The Reaction of Security Prices to Tracking Stock Announcements

Transcription

1 Journal of Economics and Finance Volume 24 Number 1 Spring 2000 Pages The Reaction of Security Prices to Tracking Stock Announcements John Elder and Peter Westra* Abstract This paper provides some empirical evidence on a relatively new and increasingly prevalent form of equity restructuring called tracking stock. We identify the effects associated with tracking stock announcements by excluding from our sample those announcement events that include other significant news announcements on the event date, such as announcements of acquisitions and earnings. For the 35 announcement events that fit this criteria, we find a mean abnormal return of over 3 percent in the two-day period surrounding the announced proposal to issue a tracking stock, with 30 of the 35 firms in the sample earning positive abnormal returns. Introduction This paper provides some empirical evidence on a relatively new and increasingly prevalent form of equity restructuring called tracking stock by examining the reaction of share prices to announcements proposing a tracking stock. By comparing returns relative to those predicted by a market model, we find a mean abnormal return of over 3 percent in the two-day period surrounding the announcement for a sample of 35 firms. Individually, 30 of the 35 firms earned positive abnormal returns. These estimates are comparable quantitatively to previous estimates of abnormal returns associated with announcements of carve-outs and spin-offs. Tracking stock, also known as targeted or lettered 1 stock in the financial press, is a class of common stock that is linked to the performance of a specific business group within a diversified firm. Tracking stock was first issued in 1984, but has been more widely implemented only very * John Elder, Department of Economics, Middlebury College, Middlebury, VT 05753, jelder@middlebury.edu; Peter Westra, Deutsche Banc Securities Inc., 31 West 52nd Street, New York, NY10019, peter.westra.99@alumni.middlebury.edu. The views expressed in this paper are that of the author(s) and do not reflect the views or opinions of Deutsche Bank Securities Inc. or any of its affiliates. The authors thank without implication Peter Matthews and seminar participants at Middlebury College for useful comments and suggestions on a previous version of this paper. 1 As noted by Hass (1996), the terms lettered and alphabet stock arose when General Motors distinguished its tracking stocks as class E and, later, class H. The term targeted stock was introduced by Lehman Brothers in the early 1990s in an attempt to establish a proprietary name for its related services. The Securities and Exchange Commission typically requires corporations to adopt the term group to identify the tracked division.

2 The Reaction of Security Prices to Tracking Stock Announcements 37 recently. Six years after its introduction, a total of two tracking stocks had been issued. In the year 1999, sixteen tracking stocks were proposed, with several more rumored to be imminent. There currently exists a relatively small literature on tracking stocks. Hass (1996) provides an excellent survey of legal issues associated with the tracking stock structure. Logue, Seward, and Walsh (1996) detail important features of tracking stocks and find some evidence of abnormal returns associated with an announcement to issue a tracking stock, but, constrained by a very small sample of nine firms, do not conduct statistical inference. Even more recent studies, such as Billett and Mauer (2000), D Souza and Jacob (2000), and Zuta (1999), are constrained by relatively small samples. For example, the former two studies include 18 and 12 non-acquisition related announcements, respectively, and, constrained by sample size, do not exclude announcements contaminated by other major simultaneous events such as earnings announcements that confound identification of the price effect associated with the announcement of the tracking stock. Moreover, the distributions of the relevant test statistics are justified only asymptotically. Drawing from a much larger population of 51 tracking stock announcements, this paper carefully identifies price effects associated with tracking stock announcements by excluding those announcement events that occur simultaneously with other significant announcements. Tracking stocks possess some very unusual features relative to the product of related, but more conventional, equity restructurings such as carve-outs and spin-offs. Both carve-outs and spin-offs create a new corporate entity by distributing equity in a subsidiary outside the parent corporation. Shareholders in the new firm typically possess the rights conventionally due shareholders: the right to elect a board of directors, the right to vote on matters of significant importance, and a claim against the firm s net assets. Shareholders of a tracking stock, however, do not elect directors to oversee management of the tracked business group, nor do they have a claim against the assets of the tracked group. The primary feature that links the tracking stock to the tracked business group is a limited claim on the future earnings of that group. These unconventional features, coupled with a lack of empirical research, may account for the disparate assessments that have recently appeared in the financial press. Recent headlines range from:... Targeted stock: Shun It, Some Experts Say 2 and On the Wrong Track: Complex Financial Innovations Like Tracking Stocks... Bring Few Benefits to Shareholders 3 to numerous claims by others, including many practitioners, that tracking stocks unlock value, 4 or, possibly, unleash value. 