Sample Employee Restructuring Announcement by uhp13786

VIEWS: 883 PAGES: 46

More Info
									                 Going-Private Restructuring and Earnings Expectations:
  A Test of the Release of Favorable Information for Target Firms and Industry Rivals*




                                         Jeremy Goh
                                Associate Professor of Finance
                              Singapore Management University


                                     Michael Gombola**
                                     Professor of Finance
                                      Drexel University


                                        Feng-Ying Liu
                                     Professor of Finance
                                       Rider University


                                         De-Wai Chou
                                Assistant Professor of Finance
                               National Chung Cheng University




                                 Printing Date: August 3, 2011



*A previous version of this paper received the award for “Outstanding Paper in Corporate
Finance” by the Eastern Finance Association at its 2002 annual meeting.

**Corresponding author: Michael Gombola, Department of Finance, Drexel University,
Philadelphia, PA 19104. Tel. (215) 895-1743. Fax (215) 895-2955.
E-mail:gombola@drexel.edu.
                     Going-Private Restructuring and Earnings Expectations:
   A Test of the Release of Favorable Information for Target Firms and Industry Rivals

                                              Abstract

        This research documents that analysts significantly revise upward their forecasts of
earnings for targets of going-private bids subsequent to the announcement of going-private
restructuring. These upward forecast revisions are significantly related to the stock price reaction
to the announcement. Potential tax savings and the reduction of agency costs associated with
going private restructuring do not appear to explain these upward forecast revisions. Analysts
also significantly revise upward their forecasts of earnings for industry rivals of targets of going-
private bids. Furthermore, these upward forecast revisions for rivals are significantly related to
the stock price reaction for rivals. Forecast revisions for rivals are not related to proxies for
agency costs or potential tax benefits of leverage. Overall, our results provide evidence
supporting the hypothesis that going-private announcements convey favorable information about
future earnings of targets of going-private bids as well as their industry rivals.

Key words: Going-private restructuring; Information transfer; Earnings expectations
                     Going-Private Restructuring and Earnings Expectations:
   A Test of the Release of Favorable Information for Target Firms and Industry Rivals


1. Introduction

       Previous empirical studies examining the announcement effects of going-private

restructuring on stock returns document a positive announcement period abnormal stock return.

(e.g., DeAngelo, DeAngelo and Rice (1984), Lehn and Poulsen (1989), Marais, Schipper and

Smith (1989) and Slovin, Sushka and Bendeck (1991)). This empirical finding is consistent with

several explanations or hypotheses. These explanations include: (i) revelation of favorable

private information about the future earnings and cash flows of the company to be taken private

(the information hypothesis); (ii) mitigation of agency problems of free cash flow to be realized

when the company is taken private (the free cash flow hypothesis); (iii) potential tax savings

associated with financing the transaction (the tax benefits hypothesis); and (iv) wealth

redistribution from bondholders to stockholders as the bondholders bear more risk of the private

company (the wealth transfer hypothesis). The observed positive abnormal stock returns could

result from any of these hypotheses or a combination of them.

       In this study, we examine the information content of going-private transactions. We test

the hypothesis that announcements of going-private transactions convey favorable information to

the market about earnings prospects for targets of going-private bids. To test the hypothesis, we

use analysts’ forecasts of future earnings as a proxy for investor expectations, and examine

monthly revisions of analysts’ forecasts of future earnings around the going private

announcement. Examining analysts’ revisions of earnings forecasts can provide direct evidence

on the information hypothesis - that the wealth effect of going-private transactions results from




                                                 1
information about earnings prospects for companies to be taken private.

       This hypothesis predicts an upward revision of earnings forecasts by analysts subsequent

to announcement of going-private restructuring. A positive correlation between the

announcement-period stock returns and earnings forecast revisions would provide further support

for the information hypothesis. A positive correlation, however, cannot completely rule out the

contribution of other explanations to the observed positive stock price return associated with a

going-private announcement. In further analyses, we control for the potential effect of mitigation

of agency problems, tax savings, and potential contaminating information.

       We find significant and positive abnormal forecast revisions for current-year earnings of

target firms. We also find a significant positive relation between abnormal earnings forecast

revisions for target firms and the abnormal stock return around the announcement of going-

private transactions. This relation remains even after controlling for the potential effects of free

cash flow and tax benefits of debt, as well as other contaminating information. Furthermore,

results show that neither the announcement period stock returns nor the earnings forecast

revisions subsequent to the announcement can be explained by the target’s free cash flow or its

potential tax benefits of debt. Our results provide strong support for the hypothesis that going-

private announcements convey favorable information about earnings prospects of target firms.

       Slovin, Shushka, and Bendeck (1991) argue that financial markets could perceive the

private information revealed in the announcement of going-private restructuring as industry-

wide, rather than firm-specific. They document a significant announcement-period abnormal

return of 1.32 percent for industry rival firms and conclude that the financial market perceives the

information conveyed in the announcement of going-private transactions as industry-wide. As

indicated by Slovin, Shushka and Bendeck (1991), their finding of intra-industry valuation effect



                                                  2
of a going-private announcement could result from any of three contributing factors: (1) private

information revealed about improved earnings and cash flows for industrial rivals (i.e., intra-

industry information effects), (2) increased probability of takeover bids for rival firms in the

same industry (i.e., intra-industry takeover probability effects), and (3) the potential for

ameliorating the agency problems of free cash flow for other firms in the industry, in response to

a takeover bid (i.e., intra-industry agency problem effects). The observed positive stock price

reaction to a going-private buyout bid for industry rivals could be consistent with any of these

effects or a combination of them.

       In this study, we test the hypothesis of intra-industry information effects by examining

revisions in analysts’ earnings forecasts for rivals of going-private bids. By testing earnings

forecast revisions of rival firms, we extend the study by Slovin, Shushka and Bendeck (1991) in

narrowing down the factors contributing to stock price effects on rival firms. Upward earnings

forecast revisions for rival firms would provide direct evidence supporting the intra-industry

information transfer explanation. An information transfer to rivals must originate with

information about target firms. Therefore, evidence in support of an intra-industry information

transfer effect would also provide support for the information hypothesis for target firms.

Upward earnings forecast revisions for rivals would not support the other two explanations of a

positive stock price reaction. An increased probability of takeover for industry rivals could

produce a positive price reaction for these firms, but should not produce an upward revision in

earnings forecasts for rivals, particularly for current-year earnings. The potential for industry-

wide reduction in agency problems of free cash flow could possibly be associated with eventual

improvements in earnings forecasts if management preemptively improves earnings performance

in response to increased takeover probability. Although eventual earnings improvements could be




                                                  3
observed, it is doubtful that analysts would immediately revise earnings forecasts upward due to

the possibility that management could potentially reduce agency problems and improve earnings

in an attempt to thwart a takeover.

        Consistent with Slovin, Shushka and Bendeck (1991), we find significant, positive

abnormal returns for rival firms around the announcement of buyout bids. We also find rival

firms experience significant upward revisions in analyst forecasts of current-year earnings and

five-year earnings growth. In addition, we find a significantly positive relation between forecast

revisions (both current-year earnings and 5-year earnings growth) and abnormal returns for rival

firms. This relation remains significant even after controlling for other variables, including the

free cash flow of rivals, debt capacity of rivals, relative size of targets and rivals, and identity of

the acquirers, as well as potentially contaminating information. Overall, our results for rival firms

provide strong support for the hypothesis that going-private announcements convey industry-

wide favorable information about earnings prospects.

2. Hypotheses and Related Literature

2.1. Hypotheses for Shareholder Gains of Target Firms

2.1.1. Mitigation of Agency Problems

        Jensen (1986) suggests that corporate restructuring, such as going private transactions,

can be motivated by value enhancement engendered by reduction of agency costs otherwise

prevalent in firms owned by atomistic shareholders. By taking a firm private, the new

concentrated ownership group imposes discipline resulting in reduced overinvestment, and

reduced management perquisites unrelated to job performance. When investors recognize the

prospect of increased efficiency in investment and management, the market reacts positively to




                                                   4
the going-private announcement.

       In a test of this hypothesis, Lehn and Poulsen (1989) report a significant relation between

a firm's free cash flow and the takeover premium. They also find that free cash flow is a predictor

of the probability of a firm being taken private. A later study by Halpern, Kieschnick, and

Rotenberg (1999) examines a sample of firms with dysfunctionally low managerial ownership

prior to going private and finds no evidence of agency problems of free cash flow. To the

contrary, this group of firms exhibits lesser investing activity and lower free cash flow than a

matched sample of public firms or a matched sample of target firms in mergers.

       Examining a sample of firms with high managerial ownership prior to going public,

Halpern, Kieschnick, and Rotenberg (1999) also find no evidence that mitigation of agency

problems is a major motivating factor in taking these firms private. Instead, managers increase

their proportionate holdings in the firm while at the same time they reduce their dollar holdings

in the firms taken private. By substituting external debt for their own equity, managers “cash out”

of the firm taken private and increase their ability to diversify personal portfolios.

2.1.2. Tax savings

       Wealth gains observed in going-private transactions could result from tax advantages

associated with this form of corporate restructuring. As detailed by Kaplan (1989), and

Lowenstein (1985), these tax advantages include the deductibility of interest payments associated

with increased debt, increased deduction for depreciation, and the tax advantages of financing the

transaction with employee stock ownership plans. Kaplan (1989) finds that tax benefits are an

important source of the wealth gains for a sample of management buyouts. He concedes,

however, that companies could obtain many of the same tax benefits without going private.




