Downloading . . . If the document you requested does not automatically load, please click here. THE EFFECT OF EQUITY-FOR-DEBT SWAPS ON SECURITY RETURNS: SOME NEW EVIDENCE Rajiv Kalra, Kam C. Chan and Gary A. Raines Abstract This study re-examines the effects of equity-for-debt swaps on security returns. Special attention is paid to the effects of the swaps across industries and calendar years. The results indicate that the market reactions to such swaps after 1984 (Deficit Reduction Act) are not significantly different than those before 1984. Moreover, we find that although there is a decrease in alpha and increase in beta of the market model during the announcement period, as expected, beta declines in the post swap period. INTRODUCTION In this paper we examine the reaction of equity-for-debt swap announcements on security price. An equity-for-debt swap is a leverage decreasing recapitalization transaction in which a swapping firm exchanges a portion of its debt for newly issued equity. It is a pure capital structure change because approximately an equal amount of stock and debt is exchanged, while the investment policies remain unchanged. Thus, these swaps provide another opportunity to study the effects of capital structure changes, resulting from new common stock issues, on shareholder wealth and the total risk of the firm. Several previous studies (Peavy and Scott [1985], Finnerty [1985], Rogers and Owers [1985], Lys and Sivaramakrishnan [1988], and Cornett and Travlos [1989]) document a negative effect of equity-for-debt swaps on security returns around their announcements. These studies focus primarily on the effect of the swap announcements on the security returns before the implementation of the Deficit Reduction Act of 1984. The Act removed the tax-exempt status of the accounting income from such swaps. These studies, however, confine their samples to swap announcements during or prior to 1984. Scott and Martin [1975] and Bowen, Daley, and Huber [1982] show that there are considerable differences in financial structure across industries. However, there have been no studies which examine possible differences in effects across industries of equity-for-debt swaps taking place subsequent to the Act. Equity-for-debt swaps are leverage-decreasing events. Since such events have an impact on the risk structure of the company, one may expect differential effects of the swaps announcements. Moreover, earlier studies of equity-for-debt swaps implicitly assume that systematic risk of the swapping firm remained unaffected after the swap announcements. The purpose of this study is to re-examine the effects of equity-for-debt swaps on security returns. Special attention is paid to the effects of the swaps across industries and calendar years. Given the rapidly changing financial environment in the 1980s, it is likely that the impact of the equity-for-debt swaps is different across industries and after the implementation of the Deficit Reduction Act of 1984. The assumption of unchanged systematic risk around announcements is also examined. The remainder of the paper is organized as follows. Section II briefly reviews the related literature. Section III summarizes the data collection procedure. Section IV discusses the methodology. The empirical results are presented in Section V which is followed by a summary in Section VI. A BRIEF LITERATURE REVIEW The pioneer studies of Modigliani and Miller [1958, 1963] lay down the theoretical environment for studies of capital structure. Modigliani and Miller [1963] suggest that corporate debt has value to the firm in terms of the present value of the interest tax shield. Since the equity-for-debt swap reduces a firm's debt level, one would expect the swap announcement to have a negative impact on the security returns. Miller [1977] claims that debt financing does not offer any tax savings when personal income taxes are considered.1 Accordingly, a swap announcement should not have any influence on the security returns. Kraus and Litzenberger [1973], Scott [1976], and Kim [1978] provide the bankruptcy cost arguments of capital structure. They suggest that a firm will have a high bankruptcy cost if it carries a high level of debt. As a firm decreases its leverage through an equity-for-debt swap, the impact of such an announcement on security returns should be positive. Ross [1977] hypothesizes the information signaling effects of capital structure. He argues that an increase