# Value creation from equity carve-outs.

J. Randall Woolridge (*)Using a large sample of equity carve-out events during the 1980s and 1990s, we find that rivals of carve-out parent firms display negative announcement-period returns. This finding distinguishes the divestiture gains hypothesis from the asymmetric information hypothesis. Additional tests provide further support for the divestiture gains hypothesis. Operating performance improvements for both parents and their carved-out subsidiaries are evident.

Announcements of equity carve-outs produce positive stock returns for parent firms (see, e.g., Schipper and Smith, 1986; Klein, Rosenfeld, and Beranek, 1991; and Mulherin and Boone, 2000). Two general explanations are suggested for these parent firm gains.

Schipper and Smith (1986) conjecture that equity carve-outs result in divestiture gains due to separate financing for the subsidiary's investment projects, a more efficient set of contracts between shareholders and managers, and the creation of pure-play stocks. The authors provide no evidence concerning this divestiture gains explanation beyond the observed announcement returns.

Nanda (1991), in an extension of Myers and Majluf (1984), develops a signaling model in which firms raise capital through equity carve-outs when parent firm shares are relatively undervalued and subsidiary shares are relatively overvalued. The good news for parent investors is that the parent shares are not overvalued, at least when compared to subsidiary valuation. Since the parent is typically several times larger than the subsidiary, this leads to an increase in parent share price.

Empirical tests of the divestiture gains hypothesis and the asymmetric information hypothesis provide some support for both hypotheses. Allen and McConnell (1998) examine the use of the carve-out stock offering proceeds for a sample of 188 equity carve-outs over 1978-1993 and find that announcement-period gains for parents are higher if the proceeds are paid out than if they are retained. These results provide some support for the divestiture gains hypothesis. Vijh (1999), using a sample of 336 equity carve-outs over the 1980 to 1997 time period, evaluates three-year returns of parent and subsidiary stocks after issue; some evidence is found that the long-term returns are related to the number of business segments before carve-out, which is used as a proxy for divestiture gains arising from the refocusing of parent and subsidiary operations. Mulherin and Boone (2000) and Allen (1998) also find divestiture gains in equity carve-outs. Mulherin and Boone (2000) study a sample of 125 equity carve-outs over 1990-1 999 and conclude that the positive wealth effects associated with equity carve-out announcements are due to synergistic gains. Allen's (1998) examination of the long-run stock performance, operating performance, and incentive-based compensation of Thermo Electron and its 11 equity carve-outs reveals improvements in all areas.

In contrast to the above studies, Slovin, Sushka, and Ferraro (1995), using a sample of 36 equity carve-outs over 1980-1991, find that share prices of equity carve-out rival firms react negatively to carve-out announcements. They interpret this finding as support for the Nanda (1991) model. The carve-out announcement signals to investors that managers believe the carve-out (and, by extension, the entire industry) is overvalued.

While all this research offers some understanding of parent announcement-period gains, no one attempts to determine whether the gains reflect divestiture gains alone, asymmetric information alone, or a combination of the two. Vijh (2002) provides the first contribution toward this end, with joint but separate tests. He asserts that divestiture gains should increase with increasing size of the subsidiary assets relative to the combined assets. Nanda's (1991) model suggests the opposite.

Vijh (2002) finds a direct relation between announcement-period returns and the ratio of subsidiary assets to non-subsidiary assets, thus lending support to the divestiture gains hypothesis. His other tests of the divestiture gains hypothesis include analysis of Wall Street Journal reports, the relatedness of parent and subsidiary operations, subsequent spin-offs and third party acquisitions, and the use of carve-out proceeds. He concludes that the combined evidence shows that the market reacts positively to the announcement of carve-outs due to divestiture gains, not asymmetric information.

We test the divestiture gains hypothesis versus the asymmetric information hypothesis more directly. The analysis focuses on the impact of carve-outs on parent rivals. The asymmetric information hypothesis implies a positive reaction by parent rivals. The carve-out signals that the parent, and, by extension, the parent's entire industry, is undervalued compared to the carve-out's industry. Parent rivals react positively to this signal. The divestiture gains hypothesis, in contrast, predicts that announcement-period returns for parent rivals will be negative. Carve-outs allow for such wealth-increasing events as the financing of new projects, the creation of more efficient sets of contracts between parents and their subsidiaries, improvements in the alignment of incentives between subsidiary managers and their shareholders, and increased corporate focus. Any divestiture gains experienced by the parent could be to the detriment of parent rivals.

The results support the divestiture gains hypothesis. Rivals of parent firms exhibit negative stock price reactions to equity carve-out announcements. This finding differs from that of Slovin, Sushka, and Ferraro (1995), who observe an insignificant (positive mean, negative median) stock price reaction by parent rivals.

The difference in the results versus the Slovin, Sushka, and Ferraro (1995) results (SSF hereafter) is most likely due to differences in the samples. SSF have a sample of 36 equity carve-outs, with 32 parent firms. They generate a sample of parent rivals using all NYSE/ AMEX firms over 1980-1991 with the same four-digit SIC code as their 32 parent firms. We have a sample of 185 equity carve-outs, with 183 parent firms. Also, a sample of parent rivals is generated using all NYSE/AMEX and Nasdaq firms over 1981-1994 with the same four-digit SIC code as the 183 parent firms. Thus, the sample is more comprehensive. We evaluate significantly more parent firms and rivals over a longer time period, and do not exclude Nasdaq firms from the sample. (1)

Tests of the divestiture gains hypothesis versus the asymmetric information hypothesis are further refined by recognizing the special case of "own-industry carve-outs," when the carve-out and the parent are in the same industry. In own-industry carve-outs, the carve-out cannot be both a negative signal for the carve-out's industry and a positive signal for the parent's industry. If the parent and the subsidiary are competitors, improvements made by one could be at the expense of the other and at the expense of some rival firms.

Predictions generated by Nanda's signaling model are particularly unclear for own-industry carve-outs. If carve-outs signal overvaluation of the subsidiary, then the overvaluation could mitigate the rise in parent share price since the parent is in the same industry as the subsidiary, and rival share price reactions could be positive, negative, or nil.

To provide clarity in the test of the divestiture gains hypothesis versus the asymmetric information hypothesis, separate event studies are reported based on whether the parent and carve-out are in the same industry. For cross-industry carve-outs, where the predictions of the two hypotheses are clear, there is support for the divestiture gains hypothesis. Parent rivals react negatively to the cross-industry equity carve-out announcement. For own-industry carve-outs, where predictions are less clear, parent rivals also react negatively.