5 Understanding the reaction of financial markets to announcements of a firm s intention to reorganize under a tracking stock structure is therefore interesting and relevant on its own merits. It would also, however, contribute directly to the established literature detailing the reaction of share prices to announcements of other restructurings. Several authors, including Schipper and Smith (1983), have estimated a mean abnormal return of about 3 percent associated with an announcement of a firm s intention to spin-off a subsidiary. 6 Schipper and Smith (1986) estimate a mean abnormal return of about 2 percent associated with an announcement of an equity carve-out. Hite, Owers, and Rogers (1987) estimate a mean abnormal return of about 1.7 percent associated with a total or partial asset sale. Several hypotheses have been offered to explain these findings. One category, broadly defined, 2 New York Times, It s Called Targeted Stock: Shun It, Some Experts Say, July 12, Financial Times, Complex Financial Innovations Like Tracking Stocks Allow Managers to Retain Control, but Bring Few Benefits for Shareholders, May 18, Boston Globe, Genzyme Tracking Stocks Are Off Track on Returns, June 24, 1999; Dow Jones Newswires, Firms Turn to Tracking Stocks to Unlock Value of Web Units, July 12, 1999; and New York Times, Shares that Track Assets Add Value at a Cost, July 18, San Francisco Chronicle, Tracking Stock Can Unleash a Unit svalue, May 11, See also Miles and Rosenfeld (1983) and Hite and Owers (1983).

3 38 JOURNALOF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 suggests that investors expect gains, possibly due to decreased agency costs, from more focused and transparent business units. For example, Schipper and Smith (1983, 1986, and 1989) suggest that because carve-outs and spin-offs require a commitment to increased financial disclosure, they may facilitate the implementation of improved incentive schemes or otherwise improve the ability of shareholders to monitor management. More formally, Milgrom (1988), Bagwel, and Zechner (1993) and Harris and Raviv (1996) show how asymmetries between lower level and top management in information, preferences, or incentives may result in an inefficient allocation of capital. Empirically, Desai and Jain (1999) find that spin-offs that increase focus earn positive long-run abnormal returns. Vijh (1999) finds some evidence that improved focus is a determinant of the long-run performance of carved-out subsidiaries. 7 Alternatively, Nanda (1991) suggests that the positive abnormal returns from carve-outs may be due to transitory differences between the market value and managers perceived value of both the subsidiary and the parent firm. Nanda shows that when such differences exist, managers may find it optimal to finance new equity through a carve-out, which signals markets that managers believe the parent firm to be undervalued. Empirically, Slovin, Sushka, and Ferraro (1995) analyze the share-price reaction of rivals to firms that announce voluntary asset divestitures and find evidence in support of this view. Another hypothesis, offered by Schipper and Smith, suggests that investors value the relatively pure-play investment opportunities created by carve-outs, since the pure-play allows investors to invest separately in the subsidiary without simultaneous investment in the parent. However, arbitrage dictates that this should matter only to the extent that the new pure-play is unique. Tax and regulatory considerations may also play a role, as both Schipper and Smith (1983) and Hite and Owers (1983) find that a substantial number of firms in their samples expect to benefit through such considerations. Given the related nature of spin-offs, carve-outs, and tracking stocks, it seems reasonable to ask whether the observed abnormal returns associated with announcements of each are consistent with one hypothesis. Unlike spin-offs, tracking stocks do not yield tax or regulatory advantages relative to the consolidated corporation. Tracking stocks also do not exploit transitory differences in valuations in the same manner as carve-outs, since tracking stocks frequently raise no new equity capital. However, tracking stocks, like carve-outs and spin-offs, do tend to result in more focused business units committed to increased financial disclosure. The empirical finding that announcements of each earn positive abnormal returns is therefore consistent with the hypothesis that investors expect gains from more focused and transparent business units. The rest of the paper is organized as follows. The second section provides more details on tracking stocks. The third section presents an overview of the methodology with additional details included in an appendix. The fourth presents the data and empirical results, and the fifth concludes. Tracking Stocks, Carve-Outs, and Spin-Offs The formation of an equity structure based on tracking stock is, in some respects, similar to a spin-off or a carve-out, but with considerable differences. In both spin-offs and carve-outs, a new corporate entity is created by distributing equity in a subsidiary, where it then begins trading publicly. A spin-off distributes the equity of the spun-off subsidiary to the shareholders of the parent as a dividend, resulting in no new equity financing. Typically the parent firm relinquishes control over the subsidiary by retaining little or no equity ownership. 7 For additional recent empirical evidence on positive effect of corporate focus on firm value, see Berger and Ofek (1995) and Comment and Jarell (1995).

