# Does an industry effect exist for leveraged buyouts?

* The process of taking a firm private (called "going
private" or "leveraged buyout") has received much
attention in the financial literature. In a leveraged buyout (LBO),
shareholders earn a substantial premium for their shares over recent
trading prices. Several studies have attempted to explain these returns
to shareholders and the motivations behind leveraged buyouts by
concentrating on both firm-specific and industry factors to explain why
LBOs occur.(1)

Theoretical factors for explaining the motivations for taking a firm private can be grouped under two competing hypotheses. The first, the firm effects hypotheses, states that firm-specific characteristics, such as management or operating inefficiency, are the primary motivation for taking a firm through a leveraged buyout. The second, called the industry effect hypothesis, states that factors in common to the firm's industry are the primary motivation for taking a firm private.

The purpose of this study is to examine the industry effect hypothesis. In Section I, we present a discussion of the theories and evidence for why leveraged buyouts occur. Section II describes the data used to test the industry effect hypothesis and Section III discusses both the methodology and the results. Finally, Section IV discusses the conclusions.

I. Theories for LBO Activity

Past research has attempted to explain the common factors that might support the industry effect hypothesis of leveraged buyout activity. One proposed factor is the potential leverage-induced tax savings involved in LBOs. Other factors are related to potential agency problems.

The tax-savings hypothesis has received considerable attention. It follows from the tax shield of interest payments on the increased debt in an LBO. Marais, Schipper, and Smith [13] present evidence that tax savings are correlated with the LBO premium. Kaplan [7] also finds that the tax benefits are a large source of wealth in leveraged buyouts.

The agency relationship between managers and shareholders has also received considerable attention. Jensen [5] proposes that leveraged buyouts reduce a special type of agency problem associated with free cash flow, i.e., "cash flow in excess of that required to fund all projects that have positive net present values" [5, p. 323]. Lehn and Poulsen [10] test the free cash flow hypothesis and find evidence consistent with shareholder gains resulting from reducing the agency costs associated with free cash flow.

The industry effect hypothesis suggests that certain industries are more likely to have LBO activity. Also, other industries, for example, those that are regulated, are less likely to have LBO activity. For there to be a concentration of LBOs in an industry, most firms within it must generate enough cash flow above their current needs to service the debt used in the LBO and have the potential to benefit from reducing agency costs. Jensen [4] points out that "some of the best examples of this have occurred in the oil, tire, and tobacco industries -- all industries that have been forced to shrink their operations in the last decade" [4, p. 44]. Lehn, Netter, and Poulsen [12] support the industry effect hypothesis. They report that LOBs occur for firms in industries that are faced with slower growth prospects and lower research and development (R&D) expenses.

To test the industry effect hypothesis, this paper examines the correlation between the number of leveraged buyouts and industries. We test the hypothesis using the general nonparametric statistics comparing the frequency distribution of LBOs and non-LBOs across industries. We examine the free cash flow, the growth rate and the debt capacity of each industry identified as having significant LBO activity.(2) Finally, we examine the cumulative excess returns occurring for each industry portfolio identified as having significant LBO activity. In general, our results provide only limited, weak support for the industry effect hypothesis.

II. Data

The sample used in this study consists of 263 successful going-private (leveraged buyout) transactions from 1980 through 1987 identified by Lehn and Poulsen [10]. This sample contains all successful leveraged buyouts reported in the Wall Street Index. Each firm in the sample is assigned an event date corresponding to the first announcement of its leveraged buyout transaction in the Wall Street Journal.

From this sample, each LBO is placed in an industry using the CRSP (Center for Research in Security Prices) master price tape. Industries are defined using the first two digits of the Standard Industrial Classification (SIC) codes. Using the first two digits allows us to place each firm into a fairly specific category, yet at the same time keep the number of firms in each category large enough to have meaningful statistical comparisons.(3)

After placing each firm in the sample into an industry, we form portfolios consisting of all firms in each two-digit SIC code industry with securities trading during the first and second leveraged buyout for that industry. In order to be included in the study, all firms had to have returns available from the CRSP tapes during the proper time period around the event date. Based on this screening procedure, the final sample contains 170 leveraged buyouts and 1,943 control firms.

III. Methodology and Results

A. Test of LBO Concentration by Industry

The first step in testing the industry effect hypothesis is to test for the concentration of LBOs in some industries. The null hypothesis is that LBOs occur equally across all industries. The alternative is that LBOs occur more frequently in some industries (an industry effect). A Chi-square test statistic computed for each industry tests whether the difference between the actual number of LBOs and the expected number of LBOs is zero. The calculation of the expected number of LBOs assumes that the leveraged buyouts are distributed equally across all industries weighted for the total number of firms in each industry.

Exhibit 1 presents the frequency distribution for the sample of leveraged buyout firms across industry classes. The first part of Exhibit 1 shows the expected and actual number of LBOs for each industry which had a significantly greater-than-expected number of leveraged buyouts at the 5% level.(4) The second part presents the results for the nonparametric general association tests.

The tests for general association show correlation between leveraged buyout activity and industry classification. The Cochran-Mantel-Haenszel test of independence is rejected at the 0.026 level and the null hypothesis of no general association between industries and LBOs can be rejected at the 0.000 level. These findings are consistent with the existence of an industry effect in LBOs.

Out of the 62 industries studied, ten (or 16.13%) had significant LBO activity (see Exhibit 1). These industries include both traditionally labor-intensive industries, such as textile and apparel products, and industries heavily dependent upon natural resources, such as paper products. As expected, the high LBO activity industries do not include high-growth industries, such as computers and high technology. This finding is consistent with LBOs appearing in low-growth industries that are anticipated to have substantial free cash flow. This result is also consistent with the findings by Lehn, Netter, and Poulsen [12] who show that low-growth firms tend to have leveraged buyouts.

At the same time, other industries normally thought of as having characteristics favorable to LBOs, such as the oil and gas industry do not appear in Exhibit 1. Thus, the question arises as to why other industries which are traditionally low-growth/labor-intensive, or rich in natural resources, do not also have significantly greater LBO activity than expected. Following this line of reasoning, Exhibit 2 presents the industries which had fewer leveraged buyouts than expected. Seven industries met the Chi-square test for significantly (at the 5% level) fewer LBOs than expected. As expected, the high technology and electronic industries (SIC codes 3500 and 3600) had fewer LBOs than expected. The investment industry and the electric and gas industry are traditionally regulated and, not surprisingly, have fewer LBOs than expected. The surprising result is that the oil and gas extraction and refining industries had fewer LBOs than expected. This finding appears to completely counter the belief than LBOs will be concentrated in low-growth industries with free cash flow.

The results of the Chi-square test provide us with two sets of industries, one with a high concentration of LBOs and the other with a low concentration of LBOs. This result suggests the existence of an industry effect in LBOs. However, the industries in each set do not match our prior expectations of low-growth/high free cash flow industries. Therefore, before any conclusion can be drawn about the existence of an industry effect for LBOs, further tests on the two sets of industries are required.