Further tests provide additional support for the divestiture gains hypothesis. If carve-outs result in stronger competition on the part of the parent and the subsidiary (as the divestiture gains hypothesis would suggest), then carve-outs should result in improvements in parent and subsidiary operating performance. Following Barber and Lyon (1996), we evaluate firms on an industry-adjusted, size-adjusted, and prior-performance-adjusted basis. There are operating performance improvements for both parents and their carved-out subsidiaries. (2)

Section I discusses the implications of the divestiture gains hypothesis and the asymmetric information hypothesis in more detail. Section II describes the sample. Section III examines stock price reactions to equity carve-out announcements and section IV presents operating performance results. Section V concludes.

I. Two Basic Hypotheses

This section first discusses the asymmetric information hypothesis and the divestiture gains hypothesis in more detail. It then presents implications and refinements of the two hypotheses.

A. Asymmetric Information Hypothesis

Leland and Pyle (1977), Allen and Faulhaber (1989), and Welch (1989) develop asymmetric information models of the initial public offering market in which fully informed managers sell to imperfectly informed investors. Myers and Majluf (1984) model seasoned equity offerings where fully informed managers sell to uninformed investors. Nanda (1991) models firms raising capital through equity carve-outs when parent firm shares are relatively undervalued and subsidiary shares are relatively overvalued. Parent investor shares are not overvalued, at least when compared to subsidiary valuation. For a parent that is typically much larger than the subsidiary, this produces an increase in parent share price.

The asymmetric information hypothesis, then, predicts that 1) parents will react positively to carve-out announcements, 2) parent rivals will react positively to carve-out announcements, and 3) subsidiary rivals will react negatively to carve-out announcements. While the asymmetric information hypothesis makes no explicit predictions about changes in operating performance, improvements in subsidiary operating performance would not seem to be consistent with subsidiaries being overvalued. Subsequent decreases in operating performance might be consistent with overvaluation.

B. Divestiture Gains Hypothesis

The divestiture gains hypothesis consists of a collection of individual, but related hypotheses. All predict that parents and their carved-out subsidiaries will, in one way or another, become more competitive in their industries. It is not the intention to test each of these hypotheses individually. Rather, the goal is to distinguish between 1) the asymmetric information hypothesis, and 2) the set of hypotheses that we generally refer to as the divestiture gains hypothesis. For completeness, each of the latter hypotheses are discussed.

1. Financing Strategy and Investment Strategy Hypotheses

Both the financing strategy hypothesis and the investment strategy hypothesis are related to the use of carve-out proceeds. The financing strategy hypothesis suggests that if proceeds are used to repay debt, parent announcement-period returns will be higher than if proceeds had been retained (see Lang, Poulsen, and Stulz, 1995 and Allen and McConnell, 1998 for a more thorough discussion). The investment strategy hypothesis suggests that if proceeds are retained and used "positively" (e.g., to finance new projects or upgrade existing projects), then the market will react positively (see Schipper and Smith, 1986 for a more thorough discussion).

Taken together, these two hypotheses suggest that carve-outs can result in reductions in agency costs. These reductions in agency costs could make the parent and the subsidiary more competitive in their industries. Any improvements made by the parent and/or the subsidiary could be to the detriment of their rivals.

Both the financing strategy hypothesis and the investment strategy hypothesis thus predict that, 1) parents will react positively to carve-out announcements, 2) parent rivals will react negatively to carve-out announcements, and 3) subsidiary rivals will react negatively to carve-out announcements. The two hypotheses further predict that carve-outs could result in operating improvements for the parent and the carved-out subsidiary.

2. Contracting Efficiency Hypothesis

Schipper and Smith (1986) provide the earliest evidence that equity carve-outs produce positive average announcement-period returns. They conjecture that equity carve-outs result in a more efficient set of contracts between shareholders and managers, but provide no evidence concerning this hypothesis beyond the observed announcement-period returns. This contracting efficiency hypothesis relies partially upon the fact that subsidiary managers receive compensation based on carve-out stock.

Equity carve-outs also initiate fundamental changes at the parent level. The parent firm must deal with its subsidiary at market prices in order to assure shareholders of the parent and subsidiary that wealth transfers will not occur from one group to the other. A more efficient set of contracts is likely to produce operational improvements. The negative reaction by rivals of the carve-out observed in this and other studies could be explained as stemming from stronger competition from the more efficient subsidiaries. Similarly, improved operations of the parents might also be detrimental to rivals of the parent firms, causing those rivals to exhibit negative share price reactions to the carve-out announcements. Nanda's (1991) signaling model would not predict a negative reaction by rivals of the parent firms to carve-out announcements.

The contracting efficiency hypothesis predicts that, 1) parents will react positively to carve-out announcements, 2) parent rivals will react negatively to carve-out announcements, and 3) subsidiary rivals will react negatively to carve-out announcements. The hypothesis also predicts that carve-outs could result in operating improvements for both the parent and the carved out subsidiary.

3. Incentive Alignment Hypothesis

The incentive alignment hypothesis is a special case of the efficient contracting hypothesis that applies only to subsidiary managers. It argues that performance improvement may arise because of improvements in the alignment of incentives between subsidiary managers and their shareholders.

Aron (1991) and Allen (1998) posit that the incentive structure for subsidiary managers may be more compatible with shareholder interests after a carve-out occurs. Logue, Seward, and Walsh (1996) make a similar argument for targeted stock. The creation of a subsidiary with publicly traded stock allows for the introduction of a variety of incentive plans for the subsidiary managers that may not have been feasible when the subsidiary was not publicly traded. In turn, the potential to write improved incentive contracts could improve the performance of the (partially) divested subsidiary.

Gaver and Gaver (1995) find that higher-growth firms pay a higher proportion of executive compensation in the form of stock and stock options. As discussed later, equity carve-outs have significantly higher growth opportunities than their parents.

The incentive alignment hypothesis predicts that 1) parents will react positively to carve-out announcements, 2) parent rivals will react negatively to carve-out announcements, and 3) subsidiary rivals will react negatively to carve-out announcements. The hypothesis also predicts that carve-outs could result in operating improvements for both the parent and the carved-out subsidiary.

4. Corporate Focus Hypothesis

Comment and Jarrell (1995) observe that improved corporate focus leads to stock valuation gains. They argue that managerial skills may be well suited to the management of a core business, but not to the management of non-core assets. Thus, performance improvement may follow focus-increasing events when managers are freed from operations unrelated to the core business.

Equity carve-outs (and divestiture, in general) can be used to test implications of the corporate focus hypothesis because cross-industry carve-outs create an increase in focus (see Daley, Mehrotra, and Sivakumar, 1997 for a discussion of spin-offs and corporate focus; see Boone, 2001 for a discussion of equity carve-outs and corporate focus).

The corporate focus hypothesis and the contracting efficiency hypothesis are not mutually exclusive; one reason focus might boost share price is that more efficient contracts may be developed between shareholders and managers. Highly diversified firms have difficulty offering stock-based compensation that is sensitive to the performance level of each division. Any observed differences in either cross-industry and own-industry share price reactions or changes in subsequent operating performance could be attributable to changes in focus.