4 The Reaction of Security Prices to Tracking Stock Announcements 39 In a carve-out, the parent corporation sells a portion of the equity in the subsidiary, either as an initial public offering by the subsidiary itself or as a seasoned offering by the parent. Either method results in new equity financing. Due to tax considerations the parent firm typically maintains at least 80 percent of the voting rights and a majority interest in the carved-out subsidiary. Under both restructurings, shareholders in the new corporate entity possess the rights conventionally due shareholders: the right to elect a board of directors to oversee management, the right to vote on matters of great importance, and a claim against the new entity s net assets. In contrast, the formation of an equity structure based on tracking stock does not create a new corporate entity. A tracking stock structure is formed by creating a new class of common stock, within an existing diversified corporation, whose value is linked to the performance of a specific business group through special provisions introduced into the firm s articles of incorporation. This link is usually strongest through a limited claim on the earnings generated by the division. Typically, the dividends paid are dependent on the earnings generated by the tracked group in a structured fashion either as a fixed percentage of the tracked group s net income or as a specific dollar amount that is adjusted over time according to net income or shareholders equity (see Logue, Seward, and Walsh 1996). The upper bound on dividends for the tracked group is therefore comparable to the amount that would be available had the division been spun-off or carved-out. This analogy is not exact because the group may share with the parent firm the use and cost of inputs that are relatively fixed, such as corporate offices and payroll services, that it otherwise would not as separate corporate entity. Shares in the tracking stock may be distributed either as a public offering, as dividends to existing shareholders, or as currency for an acquisition. The first method results in new equity financing, although it has been used less frequently. As with carve-outs and spin-offs, the physical assets of the consolidated firm are divided and attributed to the particular business groups. Unlike carve-outs or spin-offs, however, this division is solely for accounting purposes a tracking stock does not represent a legal claim on the assets of the associated business group. Rather, a tracking stock represents a claim on a fraction of the assets of the consolidated firm. For example, in the event of liquidation of the parent corporation, the claim would typically be dependent on the proportion of the total market value accounted for by each class of stock. In the event that assets of the tracked group are sold, the parent might be obligated to distribute the proceeds as a special dividend or as a share repurchase. 8 The tracked group is also not governed by a separate and independent board of directors. It is governed by the directors of the parent firm, with the shareholders typically having voting rights that float with the market value of their tracking stock relative to that of the total market capitalization of all classes of common stock for the firm. Such distribution of voting rights ensures that the votes per dollar of market value are approximately equal for each class of common stock, but they also imply that the directors will answer to at least two groups of stockholders with potentially very different and competing interests. The interests of each tracked group will therefore be subordinate to the interests of the consolidated firm. 9 There are, then, at least two related aspects in which a tracked business group will behave differently than the independent corporate entity created by a carve-out or a spin-off. First, the directors may not have an incentive to strategically manage the tracked business group in its own best interests, and, second, there exists the opportunity for considerable cross-subsidization across business groups, either through exposure to the liabilities of the consolidated firm or through purposeful redirection of resources. Both of these factors would tend to reduce the extent to which the tracked business group represents a pure-play to investors relative to a carved-out or spun-off sub- 8 Logue, Seward, and Walsh (1996) provide additional details. 9 Hass (1996) analyzes from a legal perspective the implications of such potential conflicts of interest.