in leverage implies management's optimism about the firm's future and is a good signal. A decrease in leverage, conversely, conveys 'bad news' about the firm. Consequently, when a firm decreases its leverage through a swap, we would expect (under Ross's hypothesis) a negative effect on the security returns. In summary, according to the capital structure theories, the effect of an equity-for-debt swap is obscure. The aggregate impact of corporate tax, personal tax, bankruptcy costs, and information signaling on security returns is uncertain. Nevertheless, studies of Peavy and Scott [1985], Finnerty [1985], Rogers and Owers [1985], Lys and Sivaramakrishnan [1988], and Cornett and Travlos [1989] all document a negative effect of equity-for-debt swaps on security returns around the announcement dates. DATA COLLECTION This study uses equity-for-debt swap announcements during 1981 to 1989 in the Wall Street Journal (WSJ).2 The National News Paper Index, an on-line data base, through DIALOG Information Services, Inc., was searched to assemble the equity-for-debt swap announcements. The search provided one line abstracts of leverage reduction news published in both the eastern and the western editions of the WSJ. The eastern edition of the WSJ was then read on microfilm to eliminate newspaper articles. The dollar amount involved in each swap, wherever available, was noted at this stage. The sample was then screened and firms were retained in the sample only if: 1. The firm's stock was listed on one of the national exchanges. 2. The stock return data of the firm was available on the CRSP daily return tape. 3. The stock was trading during the study period. 4. The firm made a specific equity-for-debt announcement. 5. The swap was voluntary. 6. The announcement was not associated with a bankruptcy or a liquidation. Sample sources were screened 6 months prior to the announcements to ensure that it was the first public announcement of the event. The final sample contains 139 swap announcements by 105 firms. Overall 8 firms had 3 swap announcements each during the study period; 18 firms had 2, and 79 firms had 1 announcement each. Sample firms were then classified into industry groups. The two letter standard industry code (SIC) employed by CRSP is used to classify a firm within an industry group. Table 1 summarizes industry-wise announcements. The sample is divided into 33 different industry categories. The chemicals industry had the maximum number of swap announcements (14) during the study period. This was closely followed by the primary metals industry with 13 swap announcements. Information regarding the dollar values of all the swaps in these two categories is available. The primary metal industry had $1,112 million worth of equity-for-debt swaps, whereas chemicals had $619 million worth of swaps during the period. TABLE 1 Equity-For-Debt Swaps By Industry In 1981-1989 139 equity-for-debt swap announcements are classified according to the Standard Industry Classification (SIC). Market values of the swaps in each industry are also reported. Table 2 lists the swap announcements on an yearly basis. The maximum number of swaps (54) occurred in 1982. Incidentally, 1982 is also the year when both chemicals and primary metals had maximum number of swaps. Table 2 shows that the majority of swap announcements in this study took place during period 1981-1984. The number of swap announcements dropped significantly after 1984. It appears that the passage of the Deficit Reduction Act of 1984 made it less profitable to engage in equity-for-debt swaps. Table 3 summarizes year-wise and industry-wise swap announcements. TABLE 2 Equity-For-Debt Swaps By Year In 1981-1989 139 equity-for-debt swap announcements are classified according to the calendar years of the announcements. Market value of the swaps in each year are also reported. METHODOLOGY The following model is used to test for variation in the systematic relationship of securities to the market as a consequence of the swap announcement: Equation 1 where: p = an equal weighted portfolio of all firm in the study Rp,t = the average of the daily realized returns on securities in portfolio p in day t. Daily returns on individual securities are taken from the CRSP daily return tape D0,t = a dummy variable which is equal to 1 during event interval (t-3 to t+20), and zero elsewhere (p,0 = the intercept term for portfolio p (p,1 = shift in the intercept term for portfolio p during the