The corporate focus hypothesis predicts that, 1) parents will react positively to carve-out announcements, 2) parent rivals will react negatively to carve-out announcements, and 3) subsidiary rivals will react negatively to carve-out announcements. The hypothesis also predicts that carve-outs could result in operating improvements for both the parent and the carved-out subsidiary.

C. Summary of the Hypotheses and Their Implications

The asymmetric information model predicts, 1) a positive parent stock price reaction, 2) a positive parent rival stock price reaction, and 3) a negative subsidiary rival stock price reaction to the announcement of an equity carve-out. The divestiture gains hypothesis (including financing strategy, investment strategy, contracting efficiency, incentive alignment, and corporate focus hypotheses) predicts: 1) a positive parent stock price reaction, 2) a negative parent rival stock price reaction, and 3) a negative subsidiary rival stock price reaction to the announcement of an equity carve-out.

The distinguishing feature between the two hypotheses is the parent rival reaction to equity carve-out announcements. The asymmetric information model predicts a negative reaction by parent rivals; the divestiture gains hypothesis predicts a positive one.

We can also examine the operating performance of the parent and the subsidiary to differentiate the two hypotheses. The asymmetric information hypothesis makes no explicit predictions about changes in operating performance, while the divestiture gains hypothesis predicts improvements in operating performance for the parent and/or subsidiary.

D. Refinements of the Hypotheses

While the main implications of the two hypotheses are clear, there are two refinements that need to be made. The first deals with the special case of own-industry carve-outs. The second has to do with the amount of control relinquished by the parent.

1. Own-Industry Carve-Outs

Predictions of the divestiture gains hypothesis and the asymmetric information hypothesis are less clear when parents and their subsidiaries are in the same industry (own-industry carve-outs). If the parent and the subsidiary are competitors, then improvements made by one could be at the expense of the other as well as at the expense of some rival firms. Predictions generated by Nanda's signaling model are particularly unclear for own-industry carve-outs. If carve-outs signal overvaluation of the subsidiary, then the overvaluation could mitigate the rise in parent share price since the parent is in the same industry as the subsidiary and rival share price reactions could be positive, negative, or zero.

In general, cross-industry carve-outs provide clear tests of the divestiture gains hypothesis versus the asymmetric information hypothesis. Own-industry carve-outs are less clear for several mitigating factors.

2. Spin-Offs versus Equity Carve-Outs

Hite and Owers (1983), Miles and Rosenfeld (1983), and Schipper and Smith (1983) find apparent divestiture gains in spin-offs. Spin-offs differ from carve-outs in the amount of control over assets relinquished by the parent. In spin-offs, this amount, as measured by parent holdings, is complete. Parents fully divest themselves of ownership in their spun-off subsidiaries. In carve-outs, however, the control relinquished is usually only partial. Parents tend to retain some portion of ownership in their carved-out subsidiaries.

Given that we observe divestiture gains in equity carve-outs, where relinquishment of parent control is only partial, we can test whether divestiture gains are related in some way to the amount of control relinquished by the parent, or whether the creation of a separately traded stock is enough to generate divestiture gains.

II. Sample Selection and Description

This section describes the sample selection procedure and provides descriptive statistics.

A. Sample Selection Procedure

To generate a sample of equity carve-outs, we examine data from the Securities Data Company and the Investment Dealers' Digest for initial public offerings by a corporate issuer during the period 1981-1994; we find a net total of 593 initial public offerings (IPOs) involving a corporate issuer. The Wall Street Journal, the Wall Street Journal Index, the Directory of Corporate Affiliations, and America s Corporate Families are consulted to identify parent firm names. For inclusion in the sample, we required: 1) that the event involve an initial public offering of common stock in a previously wholly owned subsidiary of a corporation traded on the NYSE/AMEX or Nasdaq, and 2) that the subsidiary be a domestic corporation. The search resulted in a final sample of 185 verified equity carve-outs for the period 1981-1994.

For the final sample, we search the Dow Jones News Retrieval and individual firm prospectuses to obtain: the announcement dates of the equity carve-out, the price per share of the IPO, the number of shares offered, the type of offering (primary, secondary, or joint), the use of the proceeds, the percentage of shares retained by the parent firm, and details of executive stock option plans. The Center for Research in Security Prices daily return files are used to obtain market values and security returns. Finally, we used Compustat to obtain operating performance measures, including sales, net income, total assets, and capital expenditures.

B. Descriptive Statistics

Table I reports descriptive statistics for the final carve-out sample. Panel A presents the distribution of equity carve-outs by year. The distribution is fairly uniform, with at least six and at most 28 events occurring in any given year.

Panel B reports the average parent and subsidiary market values in the year of the carve-out. Market value is calculated as the price per share times the number of shares outstanding. For the full sample, the average parent market value is more than $3 billion, and the average subsidiary market value is almost $544 million.

Panel C of Table I describes the percentage of carved-out common stock the parent firm retains immediately subsequent to the equity carve-out. Seventy-six percent of the parents retain a 50% or greater share in their subsidiaries, and 36% of the parents retain an 80% or greater share. (3)

Panel D identifies the use of proceeds (paid out to shareholders and/or creditors or retained) by offering type (secondary, primary or joint). Consistent with Allen and McConnell (1998), we define a secondary offering as one in which the parent alone collects the proceeds of the sale. In a primary offering, the subsidiary alone collects the proceeds of the sale. A joint offering occurs if both the subsidiary and the parent collect the proceeds of the sale. As depicted here, although carved-out subsidiaries exact at least a portion of the proceeds of sale in a large majority of cases, the carved-out subsidiaries tend to use the proceeds they collect to repay debt owed to their parents or other creditors, or to pay a special dividend to their parents.

III. Stock Price Reactions

In a simple test to distinguish between the divestiture gains hypothesis and the asymmetric information hypothesis, we examine parent, subsidiary rival, and parent rival stock price reactions to equity carve-out announcements. The divestiture gains hypothesis predicts a positive reaction by parents, a negative reaction by subsidiary rivals, and a negative reaction by parent rivals. The asymmetric information hypothesis predicts a positive reaction by parents, a negative reaction by subsidiary rivals, and a positive reaction by parent rivals. As we have noted, the parent rival's reaction differentiates the two hypotheses.

A. Methodology

We calculate excess returns on a market-adjusted basis, using the CRSP equally weighted market return (including dividends) as the measure for market returns. To calculate marketadjusted returns, we adjust daily individual firm returns, [r.sub.i,t], for the contemporaneous return of the market, [r.sub.m,t], and then average across all firms. A cumulative adjusted return, [CAR.sub.T], for the event period T is calculated by summing the average daily market-adjusted returns across all event window days. Then, t-tests are conducted by dividing the cumulative adjusted returns by their contemporaneous cross-sectional standard errors.