5 40 JOURNALOF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 sidiary, but they should not impede expected gains in managerial efficiency through the design of and implementation of improved incentive structures or the ability of shareholders to more closely monitor the management of the individual business groups. As mentioned in the previous section, the empirical evidence on carve-outs and spin-offs indicates abnormal returns in the days associated with the announcement of 2 percent to 3 percent. Given the features of tracking stocks outlined here, if investors expect gains from more focused and transparent business units, we should expect to observe abnormal returns of a similar magnitude, although tempered by the extent to which investors view the lack of autonomy by the tracked business group as a liability. Methodology The methodology we employ is relatively standard to event studies. For each firm we estimate abnormal returns beginning 51 days prior to the announced event by calculating the daily difference between the actual return on that firm s equity and a forecasted normal return. We then calculate cumulative abnormal returns (CAR) by summing the abnormal return on consecutive trading days. Normal returns are defined as the forecast from a market model conditional on the contemporaneous return on a market index, where the parameters of the market model are estimated for each firm over a 230-day estimation window, ending 51 days prior to the event. Our market index is the Standard and Poor s 500. Other authors, such as Spiess and Graves (1999), have found that single-factor models produce results comparable to those from, for example, multi-factor models, such as a three-factor model based on Fama and French (1993). In calculating our test statistics, we relax a simplifying assumption common to event studies that the sampling error of the estimated parameters is zero, even though our estimation window is large. In practice, our test-statistics are somewhat smaller in magnitude when we account for sampling error on the order of 10 percent or so. The details of the estimation methodology are included in the appendix. Data and Results Prior to December 31, 1999, there were 51 announcements by managements proposing a tracking stock, beginning with General Motors in These were identified by searching the news archives, The Wall Street Journal, the Dow Jones Newswire, The New York Times, and other major news sources; searching SEC documents; and verifying against published lists of currently trading tracking stocks. It is possible that this method may have failed to identify a firm receiving little press coverage that announced an intention to issue a tracking stock but subsequently canceled, but this number, if not zero, is likely to be very small. The announcement need not indicate a filing with the Securities and Exchange Commission, but rather indicate an intention by management to formally propose the tracking stock to shareholders. The nine announcement dates prior to February 1995 are identical to those indicated by Logue, Seward, and Walsh (1996), except one, the date for Ralston, for which we found a published announcement three days prior. 10 In order to properly identify the effects associated with the announcement of the tracking stock, we also exclude announcements contaminated by significant confounding events, as described below. Earlier studies, such as Billett and Mauer (2000) and Zuta (1999), are constrained by very small 10 This method is slightly different than that of Billett and Mauer (2000), for example. Billet and Mauer use the date of the first news reference to a tracking stock, even if the announcement indicates only that the firm is, for example, considering proposing a tracking stock structure or a spin-off. This results in our using different announcement dates for TCI, U.S. West, and Inco Limited. Our method of dating also excludes announcements for firms that consider proposing a tracking stock structure, without ever formally proposing one. During 1999, such announcements, such as those for Microsoft, have been numerous.