                                                  5
       Further analysis of tax consequences of going-private transactions by Weston, Siu, and

Johnson (2001) reveals potential tax disadvantages of going-private transactions. These

disadvantages include capital gains tax paid by selling shareholders and taxes paid by the firm on

gains from asset sales as well as higher income taxes paid by the firm if it can increase its

earnings through efficiencies. Consequently, they conclude that the overall tax consequences of

a going-private transaction could be indeterminate.

       Unused tax advantages of debt financing do not appear to be unique to firms taken

private. Examining a sample of firms taken private, Halpern, Kieschnick, and Rotenberg (1999)

find no difference either in debt utilization or in tax expenditures between firms later taken

private and a matched sample of firms that remain publicly held. Consequently, they find no

evidence that would suggest that firms taken private are unique in their ability to enjoy the tax

consequences of the transaction.

2.1.3. Wealth transfer

       Since going private transactions are often financed through large increases in debt, the

financing could have an ancillary effect of wealth transfer between bondholders and

stockholders. This wealth transfer could explain (at least in part) the wealth effect in going

private transactions. Lehn and Poulsen (1988) find that the effect of a going private restructuring

on the value of debt securities is not significant. Marais, Schipper and Smith (1989) also find an

insignificant price effect on nonconvertible debt. Asquith and Wizman (1990), however, report

losses to bondholders of firms that are taken private.

2.1.4. Undervaluation and information asymmetry

       The undervaluation or information asymmetry explanation for the positive stock price




                                                  6
effect observed in going private transactions stems from differences between insiders (“informed

shareholders”) and outsiders (“uninformed shareholders”) in their expectations of future earnings

and the true value of the firm. By tendering a premium-price takeover bid, the buyout group, who

are typically either incumbent management or buyout specialists, reveal their private information.

       Halpern, Kieschnick, and Rotenberg (1999) show that targets of going-private

transactions are underperformers in earnings and stock price performance. Due to the

underperformance, it is plausible that expectations for these companies have been revised

downwards, perhaps overly so. Overdone pessimism could result in underpricing. Insiders, who

have a better understanding of future prospects, can capitalize on temporary underpricing, by

taking the firm private. At a later date, when the true prospects for the firm are better apparent to

investors, the firm can be taken public again to the profit of the group that has not succumbed to

the pessimism. DeGeorge and Zeckhauser (1993) show that after being taken private, firms

outperform their counterparts both in terms of accounting returns and stock price performance.

       Insider trading prior to going-private announcements can provide an indirect test of the

information hypothesis. Harlowe and Howe (1993) and Kaestner and Liu (1996) find that

insiders of firms taken private engage in significant abnormal buying prior to the going-private

announcement. Kaestner and Liu also show that the insider buying is not related to either

potential tax savings or free cash flow of the target firm. It could, however, be in anticipation of a

premium-price transaction. This study provides a more direct test of the information hypothesis

by testing earnings forecast revisions associated with a going-private announcement, since

earnings forecasts are a more direct measure of prospects of the firm to be taken private.




                                                  7
2.2.   Industry-Wide Information Effects

2.2.1. Favorable Private Information about Future Earnings Prospects of Rivals

       Foster (1981) describes the information transfer from an announcing firm to other firms

within an industry. Similarly, Szewczyk (1992) demonstrates how announcement of a stock

offering can lead to price reactions for similar firms. Likewise, Asness and Smirlock (1991)

describe how REIT bankruptcies affect valuation of all firms in the industry. In all of these cases,

the announcing firm generates information that is not only firm-specific but also industry-wide.

       Industry-wide revision of expectations could transpire in response to a going-private

announcement. If a going-private bid reveals private information about earnings prospects of a

target firm, this favorable information could transfer to other firms within the same industry. The

conjecture of industry-wide information effects is supported by evidence that multiple going-

private transactions within the same industry are commonly observed, to the extent that

industries could be considered subject to “buyout waves” (Ambrose and Winters, 1992).

       Just as earnings forecasts can provide a direct test of the information effect for target

firms, it can also provide a direct test of intra-industry information transfer for rival firms. This

intra-industry information transfer effect predicts a positive revision in earnings forecasts for

rival firms. It also predicts a positive relation between the positive stock price reaction shown by

Slovin Shushka and Bendeck (1991) for rival firms and their earnings forecast revisions.

2.2.2. Industry-Wide Reduction in Agency Problems of Free Cash Flow

       A takeover aimed at reducing the agency problems of free cash flow for one firm within

the industry could provide information about industry-wide agency problems and their potential

for attenuation. The ultimate effect on stock prices is not clear, however. If investors were




                                                   8
previously unaware of agency problems within the target firm and the industry, then the

discovery of such agency problems could lead to a negative stock price reaction and industry-

wide downward revision in earnings forecasts.

       If investors view a takeover bid in the industry as an indication of the potential for

reducing known industry-wide agency problems, then the shareholder wealth effect for rivals

could be positive. The positive price response could impound the future benefits of reduced

agency problems either through other takeovers within the industry or pre-emptive reduction in

agency costs in an effort to thwart future takeovers. In either case, the agency cost reduction

might not be immediate. Even though far off in the future, the prospect of industry-wide

reduction in agency costs could result in stock price improvements.

2.2.3. Increased Probability of Takeover for Rivals

       An increased probability of future control bids within the industry could affect stock

prices industry-wide solely due to the premium price at which going-private transactions are

accomplished. For example, in the case of a twenty percent takeover premium, a ten percent

increase in the probability of takeover for rival firms would increase the expected value of

takeover premium by two percent of stock price and lead to a two percent price reaction for rival

firms. Such a price increase for rivals could be observed even if the target’s assets are liquidated

or put to use in another industry.

       If the positive stock price reaction for rival firms shown by Slovin, Shushka, and Bendeck

(1991) is due solely to increased probability of takeover bids industry-wide, there should no

evidence of significant upwards earnings forecast revisions for rival firms. Likewise, there should

be little association between the stock price reaction and earnings forecast revisions for rivals. A




                                                  9
significant stock price reaction in the absence of significant upwards earnings forecast revisions

for current-year earnings would be consistent with the hypothesis that the price reaction is due to

increased probability of takeover for rivals or the hypothesis that the price reaction is due to

industry-wide reduction in agency problem of free cash flow.

3. Sample Selection and Characteristics of Buyout Target Firms and Industry Rivals

       The sample of firms taken private was obtained by searching LEXIS/NEXIS, the Wall

Street Journal Index, Dow Jones News Retrieval Service, Mergerstat Review and the Securities

Data Corp. database for the years 1980-1996. The data represent an update of the sample for

years 1980 to 1987 used by Lehn and Poulsen (1989). To be included in the final sample, the

following screening criteria must be met:

       1. The transaction must be confirmed in the Wall Street Journal. The publication date of

          the announcement in the Wall Street Journal is used as the announcement day.

       2. The announcing firm must have sufficient data in the CRSP database for the

          calculation of abnormal stock return surrounding the announcement day.

       3. The announcing firm must be followed by security analysts and have sufficient data in

          the IBES earnings database to allow calculation of earnings forecast revisions.

       4. The announcing firm is not in the utility, real estate, insurance and financial industries.

       An initial sample of 353 announcements was verified as having an announcement

published in the Wall Street Journal. This initial sample was reduced to 323 announcements of

going private restructuring with available data in the CRSP database and estimates of current-

year earnings in the IBES database. Each announcement corresponds to a unique firm that is

subject to a going-private bid. Our final sample contains 323 announcements for 323 firms.



                                                 10
       Table 1 reports the distribution of the sample by year and provides a year-by-year

comparison with the Lehn and Poulsen (1989) sample. It also provides a year-by-year

comparison of firm size for our sample and the Lehn and Poulsen sample. The number of firms

for each year of our sample is about half the number in the Lehn and Poulsen sample. This

smaller number of firms is combined with a larger firm size for our sample, which is about twice

the size of the firms in the Lehn and Poulsen (1989) sample. The larger firm size and smaller

number of firms per year of our sample is due primarily to the requirement that firms in our

sample must be included in the IBES database. Particularly for earlier years of our sample, data

for many smaller firms was not available in IBES.

       Our sample is remarkably similar to the sample employed by Kaplan and Stein (1993),

who impose a minimum total transaction value of $100 million. With the similarity to the Kaplan

and Stein (1993) sample comes the implication of their primary finding, that the characteristics of

firms taken private changed significantly between the early years and later years of their sample.

         Table 2 presents the distribution of Standard Industrial Classification (SIC) codes for the

sample of target firms. The presence of 55 separate two-digit SIC codes, with 16 of these

representing at least one percent of the sample, indicates a wide variety of industries. A

disproportionately large percentage of the sample is represented by wholesale and retail firms,

which comprise almost 22 percent of the sample.

       To develop the sample of intra-industry rivals, we utilize the CRSP daily returns file to

identify all firms in the CRSP file that share the target’s four-digit SIC code. To be included in

the sample of intra-industry rivals, the following five criteria must be met:

       1. The firm must remain publicly held throughout the sample period.




                                                 11
       2. The firm must not be a foreign firm.

       3. The firm must not have any major corporate events announced during a period from

           three days prior to three days following the target company’s announcement. Major

           corporate events include mergers, acquisitions, divestitures, stock repurchases, stock

           offerings, and earnings announcements.