Using a two-tailed test, the null hypothesis is that returns are not statistically different from zero; the alternative hypothesis is that returns are statistically different from zero. The industry rival Firm sample is based on four-digit SIC codes.

Although we do not report them here, we also calculate excess returns on a market model basis. The results are not sensitive to the method employed.

B. Empirical Results

Table II provides mean and median share price reactions to carve-out announcements for parent firms, parent industry rivals, and subsidiary industry rivals. Parent firms, as reported in Panel A, exhibit positive and significant mean and median share price reactions of 1.92% (p-value = 0.00) and 0.75% (p-value 0.00), respectively.

We report separate event studies depending on whether the parent and the carve-out are in the same industry. As we have noted, when parent and carve-out are in the same industry, the carve-out cannot be both a negative signal for the carve-out's industry and a positive signal for the parent's industry; the industries are one and the same.

In the 153 instances when the parent and the carve-out are in different industries (denoted "cross-industry"), the mean parent abnormal return is 2.10% (p-value = 0.00). In the 30 cases when the parent and the carve-out are in the same industry (denoted "own-industry"), the mean parent abnormal return is 0.98% (p = 0.37). Consistent with Boone (2001), a difference of means test for the parent returns is insignificant (p-value = 0.43).

Panels B and C of Table II report parent rival returns and subsidiary rival returns. Returns to rivals are consistently negative. Coincidentally, for both parent and subsidiary rivals, the mean and median returns are -0.41% (p = 0.00) and -0.66% (p = 0.00), respectively. There is no significant difference in either case between own-industry events and cross-industry events.

We also examine whether divestiture gains are related in some way to the amount of control relinquished by the parent. As shown in Panels A (full sample) and B (cross-industry sample) of Table Ill, we find very little support for such a relation. Stock returns to parent rivals are negative and generally statistically significant, regardless of the level of parent control relinquished. There are no significant differences in stock returns among the various levels of relinquishment examined. At least for the sample, it appears that the creation of a separately traded stock is enough to generate divestiture gains, especially after controlling for the confounding effects of own-industry carve-outs. (4)

The results clearly support the divestiture gains hypothesis, but not the asymmetric information hypothesis. Both hypotheses predict the positive parent reaction and the negative subsidiary rival reaction. The negative parent rival reaction, however, is predicted only by the divestiture gains hypothesis. The asymmetric information hypothesis predicts the opposite reaction by parent rivals.

IV. Changes in Operating Performance

A more direct test of the divestiture gains hypothesis investigates changes in operating performance of parents and subsidiaries. If carve-outs produce a more competitive parent and subsidiary (as the divestiture gains hypothesis would suggest), then carve-outs should result in improvements in parent and subsidiary operating performance.

We examine both changes in the level of operations (as measured by growth rates in sales, income, assets, and capital expenditures) and in the efficiency of operations (as measured by return on assets, return on cash-adjusted assets, return on market value of assets, and return on sales). As Barber and Lyon (1996) note, either method will yield well-specified, powerful test statistics, provided it incorporates a firm's past performance. Research designs that incorporate only a firm's industry or a firm's size will likely yield test statistics that are misspecified, because accounting-based measures of performance tend to mean-revert over time. Adjusting for past performance takes this mean reversion tendency into account.

The first of the methods, which measures changes in growth rates, indicates a change in the scale of operations. Such scale changes would not be detected using standard ratios that multiply the numerator and denominator by approximately the same scale factor. This method may be the more appropriate for carved-out subsidiaries since they are smaller and have much higher market-to-book ratios, on average, than their parent firms.

Market-to-book ratios are frequently used as a proxy for the portion of market value attributable to growth opportunities. The carve-out firms have significantly higher market-to-book ratios, suggesting that investors expect carve-outs to experience greater growth than their parent firms. [5]

For the much larger parent firms, restructuring by equity carve-out will most likely not change the scale of operations for the entire parent firm. Rather, we believe restructuring by equity carve-out is an attempt at the parent level to achieve more efficient operations. Thus, improvements are most likely to be observed using ratios.

A. Changes in the Level of Operations

This section presents the methodology and the empirical results for our examination of changes in the level of operations.

1. Methodology

We examine changes in the level of parent and subsidiary operating performance by measuring growth rates for net sales, operating income before depreciation, total assets, and capital expenditures at the firm level. (6) We then determine whether these individual firm growth rates are statistically different from the changes in the median growth rates for the industry as a whole, as well as for two subgroups of industry rivals (rivals whose size is within 75% of the sample firm's size and those whose pre-event performance measures are within 25% of the sample firm's performance measures). Measures of net sales, operating income before depreciation, capital expenditures (property, plant, and equipment), and total assets are obtained from Compustat.

First, we calculate each sample firm's growth rates in accounting variables as:

[P.sub.i,t] - [P.sub.i,t-1]/[P.sub.i,t-1] (1)

where [P.sub.i,t] is the level of the accounting variable for firm i in time t.

Next, median industry (rival) firm growth rates in accounting variables are calculated by taking the median of the individual rival firms' growth rates, defined as:

[PI.sub.i,t] - [PI.sub.i,t-1]/[PI.sub.i,t-1] (2)

where [PI.sub.i,t] is the level of the accounting variable for rival firm i in time t. Industry (rival) is defined as firms within the same four-digit SIC code as the sample firm (excluding parent and carve-out firms) with data available on Compustat and CRSP during the appropriate time period.

The first industry rival subset consists of those firms within the same four-digit SIC code as the sample firm that have year 1 performance measures within 25% of the sample firm's performance measures. These industry rivals are used for performance-adjusting.

For completeness, size- and industry-adjusted growth rates are also re-examined. For this subset, we include firms within the same four-digit SIC code as the sample firm whose size is within 75% of the sample firm's size.

The measure of abnormal return is defined as the difference between the sample firm's growth rate and its industry's median growth rate. We calculate and report the level of statistical significance for the median values of these industry-, size-, and performance-adjusted growth rates using the Wilcoxon signed rank test. Barber and Lyon (1996) find that Wilcoxon signed-rank T* test statistics are uniformly more powerful than parametric t-statistics.

The number of sample firms in the study changes over time for several reasons. First, prior to publicly trading, only those carve-outs that issue debt are required to report their financial position. Second, the coverage of Compustat is limited. Third, some of the firms in the sample are delisted in later years. (7)

2. Empirical Results

Table IV reports parent and subsidiary median industry-adjusted, size-adjusted, and performance-adjusted operating growth rates over the (0, +1) event years, where year 0 represents the year the equity carve-out goes public. As shown in Panel A, the results for parent firms are dependent on the method of benchmarking used. Parent performance-adjusted growth rates in sales, income, total assets, and capital expenditures are all positive and

statistically significant at the 5% level or better. Parent size-adjusted growth rates are largely negative and statistically significant, and parent industry-adjusted growth rates are all negative, but not statistically significant.