6 The Reaction of Security Prices to Tracking Stock Announcements 41 TABLE 1. SUMMARY OF TRACKING STOCK ANNOUNCEMENTS Date Parent Corporation Notes Industry of Parent (by two-digit SIC) 10/1/84 General Motors a Acquisition Auto and Truck Manufacturers 11/1/85 General Motors a Acquisition Auto and Truck Manufacturers 2/1/91 USX (USX1) Primary Metal Industries 4/14/92 USX (USX2) Primary Metal Industries 1/22/93 Ralston-Purina b (RAL) Simultaneous Food and Kindred Products 3/2/93 RJR Nabisco Holdings (RN) Canceled Food and Kindred Products 3/15/93 Pittston Company (PCO) Coal Mining 7/29/93 Fletcher Challenge a Foreign Paper and Allied Products (foreign listed) 1/5/94 K-Mart (KM) Canceled General Merchandise Stores 3/14/94 Seagull Energy (SGO) Canceled Electric, Gas, and Sanitary Services 7/27/94 Genzyme a Acquisition Chemicals and Allied Products 11/16/94 Tele-Communications (TCI1) Communication 2/1/95 American Health Properties (AHP) Health Services 2/15/95 CMS Energy (CMS) Electric, Gas, and Sanitary Services 4/10/95 US West (USW) Communication 8/3/95 MCI (MCIC) Canceled Communication 9/16/95 Pittston Services Group (PZB) Coal Mining 4/3/96 Inco a Acquisition Metal Mining 8/12/96 Delmarva Power and Light a Acquisition Electric, Gas, and Sanitary Services 9/28/96 Fletcher Challenge a Simultaneous Paper and Allied Products 11/2/96 Circuit City Stores (CC) Automotive Dealers and Service Stations 11/7/96 Epitope a Simultaneous Medical Equipment and Supplies 12/20/96 Tele-Communications Inc. (TCI2) Communication 2/4/97 Genzyme a Acquisition Chemicals and Allied Products 4/23/97 Viacom a Simultaneous Communication 5/22/97 Tele-Communications Inc. (TCI3) Communication 9/18/97 Georgia Pacific (GP) Lumber and Wood Products 12/19/97 Telephone Data Systems (TDS) Canceled Holding and Other Investment Offices 5/19/98 Sprint a Acquisition Industrial Machinery and Equipment 6/25/98 AT&T a Acquisition/Canceled Communication 7/21/98 Agouron b (AGPH) Simultaneous Chemicals and Allied Products 8/18/98 Incyte a Canceled/Simultaneous Chemicals and Allied Products 9/23/98 Perkin-Elmer (PKN) Instruments and Related Products 10/8/98 Ziff-Davis a Simultaneous Printing and Publishing 12/7/98 Hemispherx Biopharma (HB) Canceled Chemicals and Allied Products 3/1/99 Quantum (QNTM) Industrial Machinery and Equipment 3/11/99 Dupont (DD) Pending Chemicals and Allied Products 3/18/99 Donaldson, Lufkin & Jenrette (DLJ) Security and Commodity Brokers 5/13/99 Snyder Communications (SNC) Engineering and Management Services 5/19/99 J. C. Penney b (JCP) Pending/Simultaneous General Merchandise Stores 6/13/99 Walt Disney a Acquisition Motion Pictures 6/17/99 Genzyme (GENZ) Chemicals and Allied Products 8/3/99 Sovereign Bancorp (SVRN) Pending Holding and Other Investment Offices 9/16/99 Staples (SPLS) Wholesale Trade/Nondurable Goods 9/30/99 Cendant (CD) Pending (1/14/99) Hotels and Other Lodging Places 11/11/99 Knight-Ridder (KRI) Pending Printing and Publishing 11/23/99 Excite at Home (ATHM) Pending Personal Services 11/23/99 Electronic Arts a Pending Amusement and Recreation Services 11/24/99 AT&T (ATT) Pending Communication 12/22/99 Andrx (ADRX) Pending Chemicals and Allied Products 12/22/99 Cablevision Systems (CVC) Pending Communication Totals 51 Notes: This table reports summary information on the announcements of tracking stocks. The note acquisition indicates the number of tracking stock announcements that were intended for use in the acquisition of another company. The note canceled indicates the number of tracking stock announcements that were subsequently canceled. The date indicates the date the announcement appeared in a printed news source, which is usually one day after the actual announcement. a Indicates firm that is excluded from the sample of 35 firms, either because the announcement was for an acquisition or because there was a significant simultaneous announcement. b Indicates firm that is excluded from the sub-sample of 32 firms because of a simultaneous announcement.

