2.523** 2.494** 2.155** 2.076 2.188** 1.918* 2.75 2.77 2.59 0.74 2.02 1.84 1996 2.191** 2.136** 1.820** 1.713 1.803* 1.553 2.36 2.34 2.14 0.61 1.66 1.49 1997 1.671* 1.611* 1.324 1.223 1.305 1.082 1.79 1.76 1.56 0.44 1.19 1.03 2005 1.630* 1.608* 1.346 1.020 1.154 0.979 1.77 1.78 1.64 0.36 1.07 0.96 R 2 0.1992 0.2020 0.2068 0.2036 0.2058 0.2088 Notes: OLS regression results explaining the level of discounting in REITs from 1994 to 2006. Discounting is the ratio of the closing price on the day before the offer to the offer price. Amount of proceeds is the total amount of the offer proceeds expressed in millions of dollars. The relative amount is the number of shares offered divided by the total number of shares outstanding before the offering, reputation is the lead underwriter s Carter-Manaster ranking, volatility is the standard deviation of excess return for the for 30 trading days that end 11 trading days prior to the offer date. Year dummy variables are included. Number of observations: Model 1 613, Models 2 and 3 602, Model 4 697, and Models 5 and 6 686. *Significant at the 10% level. **Significant at the 5% level. associated with lower levels of underpricing. This finding supports the underwriter certification hypothesis applied to underpricing. Also, underpricing is higher when the general market return in the previous month is higher, which accounts for the role of investor sentiment in the offer day return. Overall, the differences between Exhibits 6 and 7 illustrate that there are some different forces at work in determining the level of discounting and level of underpricing in REIT SEOs. The significant explanatory variables are fairly robust and consistent across model specifications in the discounting regressions, but the significance of explanatory variables is highly dependent upon the model specification in the underpricing models. Market return, UPREIT structure, and number of issues in the past year are significant in explaining both the amount of discounting and the amount of underpricing. As in Ghosh, Nag, and Sirmans (2000b), underwriter reputation is a significant factor in determining the amount of underpricing. It does not, however, explain the amount of discounting and may JRER Vol. 35 N o. 2 2 0 1 3

168 Goodwin Exhibit 7 OLS Regression Results Explaining the Level of Underpricing 1 2 3 4 5 6 Intercept 2.205* 1.940 1.938 2.915 1.989 1.857 1.65 1.44 1.42 1.02 1.57 1.41 Amount of proceeds 0.001 0.001 0.001 0.000 0.000 0.000 1.58 1.55 1.52 0.61 0.62 0.59 Relative amount 0.004 0.005 0.003 0.004 0.006 0.069 0.18 0.21 0.02 0.16 0.23 0.57 1/stock price 1.351 0.519 0.296 4.397 4.392 11.436 0.29 0.11 0.02 1.04 1.03 1.01 1/stock price 2 1.695 52.835 0.02 0.66 Underwriter reputation 0.107 0.094 0.094 0.106 0.100 0.103 1.08 0.09 0.94 1.55 1.45 1.49 Volatility 16.542 10.893 10.850 33.330 30.793 28.872 0.46 0.30 0.30 1.25 1.14 1.05 Number of issues 0.122 0.110 0.110 0.109 0.102 0.095 in past year 1.34 1.19 1.18 1.31 1.21 1.13 Market return 5.772* 5.218 5.215 5.922** 5.436* 5.378* 1.80 1.60 1.60 1.98 1.80 1.77 Firm size 0.000 0.000 0.000 0.000 0.10 0.09 0.03 0.13 UPREIT 0.673** 0.684** 0.684** 3.10 3.13 3.13 Mortgage 0.338 0.343 0.327 0.52 0.53 0.51 Diversified 0.212 0.205 0.181 0.31 0.30 0.26 Health 0.848 0.840 0.840 1.17 1.15 1.15 Hotel 0.216 0.345 0.347 0.33 0.53 0.53 Industrial 0.502 0.470 0.493 0.73 0.68 0.71 Office 0.221 0.215 0.248 0.34 0.33 0.38 Residential 0.070 0.044 0.085 0.10 0.06 0.12 Retail 0.316 0.330 0.303 0.51 0.53 0.48 Self-Storage 1.580* 1.573* 1.597* 1.68 1.67 1.68