Although we report growth rates for the parent firms, we believe restructuring by equity carve-out is an attempt to wring more efficient operations out of the parent-not to increase the scale of the parent's operations. That is, improvements in parent operating ratios are expected, but growth in parent sales, income, assets, and capital expenditures are not necessarily expected.

The results for carved-out subsidiaries are consistent across different benchmarks, as shown in Panel B of Table IV. Subsidiary growth rates are positive and statistically significant for net sales in two of the three cases (p = 0.07 and 0.00), for total assets in all three cases (p 0.02, 0.00, and 0.00), and for capital expenditures in all three cases (p = 0.00, 0.02, and 0.00). Growth rates in operating income before depreciation are positive but statistically significant in only one case (p 0.00), probably because of the typical time lag between investing and seeing an investment generate increased profits.

The results reported in Panel B of Table IV are consistent with the conjecture that equity carve-outs result in substantial operational changes for subsidiaries once they have gone public, thus providing support for the divestiture gains hypothesis. (8) These results are not driven by cash infusion from the sale of new shares. First, the cash infusions are too small relative to total assets to make a significant impact on the subsidiaries' level of total assets. (9) Second, when the subsidiary procures at least a portion of the proceeds of the sale (which is true of 78% of the events), it retains the cash in only 46% of these cases. Most important, the time period from year 0 to year 1, which shows a high level of significance for industry-adjusted and performance-adjusted growth rates, already incorporates the effect of the cash infusion upon total assets.

Mean reversion does not explain the changes in operating performance documented in Table IV, either. Performance matching takes care of the mean reversion problem.

That is, if a sample firm had poor operating performance prior to the carve-out, it is matched with a set of control firms that also had poor operating performance prior to the carve-out. Any subsequent mean reversion experienced by the sample firm will also be experienced by the control group. Thus, if we find a significant difference between the changes in operating performance of the sample firm and the changes in operating performance of the control group, the difference is not due to mean reversion. (10)

Table IV offers strong support for the conjecture that carve-outs are followed by significant levels of new investment by the carved-out subsidiary. Although not reported here, the results of Table IV are generally evident for both cross-industry and own-industry carve-out events. Carve-out growth rates in sales, total assets, income, and capital expenditures are significantly positive (p = 0.00 in all cases) for cross-industry carve-out events. Carve-out growth rates in sales, total assets, and capital expenditures are significantly positive (p = 0.03, 0.03, and 0.06, respectively) for own-industry carve-out events.

B. Changes in the Efficiency of Operations

This section presents the methodology and the empirical results for our examination of changes in the efficiency of operations.

1. Methodology

We investigate changes in the efficiency of operations by examining sample firm changes in ROA (return on assets), ROCAA (return on cash-adjusted assets), ROMVA (return on market value of assets), and ROS (return on sales). Measures of operating income before depreciation, total assets, cash balances, common equity, and net sales come from Compustat. Parent accounting data are again adjusted for any consolidation of the subsidiary for reporting purposes.

ROA is defined as operating income before depreciation divided by the average of beginning-and ending-period book value of total assets. ROCAA is defined as operating income before depreciation divided by the average of beginning- and ending-period book value of assets net of cash balances. ROMVA is defined as operating income before depreciation divided by the average of beginning- and ending-period market value of total assets. Market value of total assets is calculated as the book value of total assets less the book value of common equity plus the market value of common equity. This measure is available only for parent firms, as the subsidiaries do not have market value figures prior to the carve-out. Finally, ROS is defined as operating income before depreciation divided by net sales.

Barber and Lyon (1996) indicate that the models that are most powerful in detecting abnormal performance are those that use a firm's lagged performance in forming a measure of expected performance. They also find that the power of the test statistics erodes as the expectation model moves from a firm's performance lagged by one year to that lagged by three years. Therefore, the model of expected performance is defined as:

E ([P.sub.it]) = [P.sub.i,t-l] + ([PI.sub.it] - [PI.sub.i,t-l]) (3)

where [P.sub.it] is the performance of firm i in year t, and [PI.sub.it] is the performance of firm i's industry in year t.

Industry performance is computed using median accounting figures for firms in the same four-digit SIC code as either the parent or its carve-out. We also examine size-adjusting (within 75% of the parent size) and pre-performance measures (within 25% of the parent event year - 1 performance).

The abnormal performance of firm i in year t, [AP.sub.it], is defined as realized performance, [P.sub.it], less expected performance, E([P.sub.it]):

[AP.sub.it] = [P.sub.it] - E([P.sub.it]) (4)

To test the null hypothesis that the median abnormal performance is equal to zero, we employ non-parametric Wilcoxon signed-rank test statistics. Using lags of two or three years yields results similar to those reported.

2. Empirical Results

Table V reports parent and subsidiary median industry-adjusted, size-adjusted, and performance-adjusted changes in accounting ratios over the (0, +1) event years, where year 0 represents the year the equity carve-out goes public. As shown in Panel A, parent firms experience positive changes in ROA, ROCAA, ROMVA, and ROS. These results are statistically significant for performance-adjusted changes. The median parent performance-adjusted change in ROA is 0.0447 (p = 0.00), in ROCAA 0.0553 (p = 0.00), in ROMVA 0.0335 (p 0.00), and in ROS 0.0629 (p = 0.00).

The results in Panel A indicate that parents perform significantly better than expected after the carve-out goes public. It appears that the equity carve-out makes the parent firm able to operate more efficiently.

The results for subsidiary changes in accounting ratios, reported in Panel B of Table V, are mixed. Median performance-adjusted changes in ROA, ROCAA, and ROS are positive and statistically significant (p = 0.00 for all three). Changes in subsidiary ROMVA cannot be calculated, as market value figures are not available prior to the carve-out.

By the other benchmarks, the results for subsidiary changes in ROA, ROCAA, and ROS are non-positive and generally insignificant. Carved-out subsidiaries have relatively high market-to-book ratios, suggesting that investors expect carve-outs to experience significant growth. If the rate of operating growth for the carve-out is similar for both the numerator and the denominator of a given accounting ratio measure, it is unlikely that the change in that accounting ratio will be statistically significant.

Overall, the results of Table V support the hypothesis that equity carve-outs accompany significant improvements in operating performance. These findings are consistent with the divestiture gains hypothesis. Unreported results for both cross-industry and own-industry carve-out events generally confirm the results of Table V. Changes in parent ROA, ROCAA, ROMVA, and ROS are positive and statistically significant for cross-industry carve-out events. Changes in parent ROS are positive and statistically significant for own-industry carve-out events.

V. Conclusions

Two general explanations have been suggested for parent announcement-period gains in equity carve-outs. Schipper and Smith (1986) conjecture that equity carve-outs result in divestiture gains due to separate financing for the subsidiary's investment projects, a more efficient set of contracts between shareholders and managers, and the creation of pure-play stocks. Nanda (1991) suggests that firms raise capital through equity carve-outs when parent firm shares are relatively undervalued and subsidiary shares are relatively overvalued. For a parent typically several times larger than the subsidiary, this leads to an increase in parent share price.