7 42 JOURNALOF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 samples from before 1998 and therefore do not exclude all announcements that are contaminated by simultaneous events. The announcement day (t=0 in event time) is defined as the day that news of the announcement appeared in a printed news source. Any share price reactions to tracking stock announcements should therefore be captured at t=-1 or t=0. Table 1 details the main features of these announcements. More than half of the announcements, 23, have occurred in the 2 years prior to December During that period, there has been an increase in the concentration of parent firms in technology-oriented industries, such as the communication and chemical and allied products (which includes pharmaceutical) industries. Earlier announcements were more evenly distributed among manufacturing industries, such as auto and truck manufacturing, food products, coal mining, utilities, and paper products. Of the 51 total announcements, one was by a foreign-based company, and nine proposed introducing a tracking stock in conjunction with the acquisition of another firm. Because the simultaneous announcement of an acquisition confounds identification of the share price response to the tracking stock announcement, this study considers only those announcements by firms proposing to split their existing business groups into a tracking structure, which leaves a population of 41 announcements. Of these 41 announcements, one was by a firm, Ziff-Davis, that first went public only 113 trading days previously, which leaves a relatively short and volatile window for estimating the parameters of the market model. This firm is dropped from the sample. Five other firms made other major announcements concurrent with the announcement of the tracking stock. The first, Viacom, made three announcements on the same day: the proposed tracking structure for its Blockbuster business group, that earnings for the Blockbuster business group would be 15 percent to 20 percent FIGURE 1. MEAN CUMULATIVE ABNORMAL RETURN (CAR) FOR SAMPLES OF 32 AND 35 FIRMS

8 The Reaction of Security Prices to Tracking Stock Announcements 43 TABLE 2. DAILY MEAN ABNORMAL RETURNS AND MEAN CUMULATIVE ABNORMAL RETURNS Full Sample; N=35 Excluding All Simultaneous; N=32 Event Mean Test- Mean Mean Test- Mean Day AR[t] statistic CAR[-50,t] AR[t] statistic CAR[-50,t] Notes: This table reports mean abnormal returns and mean cumulative abnormal return for the days specified for the full sample of 35 firms and the sub-sample which excludes 3 firms that made simultaneous announcements. Mean abnormal returns and mean cumulative abnormal returns for -50 to -21 are not reported. The reported test statistic is a test for the null hypothesis that the abnormal return equals zero for day t and is asymptotically distributed Gaussian with unit variance. The market index is the Standard and Poor s 500. lower than the year prior, and that the chief executive of the Blockbuster business group was resigning unexpectedly. 11 Several sources cited by the New York Times, including the chairman of Viacom, emphasized the role of the resignation in determining the market s reaction, which was described as reeling. Incyte announced a tracking stock simultaneously with an acquisition of another firm, and some news releases are ambiguous as to whether the tracking stock was intended to finance the acquisition. Fletcher Challenge, in its second announcement in 1996, announced the tracking stock concurrently with earnings and other news. Epitope announced the tracking stock concurrently with the acquisition of another firm. Electronic Arts announced a tracking stock 11 New York Times, It s Blockbuster Chairman Out; Viacom to Issue Shares to Unit, April 23, 1997.