event period (p,0 = the systematic risk of portfolio p (p,1 = measure of shift in the systematic risk of the portfolio during the event window Rm,t = the return on value weighted market portfolio (VWRETD) in day t, taken from the CRSP daily return tape ep,t = a random disturbance term TABLE 3 Equity-For-Debt Swaps By Industry And Year In 1981-1989 139 equity-for-debt swap announcements are classified according to the Standard Industry Classification (SIC) and calendar years. We apply the standard event study methodology to investigate the impact of swap announcements on the common stock returns. The daily return generating model proposed by Karafiath (1988) is used3 as follows: Equation 2 where: Ri,t = the realized return on security i in day t Dn,t = a dummy variable which is equal to 1 on observation n (in the event interval) and zero elsewhere (i,n = estimated coefficient on dummy variable Dn,t (i = the intercept term for security i (i = the systematic risk of security i Rm,t = the return on value weighted market portfolio in day t ei,t = a random disturbance term The 'event day' (t=0) is the newspaper date on which the swap announcement is reported. Newspapers frequently carry news that is released prior to the date of the newspaper. To allow for this, the event window contains day t=-3 through day t=+20. The use of dummy variables in the above model allows estimates of prediction errors (excess returns) for each security i, for each day t, to be directly obtained from ( coefficients. Equation 3 The standardized prediction errors for each security for each day (SPEi,t) in the event interval are then computed. Equation 4 where si,t is equal to the standard error of the excess return for security i for observation t and is computed as: Equation 5 where: = the residual variance for security i from the market model regression Xi = the number of observations in the estimation period Rm,t = the return on value weighted market index (VWRETD) for day t = the average return on the value weighted market index over Ti days used as estimation period (day -250 to day -11). The prediction errors are assumed to be normally distributed with zero mean, so the standardized prediction errors, SPEi,ts, are assumed to be student 't' distributed. The number of days used for regression is large, and therefore, SPEi,ts are approximately standard normal. Cumulative prediction errors (CPEi,Ts) for various sub-period (T1 to T2) are then computed: Equation 6 For each trading day t in the event period (day -3 to day +20), the average prediction error (APEt) is computed as: Equation 7 where Nt is the number of firms in the portfolio. APEt provides the prediction error on an equally-weighted portfolio and can be treated as a single observation in statistical analysis. The APEts are summed over various intervals in the event period to give cumulative average prediction errors (CAPETs). Equation 8 where T2 and T1 are specific days in the event window. Average standardized prediction error (ASPEt) for each day in the event window is computed as: Equation 9 Individual SPEis are assumed to be standard normal and independent under the null hypothesis of no excess returns. ASPEts, therefore, are approximately normally distributed with variance 1/Nt.4 The test statistic, Z, is then found as: Equation 10 These are the test statistics used to examine the statistical significance of daily abnormal returns associated with equity-for-debt announcements. To measure the statistical significance of abnormal returns over various intervals, we compute standardized cumulative average prediction error (SCPEi,T) as: Equation 11 Interval test statistics are then computed as:6 Equation 12 EMPIRICAL RESULTS Test results of the market model parameter shifts are presented in Table 4 Panel A. The results of the ordinary least squares regression show that there was a significant decline in alpha and a significant increase in beta during the study period (day -3 to day +20). Logic would suggest that a decrease in financial leverage will reduce systematic risk which should lead to a decrease of beta. We speculate that there was a break in the portfolio's systematic relationship with the market during the study period and that the parameter shifts observed are not permanent. Further evaluation of data before and after the swap (day -200 to day +125) is presented in Table 4 Panel B. Beta decreased to .501 by 26 days after the swap announcement. Systematic risk as a