Information Transfers across Same-Sector Funds When Closed-End Funds Issue Equity

The Financial Review 37 (2002) 551--561 Information Transfers across Same-Sector Funds When Closed-End Funds Issue Equity Eric J. Higgins Kansas State University Shawn Howton Villanova University Shelly Howton Villanova University Abstract This study examines the reaction of non-issuing, same-sector funds when a closed-end fund announces a seasoned equity offering. The non-issuing, same-sector funds have a significant, negative announcement-day abnormal return. The abnormal returns for U.S. debt funds are less negative than U.S. equity and international debt funds. The abnormal returns for international debt funds are more negative than international equity funds. Announcement-day abnormal returns are directly related to the announcement-day abnormal return of the issuing fund and the premium/discount of the issuing fund. Announcement-day abnormal returns are inversely related to the premium/discount of the non-issuing, same-sector funds. Keywords: seasoned offerings, closed-end funds, intraindustry JEL Classifications: G14/G20/G23 Corresponding author: Department of Finance, College of Commerce and Finance, Villanova University, Villanova, PA 19085; Phone: (610) 519-6111; Fax: (610) 519-6881; E-mail: shelly.howton@villanova.edu 551

552 E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 1. Introduction A large body of financial research focuses on the valuation effects of a manager s decision to issue securities. Masulis and Korwar (1986) find a significant negative announcement-day return when a firm announces an equity offering. Models by Miller and Rock (1985) and Myers and Majluf (1984) are often used to explain the negative announcement-day returns. The hypothesis from the theoretical models is that the market interprets the security issuance as a signal from a firm s management of unfavorable information about the firm. Myers and Majluf suggest that the managers have information relating to firm value that shareholders do not and that their decision to issue equity is a sign that the equity trading in the market is overvalued. In an extension of the studies examining the reaction on announcements of equity offerings for industrial firms, Akhigbe and Madura (2001) and Higgins, Howton, and Howton (2002) examine the initial and long-run performance of closed-end funds at the announcement of a seasoned equity offering (SEO). These studies argue that closed-end funds are subject to fewer information asymmetries than industrial firms and should have a less negative reaction to announcements of equity issues than previously examined industrial firms. In addition, a sample of closed-end funds can be divided into groups in which managers of the funds have differing levels of asymmetric information. Akhigbe and Madura and Higgins, Howton, and Howton find evidence of both a negative announcement effect and long-run negative abnormal returns to the issuing funds. Higgins, Howton, and Howton also find that funds that they identify as having larger levels of asymmetric information have more negative abnormal returns around equity offerings and that closed-end funds in general have less negative announcement-day reactions to the equity issue than industrials. In addition to the issuing firm reaction to a security issue announcement, several papers examine the industry-wide effects on firms around announcements of equity offerings. The purpose of these papers is to examine whether the information being conveyed by the security issue is unique to the issuing firm or whether the information is transferred across same-type firms. Szewczyk (1992) performs a comprehensive examination of intraindustry effects on announcements of security issuances. He looks at announcements of SEOs and of straight and convertible debt issuances. Szewczyk finds a significant negative reaction in the industry on the announcement of an SEO. He also finds a significant industry reaction at announcement of straight and convertible debt offerings. Slovin, Sushka, and Polonchek (1992) look at equity issuance announcements for banks and industrial firms. They examine the price effects on other securities at the time of the announcement. Although they find no evidence of an intraindustry effect for the industrials, they do document a reaction to the announcements in the banking industry. The findings from both of these papers suggest that the information conveyed when a firm announces a security issue is not confined to the issuing firm. The purpose of this study is to extend the work of Szewczyk (1992), Slovin, Sushka, and Polonchek (1992), Akhigbe and Madura (2001), and Higgins, Howton, and Howton (2002) by examining the market reaction of same-sector closed-end