Empirical tests of the divestiture gains hypothesis and the asymmetric information hypothesis provide some support for both hypotheses. We test the two hypotheses more directly by focusing on the impact of carve-outs on parent rivals. The divestiture gains hypothesis predicts that announcement-period returns for parent rivals should be negative. The asymmetric information hypothesis implies the opposite.

The results support the divestiture gains hypothesis. We find that rivals of parent firms exhibit negative stock price reactions to equity carve-out announcements. Parents and their carved-out subsidiaries also exhibit improvements in operating performance, which we interpret as additional support for the divestiture gains hypothesis.

We thank Gordon Hanka, Shane Johnson, Daniel Klein, Scott Lee, Harold Mulherin, Dennis Sheehan, and Rene Stulz.

(*.) Heather M. Hulburt is an Assistant Professor at West Virginia University. James A. Miles and J. Randall Woolridge are Professors at Pennsylvania State University.

(1.) The authors and SSF construct industry portfolios and cumulative average adjusted portfolio returns in a similar manner.

(2.) Lie (2001) extends the work or Barber and Lyon (1996). Both studies find that adjusting for prior performance produces the most reliable test statistics.

(3.) By retaining at least 50%, the parent firm can consolidate the subsidiary for financial reporting purposes. By retaining at least 80%, the parent firm can consolidate the subsidiary for tax purposes, and can obtain a 100% dividend-received tax deduction.

(4.) Further study shows that regressing parent rival portfolio returns on the percentage of the carved-out subsidiary stock that the parent relinquishes yields a regression coefficient that is positive, but not significant.

(5.) The avcrage market-to-book ratio in the year after the carve-out is 2.34 for subsidiaries and 1.73 for parents; a difference of means tests yields a p-value of 0.01.

(6.) Because ARB 51 and SFAS 94 call for the consolidation of majority-owned subsidiaries for reporting purposes, we adjust the parent's accounting figures net of the subsidiary's accounting figures to get parent firm figures on a stand-alone basis. While ARB 51's general rule was to consolidate all majority-owned subsidiaries, in several exceptions that presumption was overcome. One such exception is when the companies engage in different lines of business. SFAS 94 requires consolidated financial statements for a parent company and a majority-owned subsidiary even though the subsidiary's business is unrelated to the parent company's business. Consolidated statements are required only when there is 50% or more ownership. Between 20% and 50%, or less, if there is "substantial influence," the equity method is used. Under this method if, for example, there is 20% ownership, 20% of the income of the subsidiary would be on the income statement of the parent, but there would not be a line by line consolidation of t he individual accounts (e.g., cash, receivables).

(7.) Twenty-nine carve-outs are merged or taken over within the first three years of trading.

(8.) We obtained Compustat accounting data for 15 carved-out subsidiaries with publicly traded debt in years -2 and -1. Growth rates were not high prior to the equity carve-outs for this small sample. Compustat accounting data are not available for years -2 and -l for the remainder of the sample. We know of no other electronic source of accounting data for firms that do not have publicly traded securities, except for the Securities and Exchange Commission's Electronic Data Gathering, Analysis, and Retrieval (EDGAR) system. Operating statistics of subsidiary units prior to carve-outs are reported on SEC form S-I. Companies were phased into EDGAR filing over a three-year period ending May 6, 1996.

(9.) The median offering size as a fraction of total equity carve-out market value is less than 17%.

(10.) See Barber and Lyon (1996) for a detailed description of the pre-event performance matching procedure.

References

Allen, F. and G. Faulhaber, 1989, "Signaling by Underpricing in the IPO Market," Journal of Financial Economics 23, 303-323.

Allen, J., 1998, "Capital Markets and Corporate Structure: The Equity Carve-Outs of Thermo Electron," Journal of Financial Economics 48, 99-124.

Allen, J. and J. McConnell, 1998, "Equity Carve-Outs and Managerial Discretion," Journal of Finance 53, 163-186.

Aron, D., 1991, "Using the Capital Market as a Monitor: Corporate Spinoffs in an Agency Framework," The Rand Journal of Economies 22, 505-518.

Barber, B. and J. Lyon, 1996, "Detecting Abnormal Operating Performance: The Empirical Power and Specification of Test Statistics," Journal of Financial Economics 41, 359-399.

Boone, A., 2001, "Can Focus Explain Carve-Out Gains?," Pennsylvania State University, Working Paper.

Comment, R. and G. Jarrell, 1995, "Corporate Focus and Stock Returns," Journal of Financial Economics 37, 67-87.

Daley, L., V. Mehrotra, and R. Sivakumar, 1997, "Corporate Focus and Value Creation: Evidence from Spinoffs," Journal of Financial Economics 45, 257-281.

Gaver, J. and K. Gaver, 1995, "Compensation Policy and Investment Opportunity Set," Financial Management 24, 19-32.

Hite, G. and J. Owers, 1983, "Security Price Reactions around Corporate Spin-Off Announcements," Journal of Financial Economics 12, 409-436.

Klein, A., J. Rosenfeld, and W. Beranek, 1991, "The Two Stages of an Equity Carve-Out and the Price Response of Parent and Subsidiary Stock," Managerial and Decision Economics 12, 449-460.

Lang, L., A. Poulsen, and R. Stulz, 1995, "Asset Sales, Firm Performance, and the Agency Costs of Managerial Discretion," Journal of Financial Economics 37, 3-37.

Leland, H. and D. Pyle, 1977, "Informational Asymmetries, Financial Structure, and Financial Intermediation," Journal of Finance 32, 371-387.

Lie, E., 2001, "Determining Abnormal Operating Performance: Revisited," Financial Management 30, 77-91.

Logue, D., J. Seward, and J. Walsh, 1996, "Rearranging Residual Claims: A Case for Targeted Stock," Financial Management 25, 43-61.

Miles, J. and J. Rosenfeld, 1983, "The Effect of Voluntary Spin-Off Announcements on Shareholder Wealth," Journal of Finance 38, 1597-1606.

Mulherin, H. and A. Boone, 2000, "Comparing Acquisitions and Divestitures," Journal of Corporate Finance 6, 117-139.

Myers, S. and N. Majluf, 1984, "Corporate Financing and Investment Decisions when Firms Have Information that Investors do not Have," Journal of Financial Economics 13, 187-221.

Nanda, V., 1991, "On the Good News in Equity Carve-Outs," Journal of Finance 46, 1717-1736.

Schipper, K. and A. Smith, 1983, "Effects of Recontracting on Shareholder Wealth: The Case of Voluntary Spin-Offs," Journal of Financial Economics 12, 437-467.