REIT Seasoned Equity Offers 169 Exhibit 7 (continued) OLS Regression Results Explaining the Level of Underpricing 1 2 3 4 5 6 Specialty 0.491 0.477 0.517 0.60 0.56 0.60 1995 1.615* 1.569* 1.573* 2.077 1.228 1.350 1.90 1.91 1.93 0.75 1.42 1.50 1998 1.425 1.497* 1.494* 1.144 2.010** 1.914** 1.60 1.73 1.75 0.41 2.27 2.10 R 2 0.1726 0.1706 0.1706 0.1747 0.1737 0.1742 Notes: OLS regression results explaining the level of underpricing in REITs from 1994 to 2006. Underpricing is defined as the ratio of the closing price on the offer day to the offer price. Amount of proceeds is the total amount of the offer proceeds expressed in millions of dollars. The relative amount is the number of shares offered divided by the total number of shares outstanding before the offering, reputation is the lead underwriter s Carter-Manaster ranking, volatility is the standard deviation of excess return for the for 30 trading days that end 11 trading days prior to the offer date. Year dummy variables are included. Number of observations: Model 1 613, Models 2 and 3 602, Model 4 697, and Models 5 and 6 686. *Significant at the 10% level. **Significant at the 5% level. be attributed to the role that underwriters play in price support following an equity offering. Interestingly, issue-specific characteristics such as the size of the issue (in dollar proceeds and relative amount of shares being offered) play a significant role in determining the amount of discounting but not the amount of underpricing. These are factors that REIT managers can control and consider in their capital budgeting decisions. The results of this study show that as number of issues in the past year increases, the level of discounting decreases. As a robustness check, the effect of shelf registration on the level of discounting and underpricing is examined. If shelf registration also acts as a proxy for frequent issuers with information available, shelf registration would be associated with lower levels of discounting and underpricing. Shelf registration decreases both the level of discounting and underpricing by about 0.5%. 5 Conclusion This paper examines the discounting of REIT SEOs from 1994 to 2006. Discounting represents a real dollar amount left on the table of nearly $2 million for the average REIT equity offering during this time period, so an understanding JRER Vol. 35 N o. 2 2 0 1 3

170 Goodwin as to what influences the rate of discounting is crucial. This is the first paper to explain the degree of discounting in REITs and the first to more extensively explain the amount of underpricing over this time period. There is evidence to support both the placement cost and value uncertainty hypotheses of discounting. New shares that are more difficult to place require higher rates of discounting. Also, firms associated with a higher degree of value uncertainty tend to require higher rates of discounting from the closing price on the day prior to the offering date. Evidence in this study supports that there is a greater amount of asymmetric information associated with the UPREIT structure, and this results in greater discounting and underpricing of SEOs. Finally, the results indicate that different factors influence the level of discounting and level of underpricing. The findings have several implications for researchers in this area, as well as for the parties involved in the SEO transaction. Past research in REIT SEOs used underpricing as the dependent variable and either did not measure discounting or stated the results were the same regardless of the definition. This study shows that the modern REIT SEO market has changed and the definition matters. The results imply that markets have become more efficient at pricing SEOs. Efficient markets should account for pricing changes due to the increased supply of shares on the announcement date, but any remaining uncertainty about valuation or placement costs is captured in the discounting of the offer price. Underpricing captures no remaining pricing effect on the offer date. This finding suggests a change in the nature of REIT capital markets that may extend to the broader financial markets. Investors committing to purchase a specified number of shares at the offer price have some expectation about how the stock price is going to react in the hours, weeks, and months following the offering. Their expectations should incorporate the fact that the markets have efficiently accounted for the number and pricing of the new shares and prices after the offer date will be determined by new firm and market information. Finally, Ooi, Ong, and Li (2010) show that REITs do exhibit market timing behavior in financing decisions, and the results of this study further indicate there may be an optimal strategy for issuing firms. Issuing a larger number of shares results in a higher placement cost and leads to higher a level of discounting, which means they need to issue a larger number of shares to raise a required level of proceeds. On the other hand, issuing equity more times in a given year results in more information production and less valuation uncertainty, so there is a lower level of discounting. While other underwriting fees are not considered for the sake of simplicity, the results imply that under certain circumstances it may be better for a REIT to have a few smaller equity issuances rather than one large issuance. Since REITs must frequently access the capital markets due to their lack of retained earnings, there may be an optimal issuance strategy that balances both direct underwriting fees and indirect costs of offer price discounting. Managerial efficiency does not only increase firm value but also reduces the cost of new investment thereby leading to higher profitability.