B. Tests for Industry Cash Flow and the Industry Effect

We test the link between industry cash flow and the industry effect hypothesis by examining the two sets of industries identified by the Chi-square test. In order to fully examine the relationship between industry cash flow and leveraged buyouts, our analysis includes free cash flow from operations, investment activity (proxied by asset growth), debt capacity and free cash flow stability. Industries that have low asset growth rates, produce high free cash flow from operations, have excess debt capacity, and have stable free cash flows are expected to have higher concentrations of leveraged buyouts. To test this proposition, we calculate the growth rates, the debt capacity and the free cash flow for each two-digit SIC code industry identified in the Chi-square test.

Growth Rates. The growth rate is defined as the three-year continuously compounded rate of growth in total assets of each firm, calculated for the three years preceding the first LBO in the industry. The industry mean and median growth rates are compared to the growth rates of all firms reporting data on COMPUSTAT for the appropriate time period.

Exhibit 3 contains annualized three-year growth rates for the industries with greater-than-expected LBO activity. To support the industry effect hypothesis, we expect lower growth in the high LBO activity industries than in the entire set of firms. The results show that the apparel industry (SIC 2300) has a significantly (at the 5% level) lower mean and median growth rate than the entire set of firms.(5) The textile mill products industry (SIC 2200) has a lower median. Contrary to expectations, the miscellaneous retail stores industry (SIC 5900) has a significantly higher mean growth rate than the entire set of firms. In the other comparisons of growth rates, the industry growth rates are not significantly different from the growth rates of the entire set of firms.

We also examine the mean difference of the annualized three-year growth rates year-by-year across the time covered by the sample.(6) The results show that the mean growth rate for the industries is usually less than the mean growth rate of the entire set of firms, but the difference is seldom significant. The only pattern of significantly (at 5%) lower growth by an industry is in the apparel products industry (SIC 2300). The low-growth rates in the apparel products industry occur at the time the first LBO occurred in the industry and therefore are consistent with our expectations. The overall results across time do not support that these industries are low-growth industries.

Free Cash Flow. The annual firm free cash flow (FFCF) is calculated using a modification of the free cash flow definition proposed by Lehn and Poulsen [10]. Free cash flow is calculated on an annual basis, and is (1) FFCF = INC - TAX - INT - PFDDIV, where:

PFDDIV = preferred stock dividends. FFCF is divided by the firm's total assets, calculated at the year end, to produce a weighted free cash flow estimate.(7) We do not subtract the firm's common dividends from our definition since the analysis is trying to capture the firm's discretionary cash flow that could support the high debt loads of a leveraged buyout.

To examine industry free cash flow (ICF), we calculate both the mean and median of the weighted firm free cash flow for all firms in an industry. For each industry, the industry free cash flow is compared to the overall population free cash flow in that year.

Exhibit 4 contains the results of the test on industry free cash flow for the industries with greater-than-expected LBO activity. The test uses the weighted free cash flow from the year preceding the first LBO in the industry. To support the industry effect hypothesis, we expect greater free cash flow in industries with significantly greater LBO activity than in the overall set of firms. In general, the results do not offer support for the industry effect hypothesis. The miscellaneous retail stores (SIC 5900) has a significantly (at 5%) lower mean and median free cash flow than the overall set of firms. Also, the apparel products industry (SIC 2300) has a significantly (at 1%) lower median free cash flow. The paper products industry (SIC 2600) and the food stores industry (SIC 5400) have a significantly (at 1%) higher median free cash flow. In the other comparisons of free cash flow, the industry free cash flow is not significantly different from the free cash flow of the entire set of firms.

We also examined the mean difference in weighted free cash flow for each industry and the weighted free cash flow for all firms, year-by-year, across the time period covered by the sample. The results show that the mean free cash flow of the industries is usually greater than the mean for the entire set of firms, but the difference is seldom significant.(8)

Cash Flow Stability. In order to examine the stability of free cash flows, Exhibit 4 also reports each industry's free cash flow coefficient of variation (CV). The coefficient of variation is computed as the standard deviation of industry free cash flow divided by the mean of the industry free cash flow for the year before the first LBO. Since firms which undergo a leveraged buyout must support increased debt loads, stable free cash flow is needed to meet rigid interest payments. Thus, in order to support the industry effect hypothesis, we expect industries with significant LBO activity to have a lower free cash flow coefficient of variation than the overall set of firms. As reported in Exhibit 4, with the exception of the miscellaneous retail stores industry (SIC 5900), all industries with greater-than-expected LBO activity have lower coefficients of variation than the overall set of firms.(9) These results provide support for the industry effect hypothesis.

Remaining Debt Capacity. In a leveraged buyout, the firm is taken private by replacing existing equity with debt. A firm's total debt capacity equals the firm's existing debt plus any remaining debt capacity, where the total debt capacity is less than or equal to the firm's total assets. Since the process of taking a firm private is a costly transaction, only firms with significant remaining debt capacity should expect to benefit from the LBO process. Thus, in order to support the industry effect hypothesis, we expect industries with greater-than-expected LBO activity to have greater amounts of remaining debt capacity than the overall set of firms. We use the weighted market value of equity as a proxy for the remaining debt capacity. The weighted remaining debt capacity (WDC) is calculated by dividing the market value of common equity for the year end prior to the first LBO by the firm's total assets. Larger debt capacity ratios imply greater amounts of remaining debt capacity. We then compute the mean and median weighted remaining debt capacity for each industry. For each industry, the industry remaining debt capacity is compared to the overall population remaining debt capacity in that year.

Exhibit 5 reports the mean and median weighted remaining debt capacity for the year before the first LBO in the industries with greater-than-expected LBO activity. None of the industries have a mean weighted remaining debt capacity significantly different from the population mean weighted remaining debt capacity. Using the population proportions test, we find that four industries have significantly lower median weighted debt capacity ratios implying significantly less remaining debt capacity: Textile mill products (SIC 2200) and apparel products (SIC 2300) (at the 1% level of significance); restaurants (SIC 5800) (at 5%); and miscellaneous retail stores (SIC 5900) (at 10%). These results are inconclusive because only four industries show a significant difference. In addition, the significant differences shown are in the opposite direction of our prediction for the industry effect hypothesis. Therefore, the results from the tests on debt capacity do not support the industry effect hypothesis.

The results of examining the free cash flow, growth rates, free cash flow stability and remaining debt capacity of the industries with greater-than-expected LBO activity show that none of the industries fit all of our expectations for a high LBO activity. Thus, our results cannot support the hypothesis that the expected industry effect is driving LBO activity.