Schipper, K. and A. Smith, 1986, "A Comparison of Equity Carve-Outs and Seasoned Equity Offerings," Journal of Financial Economics 15, 153-186.

Slovin, M., M. Sushka, and Steven R. Ferraro, 1995, "A Comparison of the Information Conveyed by Equity Carve-Outs, Spin-Offs, and Asset Sell-Offs," Journal of Financial Economics 37, 89-104.

Vijh, A., 1999, "Long Term Returns from Equity Carveouts," Journal of Financial Economics 51, 273-308.

Vijh, A., 2002, "The Positive Announcement-Period Returns of Equity Carveouts: Asymmetric Information or Divestiture Gains?," Journal of Business 75 (Forthcoming).

Welch, I., 1989, "Seasoned Offerings, Imitation Costs, and the Underpricing of Initial Public Offerings," Journal of Finance 44, 421-449.

Table I. Descriptive Statistics This table shows descriptive statistics for 185 equity carve-out events over 1981-1994. Panel A. Distribution of Carve-Out Events by Year 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 11 6 8 6 6 21 15 9 9 7 22 18 1981 1993 1994 11 28 19 Panel B. Average Parent and Subsidiary Market Values (a) Average Parent Market Value Average Subsidiary Market ($ Millions) Value ($ Millions) 3,134.15 543.75 Panel C. Breakdown of Percentage of Common Stock Retained by Parent Firms Percentage of Common Stock Retained by Numbers of Parent Firms 0-39% 40-49% 50-59% 60-69% 70-79% 80-89% 90-99% 35 10 28 18 28 58 8 Panel D. Breakdown of Use of Proceeds by Offering Type (b) Use of Proceeds Secondary Offering Primary Offering (c) Paid Out to Paid Out to Paid Out Retained Creditors Parent 13 28 27 42 Secondary Offering Primary Joint Offering Offering (c) Paid Out Retained Paid Out (d) Retained 13 54 9 12 (a)Market value (share price multiplied by shares outstanding) is calculated at the end of event year 0. (b)A primary offering indicates that the subsidiary received all proceeds of the sale; a secondary offering indicates that the parent firm received all proceeds of the sale; and a joint offering indicates that both the subsidiary and the parent firm received some portion of the proceeds of the sale. (c)Proceeds are deemed paid out if they are paid to either shareholders or creditors; otherwise, proceeds are deemed retained. (d)In two cases, the carve-out firms pay out their portion of the proceeds to their parent firms. Table II. Stock Performance of Parents, Parent Rivals, and Carve-Out Rivals Surrounding Announcement Date This table presents market-adjusted common stock returns for the (-1, +1) event window, where event day 0 is the earlier of 1) the filing date, or 2) the first announcement in the Wall Street Journal. (a, b) Overall Cross-Industry (e) Panel A. Parent Firms Mean 1.92% 2.10% t-statistic p-value (c) (0.00) (***) (0.00) (***) Median 0.75% 0.59% Sign rank p-value (d) (0.00) (***) (0.00) (***) N 183 153 Panel B. Parent Rival Portfolios Mean -0.41% -0.42% t-statistic p-va1ue (c) (0.00) (***) (0.00) (***) Median -0.66% -0.69% Sign rank p-value (d) (0.00) (***) (0.00) (***) N 173 143 Panel C. Subsidiary Rival Portfolios Mean -0.41% -0.42% t-statistic p-value (c) (0.00) (***) (0.00) (***) Median -0.66% -0.66% Sign rank p-va1ue (d) (0.00) (***) (0.00) (***) N 182 152 Own-Industry (t) Difference Panel A. Parent Firms Mean -0.39% t-statistic p-value (c) (0.37) (0.43) Median 0.85% Sign rank p-value (d) (0.35) (0.97) N 30 Panel B. Parent Rival Portfolios Mean -0.39% t-statistic p-va1ue (c) (0.03) (**) (0.89) Median -0.59% Sign rank p-value (d) (0.04) (**) (0.61) N 30 Panel C. Subsidiary Rival Portfolios Mean -0.39% t-statistic p-value (c) (0.03) (**) (0.89) Median -0.59% Sign rank p-va1ue (d) (0.04) (**) (0.69) N 30 (***)Significant at the 0.01 level. (**)Significant at the 0.05 level. (a)Returns are calculated on a market-adjusted basis, where the CRSP equally weighted index is used as the measure for market returns. (b)We create portfolios of rivals for each carve-out event and calculate median market-adjusted returns for each portfolio. (c)The t-statistics test the hypothesis that the mean holding-period returns equal zero. (d)The two-tailed Wilcoxon signed rank statistics (sign rank) test the null hypothesis that the median performance-adjusted rates are equal to zero. (e)Carve-out events are deemed cross-industry when the parent and its subsidiary have different four-digit SIC codes. (f)Carve-out events are deemed own-industry when the parent and its subsidiary have the same four-digit SIC code. Table III. Stock Performance of Parent Rivals Surrounding Announcement Date By Amount of Parent Control Relinquished This table presents market-adjusted common stock returns for the (-1, +1) event window, where event day 0 is the earlier of 1) the filing date, or 2) the first announcement in the Wall Street Journal. (a,b) Parent Relinquishment of Control Less than 20% 20% to 50% Panel A. Full Sample of Parent Rival Portfolios Mean -0.55% -0.45% t-statistic p-value (c) (0.00) (***) (0.00) (***) Median -0.69% -0.80% Sign rank p-value (d) (0.00) (***) (0.00) (***) N 60 70 Difference of Means p-values: Column 1 versus column 2 (0.61) Column 2 versus column 3 (0.22) Column 1 versus column 3 (0.11) Panel B. Cross-Industry Sample of Parent Rival Portfolios (e) Mean -0.57% -0.42% t-statistic p-value (c) (0.00) (***) (0.02) (**) Median -0.78% -0.77% Sign rank p-value (d) (0.00) (***) (0.01) (***) N 48 55 Difference of Means p-values: Column 1 versus column 2 (0.52) Column 2 versus column 3 (0.48) Column 1 versus column 3 (0.20) Parent Relinquishment of Control More than 50% Panel A. Full Sample of Parent Rival Portfolios Mean -0.17% t-statistic p-value (c) (0.38) Median -0.37% Sign rank p-value (d) (0.16) N 43 Difference of Means p-values: Column 1 versus column 2 (0.61) Column 2 versus column 3 (0.22) Column 1 versus column 3 (0.11) Panel B. Cross-Industry Sample of Parent Rival Portfolios (e) Mean -0.24% t-statistic p-value (c) (0.24) Median -0.43% Sign rank p-value (d) (0.07) (*) N 40 Difference of Means p-values: Column 1 versus column 2 (0.52) Column 2 versus column 3 (0.48) Column 1 versus column 3 (0.20) (***)Significant at the 0.01 level. (**)Significant at the 0.05 level. (*)Significant at the 0.10 level. (a)Returns are calculated on a market-adjusted basis, where the CRSP equally weighted index