E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 553 funds to the announcement of an equity offering. We use fund classifications that are based on the types of securities that the fund holds and then look at the market reaction to the announcement of the offering on other funds that are classified in the same category as the announcing firm. We identify specific classifications as groups of funds that are more difficult for investors to value. Based on the Myers and Majluf (1984) argument, managers of these funds could choose to issue equity at times when their funds are overvalued in the market. Although this signal of overvaluation could be the reason for a negative reaction to the announcement of an equity offering, we hypothesize that the market could also interpret the equity issue as signaling the overvaluation of similar funds. We classify the funds in our sample according to their relative levels of information asymmetries. For example, equity and international securities are more difficult to value than debt and domestic securities, respectively. We compare the industry reactions to announcements of SEOs in the equity funds to reactions to SEOs in the debt funds and those in the international funds to those in the domestic funds to determine whether differential industry reactions occur that are related to different levels of information asymmetries. We test the hypothesis that markets view announcements of SEOs in their industry as containing more information if the funds in that area are more difficult to value. We also examine the determinants of cross-sectional differences in the announcement-day abnormal returns for non-issuing, same-sector funds. The remainder of the paper is organized as follows. Section 2 contains a description of the sample used in the paper. Section 3 describes the methodology and presents the results. Section 4 provides a summary of the paper and conclusions. 2. Data We used the Securities Data Corporation database to obtain an initial sample of SEOs made by closed-end funds that occur between January 1992 and December 1997. The categories of closed-end funds included in the sample are restricted to debt funds and equity funds. 1 The funds included hold both domestic and international securities. All funds must be listed on the NYSE, Amex, or Nasdaq and are required to have prices that appear on the Center for Research in Security Prices (CRSP) files. This sample represents a subset of the samples used in Akhigbe and Madura (2001) and Higgins, Howton, and Howton (2002). The previous studies used SEOs beginning in 1986. The Wall Street Journal (WSJ) classifications of closed-end funds began in 1992, requiring the truncation of the sample. We collect several variables related to the equity offering from the Securities Data Corporation database. For each issuing firm in our sample, we obtain the principle amount of the offering, the number of shares offered, the offer price, and the date of 1 Real Estate Investment Trusts and closed-end funds holding other types of assets besides debt or equity are not included in the initial sample.

554 E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 the offer. Daily price information for each of the issuing firms is obtained using the CRSP database. We use daily prices to find an announcement-day abnormal return. We also use CRSP to find returns for the issuing funds for the year prior to issuance. On the announcement day, we find the net asset value (NAV) of the issuing fund from the WSJ and the premium/discount to NAV for the issuing funds. We classify each of the issuing funds in our sample into groups based on the types of securities held in the fund. The funds are classified as either debt or equity funds and also as either domestic or international funds. This classification technique gives us four categories of issuing funds for our sample: domestic equity funds, international equity funds, domestic debt funds, and international debt funds. These four categories are similar to ones used by the WSJ in its listings of closed-end funds. Each of these four categories of funds make up the four types of sectors that we use to form same-sector groups of non-issuing funds to test for reactions to announcements of the equity issuance. Over the sample period used in this study, the WSJ categorizes all closed-end funds based on the main type of security held by the fund. On the announcement date for the issuing firm, we collect data for a group of same-sector funds from the WSJ. The issuing fund is classified in one of the four categories described previously. On the day that the fund makes the announcement of an SEO, we use the WSJ classification of funds to find all other funds that are classified as that type on that day. This sample of same-sector funds is our matching sample that we observe for reaction to the offering announcement. For each of these same-sector funds, we obtain the NAV on the announcement day from the WSJ and also find the returns to the same-sector funds for the three-day window using the CRSP database. Our final sample contains 63 closed-end funds that announced seasoned equity issues over the sample period. There are a total of 3,910 non-issuing, same-sector funds in the matching sample. Table 1 contains summary statistics for the issuing funds as well as the nonissuing matching samples. The summary statistics are compiled for the full sample and the four subsamples. Issuing firms for the full sample trade at a 2% premium to NAV whereas the non-issuing funds trade at a slight discount. There are on average 48 matching firms for each issuing firm with a range of between 7 and 92 matching firms. The average issue represents 1.13% of total assets represented in the matching sample. International equity funds represent over half of the sample. These funds have the highest number of matching funds, and the issues represent the smallest percentage of total assets for the matching funds of the four subsamples, on average. 3. Methodology and results 3.1. Event study methodology and results Event study methodology is used to examine the stock price reaction of firms in the same-sector fund sample when an SEO is announced. Day 0 is the day that a fund in the same sector, as defined by the WSJ, announces an SEO. Because the funds