       4. The firm must have sufficient data in the CRSP database for the calculation of

           abnormal stock return surrounding the announcement day.

       5. The firm must have sufficient data in the IBES earnings database to allow calculation

           of earnings forecast revisions.

           Application of these criteria produces a sample of 3,145 rival firms

4. Estimating Abnormal Returns and Abnormal Forecast Revisions

       To estimate the stock price reaction to going private announcements, we use the event

study technique. We calculate abnormal returns using the market model, with the CRSP equally

weighted index as a proxy for the market return. We estimate the coefficients of the market-

model by ordinary least squares regression using an estimation period that begins 345 trading

days before the announcement and ends 90 trading days before the announcement. To test the

statistical significance of abnormal returns, we use the t-test for the mean of the abnormal

returns.

       We estimate revisions of earnings forecasts according to the Brous (1992) methodology

to calculate the abnormal forecast revision (AFR), which adjusts for serial correlation in earnings

forecasts and a tendency for successive downward revision of analysts' forecasts in unadjusted




                                                 12
forecast revisions (FRs).1

           We define FR as the monthly change in the mean of earnings forecasts by analysts,

standardized by stock price per share:

(1)                  FRi,t = (Fi,t - Fi,t-1)/Pi

Where FRi,t is the forecast revision of earnings for firm i from month t-1 to month t, Fi,t and Fi,t-1

are the mean values of analysts' earnings forecasts for firm i at months t and t-1, respectively, as

reported in the IBES database. Pi is the stock price for firm i at the end of the month prior to the

announcement of going private restructuring.2

           O'Brien (1988) shows that forecast revisions as defined in Equation (1) might be subject

to an optimism bias. Optimistic forecasts at the beginning of the fiscal year are systematically

revised downward as the year progresses. Brous (1992) provides evidence of serial correlation in

earnings forecast revisions that might be due to the fact that not all analysts revise their forecasts

on a monthly basis. Consequently, information released during one month can be reflected in

changes in analysts' earnings forecasts during several subsequent months. Brous and Kini (1993)

report that approximately 20 percent of analysts revise their earnings forecasts each month. This

percentage implies that information will be fully reflected in analysts' forecast revisions during

the four- to five-month period following the announcement.

           Since there is the potential for both optimism bias and serial correlation in the forecast


  1
      A detailed description of this methodology is available in the Appendix provided by Brous and Kini (1994).
  2
    Christie (1987) presents the theoretical and empirical benefits of normalizing by price per share rather than earnings
per share. Pound (1988) provides an example to demonstrate the merits of normalizing by price per share instead of
earnings per share. This example considers a firm with a stock price of $30 per share. An earnings change from $-0.01
to $0.03 per share would be represented as .04/30 and a change from $0.01 to $0.03 would be represented by .02/30.
Normalizing by price per share maintains the relative importance of these revisions. Normalizing by earnings might not.




                                                           13
revisions, we estimate the abnormal forecast revision in a manner consistent with Brous (1992)

to adjust for these two potential problems. Using data for all IBES firms, Brous and Kini (1993)

show evidence that their forecast adjustment process used to determine AFRs induces no new

bias for a random month.

       We estimate the abnormal forecast revision for each firm as the difference between the

forecast revision, FR, and the expected forecast revision, E[FR]:

(2)            AFRi,t = FRi,t - E[FRi,t]

       We estimate the expected forecast revision for firm i in month t, E[FRi,t], as:
                                    n-1

(3)            E[FRi,t] = ki + (1/n)   e    i,t-c

                                       c=1


       The expected forecast revision equals the forecastable component (ki) plus the equally

weighted average of the n previous months' unexpected component. We estimate the forecastable

component, ki, for each company in our sample. We use all months of earnings forecasts in the

IBES database for the company except the 12-month period beginning six months prior to the

going private announcement and ending six months after the announcement. The unforecastable

component, ei,t-c, is equal to the difference between the ki and the actual monthly forecast

revision for firm i in month t-c.

       Consistent with Brous (1992), the value of n in Equation (3) is taken as five, which is

equivalent to assuming that analysts' forecasts follow a fourth-order moving average process. A

fourth-order process is consistent with a four-month lag in analysts' revisions.

       We estimate cumulative AFRs beginning with the announcement month (month 0),




                                                     14
which we define as the first month in which IBES data could partially reflect information

conveyed by the going private announcement. We calculate cumulative AFRs over two-month

(0,1) and four-month (0,3) periods. A four-month (0,3) cumulation period should capture all

revisions associated with an announcement.3

         We test the statistical significance of AFRs and cumulative AFRs, using the parametric t-

test for the mean of the AFRs and cumulative AFRs and the nonparametric Wilcoxon signed-

rank test for the median of the AFRs and cumulative AFRs.

5. Results

5.1. Abnormal stock returns for Target Firms and Industry Rivals

         Table 3 reports average abnormal stock returns for target firms in Panel A and for

industry rivals in Panel B. The results provide strong evidence of positive and significant

abnormal returns for both target firms and industry rivals during the announcement period. For

targets, the (0,1) announcement interval has a cumulative 2-day return of 12.68 percent, which is

significant at the 0.01 level (t=51.98). Each of the two days within that interval is also significant

at the .01 level. Evidence of significant returns is also shown for the days prior to the

announcement. The abnormal return for the interval beginning 20 days prior to the announcement

and ending 1 day prior to the announcement is 8.63 percent, which is significant at the .01 level

(t=11.02). Combining this pre-announcement interval with the two-day announcement interval

produces a cumulative abnormal stock return of 21.31%. The size of this cumulative return is

similar to that shown in previous studies by DeAngelo, DeAngelo and Rice (1984) and Slovin,


  3
     The IBES database records analysts' earnings forecasts on the third Thursday of each month. If a forecast is made
after the third Thursday of the month, it is included in the forecasts for the following month. Announcements days are
therefore categorized into IBES months, rather than calendar months.




                                                         15
Sushka and Bendeck (1991).

         The abnormal return for rivals over the (0,1) announcement interval is 0.33 percent,

which is statistically significant at the .01 level (t=3.01). The abnormal return for rivals for the

interval beginning 20 days prior to the announcement and ending one day prior to the

announcement is 1.57 percent, which is significant at the .01 level (t=4.48). The overall abnormal

return over the entire (-20, +1) event period is 1.91 percent, which is statistically significant at

the .01 level (t=5.18). This cumulative abnormal stock return is about one tenth of the cumulative

average abnormal stock return for target firms shown in Panel A.

         Although on a per-firm basis, the reaction for rival is only one-tenth the reaction for

targets, the aggregate effect is almost the same. Based on market values, the aggregate value

effect for the entire sample of rival firms is about $50 billion, compared to an aggregate of $58

billion for the entire sample of target firms.4 The aggregate shareholder gain for rival firms is

about 86% of aggregate shareholder gain for target firms5.

         The sample of rival firms also displays some evidence of positive abnormal returns after

the announcement. The cumulative abnormal return of 0.32% over the (+2,+5) post-

announcement period is significant (t=1.99, not shown in Table 3), with much of the return due

to a return of 0.18% that is significant at the .05 level (t=2.24) in day +4 alone. None of the other

daily post-announcement returns is significant. A significant post-announcement return could

result either from a lag in the information effect or to some contaminating information for rivals.




4
   The aggregate market value for the 323 target firms is $171,949 million, as shown in Table 1. The aggregate
market value of the 3145 rival firms is $2,679,540 million, or more than 10 times larger.
5
   Slovin, Shushka and Bendeck (1991) find an aggregate value effect for their sample of rival firms that is about 83
percent of the aggregate value effect for their sample of target firms.



                                                         16
5.2. Abnormal Forecast Revisions of Earnings for Target Firms and Industry Rivals

        Table 4 presents monthly and cumulative abnormal forecast revisions (AFRs) of current-

year earnings and five-year earnings growth around the announcement of going private

restructuring for target firms in Panel A and for rival firms in Panel B. The table includes results

for the announcement month and each of the three months before and after the announcement.

        Panel A provides strong evidence of significant upward abnormal revisions in analysts'

forecasts of current-year earnings for target firms subsequent to going-private announcements.

The average AFR of current-year earnings for the announcement month (Month 0) is positive and

significant at the 0.01 level (t=3.17). For a firm with a price/earnings (P/E) multiple of 13.75,

which is the median P/E for the sample, the AFR of 0.0051 for the announcement month

corresponds to an increase of 7% (=0.0051*100*13.75) in forecasted earnings for that single

month. The cumulative AFR of current-year earnings for the 2-month (0,1) period is also positive

and significant at the 0.01 level (t=2.41). The cumulative AFR for the 4-month (0,3) period,

however, is positive, but not statistically significant (t=1.59).

        For each of the three months immediately preceding the announcement, the AFR is not

significant. Month –3 has a positive, but insignificant AFR, and Months –2 and –1 have negative

and insignificant AFRs. Whereas stock price behavior shown in Table 3 provides evidence of

information leakage prior to the announcement, earnings forecast behavior does not. Either

analysts are not aware of an impending going-private transaction prior to its announcement or

they refrain from employing private information to update earnings forecasts until the

information is made public.