percent of total risk also decreased substantially. This further suggests that relative systematic risk, as measured by the market, is not stable during the swap announcement period. TABLE 4 Parameter Shifts During The Study Period This table shows the test statistics for tests to evaluate portfolio parameter shifts during and after the 139 equity for debt swap announcements during 1981 to 1989. Panel A Market Model Parameters And Parameter Shifts During Day -3 To Day +20 Panel B Total Risk, Systematic Risk, And Unsystematic Risk Before And After The Announcement Period (Day -2 To Day +25) The effects of swap announcements in various intervals in the event period (day -3 to day +20) are also evaluated. However, the analysis, is presented primarily for two intervals(a 2-day window (day -1 to day 0) and a 22-day window (day -1 to day +20). These two windows represent the announcement period (two-day) and the event period (22-day). Table 5 presents the summary statistics for the cumulative prediction errors (CPEs) of the entire sample in 2-day (day -1 to 0) and 22-day (day -1 to day 20) windows. The mean CPEs for both these windows are negative. Only 35% of the firms experienced a positive CPE during the 2-day announcement period while 47% firms experienced positive share price reaction to swap announcements during the 22-day period. Over the two-day period the maximum CPE was 12.9% (t value 1.18) (Energy Exchange Corp.) and the minimum was -32.9% (t value 3.58) (Instruments Systems Corp.). Over the 22-day period the maximum CPE was 53.5% (t value 31) (Cities Services Co.) and the minimum was -44.8% (t value 1.87) (Michigan General Corp.). TABLE 5 Summary Statistics For The Cumulative Prediction Errors (CPEs) Summary statistics for the 2 day (day -1 to day 0) and 22 day (day -1 to day +20) periods cumulative prediction errors (CPEs) are presented. Industry-wise analysis is examined next. However, a number of industries represented in the sample had very few equity-for-debt swap announcements. No meaningful industry-wise analysis could be done for these industries. Therefore, industries with three or fewer swaps are clubbed together into a 'miscellaneous' category. This grouping results in a total of 16 industry categories, including the miscellaneous category. Table 6 presents industry-wise minimum, maximum, and average CPEs for 2-day and 22-day windows. Panel A presents the statistics for the 2-day window while the 22-day window results are contained in Panel B. For the 2-day announcement period, both, Energy Exchange Corp., which had the maximum share price reaction, and Instruments Systems Corp., which experienced the minimum share price reaction, fall under the miscellaneous category. Stone, clay and glass industry had the maximum average CPE of 0.8%. The oil and gas extraction industry, on the other hand, had the minimum average share price reaction(CPE -3.4%. Cities Services Co., which experienced the maximum share price reaction of 53.5% (t value 3.51) for the 22-day announcement period is in the petroleum refining industry. Michigan General Corp. which experienced the minimum share price reaction of -44.8% (t value 1.87) falls under stone, clay and glass industry category. The primary metals industry had the maximum average CPE of 6.0%, while Stone, Clay and Glass industry had the minimum average share price reaction of -12.3%. Table 7 presents the year-wise cumulative prediction errors (CPEs) of the 2-day and 22-day announcement periods. The average CPEs for the two-day period for nearly all the years are negative. For the 22-day period, only half of the calendar years have negative average CPEs. TABLE 6 Summary Statistics For The Cumulative Prediction Errors (CPEs) By Industry Summary statistics of the 2 day (day -1 to day 0) and 22 day (day -1 to day +20) periods cumulative prediction errors (CPEs) by industry. Panel A: Two-Day CPEs By Industry Panel B: Twenty-Two-Day CPEs By Industry TABLE 7 Summary Statistics For The Cumulative Prediction Errors (CPEs) By Calendar Year Summary statistics for the 2 day (day -1 to day 0) and 22 day (day -1 to day +20) periods cumulative prediction errors (CPEs) by calendar year. Panel A: Two-Day CPEs Panel B: Twenty-Two-Day CPEs The cumulative average prediction errors (CAPEs) and related Z statistics for various sub-periods are shown in Table 8. Of the ten event windows examined, six of the sub-periods show significant negative security price reaction around the