E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 555 Table 1 Summary statistics for issuing funds and non-issuing same-type funds The table contains sample statistics for a sample of 63 closed-end funds that had seasoned equity offerings from 1992 to 1997. Sample statistics for non-issuing funds that are in the same fund category as listed in the Wall Street Journal are also included. Statistics for the full sample as well as subsamples separated by fund type are reported. The % premium/discount is measured as the difference between the net asset value per share of the fund and the funds price divided by the funds price. Panel A: Full sample (N = 63) Mean Std. Dev. Minimum Maximum Non-issuing matching funds Percent premium/discount 0.0094 0.0489 0.0997 0.0964 Average mkt. value (in 100,000s) 202.411 92.937 103.128 462.264 Issue size/average mkt. value 0.3538 0.4112 0.0163 2.0189 Total market value (in 100,000s) 9196.988 5542.432 618.765 25749.311 Issue size/total market value 0.0113 0.0186 0.0008 0.0781 Number of matching firms 48.0317 23.3990 7.0000 92.0000 Issuing funds Percent premium/discount 0.0216 0.1056 0.1585 0.2214 Issue size (in 100,000s) 65.973 90.321 7.300 574.100 Issue size/market value of issuer 0.2412 0.1371 0.4760 0.9970 Panel B: Domestic debt funds (N = 8) Non-issuing matching funds Percent premium/discount 0.0091 0.0326 0.0604 0.0526 Average mkt. value (000,000s) 213.610 90.665 103.127 343.324 Issue/average mkt. Value 0.3870 0.5470 0.0700 1.6722 Total market value (000,000s) 13608.135 12006.721 618.765 25749.311 Issue/total market value 0.0174 0.0261 0.0008 0.0773 Number of matching firms 52.0000 36.7112 7.0000 92.0000 Issuing funds Percent premium/discount 0.0448 0.0783 0.0663 0.1597 Issue size (in 100,000s) 98.100 193.015 18.100 574.100 Issue size/market value of issuer 0.2688 0.1201 0.1581 0.5019 Panel C: International debt funds (N = 7) Non-issuing matching funds Percent premium/discount 0.0471 0.0341 0.0997 0.0057 Average mkt. value (in 100,000s) 188.238 48.940 133.754 268.514 Issue/average mkt. Value 0.6270 0.6358 0.0880 1.5629 Total market value (in 100,000s) 3540.230 883.491 2675.090 4773.149 Issue/total market value 0.0346 0.0326 0.0037 0.0781 Number of matching firms 20.1429 4.4132 11.0000 25.0000 (continued) represent very specific asset types, we do not use a market model to estimate abnormal return. Instead, we use the mean adjusted returns described in Brown and Warner (1985) to measure abnormal returns. The announcement-day return is adjusted by subtracting the mean return over a 239-day pre-announcement period window ending