        Once the impending going-private announcement is made public, however, analysts react




                                                  17
quickly with revisions of earnings forecasts. Virtually all of the four-month (0,3) AFR and the

two-month (0,1) AFR is provided by revisions during month 0. The AFR for Month 1 is positive,

but not significantly different from zero. The AFRs for Months 2 and 3 are also not significant.6

This concentration of the AFR in the month immediately after the announcement is unusual.

Typically, only about 20 percent of analyst forecasts are updated in any single month, according

to Brous and Kini (1993), which would result in announcement effects spread out over several

months. For going-private transactions, however, the transaction is often executed within a

month or two after the bid announcement. Analysts are unlikely to prepare forecast revisions for

companies that are no longer publicly traded. Consequently, earnings forecast revisions are

unlikely after the company has been taken private.

         Panel A of Table 4 also contains monthly and cumulative AFRs of five-year earnings

growth rates for target firms around the announcement of going private restructuring. The results

in this table do not provide evidence of significant revisions of long-term earnings by analysts.

None of the months examined, from three months before to three months following the

announcement month, shows significant revisions, either positive or negative.

         Finding significant revision of current-year earnings but not of five-year earnings growth

is common in other studies of information effects on earnings expectations [e.g. Brous (1992),

Gombola and Liu (1999)]. In the specific instance of going-private transaction, the question of

five-year earnings growth may become moot if the company is not expected to remain public

over the forecasting period. If analyst earnings forecasts are provided in response to investor

interest, then a lack of investor interest in earnings growth of companies becoming private could


6
   We also extend the examination of AFRs to Months 4, 5, and 6. There is no evidence of significant forecast
revisions of either current-year earnings or five-year earnings growth for these months. Results for theses months are
not shown in Table 4 but are available from the authors.



                                                         18
limit the interest of analyst in providing such forecasts. Consequently, we would be reluctant to

conclude from the results shown in Table 4 that only current-year earnings forecasts and not five-

year growth estimates are affected by information released in going-private announcements.

        Panel B of Table 4 presents abnormal forecast revisions for industry rival firms for the

announcement month as well as three months before the announcement to three months after the

announcement. The current-year earnings AFR for the announcement month (Month 0) is

positive and significant at the 0.05 level (t=2.01). Likewise, the AFR for the following month

(Month 1) is also positive and significant at the 0.05 level (t=2.14). The following two months (2

and 3) also show positive AFRs, but the revision is not significant. The cumulative AFR of

current-year earnings for the 2-month (0,1) period is also positive and significant at the 0.01 level

(t=3.00), as is the AFR for the four-month (0,3) interval (t=3.21).7

        Results shown in Table 4 for current-year AFRs of industry rival firms differ from those

for target firms in the time and the size response to the announcement. For target firms, analysts

appear to revise earnings forecasts immediately upon the announcement. For rival firms, the

response is not immediate. Instead, analysts might wait until a regular quarterly update before

revising earnings forecasts.

        The abnormal forecast revisions for target firms are about ten times as large as those for

rival firms. For example, the two-month AFR for target firms is 0.007 as compared to a value of

0.0007 for rival firms. For a rival firm with a P/E of 14.91, which is the median P/E for the rival

sample, there is an increase of 1.04% (=0.0007*100*14.91) in forecasted earnings for the two-


7
   We also extend the examination of AFRs to Months 4, 5, and 6. We find some evidence of upward forecast
revisions of current-year earnings for Month 5. There is no evidence of upward forecast revisions of five-year
earnings growth for any of these months. Results for theses months are not shown in Table 4 but are available from
the authors.



                                                        19
month (0,1) interval. For a target firm with a P/E of 13.75, which is the median P/E for the target

sample, there is an increase of 9.8% (=0.00716*100*13.75) in forecasted earnings for the two-

month (0,1) interval. The disparity in size of earnings forecast revisions, which differs by a factor

of ten between targets and rivals is consistent with the difference in size of the abnormal stock

returns between target firms and their industry rivals, which also differs by a factor of ten, as

shown in Table 3. Since the number of rival firms exceeds the number of targets by a factor of

ten, the aggregate change in earnings forecasts for rivals approaches the change for targets. .

       Panel B of Table 4 also presents monthly and cumulative AFRs of five-year earnings

growth rates for rival firms. The AFR for the announcement month is significant at the 0.10 level

(t=1.76). The cumulative forecast revision for the two-month (0,1) interval including the

announcement month and the following month is positive and significant at the .01 level

(t=2.60). Evidence of significant upward revision in long-term earnings growth estimates for

rival firms is particularly notable since parallel results are not observed for target firms. One

explanation we can provide for this difference is that analysts no longer provide estimates of

long-term growth for firms to be taken private. For their industry rivals, however, the question of

long-term earnings growth is not moot, but is met by analysts.

5.3. Regression Analyses for Target Firms

       Support for the hypothesis that going-private announcements convey favorable

information about future earnings of targets requires a positive association between earnings

forecast revisions and the stock price reaction to the announcement of a going-private

announcement. Tables 3 and 4 show a positive stock price reaction and upward revision of

earnings forecasts for target firms, but not the necessary relation between the two. In this section

we present results of regression analyses examining the relation between earnings forecast



                                                  20
revisions and stock price reaction to a going-private announcement for target firms while

controlling for contaminating effects on either variable.

         The significant positive stock price reaction for target firms and the significant positive

earnings forecast revisions for target firms could both result from ancillary financing effects of a

going-private transaction. The debt employed to finance many going-private transactions could

simultaneously increase both the stock price and expected future earnings per share. For a

profitable firm, substitution of debt for equity could increase earnings per share even if operating

earnings do not improve.8 The increased leverage could also produce tax savings, as Kaplan

(1989) explains.

         To untangle the relation between leverage, earnings, and share price, we examine the

leverage change for each going-private transaction and later include this measure as a control

variable in the regression analyses. Table 5 presents the leverage (measured by the ratio of debt

to assets) of firms taken private for each year from 3 years prior to the going-private

announcement to three years following the announcement. This year-by-year change in financial

leverage is presented in Panel A for all firms in the sample and in Panel B for the set of 65 firms

with available data for every year during the seven-year period.

         The full sample shows a slight decrease in leverage for the years immediately prior to the

going-private announcement, followed by a moderate increase for the years following the

announcement. A similar pattern is shown by the smaller sample of 65 firms. In either case, the

leverage ratio increases from about 50% in the announcement year to about 65% for the third

year following the announcement. Most of this increase occurs during the year immediately

following the announcement. The changes are not significantly different from zero at any


8
    The increase in EPS would not necessarily increase share price since a risk increase accompanies the earnings




                                                         21
standard level of significance for either sample.

            Somewhat notable is the tendency of firms taken private to maintain higher levels of debt

years after the transaction. Although the reputation for leveraged transactions is that the new

owners attempt to unwind the leverage as quickly as possible, the process for firms in our sample

appears quite slow. One explanation for this behavior follows from the analysis of tax benefits of

debt by Graham (2000). Once debt is added in the going-private process, previously under-

leveraged firms could become comfortable at the new higher level of debt as they enjoy greater

tax advantages. Another explanation is that the firm, although intending to reduce debt after the

transaction is unable to reduce debt as quickly as desired. In either case, the increase in debt is

more than just transitory.

            In addition to leverage changes, the firm’s free cash flow position could also affect the

results shown for stock price reaction and earnings forecast revisions for target firms. Firms with

severe agency problems of free cash flow could enjoy both an increase in share value and an

increase in earnings if the agency problems are reduced or eliminated by a new management team

once the firm is taken private. Greater free cash flow and greater agency problems would allow

greater opportunity to reduce these agency problems and increase earnings and share price. The

opportunity to reduce agency problems of free cash flow should be reflected immediately in the

stock price reaction to a going-private announcement. The stock price reaction is primarily due to

the premium paid in order to take the firm private. This premium paid should impound any

benefits of reduction in agency problems of free cash flow anticipated by the buyer group. The

reduction in agency problems may not be reflected in increased earnings until long after the firm

becomes private.


increase.



                                                     22
       Examining the association between earnings forecast revisions and the stock price

reaction in a regression model is further complicated by measurement problems for these two

variables. First, the measurement period for stock return effect differs from the measurement

period for forecast revisions, with the latter extending over at least a month. Second, earnings

forecast revisions are much more susceptible to information contamination than are stock price

changes. The danger of information contamination for stock returns is limited to information

released during the short announcement period for measuring stock price effect. Because

earnings forecasts are not immediately updated, an information release may not be captured in

earnings forecast revisions until months after the announcement. Consequently, the observed

abnormal forecast revisions, such as shown in Table 4, could be contaminated by effects of

information released outside of the announcement period for measuring stock returns but within

the period for measuring earnings forecast revisions.

       To control for the contaminating effect of prior information releases on earnings forecast

revisions, previous studies [e.g. Hertzel and Jain (1991) and Gombola and Liu (1999)] employ

cumulative abnormal stock returns outside of the abnormal return event period in examining the

association between earnings forecast revisions and market reaction. Information effects revealed

outside the abnormal return period that could contaminate the forecast revision are then captured

in these abnormal returns.

       The information effect could also depend upon the identity of the acquirer. Slovin,

Shushka, and Bendeck (1991), while noting that this effect on prices is an empirical question,

demonstrate a significantly greater price reaction for targets acquired by groups including a

buyout specialist than for management-led buyout teams.