Industries With Fewer Than Expected LBOs. We also examined the free cash flow, debt capacity, and the growth rates of the industries with less-than-expected LBO activity.(10) Exhibit 6 reports the mean annualized three-year growth rates across time. Excluding the regulated industries (SIC 4900 and SIC 6700), three of the remaining four industries (SIC 1300, SIC 3500 and SIC 3600) have higher growth rates than the entire set of firms. This result is consistent with the low LBO activity in these industries. Exhibit 7 contains the mean industry free cash flow and coefficients of variation across time. The means difference tests for the weighted free cash flow do not show any consistent pattern, while the industry coefficients of variation are generally lower than the overall set of firms. When these results are examined in conjunction with the growth rate results, it is apparent that the nonregulated industries do not have the cash flow necessary to support LBO activity with their rate of growth. Exhibit 8 reports the mean weighted remaining debt capacity for industries with less-than-expected LBO activity. The results show that three of the four nonregulated industries tend to have weighted remaining debt capacity significantly greater than the overall set of firms, which is not consistent with low LBO activity. These results are mixed as to our expectations for low LBO activity and therefore, do not support the industry effect hypothesis.

C. Tests for Industry Excess Returns and Variance Changes

The industry effect hypothesis states that some industries are more likely to have LBOs than others due to the operating characteristics of the industry. If an industry effect does exist, we would expect the market to identify industry characteristics that lend themselves to the existence of LBOs and to embed that information in the stock price of each firm in the industry. However, until an LBO occurs in an industry, the probability that the industry has been correctly identified is less than 100%. Therefore, if the first LBO in an industry provides verification of an expected industry effect, then stock prices in the industry will react to the information. We expect a positive price reaction because it is well documented that target firms of takeovers in general, as well as LBOs, earn positive excess returns.(11) If an industry effect does exist, we would expect the stock prices of all companies in the industry to react to the first LBO in the industry. However, with the potential for noise when dealing with industries and with the information the market will incorporate about the industry from the first LBO, the second LBO in each industry also needs to be tested. We conclude that no abnormal market reaction to the first two LBOs in an industry is support that an industry effect does not exist. To test this proposition about the industry effect hypothesis, we calculate excess returns occurring in each industry with significantly greater actual leveraged buyouts than expected. The event study methodology is a portfolio approach for industries proposed by Bruner and Simms [2]. The Bruner and Simms methodology corrects for the problem of cross-sectional correlation by examining standardized portfolio residuals instead of aggregate individual residuals. This methodology is consistent with the Jaffe-Mandelker metholodoly discussed in Brown and Warner [1].(12)

The cumulative average residuals (CARs) are tested for a three-day (- 1 to + 1) and seven-day (- 3 to + 3) event window, where the event dates are the first and second announcement of an LBO for the industry. Along with the calculation of excess returns, we test for a variance shift for the industry returns after the first and second LBO date. The variance shift is tested using an F-statistic comparing the variance in the estimation period to the variance during the event window using the methodology proposed by Kalay and Loewenstein [6], and Ohlson and Penman [15].(13)

The results from the free cash flow, growth rate and debt capacity tests show that none of the industries with greater-than-expected LBO activity fit our expectations. Therefore, a reasonable expectation exists that the market needed to learn about these industries and their potential for LBOs. Accordingly, the market may tend to react more to the second LBO than the first LBO. However, we believe than if an industry effect does exist, the market is not inefficient to the point of needing more than two events to put an industry in play for LBOs.

Exhibit 9 contains the results from the event study on the industries with greater-than-expected LBO activity. We report test results for the first and second LBO in each industry and a cumulative abnormal return for both dates. The results show significantly positive (at 10%) returns for six industries, apparel products (SIC 2300), paper products (SIC 2600), miscellaneous retail (SIC 5900), textile mill products (SIC 2200), communication (SIC 4800), and personal services (SIC 7200). At the more generally accepted 5% significantly level, only the last three industries have positive returns. With only three of ten industries showing any significant (at 5%) abnormal returns, the event study results suggest again that the industry effect is weak.

The Kalay and Loewenstein test was done in conjunction with the event study. The procedure tests for variance shifts around an event. A shift in the variance indicates a shift in the return generating process which is consistent with increased trading due to speculation that the industry is in play for LBO activity. The results of the test show that only apparel stores (SIC 5600) had a significant (at 10%) 20-day variance shift for the first event date and textile mill products (SIC 2200) had a significant (at 10%) 41-day variance shift for the first event date. None of the other industries had significant variance shifts for any event dates. This suggests the industries were not in play, which further supports the lack of an industry effect for LBOs.

D. A. Test of Industry LBO Excess Returns

Finally, in order to control for other potential effects which could drive the results found above, we construct a test of the relationship between industry free cash flows and industry excess returns occurring around the first LBO date for all industries. Specifically, for each industry with a leveraged buyout we test the model: (2) [Mathematical Expression Omitted] where:

[CV(ICF.sub.i] = the coefficient of variation is the standard

Estimating the model parameters provides a direct test for the effect of free cash flow and growth rate on the industry excess returns occurring around the announcement of the first LBO in the industry. Finding a significantly positive parameter estimate for the variable for the level of LBO activity would support the industry effect hypothesis.

From the sample described in Section II, 47 industries had at least one LBO during the period 1980 to 1987 and have sufficient data available for inclusion in the model estimation. For each of the 47 industries, the CARs, free cash flow, growth rates and debt capacity are calculated using the methodology described above with the first LBO in the industry as the event date. The model is estimated using ordinary least squares (OLS) and the results are as follows (t-statistics in parenthesis): (3) [CAR.sub.i] = -0.0142 + [0.051ICF.sub.i] - 0.00183CV([ICF.sub.i])

[R.sup.2] = 0.164; Adj. [R.sup.2] = 0.06; F-statistic = 1.57. (*)Significant at 10%.

The results of the regression show that the thee-year industry growth rate is significant at the 10% level and positively related to the industry CAR. This result indicates larger premiums were paid in industries with more upside potential for the remaining shareholders. Thus, even though LBO's will occur in industries with lower growth rates the bidders must compensate shareholders for the growth opportunities within these industries. The results also show insignificant parameters on the industry free cash flow variable, the coefficient of variation in industry free cash flow, and the LBO activity dummy variable, indicating a lack of an industry effect. The lack of a significant parameter for the LBO activity variable supports the results reported earlier suggesting that the industry effect is not present.

IV. Conclusions

Based on the results reported above, we find only weak evidence to support the industry effect hypothesis. Only statistically weak nonparametric tests support the finding of an industry effect. After examining the actual industries which had both fewer- and greater-than-expected leveraged buyouts, the industry effect does not hold up.

We directly test the free cash flow, growth rates and debt capacity of the industries. Our results show little correlation between positive free cash flow, low growth rates and leveraged buyout activity in industries. The results do show that industries cannot support leveraged buyout activity without the necessary free cash flow.

The industry effect hypothesis is directly tested by examining the portfolio returns and variance shifts for both the first and second LBOs in each industry identified as having significantly greater LBO activity. The results for the test of excess returns occurring to the industry portfolios are inconclusive at best. The findings of no significant variance shift do not support the industry effect hypothesis.