is used as the measure for market returns. (b)We create portfolios of rivals for each carve-out event and calculate median market-adjusted returns for each portfolio. (c)The t-statistics test the hypothesis that the mean holding-period returns equal zero. (d)The two-tailed Wilcoxon signed rank statistics (sign rank) test the null hypothesis that the median performance-adjusted rates arc equal to zero. (e)Carve-out events are deemed cross-industry when the parent and its subsidiary have different four-digit SIC codes. Table IV. Parent and Subsidiary Operating Growth Rates This table shows median accounting variable growth rate figures over event years (0, +1) for all available parents and subsidiaries industry-adjusted, size-adjusted, and performance-adjusted. In each case, adjusting is defined as the sample firm's raw growth rate minus some measure of its industry rivals' median growth rate. Industry is defined as those firms with the same four-digit SIC code. Size-adjusting includes all industry rivals within 75% of the sample firm's size. Performance-adjusting includes all industry rivals with pre-performance (event years (-1, 0)) measures within 25% of the sample firm. (*) Total Sales Income Assets Panel A. Parent Firms Median Indusry-Adjusted Change -8.24% -1.76% -4.47% Sign-rank p-value (b) (0.08) (*) (0.25) (0.12) N 124 127 135 Median Size-Adjusted Change -2.63% -8.16% -6.74% Sign-rank p-value (0.01) (***) (0.02) (**) (0.00) (***) N 121 125 132 Median Performance- Adjusted Change 1.25% 6.52% 1.37% Sign-rank p-value (b) (0.00) (***) (0.00) (***) (0.00) (***) N 75 68 63 Panel B. Subsidiary Firms Median Industry-Adjusted Change 2.22% 4.47% 4.72% Sign-rank p-value (b) (0.07) (*) (0.85) (0.02) (**) N 154 154 168 Median Size-Adjusted Change 0.75% 0.00% 5.46% Sign-rank p-value (b) (0.20) (0.48) (0.00) (***) N 152 144 166 Median Performance- Adjusted Change 1.56% 0.01% 1.66% Sign-rank p-value (b) (0.00) (***) (0.00) (***) (0.00) (***) N 133 106 140 Capital Expenditures Panel A. Parent Firms Median Indusry-Adjusted Change -7.93% Sign-rank p-value (b) (0.48) N 128 Median Size-Adjusted Change 0.00% Sign-rank p-value (0.39) N 124 Median Performance- Adjusted Change 0.00% Sign-rank p-value (b) (0.05) (**) N 61 Panel B. Subsidiary Firms Median Industry-Adjusted Change 20.33% Sign-rank p-value (b) (0.00) (***) N 140 Median Size-Adjusted Change 3.42% Sign-rank p-value (b) (0.02) (**) N 139 Median Performance- Adjusted Change 4.75% Sign-rank p-value (b) (0.00) (***) N 92 (**)Significant at the 0.01 level. (**)Significant at the 0.05 level. (*)Significant at the 0.10 level. (a)The more defined measure results in a significant reduction in sample size. (b)The two-tailed Wilcoxon signed rank statistics (sign rank) test the null hypothesis that the median performance-adjusted rates are equal to zero. Table V. Parent and Subsidiary Operating Ratios This table shows median changes in operating performance over event years (0, +1) for all available parents and subsidiaries industry-adjusted, size-adjusted, performance-adjusted. In each case, adjusting is defined as the sample firm's raw growth rate minus some measure of its industry rivals' median growth rate. Industry is defined as firms with the same four-digit SIC code. Size-adjusting includes all industry rivals within 75% of the sample firm's size. Performance-adjusting includes all industry rivals with pre-performance (event years [-1, 0]) measures within 25% of the sample firm. (a) ROA (return on assets) is defined as operating income before depreciation divided by the average of beginning- and ending-period book value of total assets. ROCAA (return on cash-adjusted assets) is defined as operating income before depreciation divided by the average of beginning- and ending-period book value of assets net of cash balances. ROMVA (return on market value of assets) is defined as operating income before depreciation divided by the average of beginning- and ending-period market value of total assets. Market value of total assets is calculated as the book value of total assets less the book value of common equity plus the market value of common equity. This measure is not available for the subsidiary. ROS (return on sales) is defined as operating income before depreciation divided by net sales. ROA ROCAA Panel A. Parent Firms Median Industry-Adjusted Change 0.0017 0.0059 Sign-rank p-value (b) (0.66) (0.30) N 110 110 Median Size-Adjusted Change 0.0032 0.0047 Sign-rank p-value (b) (0.42) (0.11) N 106 106 Median Performance-Adjusted Change 0.0447 0.0553 Sign-rank p-value (b) (0.00) (***) (0.00) (***) N 60 60 Panel B. Subsidiary Firms Median Industry-Adjusted Change -0.0073 -0.0090 Sign-rank p-value (b) (0.15) (0.14) N 132 132 Median Size-Adjusted Change -0.0039 -0.0061 Sign-rank p-value (b) (0.28) (0.20) N 127 127 Median Performance-Adjusted Change 0.0802 0.0968 Sign-rank p-value (b) (0.00) (***) (0.00) (***) N 122 122 ROMVA ROS Panel A. Parent Firms Median Industry-Adjusted Change 0.0020 0.0023 Sign-rank p-value (b) (0.60) (0.90) N 110 110 Median Size-Adjusted Change 0.0006 0.0025 Sign-rank p-value (b) (0.76) (0.19) N 94 106 Median Performance-Adjusted Change 0.0335 0.0629 Sign-rank p-value (b) (0.00) (***) (0.00) (***) N 55 72 Panel B. Subsidiary Firms Median Industry-Adjusted Change -0.0092 Sign-rank p-value (b) NA (0.05) (**) N 140 Median Size-Adjusted Change 0.0000 Sign-rank p-value (b) NA (0.80) N 135 Median Performance-Adjusted Change 0.0692 Sign-rank p-value (b) NA (0.00) (***) N 128 (***)Significant at the 0.01 level. (**)Significant at the 0.05 level. (a)Closing this gap results in a significant reduction in sample size. (b)The two-tailed Wilcoxon signed rank statistics (sign rank) test the null hypothesis that the median performance-adjusted rates are equal to zero.

Printer friendly Cite/link Email Feedback | |

Author: | Hulburt, Heather M.; Miles, James A.; Woolridge, J. Randall |
---|---|

Publication: | Financial Management |

Article Type: | Statistical Data Included |

Geographic Code: | 1USA |

Date: | Mar 22, 2002 |

Words: | 9478 |

Previous Article: | Why firms diversify: An empirical examination. |

Next Article: | The Modigliani and Miller theorem and integration of financial markets. |

Topics: |