       The resulting regression model then can be expressed as:



                                                23
(4)     AFRi = b0 + b1ARi + b2CH_DEBTi + b3CFLOWi + b4PAARi + b5MBO + b6SIZE + ei

Where AFRi is the abnormal earnings forecast revision for target firm I for the two-month (0,1)

interval, ARi is the cumulative abnormal return for firm i during the 22 day, (-20, +1), interval,

CH_DEBTi is the change in leverage between three years prior to and three years following the

announcement for firm i, as shown in Table 5. CFLOWi is the ratio of cash flow to assets for the

year immediately prior to the going-private announcement for firm i. It is constructed according

to the definition of free cash flow employed by Lehn and Poulsen (1989) and Lang, Stulz and

Walkling (1991), using data from the annual Compustat files. PAARi is the pre-announcement

cumulative abnormal return for firm i during the (-91, -21) day interval. MBOi is a dummy

variable that takes a value of one if the buyout team for the ith target firm is management-led,

and zero otherwise. SIZEi is the natural logarithm of equity market value for the ith target firm,

the product of shares outstanding and closing price per share for the announcement month.

       Table 6 contains the results of estimating Equation (4) using two alternative

specifications of the dependent variable. In the first specification, shown in Panel A, AFR shown

is for current-year earnings. In the second specification, shown in Panel B, AFR shown is for

five-year earnings growth. The results in Panel A show that the relation between AFR for

current-year earnings and announcement-period abnormal return (AR) is positive and significant

at the 0.01 level (t=2.72). This result is consistent with the prediction of the information

hypothesis.

       The coefficient for the variable representing the change in debt (CH_DEBT) is not

statistically significant. Also, the coefficient is negative, which is opposite to the predictions of

the tax benefit hypothesis. The sign for the variable representing free cash flow (CFLOW) is also

negative, but is significant. The free cash flow hypothesis predicts a positive relation between



                                                  24
free cash flow and future earnings improvement, which is opposite to the observed sign.

        The coefficient for pre-announcement abnormal stock return (PAAR) is significant and

negatively related to the AFR for current-year earnings. A possible explanation for the negative

sing is that greater market underperformance prior to the announcement is reversed by the

announcement. It is consistent with a reversal in investor expectations about the future prospects

of the firm to be taken private.

        The coefficient for dummy variable representing the presence of a management-led

takeover (MBO) is negative and significant at the .05 level (t=-2.10). This result indicates that

upwards revisions of earnings forecasts are smaller for management buyouts that for buyouts by

outsiders. Analysts might be able to understand the intentions of outsiders in a going-private

transaction, whereas intentions of management-led groups could be less transparent. The

coefficient for firm size (SIZE) is also negative and significant(t=-2.34). The significance

underscores the need to incorporate this control variable in studies of earnings forecast revisions,

just as in event studies.

        The results shown in Panel B of Table 6 show little relation between AFR for five-year

earnings growth and any of the independent variables. The relation between abnormal stock

return and AFR for five-year earnings growth is not statistically significant (t=-0.55). This lack of

statistical significance is not surprising in light of the insignificant revisions of five-year earnings

growth forecasts, shown in Table 4. Two possible explanations might underlie these findings.

First, the results could indicate that the favorable information conveyed at the going private

announcements concerns only transitory, but not permanent, changes in earnings. Secondly, the

lack of significant revisions in long-term earnings growth forecasts might stem from limited

incentive for analysts to generate such forecasts for firms that are about to become private.



                                                  25
        An alternative examination of the relation between earnings forecasts revisions and the

stock price effect can be performed with the forecast revisions employed as an independent

variable to explain the stock price reaction as the dependent variable. The resulting regression

model becomes an examination of the relation between abnormal returns and earnings forecast

revisions, which tests the information hypothesis, while controlling for free cash flow and debt

change. These two variables serve as proxies for testing effects of the free cash flow hypothesis

and tax benefit hypothesis. This alternative regression model then can be expressed as:

(5)     ARi = b0 + b1AFRi + b2CH_DEBTi + b3CFLOWi + b4PAARi + b5MBOi + b6SIZEi + ei

Although all variables are as previously defined in Equation (4), interpretation of the independent

variables changes. The pre-announcement abnormal return, (PAAR) instead of serving as a

means to control for information contamination in the dependent variable, becomes a measure of

prior stock price “run-up”.9 The model is estimated alternately with AFR for current-year

earnings forecasts and then as AFR for five-year growth forecasts.

        Table 7 contains the results of cross-sectional regressions expressed in Equation (5). The

results shown in Model 1, indicate that the relation between AFR for current-year earnings and

AR is positive and significant at the 0.01 level (t=2.72). This result confirms the finding in Table

6 of a positive relation between AFR and AR and is consistent with the information hypothesis.

The relation does not extend to five-year earnings growth forecasts shown in Model 2. Neither

the variable representing free cash flow (CFLOW) nor the variable representing the impact of

potential tax savings (CH_DEBT) significantly adds to the explanation of abnormal stock return.

Neither is significant in either model. The variable measuring pre-announcement price run-up


9
 Similar measures of “run-up” have been employed in other studies of information effects, such as the examination
of the price effects of stock offering announcements by Comment and Jarrell (1991).



                                                        26
(PAAR) is significantly related to AR in Model 1, suggesting some information leakage prior to

the announcement.

       The coefficient for the dummy representing presence of a management buyout (MBO) is

positive and significant at the .05 level (t=2.52) for Model 1 and significant at the .10 level

(t=1.75) for Model 2. These results are consistent with the findings of Slovin, Shushka and

Bendeck (1991) who also find that management-led buyouts are associated with higher abnormal

returns in response to going-private announcements.

       The difference in sign for MBO between results in Table 6 and results in Table 7 should

not be disconcerting. A management-led buyout can be associated with higher abnormal stock

returns, but not necessarily higher earnings forecast revisions than a buyout by outsiders. The

abnormal stock return from a buyout announcement is primarily due to the premium offered by

the takeover group. A management-led group may be confident in offering a higher premium

over the previous market price for the same level of earnings, or earnings improvement.

       Results for target firms shown in Tables 6 and 7 provide strong evidence of a significant

and positive relation between the earnings forecast revisions for current-year earnings for firms

subject to going-private bids and the stock price reaction to the announcement. This result is

consistent with the hypothesis that going-private announcements convey information about

earnings prospects for targets of going-private bids. There is very little evidence that changes in

debt or the free cash flow position of targets firms is related to either improvement in earnings

forecasts or increases in shareholder wealth associated with the going-private announcement.

5.4. Regression Analyses for Rival Firms

       Testing for an intra-industry information transfer for rival firms requires examining the




                                                 27
association between earnings forecast revisions for rival firms and their stock price reaction to a

going-private bid within the same industry. A positive association between earnings forecast

revisions and the stock price reaction for industry rivals would provide support for the hypothesis

that going-private announcements contain industry-wide favorable information about future

earnings prospects. This section presents results of testing this association in a regression model.

       Controlling for cash flow effects and change in leverage is necessary in examining the

relation between earnings forecast revisions and stock price reaction in a regression model for

rivals, just as for targets. To the extent that agency problems of free cash flow are industry-wide,

a going-private transaction could motivate managers of rival firms to improve earnings

performance in an effort to thwart potential future takeover bids. Managers could also try to

improve earnings through increasing leverage to take advantage of tax benefits and improve

earnings per share. Levering up the firm would also further protect against unwanted going-

private bids by reducing the attractiveness as a takeover target. The lower the rival’s debt, the

greater is the opportunity to increase leverage and tax benefits.

       Since the intra-industry information transfer from targets to industry rivals should be

greater for more prominent targets in the industry, we also include a measure of relative firm size

between target firms and rival firms in the regression models. According to the intra-industry

information transfer hypothesis, we should expect a greater effect on earnings for a target firm

larger than its corresponding rival.

       The information transfer effect could also depend upon the identity of the acquirer.

Slovin, Shushka, and Bendeck (1991) and our results shown previously in Table 7 demonstrate a

significantly greater price reaction for targets acquired by management-led buyout teams than

outsiders. The effect of the identity of the acquirer on information transferred to rival firms is



                                                  28
uncertain. If management-led teams have access to superior information about industry prospects

than outsiders, a greater earnings forecast revision for rivals should be observed. Alternatively,

management-led going-private transactions could be based more on company-specific

information whereas investor-led transactions could be based on valuations within an industry

and less on company-specific characteristics. Consequently, the effect of this variable is an

empirical question.

       Similar to targets, the pre-announcement stock return, or price “run-up” can be used to

control for announcements prior to the stock return announcement period. In addition, for rivals,

the post-announcement stock return can be used to control for events after the announcement

period that could be reflected in earnings forecast revisions but not in the announcement-period

return. In the case of targets, controlling for post-announcement events is unnecessary since the

company has been taken private quickly.

        The resulting regression model is:

(6)    AFRi = b0+ b1ARi + b2CFLOWi + b3DEBTi + b4RELSIZEi + b5MBOi + b6PAARi +

               b7PSARi + ei

where AFRi , ARi , CFLOWi, PAARi and MBOi are as defined previously for target firms, but

constructed for the sample of rival firms. DEBTi is the ratio of debt to assets for the ith rival

firm, taken from COMPUSTAT for the year immediately prior to the going-private

announcement, RELSIZEi is a dummy variable that takes a value of one if the target is larger

than the ith rival firm, and zero otherwise, and PSARi is the post-announcement abnormal return

for the ith rival firm during the +2, +90) day period.