The use of two-digit SIC codes weakens the power of the tests used in this paper. However, the fact that narrower industry classifications could not be used due to the lack of LBO concentrations is further support for our conclusion that an industry effect does not drive LBO activity.

Based on these findings, we conclude that firm-specific factors are the primary motivating forces for most leveraged buyouts. The industry in which a firm operates is, at best, only a secondary factor contributing to whether a firm becomes the target of a leveraged buyout. Thus, these results suggest that future research in understanding the motivations for leveraged buyouts should concentrate on firm-specific factors. (1)See Kaplan [7], Marais, Schipper, and Smith [13], Lehn and Poulsen [10], [11], and Lehn, Netter, and Poulsen [12]. (2)Lang, Stulz and Walkling [8] use Tobin's q to examine tender offers. They find low q targets provide the highest returns, which means q could be used to identify takeover targets. They state that a low q is consistent with either poor investment opportunities or poor management. We do not believe poor management is an industry characteristic. Poor investment opportunities are necessary for free cash flow, but not sufficient. Therefore, while we agree q can be used to identify individual takeover targets, we feel it is not useful in examining the industry effect hypothesis for LBOs. (3)We attempted to use a three-digit SIC code, but the resulting industries were to small to allow drawing any reliable conclusions from the nonparametric tests used to identify the industries with significant LBO activity. Therefore, we use the two-digit SIC codes to form the industries in all of our tests. (4)Due to space limitations, only those industries which have a significantly greater number of leveraged buyouts than expected are reported. The complete distribution table is available upon request from the authors. (5)We use a means difference test to test for industries with growth rates greater than the mean and a population proportions test to test for differences in the median values. The population proportions test determines if the percentage of two populations above a stated value is significantly different (see Neter, Wasserman and Whitmore [14, p. 412]). The stated value is the median of all firms. These same tests are applied to the set of industries with less-than-expected LBO activity. (6)A table summarizing the mean differences of the annualized three-year growth rates is available from the authors. (7)The absolute amount of free cash flow is not the relevant amount in the case of LBOs. Instead, it is the amount of free cash flow relative to the amount of debt it must service in an LBO. However, we need to examine free cash flow and debt capacity as separate issues, which means we must weight free cash flow by something other than the remaining debt capacity. We weight free cash flow by total assets and to be consistent, we also weight debt capacity by total assets. (8)A table detailing these results is available upon request from the authors. (9)The coefficients of variation reported are calculated using one year of data from the year before the first LBO in the industry. We also calculate a three-year average coefficient of variation for the three years before the first LBO. The three averages show similar results to the one-year coefficients of variation. (10)We do not report exhibits similar to Exhibits 3 and 5 for the less-than-expected LBO industries, since some of these industries did not have any LBOs. (11)DeAngelo, DeAngelo, and Rice [3] report positive excess returns to target shareholders of 22.3%, and Lehn and Poulsen [9] report abnormal gains of 13.9%. (12]The excess returns for each firm are calculated using a one factor model estimated using ordinary least squares. The excess returns are averaged cross-sectionally to arrive at an average excess return for each day. The average returns are accumulated across days to form cumulative average residuals (CARs) for the test windows. The standard deviation used in the t-statistic is the standard deviation of the average residuals during a holdout period before the event window. (13]The variances of the portfolio returns during the event window for both the 20-and 41-day period are compared to the portfolio variance during the estimation period. The F-statistic for unequal is calculated by dividing the variance of the event window by the variance of the estimation period.

References

[1]S.J. Brown and J.B. Warner, "Measuring Security Price Performance," Journal of Financial Economics (September 1980), pp. 205-258. [2]R.F. Bruner and J.M. Simms, Jr., "The International Debt Crisis and Bank Security Returns in 1982," Journal of Money, Credit, and Banking (February 1987), pp. 46-55. [3]H. DeAngelo, L. DeAngelo, and E. Rice, "Going Private: Minority Freezeouts and Shareholder Wealth," Journal of Law and Economics (October 1984), pp. 367-401. [4]M.C. Jensen, "Active Investors, LBOs, and the Privatization of Bankruptcy," Journal of Applied Corporate Finance (1989/1990), pp. 35-44. [5]M.C. Jensen, "Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers," American Economic Review (May 1986), pp. 323-329. [6]A Kalay and U. Loewenstein, "Predictable Events and Excess Returns: The Case of Dividend Announcements," Journal of Financial Economics (September 1985), pp. 423-449. [7]S. Kaplan, "Management Buyouts: Evidence on Taxes as a Source of Value," Journal of Finance (July 1989), pp. 622-632. [8]L.H. Lang, R.M. Stulz, and R.A. Walking, "Managerial Performance, Tobin's Q, and the Gains from Successful Tender Offers," Journal of Financial Economics (September 1989), pp. 137-154. [9]K. Lehn and A. Poulsen, "Leveraged Buyouts: Wealth Created or Wealth Redistributed?" in Public Policy Toward Corporate Mergers, M. Wiedenbaum and K. Chilton (eds.), New Brunswick, NJ, Transition Books, 1988. [10]K. Lehn and A Poulsen, "Free Cash Flow and Stockholder Gains in Going Private Transactions," Journal of Finance (July 1989), pp. 771-787. [11]K. Lehn and A. Poulsen, "The Economics of Event Risk: The Case of Bondholders in Takeovers," Working Paper, University of Georgia, 1990. [12]K. Lehn, J. Netter, and A Poulsen, "Consolidating Corporate Control: The Choice Between Dual-Class Recapitalizations and Leveraged Buyouts," Working Paper, University of Georgia, 1990. [13]L. Marais, K. Schipper, and A. Smith, "Wealth Effects of Going Private of Senior Securities," Journal of Financial Economics (June 1989), pp. 155-191. [14]J. Neter, W. Wasserman, and G.A. Whitmore, Applied Statistics, Boston, MA, Allyn Bacon, 3rd edition, 1988. [15]J. A. Ohlson and S.H. Penman, "Volatility Increases Subsequent to Stock Splits," Journal of Financial Economics (June 1985), pp. 251-266. Brent W. Ambrose is an Assistant Professor of Real Estate at the School of Business Administration, University of Wisconsin-Milwaukee and is presently a visiting researcher at the Department of Housing and Urban Development, Office of Policy Development and Research, Washingto D.C. Drew B. Winters is an Assistant Professor of Finance at the School of Business Administration, University of Wisconsin-Milwaukee, Milwaukee, Wisconsin.

Theoretical factors for explaining the motivations for taking a firm private can be grouped under two competing hypotheses. The first, the firm effects hypotheses, states that firm-specific characteristics, such as management or operating inefficiency, are the primary motivation for taking a firm through a leveraged buyout. The second, called the industry effect hypothesis, states that factors in common to the firm's industry are the primary motivation for taking a firm private.