556 E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 Table 1 (continued) Mean Std. Dev. Minimum Maximum Issuing funds Percent premium/discount 0.0505 0.0915 0.1548 0.1111 Issue size (in 100,000s) 111.714 105.796 17.500 261.800 Issue size/market value of issuer 0.3152 0.3032 0.1432 0.9970 Panel D: Domestic equity funds (N = 11) Non-issuing matching funds Percent premium/discount 0.0328 0.0257 0.0876 0.0069 Average mkt. value (in 100,000s) 327.767 141.934 143.296 462.263 Issue/average mkt. value 0.3735 0.6399 0.0163 2.0189 Total market value (in 100,000s) 6450.617 2007.044 3341.391 8741.100 Issue/total market value 0.0137 0.0208 0.0010 0.0701 Number of matching firms 23.8182 13.5116 17.0000 63.0000 Issuing funds Percent premium/discount 0.0032 0.0742 0.1101 0.0931 Issue size (in 100,000s) 77.518 102.987 7.300 289.300 Issue size/market value of issuer 0.1649 0.1550 0.0476 0.5000 Panel E: International equity funds (N = 37) Non-issuing matching funds Percent premium/discount 0.0046 0.0538 0.0981 0.0964 Average mkt. value (in 100,000s) 165.402 28.291 117.179 231.965 Issue/average mkt. value 0.2892 0.1807 0.1009 1.0029 Total market value (in 100,000s) 10129.912 3143.008 5976.161 16933.450 Issue/total market value 0.0049 0.0028 0.0013 0.0129 Number of matching firms 59.6486 12.4236 19.0000 79.0000 Issuing funds Percent premium/discount 0.0365 0.1172 0.1585 0.2214 Issue size (in 100,000s) 46.940 30.738 18.000 180.000 Issue size/market value of issuer 0.2363 0.0541 0.0828 0.3695 5 days before the issue announcement. The event study uses a portfolio approach suggested by Szewczyk (1992) to account for the clustering of event days resulting from the use of a matching sample. The portfolio approach involves calculating an equally weighted portfolio return for the non-issuing matching firms associated with each issue. The event study analysis is then performed with the portfolio returns resulting in a sample size of 63 for the full sample analysis. Results of the event study are presented in Table 2 for the full sample as well as for subsamples defined by the issuing fund sector. In addition, event study results for the 99 issuing firms in the Higgins, Howton, and Howton (2002) study are presented for comparison purposes. Panel A of Table 2 summarizes the event study results. The full sample and each sector subsample exhibit negative abnormal returns over the announcement window and on the announcement day. Only the announcement-day return for the

E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 557 Table 2 Announcement-day abnormal returns for non-issuing, same-sector funds and issuing funds The following table contains abnormal returns for closed-end mutual funds when there is an announcement of a seasoned equity offering made by a fund in the same sector. Sectors are classified as U.S., international, equity, and debt. Panel A shows the abnormal return for both the non-issuing and issuing funds. To find non-issuing firm abnormal returns, matching sample firms are found for each original issuing firm and are combined into a portfolio. The event study is then conducted on the portfolio returns. The mean adjusted methodology is used to find abnormal returns for the non-issuing firms. Panel B makes pairwise comparisons of the abnormal returns across the different fund types for the sample of non-issuing, samesector funds. Panel A: Abnormal returns for non-issuing, same-sector funds and corresponding issuing firms Non-issuing Firms Issuing Firms Announcement-day Announcement-day Sample N abnormal return N abnormal return All 63 0.0024 99 0.0065 U.S. funds 19 0.0028 31 0.0049 International funds 44 0.0023 68 0.0071 Equity funds 48 0.0025 59 0.0073 Debt funds 15 0.0020 26 0.0041 U.S. equity funds 11 0.0036 8 0.0043 U.S. debt funds 8 0.0017 15 0.0054 International equity funds 37 0.0022 51 0.0080 International debt funds 7 0.0023 8 0.0015 Panel B: Comparison of abnormal returns across sectors for non-issuing, same-sector funds Difference between announcement-day Sector comparison abnormal returns t-statistic U.S. vs. International 0.0006 0.27 Debt vs. Equity 0.0006 0.19 U.S. debt vs. U.S. equity 0.0019 0.77 International debt vs. International equity 0.0001 0.02 U.S. equity vs. International equity 0.0014 0.51 U.S. debt vs. International debt 0.0006 0.14 Indicates statistical significance at the 0.05 level. Indicates statistical significance at the 0.10 level. international debt sector is not significant. U.S. equity funds have the most negative reaction, whereas U.S. debt funds have the least negative abnormal return in absolute terms. These results are consistent with the argument that an announcement of closedend fund SEOs conveys negative information about the values of firms in the same fund sector. The returns are between one-third and one-half of the negative abnormal returns for the issuing funds, thus the market does not appear to interpret the issuance as negatively for the matching funds as it does for the issuers. Economically, the announcement-day abnormal returns for the non-issuing firms are, in many cases, quite significant. For example, the announcement-day abnormal