       Table 8 presents results of estimating the regression model expressed in Equation (6).




                                                 29
Model 1, constructed for AFRs for current-year earnings, shows evidence of a positive and

significant (t=3.98) relation between the AFR and the AR for industry rival firms. The highly

significant association between AFR and AR provides very strong support for the hypothesis that

the market reaction for rivals is due to favorable information about their earnings prospects.

Neither the proxy for free cash flow (CFLOW) nor the proxy for tax benefits of unused debt

capacity (DEBT) are statistically significant in this model. The measure of relative size

(RELSIZE) is positive and statistically significant at the 0.01 level (t=2.56). This result indicates

that larger size of targets is associated with a greater earnings forecast revision for rivals.

Information about larger firms should have greater industry-wide information effects.

        The coefficient for the dummy variable for management-led buyouts (MBO) is not

significant at any conventional level. Also, neither the coefficient for pre-announcement

abnormal return (PAAR) nor post-announcement abnormal return (PSAR) is statistically

significant. The sign of the PAAR variable is opposite to one that would indicate information

contamination.

        In Model 2, the relation between AFR for 5-year growth rates and AR for industry rival

firms is positive, but significant only at the 0.10 level. There is no evidence of a relation between

the AFR for 5-year growth rates and either the pre-announcement abnormal return (PAAR)or

post-announcement abnormal return (PSAR).

        The regression model presented in Equation (6) allows considerable control for

confounding effects on earnings forecast revisions, but provides little inference about abnormal

returns since abnormal returns are used only as an independent variable. In order to allow

inference about abnormal returns, an alternative examination of the relation between abnormal

returns and earnings forecast revisions for rivals can be performed by employing earnings



                                                  30
forecast revisions as an independent variable and abnormal stock returns as the dependent

variable. With market reaction as the dependent variable, a regression model can be built that

incorporates alternative explanations for the stock price reaction for rival firms.

        Slovin, Shushka and Bendeck (1991) point out that the positive stock price reaction for

rivals could result from: (1) buyout bids may reveal private information about the future cash

flows of target firms and their intra-industry rivals (2) buyout bids could produce an increase in

the probability that rival firms may become the target of future control bids and (3) buyout bids

could be an indicator of industry-wide agency problems subject to reduction either by current

management or by new management after a successful buyout bid.

        Earnings forecasts can serve as a proxy for the future cash flows of inter-industry rivals,

the probability that rival firms could become targets of future control bids can be proxied by the

rival firm’s debt and cash flow position. Firms with less debt can support greater takeover

financing in a going-private transaction and are more likely to be future targets of going-private

bids. Firms with greater cash flow can also support greater takeover financing and are also more

likely to become future targets. In addition to serving as a proxy for probability of takeover, cash

flow serves as a direct measure of the agency problems of free cash flow. The greater the cash

flow of rivals, the greater are the potential benefits of their reduction.

        Incorporating measures for the identity of the buyer (management vs. outsider) and pre-

announcement stock return, the regression model becomes:

(7)    ARi = b0 + b1AFRi + b2CFLOWi + b3DEBTi + b4MBOi + b5PAARi + ei

Where all of the variables are as defined for Equation (6). The regression model does not include

the variable for relative firm size of targets and rivals (RELSIZE). Relative firm size is




                                                  31
significantly related to AFRs for rivals, as shown in Table 8. The inclusion of this variable would

induce multicollinearity in the model.

       Table 9 reports results of estimating the regression model expressed in Equation (7).

Model 1, which incorporates current-year earnings as an independent variable, shows a positive

relation that is significant at the 0.01 level (t=3.77) between AFR and abnormal return (AR) for

rival firms. This level of significance is very similar to that found when AFR for the current-year

earnings is employed as the dependent variable shown in Table 8, indicating robustness of the

association between AFR for current –year earnings and AR for rival firms. Also similar to table

8, neither the proxy for free cash flow (CFLOW) nor the proxy for tax benefits of unused debt

capacity (DEBT) is statistically significant in this model.

       There is some weak evidence (significant at the 0.10 level) of greater intra-industry

information transfer for investor-led buyout bids than management-led bids. This result could be

due to management-led bids implying more company-specific information on the target firm’s

value whereas outsider-led bids imply industry-wide information. Pre-announcement abnormal

return (PAAR) is significantly related to the abnormal return for rivals.

       The relation between AFR for 5-year growth rates and AR for industrial rivals, shown in

Model 2 in Table 9, is positive, but significant only at the 0.10 level. Also significant at the 0.10

level is the relation between the bidder’s identity (MBO) and AR in this model. Model 2 also

shows a highly significant relation between pre-announcement abnormal return (PAAR) and

announcement-period abnormal return (AR) for rivals.

       Results shown in Table 9 provide support for the hypothesis that the positive market

reaction to going-private announcements observed for industry rivals is due to inter-industry




                                                 32
information transfer about future earnings prospects for target firms and their inter-industry

rivals. This hypothesis is the first of the three hypotheses presented by Slovin, Shushka and

Bendeck (1991) to explain the positive abnormal returns of rival firms to a going-private

announcement. There is no evidence that the rival’s abnormal return is related to its free cash

flow position or its leverage and therefore no support for the other two hypotheses.

6. Conclusions

       This study provides direct evidence supporting the hypothesis that going-private

announcements convey favorable information about future earnings prospects not only for target

firms but also for industry rivals. It documents a significant upward revision in analysts' forecasts

of current-year earnings subsequent to an announcement of going private restructuring for target

firms and for industry rivals. It also documents a significantly positive relation between revisions

of analysts' forecasts of current-year earnings and the abnormal stock return associated with the

going private announcement for target firms and for industry rivals. This relation remains even

after controlling for leverage changes and free cash flow. There is also some evidence of upward

revisions in analyst forecasts of five-year earning growth for rival firms and some evidence of a

positive relation between these revisions and the abnormal stock return for industry rivals. There

is little evidence that the upward revisions of analysts' earnings forecasts or the positive stock

price reaction can be attributed to either free cash flow or potential tax benefits of increased

leverage for target firms. Likewise, neither the upward revision of analysts’ earnings forecasts

nor the positive stock price reaction is related to the cash flow or leverage position of rival firms.




                                                  33
References

Ambrose, Brent and Drew B. Winters, 1992, Does an industry effect exist for leveraged
buyouts?, Financial Management, 21, 89-101.

Asness, C. and M. Smirlock, 1991, A note on REIT bankruptcy and intra-industry information
transfer: An empirical analysis, Journal of Banking and Finance 15(6), 1171-1182.

Asquith, Paul, and Thierry Wizman, 1990, Event risk, wealth redistribution, and the return to
existing bondholders in corporate buyouts, Journal of Financial Economics 27, 195-214.

Brous, Peter, 1992, Common stock offerings and earnings expectations: a test of the release of
unfavorable information, Journal of Finance 47, 1517-1536.

Brous, Peter and Omesh Kini, 1993, A reexamination of analysts’ earnings forecasts for takeover
targets, Journal of Financial Economics 33, 201-226.

Brous, Peter and Omesh Kini, 1994, The valuation effects of equity issues and the level of
institutional ownership: Evidence from analysts’ earnings forecasts, Financial Management 23
(1), 33-46.

Christie, Andrew, 1987, On cross-sectional analysis in accounting research, Journal of
Accounting and Economics 9, 231-258.

Comment, Robert, and Gregg A. Jarrell, 1991, The relative signaling power of Dutch-auction and
fixed-price self-tender offers and open-market share repurchases, Journal of Finance 46(4), 1243-
1271.

DeAngelo, Harry, Linda DeAngelo, Linda, and Edward Rice, 1984, Going private: Minority
freezeouts and stockholder wealth, Journal of Law and Economics 27, 367-402.

DeGeorge, Francois and Richard Zeckhauser, 1993, The reverse LBO decision and firm
performance: Theory and evidence, Journal of Finance 48(4), 1323-1348.

Foster, George, 1981, Intra-industry information transfers associated with earnings releases,
Journal of Accounting and Economics 3(3), 201-232.

Gombola, Michael J. and Feng-Ying Liu, 1999, The signaling power of specially designated
dividends, Journal of Financial and Quantitative Analysis, 34(3), 409-424.

Graham, John R., 2000, How big are the tax benefits of debt? Journal of Finance 55, 1901-1941.

Halpern, Paul, Robert Kieschnick, and Wendy Rotenberg, 1999. On the heterogeneity of leverage
going private transactions, Review of Financial Studies 12, 281-309.

Harlow, W.V. and John S. Howe, 1993, Leveraged buyouts and insider nontrading, Financial



                                                34
Management 22, 109-118.

Hertzel, Michael, and Prem C. Jain, 1991, Earnings and risk changes around stock repurchase
tender offers, Journal of Accounting and Economics, 14, 253-274.

Jensen, Michael, 1986, Agency costs of free cash flow, corporate finance, and takeover,
American Economic Review 76, 323-329.

Kaplan, Steven N., 1989, Management buyouts: Evidence on taxes as a source of value, Journal
of Finance 44, 611-632.

Kaplan, Steven N. and Jeremy C. Stein, 1993, The evolution of buyout pricing and financial
structure in the 1980s, Quarterly Journal of Economics, May, 313-357

Kaestner, Robert, and Feng-Ying Liu, 1996, Going private restructuring: The role of insider
trading, Journal of Business Finance and Accounting 23 (July), 779-806.