The purpose of this study is to examine the industry effect hypothesis. In Section I, we present a discussion of the theories and evidence for why leveraged buyouts occur. Section II describes the data used to test the industry effect hypothesis and Section III discusses both the methodology and the results. Finally, Section IV discusses the conclusions.

I. Theories for LBO Activity

Past research has attempted to explain the common factors that might support the industry effect hypothesis of leveraged buyout activity. One proposed factor is the potential leverage-induced tax savings involved in LBOs. Other factors are related to potential agency problems.

The tax-savings hypothesis has received considerable attention. It follows from the tax shield of interest payments on the increased debt in an LBO. Marais, Schipper, and Smith [13] present evidence that tax savings are correlated with the LBO premium. Kaplan [7] also finds that the tax benefits are a large source of wealth in leveraged buyouts.

The agency relationship between managers and shareholders has also received considerable attention. Jensen [5] proposes that leveraged buyouts reduce a special type of agency problem associated with free cash flow, i.e., "cash flow in excess of that required to fund all projects that have positive net present values" [5, p. 323]. Lehn and Poulsen [10] test the free cash flow hypothesis and find evidence consistent with shareholder gains resulting from reducing the agency costs associated with free cash flow.

The industry effect hypothesis suggests that certain industries are more likely to have LBO activity. Also, other industries, for example, those that are regulated, are less likely to have LBO activity. For there to be a concentration of LBOs in an industry, most firms within it must generate enough cash flow above their current needs to service the debt used in the LBO and have the potential to benefit from reducing agency costs. Jensen [4] points out that "some of the best examples of this have occurred in the oil, tire, and tobacco industries -- all industries that have been forced to shrink their operations in the last decade" [4, p. 44]. Lehn, Netter, and Poulsen [12] support the industry effect hypothesis. They report that LOBs occur for firms in industries that are faced with slower growth prospects and lower research and development (R&D) expenses.

To test the industry effect hypothesis, this paper examines the correlation between the number of leveraged buyouts and industries. We test the hypothesis using the general nonparametric statistics comparing the frequency distribution of LBOs and non-LBOs across industries. We examine the free cash flow, the growth rate and the debt capacity of each industry identified as having significant LBO activity.(2) Finally, we examine the cumulative excess returns occurring for each industry portfolio identified as having significant LBO activity. In general, our results provide only limited, weak support for the industry effect hypothesis.

II. Data

The sample used in this study consists of 263 successful going-private (leveraged buyout) transactions from 1980 through 1987 identified by Lehn and Poulsen [10]. This sample contains all successful leveraged buyouts reported in the Wall Street Index. Each firm in the sample is assigned an event date corresponding to the first announcement of its leveraged buyout transaction in the Wall Street Journal.

From this sample, each LBO is placed in an industry using the CRSP (Center for Research in Security Prices) master price tape. Industries are defined using the first two digits of the Standard Industrial Classification (SIC) codes. Using the first two digits allows us to place each firm into a fairly specific category, yet at the same time keep the number of firms in each category large enough to have meaningful statistical comparisons.(3)

After placing each firm in the sample into an industry, we form portfolios consisting of all firms in each two-digit SIC code industry with securities trading during the first and second leveraged buyout for that industry. In order to be included in the study, all firms had to have returns available from the CRSP tapes during the proper time period around the event date. Based on this screening procedure, the final sample contains 170 leveraged buyouts and 1,943 control firms.

III. Methodology and Results

A. Test of LBO Concentration by Industry

The first step in testing the industry effect hypothesis is to test for the concentration of LBOs in some industries. The null hypothesis is that LBOs occur equally across all industries. The alternative is that LBOs occur more frequently in some industries (an industry effect). A Chi-square test statistic computed for each industry tests whether the difference between the actual number of LBOs and the expected number of LBOs is zero. The calculation of the expected number of LBOs assumes that the leveraged buyouts are distributed equally across all industries weighted for the total number of firms in each industry.

Exhibit 1 presents the frequency distribution for the sample of leveraged buyout firms across industry classes. The first part of Exhibit 1 shows the expected and actual number of LBOs for each industry which had a significantly greater-than-expected number of leveraged buyouts at the 5% level.(4) The second part presents the results for the nonparametric general association tests.

The tests for general association show correlation between leveraged buyout activity and industry classification. The Cochran-Mantel-Haenszel test of independence is rejected at the 0.026 level and the null hypothesis of no general association between industries and LBOs can be rejected at the 0.000 level. These findings are consistent with the existence of an industry effect in LBOs.

Out of the 62 industries studied, ten (or 16.13%) had significant LBO activity (see Exhibit 1). These industries include both traditionally labor-intensive industries, such as textile and apparel products, and industries heavily dependent upon natural resources, such as paper products. As expected, the high LBO activity industries do not include high-growth industries, such as computers and high technology. This finding is consistent with LBOs appearing in low-growth industries that are anticipated to have substantial free cash flow. This result is also consistent with the findings by Lehn, Netter, and Poulsen [12] who show that low-growth firms tend to have leveraged buyouts.

At the same time, other industries normally thought of as having characteristics favorable to LBOs, such as the oil and gas industry do not appear in Exhibit 1. Thus, the question arises as to why other industries which are traditionally low-growth/labor-intensive, or rich in natural resources, do not also have significantly greater LBO activity than expected. Following this line of reasoning, Exhibit 2 presents the industries which had fewer leveraged buyouts than expected. Seven industries met the Chi-square test for significantly (at the 5% level) fewer LBOs than expected. As expected, the high technology and electronic industries (SIC codes 3500 and 3600) had fewer LBOs than expected. The investment industry and the electric and gas industry are traditionally regulated and, not surprisingly, have fewer LBOs than expected. The surprising result is that the oil and gas extraction and refining industries had fewer LBOs than expected. This finding appears to completely counter the belief than LBOs will be concentrated in low-growth industries with free cash flow.

The results of the Chi-square test provide us with two sets of industries, one with a high concentration of LBOs and the other with a low concentration of LBOs. This result suggests the existence of an industry effect in LBOs. However, the industries in each set do not match our prior expectations of low-growth/high free cash flow industries. Therefore, before any conclusion can be drawn about the existence of an industry effect for LBOs, further tests on the two sets of industries are required.

B. Tests for Industry Cash Flow and the Industry Effect

We test the link between industry cash flow and the industry effect hypothesis by examining the two sets of industries identified by the Chi-square test. In order to fully examine the relationship between industry cash flow and leveraged buyouts, our analysis includes free cash flow from operations, investment activity (proxied by asset growth), debt capacity and free cash flow stability. Industries that have low asset growth rates, produce high free cash flow from operations, have excess debt capacity, and have stable free cash flows are expected to have higher concentrations of leveraged buyouts. To test this proposition, we calculate the growth rates, the debt capacity and the free cash flow for each two-digit SIC code industry identified in the Chi-square test.

Growth Rates. The growth rate is defined as the three-year continuously compounded rate of growth in total assets of each firm, calculated for the three years preceding the first LBO in the industry. The industry mean and median growth rates are compared to the growth rates of all firms reporting data on COMPUSTAT for the appropriate time period.