558 E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 returns for international equity funds are.22% for the non-issuing firms. With an average market value of $165 million and 37 funds, the total market value is $6.1 billion. Losing 22 basis points on $6.1 billion corresponds to a loss in market value of $13 million. This is about one quarter of the average issue size of $46.9 million. Panel B of Table 2 summarizes tests of differences in abnormal returns across fund sectors. If asymmetric information between closed-end fund managers and shareholders is driving the negative valuation effects observed in Panel A, fund sectors with higher levels of information asymmetries should have more negative announcementday and announcement-window abnormal returns than funds in sectors with lower levels of information asymmetries. The lack of significant differences between all fund types does not support the information asymmetry hypothesis. These results are not consistent with the results for issuing firms reported in Higgins, Howton, and Howton (2002). 3.2. Regression methodology and results Regression methodology is used to examine the determinants of cross-sectional differences in announcement-day abnormal returns for funds in the same sector as a fund announcing an SEO. The same-sector fund announcement-day returns are used as the dependent variable, and several variables measuring characteristics of both the issuing fund and the same-sector funds are used as explanatory variables. The regression equation to be estimated takes the form: SAR i = γ 0 + γ 1 %PREM i + γ 2 ORIGAR i + γ 3 %ISSUE-TOT i + γ 4 %PREM-ORIG i + γ 5 PREADJ i + ε I (1) where SAR i is the standardized abnormal return 2 for each portfolio of non-issuing same-sector funds associated with an issue announcement, %PREM is the average premium or discount for the non-issuing, same-sector funds, ORIGAR is the announcement-day abnormal return for the issuing firm, %ISSUE-TOT is the size of the issue divided by the total market value of the non-issuing funds, %PREM-ORIG is the premium or discount for the issuing firm, PREADJ is the one-year pre-issue adjusted return 3 for the issuing firm, γ 0 γ 5 are parameters to be estimated, and ε I is an error term. %PREM measures the percentage difference between the stock price and the NAV of the same-sector funds. The coefficient estimate for %PREM should be negative because firms trading at a premium to NAV are more likely to be overvalued and thus more likely to have a negative reaction to any sector-wide information release. 2 The standardized abnormal return is equal to the abnormal return standardized by the standard error of the returns over the event period. 3 PREADJ is adjusted by returns to an equally weighted portfolio of non-issuing, same-sector funds over the period.

E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 559 Table 3 Cross-sectional determinants of abnormal returns for non-issuing, same-sector fund sample The following table contains the results of the estimation of a regression model to examine the crosssectional determinants of abnormal returns for non-issuing closed-end funds when there is an announcement of a seasoned equity offering made by a fund in the same sector. Sectors are classified as U.S., international, equity, and debt. The following variables are used to explain abnormal returns: %PREM is the average percentage premium/discount relative to net asset value for the non-issuing closed-end funds at the time of the announcement, ORIGAR is the abnormal return for the closed-end fund announcing a seasoned equity offering, %ISSUE-TOT is the ratio of the size of the closed-end fund issue to the total market value of the funds in the same sector, %PREM-ORIG is the percentage premium/discount relative to net asset value for the closed-end fund announcing a seasoned equity offering, and PREADJ is the abnormal return in the year prior to the announcement for the closed-end fund announcing a seasoned equity offering. Variable Predicted Sign of Coefficient Parameter Estimate t-statistic Intercept 0.0041 2.50 % PREM negative 0.0671 2.00 ORIGAR positive 0.2368 4.27 % ISSUE-TOT negative 0.0543 0.68 % PREM-ORIG negative 0.0330 2.81 PREADJ negative 0.0091 1.65 Indicates statistical significance at the 0.05 level. Indicates statistical significance at the 0.10 level. Note: F-test for model: 5.40 ; R 2 = 0.3749. The coefficient estimate for ORIGAR should be positive. ORIGAR measures the response of the issuing fund s stock price to the SEO announcement. The direction of this response should be directly related to the direction of the response of same-sector funds if information about over- or undervaluation of same-sector funds is being conveyed by the equity offering. %ISSUE-TOT measures the relative size of the equity issue. Larger issues might induce supply-related decreases in market values, and thus the coefficient estimate for %ISSUE-TOT is expected to be negative. %PREM-ORIG measures the premium/discount for the issuing funds. If a fund trading at a premium issues equity, this is more likely to be interpreted as a negative signal by the market and thus the coefficient estimate should be negative in equation 1. PREADJ measures the stock-price run-up for the issuing firm. If the market interprets a large pre-issue run-up as a sign of potential overvaluation, the coefficient estimate for PREADJ should be negative. Table 3 contains results for estimation of the model described in equation 1. The coefficient estimate for %PREM is negative and significant, and the coefficient estimate for ORIGAR is positive and significant as hypothesized. The coefficient estimate for %PREM-ORIG is also positive and significant, which is opposite the hypothesized sign of the coefficient. The results are consistent with the argument that the response of the samesector funds is inversely related to the premium/discount of the fund. Funds trading