Lang, Larry H.P., Rene M. Stulz, and Ralph A. Walkling, 1991, Managerial performance, Tobin's
Q, and the gains from successful tender offers, Journal of Financial Economics 24, 137-154.

Lehn, Kenneth, and Annette Poulsen, 1988, Leveraged buyouts: wealth created or wealth
redistributed, in Murray Weidenbaum and Kenneth Chilton, eds.: Public Policy Towards
Corporate Takeovers (Transaction Publishers, New Brunswick, N.J.)

Lehn, Kenneth, and Annette Poulsen, 1989, Free cash flow and stockholder gains in going
private transactions, Journal of Finance 44 (July), 771-787.

Lowenstein, Louis, 1985, Management buyouts, Columbia Law Review 85, 730-784.

Marais, Laurentius, Katherine Schipper, and Abbie Smith, 1989, Wealth effects of going-private
for senior securities, Journal of Financial Economics 23, 155-191.

O'Brien, Patricia, 1998, Analysts' forecasts as earnings expectations, Journal of Accounting and
Economics 10, 187-221.

Pound, John, 1988, The information effects of takeover bids and resistance, Journal of Financial
Economics 22, 207-227.

Slovin, Myron B., Marie E. Sushka, and Yvette M. Bendeck, 1991, The intra-industry effects of
going-private transactions, Journal of Finance 46(September), 1537-1550.

Szewczyk, Samuel H., 1992, The intra-industry transfer of information inferred from
announcements of corporate security offerings, Journal of Finance 47(5), 1935-1945.

Weston, J. Fred, Juan Siu, and Brian Johnson, 2001, Takeovers, Restructuring, and Corporate
Governance: (Prentice-Hall, Englewood Cliffs, NJ) Third Edition.



                                               35
Table 1

Characteristics of the Going-Private Sample

The sample includes 323 going private announcements for the period from 1980 to 1996.
Equity values are computed as the product of common shares outstanding and the closing
price of common stock at the end of fiscal year immediately preceding the calendar year of
the announcement date.

             Lehn & Poulsen                             Our Sample
             (1989)
                        Mean                    Mean       Median               Total
Year         Number of Equity       Number      Equity     Equity     Standard Equity
             firms      Value       of Firms    Value      Value      Deviation Value
                        ($mil)                  ($mil)     ($mil)               ($mil)
1980         15         33.1        5           94.1       110.5      56.8      470.4
1981         21         87.8        8           172.4      117.9      179.0     1379.4
1982         27         79.1        13          130.3      88.1       84.8      1693.3
1983         35         115.4       17          187.9      115.8      236.2     3195.4
1984         48         173.9       25          342.6      321.5      313.7     8565.4
1985         36         371.7       22          770.4      339.7      1221.3    16949.8
1986         39         285.9       22          682.3      196.7      1108.0    15011.2
1987         42         224.3       26          397.9      164.7      558.5     10346.5
1988                                32          373.8      183.1      591.9     11960.1
1989                                19          653.3      273.1      971.7     12412.8
1990                                26          1277.9     154.8      1277.8    33224.0
1991                                10          262.9      57.7       486.1     2629.1
1992                                9           110.6      47.4       139.5     995.4
1993                                9           215.9      68.8       215.9     1943.0
1994                                20          340.5      101.4      609.4     6810.8
1995                                27          1173.2     118.8      4072.7    31677.6
1996                                33          384.4      108.4      719.2     12685.7

1980-1987 263             193.2     138         417.5      171.0      744.3       57611.5
1988-1996                           185         618.0      128.3      1866.7      114338.4
Full                                323         532.4      140.6      1495.6      171949.9
sample




                                               36
Table 2

Distribution of the Standard Industrial Classification (SIC) code for the Going-Private
Sample

This table shows the distribution of the SIC code for the going-private sample. The presence of
fifty-five separate two-digits SIC codes, with sixteen of these representing at least 1% of the
sample (thirty-four going-private), indicates a wide selection of industries.

           Industry                        Two-Digit SIC Code          Number of      Percentage
                                                                       Firms
Food Products                           20                                19            5.88
Chemical products                       28                                16            4.95
Electronic Equipment                    36                                13            4.02
Scientific Instruments                  38                                11            3.41
Communications                          48                                11            3.41
Eating and Drinking Establishments      58                                11            3.41
Health                                  80                                11            3.41
Fabric and Clothes                      22,23                             15            4.64
Paper and Paper Products                24,25,26,27                       17            5.26
Manufacturing                           30,31,32,33,34                    33           10.22
Computer Hardware and Software          35,73                             33           10.22
Transportation                          37,39,40,42,45                    27            8.36
Wholesale and Retail                    50,51,52,53,54,56,57,59           70           21.67
SIC Missing                                                                2            0.62
All Others                              13,15,16,17,21,47,55,70,75,       35           10.84
                                        78,79,82,83,87,89,95,322

Total                                                                    323          100.00




                                               37
Table 3

Average Abnormal Stock Returns Surrounding Going-Private Announcements
for the Going-Private Sample

We estimate average abnormal returns based on the market model around the
announcement day (Day 0) of going private. The market model is estimated over the
(-345, -91) period. The going-private sample includes 323 going private
announcements for the period from 1980 to 1996.
                   Panel A: Target Firms                  Panel B: Rival Firms
Day(s)       Abnormal Return (%) t-statistic       Abnormal Return (%) t-statistic
-20           0.28                     1.68         0.11                   1.35
-19           0.02                     0.15         0.21***                2.73
-18           0.06                     0.36         0.02                   0.25
-17           0.13                     0.8          0.05                   0.68
-16           0.61***                  3.61        -0.05                  -0.64
-15           0.50***                  2.99        -0.04                  -0.51
-14           0.09                     0.57        -0.01                  -0.07
-13           0.13                     0.76         0.21***                2.66
-12           0.30                     1.77         0.04                   0.49
-11           0.29                     1.71         0.00                   0.00
-10           0.11                     0.63         0.21***                2.66
-9            0.24                     1.42         0.15*                  1.86
-8            0.18                     1.06         0.26***                3.25
-7            0.35**                   2.10         0.29***                3.67
-6            0.57***                  3.41         0.01                   0.16
-5            0.78***                  4.66         0.10                   1.22
-4            0.03                     0.18         0.08                   1.04
-3            0.31                     1.88         0.13*                  1.65
-2            0.81***                  4.86        -0.01                  -0.10
-1            2.84***                16.96         -0.18***               -2.30
0             9.94***                59.32          0.14*                  1.79
1             2.74***                16.34          0.19***                2.46
2             0.21                     1.28         0.10                   1.23
3            -0.07                    -0.39         0.00                   0.06
4            -0.01                    -0.05         0.18**                 2.24
5             0.01                     0.03         0.04                   0.45
(0,1)        12.68***                51.98          0.33***                3.01
(-20, -1)     8.63***                11.02          1.57***                4.48
(-20, +1) 21.31***                   35.07          1.91***                5.18
*Significant at the 0.1 level.
**Significant at the 0.05 level.
***Significant at the 0.01 level.




                                                 38
Table 4

Average Adjusted Forecast Revisions of Current-Year Earnings and Five-Tear Earnings Growth
Rates for Months Surrounding Going-Private Announcements for the Going-Private Sample

We define the forecast revision for Month t as the mean of analysts’ forecasts reported in the IBES database
in Month t less the mean of analysts’ forecasts in Month t-1, scaled by stock price at the end of the month
preceding the going private announcement. We define the adjusted forecast revision for Month t as the scaled
forecast revision for Month t less the expected forecast revision for Month t, which is estimated based on a
fourth-order moving average expectations model. The t-statistic tests the null hypothesis that the mean
earnings forecast revision differs from 0. The sample includes 323 announcements of going-private
restructuring.

Panel A: Adjusted forecast revisions (AFR) for target firms


                                    AFR for Current-Year Earnings      AFR for 5-Year Earnings Growth

                                    Adjusted        t-test             Adjusted           t-test
Forecast                            Forecast        statistic          Forecast           statistic
Month                               Revisions                          Revision
-3                                  -0.00204        -0.73              -0.01025           -1.76
-2                                  -0.00439        -0.97               0.01434            1.83
-1                                   0.00061         0.35              -0.00656           -1.03
0                                    0.00512***      3.17              -0.00621           -0.96
1                                    0.00275         1.53              -0.00293           -0.38
2                                   -0.00043        -0.52              -0.00353           -0.44
3                                   -0.00036        -0.44              -0.00442           -0.52
Cumulative forecast windows:
0 to 1                              0.00716**        2.41              -0.00944           -0.81
0 to 3                              0.00509          1.59              -0.01518           -0.82

Panel B: Adjusted forecast revisions (AFR) for rival firms

-3                                  -0.0003        -1.64                0.0042             0.80
-2                                  -0.0002        -0.87               -0.0021            -0.41
-1                                  -0.0000        -0.05               -0.0007            -0.17
0                                    0.0003**       2.01                0.0085*            1.76
1                                    0.0004**       2.14               -0.0016            -0.12
2                                    0.0001         0.94               -0.0055            -0.84
3                                    0.0003*        1.87               -0.0044            -0.82
Cumulative forecast windows:
0 to 1                              0.0007***       3.00               0.0161***           2.59
0 to 3                              0.0010***       3.21               0.0057              0.72
*Significant at the 0.1 level.
**Significant at the 0.05 level.
***Significant at the 0.01 level.