Exhibit 3 contains annualized three-year growth rates for the industries with greater-than-expected LBO activity. To support the industry effect hypothesis, we expect lower growth in the high LBO activity industries than in the entire set of firms. The results show that the apparel industry (SIC 2300) has a significantly (at the 5% level) lower mean and median growth rate than the entire set of firms.(5) The textile mill products industry (SIC 2200) has a lower median. Contrary to expectations, the miscellaneous retail stores industry (SIC 5900) has a significantly higher mean growth rate than the entire set of firms. In the other comparisons of growth rates, the industry growth rates are not significantly different from the growth rates of the entire set of firms.

We also examine the mean difference of the annualized three-year growth rates year-by-year across the time covered by the sample.(6) The results show that the mean growth rate for the industries is usually less than the mean growth rate of the entire set of firms, but the difference is seldom significant. The only pattern of significantly (at 5%) lower growth by an industry is in the apparel products industry (SIC 2300). The low-growth rates in the apparel products industry occur at the time the first LBO occurred in the industry and therefore are consistent with our expectations. The overall results across time do not support that these industries are low-growth industries.

Free Cash Flow. The annual firm free cash flow (FFCF) is calculated using a modification of the free cash flow definition proposed by Lehn and Poulsen [10]. Free cash flow is calculated on an annual basis, and is (1) FFCF = INC - TAX - INT - PFDDIV, where:

INC = operating income before depreciation; TAX = total income tax; INT = gross interest expense;

PFDDIV = preferred stock dividends. FFCF is divided by the firm's total assets, calculated at the year end, to produce a weighted free cash flow estimate.(7) We do not subtract the firm's common dividends from our definition since the analysis is trying to capture the firm's discretionary cash flow that could support the high debt loads of a leveraged buyout.

To examine industry free cash flow (ICF), we calculate both the mean and median of the weighted firm free cash flow for all firms in an industry. For each industry, the industry free cash flow is compared to the overall population free cash flow in that year.

Exhibit 4 contains the results of the test on industry free cash flow for the industries with greater-than-expected LBO activity. The test uses the weighted free cash flow from the year preceding the first LBO in the industry. To support the industry effect hypothesis, we expect greater free cash flow in industries with significantly greater LBO activity than in the overall set of firms. In general, the results do not offer support for the industry effect hypothesis. The miscellaneous retail stores (SIC 5900) has a significantly (at 5%) lower mean and median free cash flow than the overall set of firms. Also, the apparel products industry (SIC 2300) has a significantly (at 1%) lower median free cash flow. The paper products industry (SIC 2600) and the food stores industry (SIC 5400) have a significantly (at 1%) higher median free cash flow. In the other comparisons of free cash flow, the industry free cash flow is not significantly different from the free cash flow of the entire set of firms.

We also examined the mean difference in weighted free cash flow for each industry and the weighted free cash flow for all firms, year-by-year, across the time period covered by the sample. The results show that the mean free cash flow of the industries is usually greater than the mean for the entire set of firms, but the difference is seldom significant.(8)

Cash Flow Stability. In order to examine the stability of free cash flows, Exhibit 4 also reports each industry's free cash flow coefficient of variation (CV). The coefficient of variation is computed as the standard deviation of industry free cash flow divided by the mean of the industry free cash flow for the year before the first LBO. Since firms which undergo a leveraged buyout must support increased debt loads, stable free cash flow is needed to meet rigid interest payments. Thus, in order to support the industry effect hypothesis, we expect industries with significant LBO activity to have a lower free cash flow coefficient of variation than the overall set of firms. As reported in Exhibit 4, with the exception of the miscellaneous retail stores industry (SIC 5900), all industries with greater-than-expected LBO activity have lower coefficients of variation than the overall set of firms.(9) These results provide support for the industry effect hypothesis.

Remaining Debt Capacity. In a leveraged buyout, the firm is taken private by replacing existing equity with debt. A firm's total debt capacity equals the firm's existing debt plus any remaining debt capacity, where the total debt capacity is less than or equal to the firm's total assets. Since the process of taking a firm private is a costly transaction, only firms with significant remaining debt capacity should expect to benefit from the LBO process. Thus, in order to support the industry effect hypothesis, we expect industries with greater-than-expected LBO activity to have greater amounts of remaining debt capacity than the overall set of firms. We use the weighted market value of equity as a proxy for the remaining debt capacity. The weighted remaining debt capacity (WDC) is calculated by dividing the market value of common equity for the year end prior to the first LBO by the firm's total assets. Larger debt capacity ratios imply greater amounts of remaining debt capacity. We then compute the mean and median weighted remaining debt capacity for each industry. For each industry, the industry remaining debt capacity is compared to the overall population remaining debt capacity in that year.

Exhibit 5 reports the mean and median weighted remaining debt capacity for the year before the first LBO in the industries with greater-than-expected LBO activity. None of the industries have a mean weighted remaining debt capacity significantly different from the population mean weighted remaining debt capacity. Using the population proportions test, we find that four industries have significantly lower median weighted debt capacity ratios implying significantly less remaining debt capacity: Textile mill products (SIC 2200) and apparel products (SIC 2300) (at the 1% level of significance); restaurants (SIC 5800) (at 5%); and miscellaneous retail stores (SIC 5900) (at 10%). These results are inconclusive because only four industries show a significant difference. In addition, the significant differences shown are in the opposite direction of our prediction for the industry effect hypothesis. Therefore, the results from the tests on debt capacity do not support the industry effect hypothesis.

The results of examining the free cash flow, growth rates, free cash flow stability and remaining debt capacity of the industries with greater-than-expected LBO activity show that none of the industries fit all of our expectations for a high LBO activity. Thus, our results cannot support the hypothesis that the expected industry effect is driving LBO activity.

Industries With Fewer Than Expected LBOs. We also examined the free cash flow, debt capacity, and the growth rates of the industries with less-than-expected LBO activity.(10) Exhibit 6 reports the mean annualized three-year growth rates across time. Excluding the regulated industries (SIC 4900 and SIC 6700), three of the remaining four industries (SIC 1300, SIC 3500 and SIC 3600) have higher growth rates than the entire set of firms. This result is consistent with the low LBO activity in these industries. Exhibit 7 contains the mean industry free cash flow and coefficients of variation across time. The means difference tests for the weighted free cash flow do not show any consistent pattern, while the industry coefficients of variation are generally lower than the overall set of firms. When these results are examined in conjunction with the growth rate results, it is apparent that the nonregulated industries do not have the cash flow necessary to support LBO activity with their rate of growth. Exhibit 8 reports the mean weighted remaining debt capacity for industries with less-than-expected LBO activity. The results show that three of the four nonregulated industries tend to have weighted remaining debt capacity significantly greater than the overall set of firms, which is not consistent with low LBO activity. These results are mixed as to our expectations for low LBO activity and therefore, do not support the industry effect hypothesis.