560 E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 at a premium to NAV have a more negative response to the SEO announcement than firms trading at a discount. In addition, the return for the same-sector funds is directly related to the announcement-day return of the issuing funds. Funds in the same sector as the issuing fund have a more negative response to the issue announcement when the issuing firm has a negative announcement-day return. This result supports the contention that the announcement is conveying valuation information about the securities held by funds in the sector. The positive coefficient estimate for %PREM-ORIG is opposite that hypothesized. Same-sector funds have a more negative reaction to the issue when the premium/discount for the issuing fund is small. One potential explanation for this result is that the market interprets equity issues for high premium funds as a firm-specific event, and as the premium/discount becomes smaller, the information being conveyed about the sector as a whole becomes more and more negative. Managers of funds that issue equity when the premium is high could be conveying information that the market is overvaluing that specific firm. As the premium gets smaller and smaller, the managers of issuing firms could be conveying information about securities held by funds in the whole sector. 4. Conclusions Previous research finds that issuing funds have a negative reaction to the SEO announcement and that this reaction is directly related to the level of information asymmetry for the issuing fund. This study examines the stock price reaction of nonissuing, closed-end funds in the same sector as the fund that announces a SEO. If a SEO conveys negative information about the issuing fund, it is possible that funds in the same sector that invest in similar type of securities but do not issue equity might also react negatively to this announcement. The results of the study are consistent with the argument that a closed-end fund SEO contains information about other funds in the sector as well as the fund itself. Non-issuing, same-sector funds have a negative and significant average abnormal return on the announcement day and over a three-day window from the day before to the day after the announcement day. These abnormal returns are not significantly different across fund types. These results are generally inconsistent with the issuingfirm results from previous research, including Higgins, Howton, and Howton (2002), as the price reactions are unrelated to information asymmetries. The determinants of cross-sectional differences in announcement-day abnormal returns are examined using regression estimation. The announcement-day abnormal returns for non-issuing, same-sector funds are directly related to the announcementday abnormal return and the premium/discount of the issuing funds and inversely related to their own premium/discount. These results suggest that the market reaction to an announcement is in the same direction for the issuing and non-issuing funds in the sector. The information being released about same-sector funds is more negative as the premium/discount of the issuing fund gets smaller. Same-sector funds with

E. J. Higgins, S. Howton and S. Howton/The Financial Review 37 (2002) 551 561 561 a high premium are more negatively affected than same-sector funds with lower premiums/discounts. References Akhigbe, A. and J. Madura, 2001. Motivation and performance of seasoned offerings by closed-end funds, Financial Review 36, 101 122. Brown, S. and J. Warner, 1985. Using daily stock returns: The case of event studies, Journal of Financial Economics 14, 3 31. Higgins, E., S.D. Howton, and S.W. Howton, 2002. Information asymmetries in the seasoned equity offerings of closed-end funds. Forthcoming in Journal of Financial Research. Masulis, R. and A. Korwar, 1986. Seasoned equity offerings: An empirical investigation, Journal of Financial Economics 15, 91 118. Miller, M. and K. Rock, 1985. Dividend policy under asymmetric information, Journal of Finance 40, 1031 1051. Myers, S. and N. Majluf, 1984. Corporate financing and investment decisions when firms have information that investors do not have, Journal of Financial Economics 13, 187 221. Slovin, M., M. Sushka, and J. Polonchek, 1992. Information externalities of seasoned equity issues: Differences between banks and industrial firms, Journal of Financial Economics 32, 87 102. Szewczyk, S., 1992. The intra-industry transfer of information inferred from announcements of corporate security offerings, Journal of Finance 47, 1935 1945.