announcements. For others, the reactions, though negative, are not significant. Table 9 presents the daily average prediction errors (APEs), cumulative average prediction errors (CAPEs), and the corresponding Z statistics around the swap announcements. Abnormal returns on days -2, -1, and 0 were significantly negative. However, the cumulative abnormal returns were negative over the entire study period. Almost all the Z statistics for CAPEs indicate significant cumulative negative reaction of the security prices to the swap announcements. A graph of CAPEs against time visualizes these results. Figure 1 shows the negative effects of the swap announcements. The evidence suggests that the swap announcements, on average, had a negative influence on the security prices. TABLE 8 Cumulative Average Prediction Errors The table shows the mean cumulative average prediction errors (CAPEs) and corresponding Z statistics for various intervals in the event window around swap announcements. Tests For Effects Of Industry Classification And Calendar Year To evaluate whether industry classification and the incident of swap in a particular year have any influence on the magnitude and direction of effects, we compare share price reactions across industries and across years. Because we need to do multiple comparisons, ordinary t-tests are not appropriate here. We conduct two multiple comparison tests-Wallar-Duncan K-ratio t-test and Tukey's studentized range tests. Wallar-Duncan procedure assumes homoskedasticity and uncorrelated means. The observed 'F' value is used to compute critical 't' value for comparing two (or more) means. A large F-value indicates heterogeneous means. Wallar-Duncan test uses the ratio, denoted by K, of the relative seriousness of Type I to Type II errors. We use K value of 100 for a significance level of 5%. Tukey's method of multiple comparisons is exact when the sample sizes are equal. It is conservative, with a confidence of at least (1-() , when the sample sizes are not equal. Tukey's multiple range test uses studentized range, q7, to compute simultaneous multiple confidence intervals. Again we perform Wallar-Duncan and Tukey procedures for the 2-day and 22-day CPEs, both across industries and across calendar years. To conserve space, the results are not presented here. The results indicate that there is no statistically significant difference in CPEs across industries and/or across calendar years. Testing for differences in market responses by multiple comparison tests before and after the Deficit Reduction Act of 1984 may be difficult as only 11 of 139 swaps occurred after 1984. To secure robust results, we also conduct the Smith-Satterthwaite t tests, which do not require equal variance assumption, on the following hypothesis: where: The results are presented in Table 10. None of the t-statistics are statistically significant. Thus, the impacts of industry classification and the Deficit Reduction Act of 1984 do not appear to be significant. Whether or not a firm receives tax-free accounting income from the swaps does not affect the security price reactions. TABLE 9 Daily Average Prediction Errors (APEs) And Cumulative Average Prediction Errors (CAPEs) The table shows the daily average prediction errors (APEs) and related Z statistics for the average prediction errors (APEs) and cumulative average prediction errors (CAPEs) in the event window around swap announcements. TABLE 10 Smith-Satterthwaite t Tests For The Average Cumulative Prediction Errors Before And After the Deficit Reduction Act of 1984 The table shows the average cumulative average prediction errors (ACPEs) in 1981-1984 and 1985-1989 and corresponding t statistics for the differences in the APCEs for various event windows around swap announcements. FIGURE 1 CAPEs For Equity For Debt Swap For Period Day -3 To Day +20 SUMMARY Like the earlier studies, we find an overall negative share price reaction to swap announcements. However, we also find that there exists no significant difference of such negative effects across industries and calendar years. Although the Deficit Reduction Act of 1984 eliminated the tax advantage of equity-for-debt swaps, the market reactions to such swaps after 1984 are not significantly different from those before 1984. Moreover, we find that there is a significant decrease in alpha and an increase in beta during the announcement period. We also find that these shifts in the market model parameters are not permanent. In fact, as expected, we find that in the aggregate systematic risk, as measured by beta, actually declines in the post swap periods. ENDNOTES 1. Miller's argument is based on the reasoning that personal income is taxed at a higher rate than capital gains. However, the Tax Reform Act of 1986 has eliminated this preferential tax treatment of capital gains. 