C. Tests for Industry Excess Returns and Variance Changes

The industry effect hypothesis states that some industries are more likely to have LBOs than others due to the operating characteristics of the industry. If an industry effect does exist, we would expect the market to identify industry characteristics that lend themselves to the existence of LBOs and to embed that information in the stock price of each firm in the industry. However, until an LBO occurs in an industry, the probability that the industry has been correctly identified is less than 100%. Therefore, if the first LBO in an industry provides verification of an expected industry effect, then stock prices in the industry will react to the information. We expect a positive price reaction because it is well documented that target firms of takeovers in general, as well as LBOs, earn positive excess returns.(11) If an industry effect does exist, we would expect the stock prices of all companies in the industry to react to the first LBO in the industry. However, with the potential for noise when dealing with industries and with the information the market will incorporate about the industry from the first LBO, the second LBO in each industry also needs to be tested. We conclude that no abnormal market reaction to the first two LBOs in an industry is support that an industry effect does not exist. To test this proposition about the industry effect hypothesis, we calculate excess returns occurring in each industry with significantly greater actual leveraged buyouts than expected. The event study methodology is a portfolio approach for industries proposed by Bruner and Simms [2]. The Bruner and Simms methodology corrects for the problem of cross-sectional correlation by examining standardized portfolio residuals instead of aggregate individual residuals. This methodology is consistent with the Jaffe-Mandelker metholodoly discussed in Brown and Warner [1].(12)

The cumulative average residuals (CARs) are tested for a three-day (- 1 to + 1) and seven-day (- 3 to + 3) event window, where the event dates are the first and second announcement of an LBO for the industry. Along with the calculation of excess returns, we test for a variance shift for the industry returns after the first and second LBO date. The variance shift is tested using an F-statistic comparing the variance in the estimation period to the variance during the event window using the methodology proposed by Kalay and Loewenstein [6], and Ohlson and Penman [15].(13)

The results from the free cash flow, growth rate and debt capacity tests show that none of the industries with greater-than-expected LBO activity fit our expectations. Therefore, a reasonable expectation exists that the market needed to learn about these industries and their potential for LBOs. Accordingly, the market may tend to react more to the second LBO than the first LBO. However, we believe than if an industry effect does exist, the market is not inefficient to the point of needing more than two events to put an industry in play for LBOs.

Exhibit 9 contains the results from the event study on the industries with greater-than-expected LBO activity. We report test results for the first and second LBO in each industry and a cumulative abnormal return for both dates. The results show significantly positive (at 10%) returns for six industries, apparel products (SIC 2300), paper products (SIC 2600), miscellaneous retail (SIC 5900), textile mill products (SIC 2200), communication (SIC 4800), and personal services (SIC 7200). At the more generally accepted 5% significantly level, only the last three industries have positive returns. With only three of ten industries showing any significant (at 5%) abnormal returns, the event study results suggest again that the industry effect is weak.

The Kalay and Loewenstein test was done in conjunction with the event study. The procedure tests for variance shifts around an event. A shift in the variance indicates a shift in the return generating process which is consistent with increased trading due to speculation that the industry is in play for LBO activity. The results of the test show that only apparel stores (SIC 5600) had a significant (at 10%) 20-day variance shift for the first event date and textile mill products (SIC 2200) had a significant (at 10%) 41-day variance shift for the first event date. None of the other industries had significant variance shifts for any event dates. This suggests the industries were not in play, which further supports the lack of an industry effect for LBOs.

D. A. Test of Industry LBO Excess Returns

Finally, in order to control for other potential effects which could drive the results found above, we construct a test of the relationship between industry free cash flows and industry excess returns occurring around the first LBO date for all industries. Specifically, for each industry with a leveraged buyout we test the model: (2) [Mathematical Expression Omitted] where:

[CAR.sub.i] = the cumulative average residual for industry i (-3 to +3 day event window); [ICF.sub.i] = the mean free cash flow for the year before the first LBO for industry i;

[CV(ICF.sub.i] = the coefficient of variation is the standard

deviation of industry free cash flow standardized by the mean of the industry free cash flow for the year before the first LBO; [IGR.sub.i] = the three-year growth rate in total assets for industry i; [IDC.sub.i] = the mean weighted remaining debt capacity for the year before the first LBO for industry i; [ADUM.sub.i] = a trichotomous variable on the level of LBO activity by industry, where 1 equals more than expected LBO activity in an industry, -1 equals less than expected LBO activity in an industry, and 0 for all others.

Estimating the model parameters provides a direct test for the effect of free cash flow and growth rate on the industry excess returns occurring around the announcement of the first LBO in the industry. Finding a significantly positive parameter estimate for the variable for the level of LBO activity would support the industry effect hypothesis.

From the sample described in Section II, 47 industries had at least one LBO during the period 1980 to 1987 and have sufficient data available for inclusion in the model estimation. For each of the 47 industries, the CARs, free cash flow, growth rates and debt capacity are calculated using the methodology described above with the first LBO in the industry as the event date. The model is estimated using ordinary least squares (OLS) and the results are as follows (t-statistics in parenthesis): (3) [CAR.sub.i] = -0.0142 + [0.051ICF.sub.i] - 0.00183CV([ICF.sub.i])

(-0.65) (0.25) (-1.34) + [0.0245IGR.sub.i] -[ 0.0138IDC.sub.i] - [0.0173ADUM.sub.i] (1.98)(*) (-0.97) (-1.36)

[R.sup.2] = 0.164; Adj. [R.sup.2] = 0.06; F-statistic = 1.57. (*)Significant at 10%.

The results of the regression show that the thee-year industry growth rate is significant at the 10% level and positively related to the industry CAR. This result indicates larger premiums were paid in industries with more upside potential for the remaining shareholders. Thus, even though LBO's will occur in industries with lower growth rates the bidders must compensate shareholders for the growth opportunities within these industries. The results also show insignificant parameters on the industry free cash flow variable, the coefficient of variation in industry free cash flow, and the LBO activity dummy variable, indicating a lack of an industry effect. The lack of a significant parameter for the LBO activity variable supports the results reported earlier suggesting that the industry effect is not present.

IV. Conclusions

Based on the results reported above, we find only weak evidence to support the industry effect hypothesis. Only statistically weak nonparametric tests support the finding of an industry effect. After examining the actual industries which had both fewer- and greater-than-expected leveraged buyouts, the industry effect does not hold up.

We directly test the free cash flow, growth rates and debt capacity of the industries. Our results show little correlation between positive free cash flow, low growth rates and leveraged buyout activity in industries. The results do show that industries cannot support leveraged buyout activity without the necessary free cash flow.

The industry effect hypothesis is directly tested by examining the portfolio returns and variance shifts for both the first and second LBOs in each industry identified as having significantly greater LBO activity. The results for the test of excess returns occurring to the industry portfolios are inconclusive at best. The findings of no significant variance shift do not support the industry effect hypothesis.