REIT Seasoned Equity Offers 171 Endnotes 1 An alternative characterization of the question is to consider discounting as the price differential between selling shares in the primary market and selling shares in the secondary market. In this way the discount can be viewed less as a loss of capital proceeds by the issuer and more as an issuance cost associated with placing new shares in the primary market. 2 Interestingly, however, Chan, Stohs, and Wang (2001) do not find this systematic difference in underpricing between real estate and industrial firms when examining data from Hong Kong. 3 Tobit regression results are not reported for the sake of brevity but support the same conclusions as the OLS regressions. 4 Tobit regression results are not reported for the sake of brevity but support the same conclusions as the OLS regressions. 5 Results are not reported but available upon request from the author. References Altinkilic, O. and R.S. Hansen. Discounting and Underpricing in Seasoned Equity Offers. Journal of Financial Economics, 2003, 69, 285 323. Asquith, P. and D. Mullins. Equity Issues and Offering Dilution. Journal of Financial Economics, 1986, 15, 61 89. Beatty, R.P. and J.R. Ritter. Investment Banking Reputation and the Underpricing of Initial Public Offerings. Journal of Financial Economics, 1986, 15, 213 32. Bhagat, S. and P.A. Frost. Issuing Costs to Existing Shareholders in Competitive and Negotiated Underwritten Public Utility Equity Offerings. Journal of Financial Economics, 1986, 15, 233 59. Chan, S.H., M.H. Stohs, and K. Wang. Are Real Estate IPOs a Different Species? Evidence from Hong Kong IPOs. Journal of Real Estate Research, 2001, 21, 337 56. Chan, S.H., K. Wang, and J. Yang. IPO Pricing Strategies with Deadweight and Search Costs. Journal of Real Estate Research, 2009, 31, 481 539. Corwin, S.A. The Determinants of Underpricing for Seasoned Equity Offers. Journal of Finance, 2003, 58, 2249 79. Dolvin, S.D. and M.K. Pyles. REIT IPOs and the Cost of Going Public. Journal of Real Estate Finance and Economics, 2009, 39, 92 106. Dunbar, C. Factors Affecting Investment Bank Initial Public Offering Market Share. Journal of Financial Economics, 2000, 55, 3 41. Eckbo, B.E. and R.W. Masulis. Adverse Selection and the Rights Offer Paradox. Journal of Financial Economics, 1992, 32, 293 332. Ghosh, C., R. Nag, and C.F. Sirmans. Is There A Window of Opportunity? Stock Market Performance of REITs Around Secondary Equity Offerings. Real Estate Finance, 1997, 13, 23 30.. An Analysis of Seasoned Equity Offerings by Equity REITs, 1991 to 1995. Journal of Real Estate Finance and Economics, 1999, 19, 175 92. JRER Vol. 35 N o. 2 2 0 1 3

172 Goodwin. A Test of the Signaling Value of IPO Underpricing with REIT IPO-SEO Pairs. Journal of Real Estate Finance and Economics, 2000a, 20, 137 54.. The Pricing of Seasoned Equity Offerings: Evidence from REITs. Real Estate Economics, 2000b, 28, 363 84. Howe, J.S. and J.D. Shilling. Capital Structure Theory and REIT Security Offerings. Journal of Finance, 1988, 43, 983 93. Lease, R.C., R.W. Masulis, and J.R. Page. An Investigation of Market Microstructure Impacts on Event Study Returns. Journal of Finance, 1991, 46, 1523 36. Loderer, C.F., D.P. Sheehan, and G.B. Kadlec. The Pricing of Equity Offerings. Journal of Financial Economics, 1991, 29, 35 57. Masulis, R.W. and A.N. Korwar. Seasoned Equity Offerings: An Empirical Investigation. Journal of Financial Economics, 1986, 15, 91 118. Mikkelson, W.H. and M. Partch. Stock Price Effects and Costs of Secondary Distributions. Journal of Financial Economics, 1985, 14, 165 94. Nanda, V. and Y. Yun. Reputation and Financial Intermediation: An Empirical Investigation of the Impact of IPO Mispricing on Underwriter Market Value. Journal of Financial Intermediation, 1997, 6, 39 63. Ooi, J.T.L., S. Ong, and L. Li. An Analysis of the Financing Decisions of REITs: The Role of Market Timing and Target Leverage. Journal of Real Estate Finance and Economics, 2010, 40, 130 60. Rock, K. Why New Issues are Underpriced. Journal of Financial Economics, 1986, 15, 187 212. Safieddine, A. and W.J. Wilhelm Jr. An Empirical Investigation of Short-selling Activity Prior to Seasoned Equity Offerings. Journal of Finance, 1996, 46, 729 49. Smith, C.W. Alternative Methods for Raising Capital: Rights versus Underwritten Offerings. Journal of Financial Economics, 1977, 5, 273 307. Yeoman, J.C. The Optimal Spread and Offering Price for Underwritten Securities. Journal of Financial Economics, 2001, 62, 169 98. The author would like to thank James Ligon for his guidance at the inception of this paper, as well as two anonymous reviewers and Ko Wang for their comments and suggestions that created a much improved finished product. Kimberly R. Goodwin, University of Southern Mississippi, Hattiesburg, MS 39406-0001 or kimberly.goodwin@usm.edu.