                                                    39
Table 5

Ratio of Total Debt to Total Assets for Years Surrounding Going Private for the
Going-Private Sample

This table reports the ratio of total debt to total assets for Year -3 through Year +3. The
ratio is calculated as total debt divided by total assets and then multiplied by 100. Data for
total debt and total assets is obtained from the COMPUSTAT files. Year 0 is the year of
the going-private announcement.


                                                      Standard
Year                            Mean (%)              Deviation (%)       Number of Firms

Panel A: All Available Firms

-3                              46.34                 22.35               201
-2                              43.28                 20.71               211
-1                              42.56                 20.88               213
0                               48.16                 26.71               143
1                               60.49                 28.91               102
2                               63.20                 30.28               92
3                               66.09                 32.18               77


Panel B: Firms with data for all years

-3                              47.80                 22.28               65
-2                              45.38                 21.61               65
-1                              44.77                 21.96               65
0                               51.71                 26.73               65
1                               60.43                 27.53               65
2                               63.35                 28.18               65
3                               64.93                 31.60               65




                                                 40
Table 6

Cross-Sectional Regressions of Abnormal Forecast Revisions of Earnings Surrounding the
Going-Private Announcement for the Going-Private Sample

       AFRi = b0 +b1ARi +b2 CH_DEBTi +b3CFLOWi +b4PAARi +b5MBOi +b6SIZEi +ei

Where AFR is the cumulative adjusted forecast revisions of current-year earnings per share and
5-year earnings growth rates for the 2-month (0,1) period for Models 1 and 2, respectively, AR
is the announcement-period abnormal return, as measured by the cumulative average abnormal
return for the (-20,1) window based on the market model, CH_DEBT is the change in total debt
to total asset ratio from year-3 to year +3, CFLOW is the ratio of free cash flow to assets for the
year immediately preceding the going-private transaction, PAAR is the pre-announcement
abnormal return, as measured by the cumulative average abnormal return for the (-90,-21)
period based on the market model, MBO is a dummy variable that takes a value of one if the
buyout team for the target firm is led by management, otherwise zero, and SIZE is natural
logarithm of the market value of equity. The t-statistics appear in parentheses.

                                              Model 1:                         Model 2:
     Independent Variables                Current-Year AFR                  Five-Year AFR
                                        as Dependent Variable            as Dependent Variable
Intercept                                        0.244***                          0.017
                                                (2.87)                            (0.09)
AR                                               0.177***                         -0.083
                                                (2.72)                           (-0.55)
CH_DEBT                                         -0.035                             0.045
                                               (-1.05)                            (0.61)
CFLOW                                           -0.298***                         -0.216
                                               (-3.73)                           (-1.08)
PAAR                                            -0.133***                         -0.072
                                               (-2.99)                           (-0.70)
MBO                                             -0.052**                           0.035
                                               (-2.10)                            (0.62)
SIZE                                            -0.015**                           0.000
                                               (-2.34)                            (0.02)
F-Statistic                                      6.93***                           0.66
N                                               65                                54
R-Square                                          0.41                              0.08
*Significant at the 0.1 level.
**Significant at the 0.05 level.
***Significant at the 0.01 level.




                                                   41
Table 7

Cross-Sectional Regressions of Announcement-Period Abnormal Stock Return for the
Going-Private Sample

       ARi = b0 +b1AFRi +b2 CH_DEBTi +b3CFLOWi +b4PAARi +b5MBOi +B6SIZEi +ei

Where AR is the announcement-period abnormal return, as measured by the cumulative average
abnormal return for the (-20,1) window based on the market model, AFR is the cumulative
adjusted forecast revisions of current-year earnings per share and 5-year earnings growth rates
for the 2-month (0,1) period for Models 1 and 2, respectively, CH_DEBT is the change in total
debt to total asset ratio from year-3 to year +3, CFLOW is the ratio of free cash flow to assets
for the year immediately preceding the going-private transaction, PAAR is the pre-
announcement abnormal return, as measured by the cumulative average abnormal return for the
(-90,-21) day period based on the market model, MBO is a dummy variable that takes a value of
one if the buyout team for the target firm is led by management, otherwise zero, and SIZE is
natural logarithm of the market value of equity. The t-statistics appear in parentheses.

                                            Model 1:                         Model 2:
     Independent Variables              Current-Year AFR                  Five-Year AFR
                                     as Independent Variable         as Independent Variable
Intercept                                     -0.051                             0.135
                                             (-0.30)                            (0.74)
AFR                                            0.631***                         -0.072
(Current-Year or Five-Year)                   (2.72)                           (-0.52)
CH_DEBT                                        0.057                             0.076
                                              (0.90)                            (1.12)
CFLOW                                          0.139                            -0.080
                                              (0.83)                           (-0.42)
PAAR                                           0.253***                          0.150
                                              (3.02)                            (1.56)
MBO                                            0.115**                           0.088*
                                              (2.52)                            (1.75)
SIZE                                           0.010                            -0.002
                                              (0.79)                           (-0.12)
F-Statistic                                    3.10***                           1.45
N                                              65                              56
R-Square                                        0.24                             0.15
*Significant at the 0.1 level.
**Significant at the 0.05 level.
***Significant at the 0.01 level.




                                                 42
Table 8

Cross-Sectional Regressions of Abnormal Forecast Revisions of Earnings Surrounding the
Going-Private Announcement for the Sample of Industry Rivals

     AFRi = b0+b1ARi+b2CFLOWi+b3DEBTi+b4RELSIZEi+b5MBOi+b6PAARi +b7PSARi+ei

Where AFR is the cumulative adjusted forecast revisions of current-year earnings per share and
5-year earnings growth rates for the 2-month (0,1) period for Models 1 and 2, respectively, AR is
the cumulative average abnormal return for the (-20,1) window based on the market model,
CFLOW is the ratio of free cash flow to assets for the year immediately preceding the going-
private transaction, DEBT is total debt over total assets for the year immediately preceding the
announcement, RELSIZE is a dummy variable that takes a value of one if the target is larger than
the rival firm in market value of equity, and zero otherwise, MBO is a dummy variable that takes
a value of one if the buyout team for the target firm is led by management, otherwise zero,
PAAR is the cumulative average abnormal return for the (-90,-21) day period based on the
market model, and PSAR is the cumulative average abnormal return for the (+2, +90) day period
based on the market model. The t-statistics appear in parentheses.

                                  Model 1:                             Model 2:
    Independent               Current-Year AFR                      Five-Year AFR
     Variables              as Dependent Variable               as Dependent Variable
Intercept                            0.0012***                      0.0237*
                                    (3.50)                         (1.93)
AR                                   0.0055***                      0.0886*
                                    (3.82)                         (1.95)
CFLOW                               -0.0000                        -0.0100
                                   (-0.42)                        (-0.80)
DEBT                                 0.0001                         0.0209
                                    (0.33)                         (0.61)
RELSIZE                              0.0010***                     -0.0100
                                    (2.56)                        (-0.80)
MBO                                  0.0002                        -0.0126
                                    (0.58)                        (-1.03)
PAAR                                -0.0008                         0.0031
                                   (-1.54)                         (0.20)
PSAR                                 0.0004                        -0.0116
                                    (0.96)                        (-0.80)
F-Statistic                          3.49***                        1.29
N                                1173                            935
R-Square                             0.02                           0.01
*Significant at the 0.1 level.
**Significant at the 0.05 level.
***Significant at the 0.01 level.




                                                    43
Table 9

Cross-Sectional Regressions of Announcement-Period Abnormal Stock Return for the
Sample of Industry Rivals

              ARi = b0 + b1AFRi + b2CFLOWi + b3DEBTi + b4 MBOi + b5PAARi + ei

Where AR is the announcement-period abnormal return, as measured by the cumulative average
abnormal return of the (-20, +1) window based on the market model, AFR is the cumulative
adjusted forecast revisions of current-year earnings per share and 5-year earnings growth rates
for the 2-month (0,1) period for Models 1 and 2, respectively, CFLOW is the ratio of free cash
flow to assets for the year immediately preceding the going-private announcement, DEBT is
total debt over total assets for the year immediately preceding the announcement, MBO is a
dummy variable that takes a value of one if the buyout team for the target firm is led by
management, otherwise zero, and PAAR is the pre-announcement abnormal return, as measured
by the cumulative average abnormal return for the (-90,-21) day period based on the market
model. The t-statistics appear in parentheses.

                                            Model 1:                      Model 2:
         Independent                   Current-Year AFR                Five-Year AFR
          Variables                 as Independent Variable        As Independent Variable
Intercept                                      0.0176***                     0.0242***
                                              (3.35)                        (3.41)
AFR                                            2.2208***                     0.0430*
(Current Year or Five Year)                   (3.77)                        (1.79)
CFLOW                                         -0.0000                        0.0000
                                             (-0.24)                        (0.26)
DEBT                                           0.0005                       -0.0023
                                              (0.09)                       (-0.09)
MBO                                           -0.0148*                      -0.0160*
                                             (-1.84)                       (-1.79)
PAAR                                           0.1119***                     0.1014***
                                            (15.23)                       (12.02)
F-Statistic                                  50.81***                      30.50***
N                                          1173                           935
R-Square                                       0.18                          0.14
*Significant at the 0.1 level.
**Significant at the 0.05 level.
***Significant at the 0.01 level.




                                               44

								
To top