The use of two-digit SIC codes weakens the power of the tests used in this paper. However, the fact that narrower industry classifications could not be used due to the lack of LBO concentrations is further support for our conclusion that an industry effect does not drive LBO activity.

Based on these findings, we conclude that firm-specific factors are the primary motivating forces for most leveraged buyouts. The industry in which a firm operates is, at best, only a secondary factor contributing to whether a firm becomes the target of a leveraged buyout. Thus, these results suggest that future research in understanding the motivations for leveraged buyouts should concentrate on firm-specific factors. (1)See Kaplan [7], Marais, Schipper, and Smith [13], Lehn and Poulsen [10], [11], and Lehn, Netter, and Poulsen [12]. (2)Lang, Stulz and Walkling [8] use Tobin's q to examine tender offers. They find low q targets provide the highest returns, which means q could be used to identify takeover targets. They state that a low q is consistent with either poor investment opportunities or poor management. We do not believe poor management is an industry characteristic. Poor investment opportunities are necessary for free cash flow, but not sufficient. Therefore, while we agree q can be used to identify individual takeover targets, we feel it is not useful in examining the industry effect hypothesis for LBOs. (3)We attempted to use a three-digit SIC code, but the resulting industries were to small to allow drawing any reliable conclusions from the nonparametric tests used to identify the industries with significant LBO activity. Therefore, we use the two-digit SIC codes to form the industries in all of our tests. (4)Due to space limitations, only those industries which have a significantly greater number of leveraged buyouts than expected are reported. The complete distribution table is available upon request from the authors. (5)We use a means difference test to test for industries with growth rates greater than the mean and a population proportions test to test for differences in the median values. The population proportions test determines if the percentage of two populations above a stated value is significantly different (see Neter, Wasserman and Whitmore [14, p. 412]). The stated value is the median of all firms. These same tests are applied to the set of industries with less-than-expected LBO activity. (6)A table summarizing the mean differences of the annualized three-year growth rates is available from the authors. (7)The absolute amount of free cash flow is not the relevant amount in the case of LBOs. Instead, it is the amount of free cash flow relative to the amount of debt it must service in an LBO. However, we need to examine free cash flow and debt capacity as separate issues, which means we must weight free cash flow by something other than the remaining debt capacity. We weight free cash flow by total assets and to be consistent, we also weight debt capacity by total assets. (8)A table detailing these results is available upon request from the authors. (9)The coefficients of variation reported are calculated using one year of data from the year before the first LBO in the industry. We also calculate a three-year average coefficient of variation for the three years before the first LBO. The three averages show similar results to the one-year coefficients of variation. (10)We do not report exhibits similar to Exhibits 3 and 5 for the less-than-expected LBO industries, since some of these industries did not have any LBOs. (11)DeAngelo, DeAngelo, and Rice [3] report positive excess returns to target shareholders of 22.3%, and Lehn and Poulsen [9] report abnormal gains of 13.9%. (12]The excess returns for each firm are calculated using a one factor model estimated using ordinary least squares. The excess returns are averaged cross-sectionally to arrive at an average excess return for each day. The average returns are accumulated across days to form cumulative average residuals (CARs) for the test windows. The standard deviation used in the t-statistic is the standard deviation of the average residuals during a holdout period before the event window. (13]The variances of the portfolio returns during the event window for both the 20-and 41-day period are compared to the portfolio variance during the estimation period. The F-statistic for unequal is calculated by dividing the variance of the event window by the variance of the estimation period.

References

[1]S.J. Brown and J.B. Warner, "Measuring Security Price Performance," Journal of Financial Economics (September 1980), pp. 205-258. [2]R.F. Bruner and J.M. Simms, Jr., "The International Debt Crisis and Bank Security Returns in 1982," Journal of Money, Credit, and Banking (February 1987), pp. 46-55. [3]H. DeAngelo, L. DeAngelo, and E. Rice, "Going Private: Minority Freezeouts and Shareholder Wealth," Journal of Law and Economics (October 1984), pp. 367-401. [4]M.C. Jensen, "Active Investors, LBOs, and the Privatization of Bankruptcy," Journal of Applied Corporate Finance (1989/1990), pp. 35-44. [5]M.C. Jensen, "Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers," American Economic Review (May 1986), pp. 323-329. [6]A Kalay and U. Loewenstein, "Predictable Events and Excess Returns: The Case of Dividend Announcements," Journal of Financial Economics (September 1985), pp. 423-449. [7]S. Kaplan, "Management Buyouts: Evidence on Taxes as a Source of Value," Journal of Finance (July 1989), pp. 622-632. [8]L.H. Lang, R.M. Stulz, and R.A. Walking, "Managerial Performance, Tobin's Q, and the Gains from Successful Tender Offers," Journal of Financial Economics (September 1989), pp. 137-154. [9]K. Lehn and A. Poulsen, "Leveraged Buyouts: Wealth Created or Wealth Redistributed?" in Public Policy Toward Corporate Mergers, M. Wiedenbaum and K. Chilton (eds.), New Brunswick, NJ, Transition Books, 1988. [10]K. Lehn and A Poulsen, "Free Cash Flow and Stockholder Gains in Going Private Transactions," Journal of Finance (July 1989), pp. 771-787. [11]K. Lehn and A. Poulsen, "The Economics of Event Risk: The Case of Bondholders in Takeovers," Working Paper, University of Georgia, 1990. [12]K. Lehn, J. Netter, and A Poulsen, "Consolidating Corporate Control: The Choice Between Dual-Class Recapitalizations and Leveraged Buyouts," Working Paper, University of Georgia, 1990. [13]L. Marais, K. Schipper, and A. Smith, "Wealth Effects of Going Private of Senior Securities," Journal of Financial Economics (June 1989), pp. 155-191. [14]J. Neter, W. Wasserman, and G.A. Whitmore, Applied Statistics, Boston, MA, Allyn Bacon, 3rd edition, 1988. [15]J. A. Ohlson and S.H. Penman, "Volatility Increases Subsequent to Stock Splits," Journal of Financial Economics (June 1985), pp. 251-266. Brent W. Ambrose is an Assistant Professor of Real Estate at the School of Business Administration, University of Wisconsin-Milwaukee and is presently a visiting researcher at the Department of Housing and Urban Development, Office of Policy Development and Research, Washingto D.C. Drew B. Winters is an Assistant Professor of Finance at the School of Business Administration, University of Wisconsin-Milwaukee, Milwaukee, Wisconsin.

Printer friendly Cite/link Email Feedback | |

Title Annotation: | Leveraged Buyouts Special Issue |
---|---|

Author: | Ambrose, Brent W.; Winters, Drew B. |

Publication: | Financial Management |

Date: | Mar 22, 1992 |

Words: | 5618 |

Previous Article: | Board composition and shareholder wealth: the case of management buyouts. |

Next Article: | Bondholder wealth effects of management buyouts. |

Topics: |