2. The data analysis for this study started in July 1991. Center for Research in Security Prices (CRSP) daily returns for 1990 were not available in time for 1990 swap announcements to be included in this study. 3. Binder (1985) and Thompson (1985) use similar return generating models. 4. See Brown and Warner (1985) and Kim and Schatzberg (1987). 5. Karafiath and Spencer (1989) demonstrated that the Dodd and Warner (1983) test statistic (Z) does not, under the assumptions made by Dodd and Warner, have unit normal distribution. They argue that, for a relatively large event window with a small estimation period, the tests can be biased. However, there is no significant risk of bias in the present case because the estimation period is relatively large (240 days). 6. See Hite, Owers and Rogers (1987, pp. 238-239) 7. q is the ratio of the range (maximum - minimum values) and standard deviation of the sample. REFERENCES [1] Binder, John J., "On the Use of the Multivariate Regression Model in Event Studies," Journal of Accounting Research 23, 1985, pp. 370-383. [2] Bowen, Robert M., Lane A. Daley, and Charles C. Huber, Jr., "Evidence on the Existence and Determinants of Inter-Industry Differences in Leverage," Financial Management 11, 1982, pp. 10-20. [3] Brown, Stephen J. and Jerold B. Warner, "Using Daily Stock Returns: The Case of Event Studies," Journal of Financial Economics 14, 1985, pp. 3-31. [4] Cornett, Marcia M. and Nicholas G. Travlos, "Information Effects Associated with Debt-for-Equity and Equity-for-Debt Exchange Offers," Journal of Finance 44, 1989, pp. 451-468. [5] Dodd, Peter and Jerold B. Warner, "On Corporate Governance: A Study of Proxy Contests," Journal of Financial Economics 11, 1983, pp. 401-404. [6] Finnerty, John D., "Stock-for-Debt Swaps and Shareholder Returns," Financial Management 14, 1985, pp. 5-17. [7] Hite, Gailen L., James E. Owers and Ronald C. Rogers, "The Market for Interfirm Asset Sales: Partial Sell-offs and Total Liquidations," Journal of Financial Economics 18, 1987, pp. 229-292. [8] Karafiath, Imre K., "Using Dummy Variables in the Event Methodology," Financial Review 23, 1988, pp. 351-357. [9] Karafiath, Imre and David E. Spencer, "Statistical Inference in Multi-period Event Studies: An Assessment," Working Paper, 1989, University of North Texas. [10] Kim, E. Han and John D. Schatzberg, "Voluntary Corporate Liquidations," Journal of Financial Economics 19, 1987, pp. 311-328. [11] Kim, H., "A Mean-Variance theory of Optimal Capital Structure and Corporate Debt Capacity," Journal of Finance, 1978, pp. 45-63. [12] Kraus, A. and L. Litzenberger, "A State Preference Model of Optimal Finance Leverage," Journal of Finance 28, 1973, pp. 911-921. [13] Lys, Thomas and Kondurur Sivaramakrishnan, "Earnings Expectations and Capital Restructuring: The Case of Equity-for-Debt Swaps," Journal of Accounting Research 26, 1988, pp. 273-299. [14] Miller, Merton, "Debt and Taxes," Journal of Finance 32, 1977, pp. 261-275. [15] Modigliani, Franco and Merton Miller, "The Cost of Capital, Corporation Finance and the Theory of Investment," American Economic Review 48, 1958, pp. 261-297. [16] Modigliani, Franco and Merton Miller, "Corporate Income Taxes and the Cost of Capital: A Correction," American Economic Review 53, 1963, pp. 433-443. [17] Peavy, John W. III and Jonathan A. Scott, "The Effect of Stock for Debt Swaps on Security Returns," Financial Review 20, 1985, pp. 303-327. [18] Rogers, R.C. and J.E. Owers, "Equity for Debt Exchanges," Financial Management 14, 1985, pp. 18-26. [19] Ross, S.A., "The Determination of Financial Leverage: The Incentive-Signalling Approach," Bell Journal of Economics, 1977, pp. 23-40. [20] Scott, David F., Jr and John D. Martin, "Industry Influence On Financial Structure," Financial Management 4, 1975, pp. 67-73. [21] Scott, J., "A Theory of Optimal Capital Structure," Bell Journal of Economics, 1976, pp. 33-54. [22] Thompson, Rex, "Conditioning the Return-Generating Process of Firm-Specific Events: A Discussion of Event Study Methods," Journal of Financial and Quantitative Analysis 20, 1985, pp. 151-168.