9 44 JOURNALOF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 TABLE 3. MEAN CUMULATIVE ABNORMAL RETURNS FOR SELECTED INTERVALS Full Sample; N=35 Excluding All Simultaneous; N=32 Mean Test- Mean Test- Interval CAR[t 1,t 2 ] statistic CAR[t 1,t 2 ] statistic -50 to to to to to to to to to to to to to Notes: This table reports mean CARs over the intervals specified for the full sample of 35 firms and the sub-sample, which excludes the 3 firms that made simultaneous announcements. The reported test statistic is a test for the null hypothesis that the mean cumulative abnormal return equals zero and is asymptotically distributed Gaussian with unit variance. simultaneously with an agreement to be the dominant provider of games to American Online and to acquire another company. Because these other news events clearly confound identification of the effects associated with the tracking announcement, these firms are also dropped from the sample. This leaves a sample of 35 announcements. In addition, three other firms, Ralston, Agouron, and J. C. Penney, made other news announcements on the same day as their proposed tracking stock structure. Although it is difficult to gauge precisely the expected impact of these other news announcements, there is some evidence to suggest it may be small. Ralston and Agouron reported earnings that, according to Valueline forecasts, may have been slightly less than expected. J. C. Penney made a restructuring announcement that was unrelated to the tracking stock and featured much less prominently in news accounts. We recognize that these simultaneous announcements, even if their impact is small, confound identification of the reaction to the tracking stock announcement, so we report results for the sub-sample of 32 firms that omits all firms with simultaneous announcements, although we also include results for the sample of 35 announcements that includes these three announcements. Six announcements proposing a tracking stock structure were subsequently canceled, and several others are pending shareholder approval. 12 The cancellations usually occurred either because the proposal was rejected by shareholder vote, or because management withdrew the proposal prior to a vote by shareholders due to negative feedback. These firms are included in the sample because we are interested in measuring the share price effects associated with the announcement of an intention to issue a tracking stock, regardless of whether the tracking stock was subsequently sample. 12 Although there were seven total cancellations, one was for a proposed acquisition, leaving six cancellations in our

10 The Reaction of Security Prices to Tracking Stock Announcements 45 issued. Deleting these firms from the sample would introduce a positive survivorship bias, since our sample would include only those announcements that, ex-post, were well received by shareholders. However, for comparison, we also report summary results with these firms deleted from the full sample. Estimation Results We estimate the parameters of the market model for each firm over a 230-day interval, ending 51 days prior to the announcement, using the Standard and Poor s 500 as the market index. We then compute abnormal returns beginning 50 days prior to the published announcement, although we focus attention on the event interval from -1 to 0, given our definition of day 0 as the day the announcement appeared in a printed source. Figure 1 reports the mean cumulative abnormal return averaged across firms beginning 50 days prior to the announcement. There is relatively little movement from zero for dates -50 to -35, then the mean CAR slowly rises, with a drop beginning about 15 days prior to the announcement. About four days prior to the event, the mean CAR rises more rapidly, with a noticeably sharp spike upward during the two-day announcement period, before leveling out afterwards. This is clearly consistent with positive share price reactions to tracking stock announcements. Table 2 reports the corresponding mean CAR, along with the mean abnormal return and a test statistic for the null hypothesis that the mean abnormal return for each individual day is zero. Results are reported for both the full sample of 35 firms and the sub-set of 32 firms that made no simultaneous news announcement. In the 19 days from -20 to -2 for the full sample, there are zero occurrences of statistically significant abnormal returns. The two greatest abnormal returns occur at days -1 and 0, with returns of 2.1 percent and 1.0 percent, respectively. They are also associated with the two greatest test statistics, 4.31 and 2.15, strongly rejecting the null of zero abnormal return on each day. This is very strong evidence in support of the hypothesis that the share price reactions to tracking stock announcements have been significantly positive and economically meaningful about 3.1 percent over the two-day event period. The results for the 32-firm sub-sample are qualitatively similar, although slightly less in magnitude. The two-day return from -1 to 0 is 2.6 percent, and the abnormal returns on days -1 and 0 have test-statistics of 3.35 and To further test whether the behavior of stock prices around the interval -1 to 0 is different from other intervals, we analyze the mean CAR over a sequence of intervals throughout the event period. These are reported in Table 3. The test-statistic, asymptotically distributed N(0,1), for the full sample over the interval [-1,0] is 4.72, which is highly significant. The mean CAR from -50 to -2 slowly drifts upward to 1.4 percent, although not significantly so, before spiking upward on day -1. The total combined effect over the [-50,0] interval is 4.4 percent. Again, the results for the sub-sample of 32 firms are qualitatively comparable although slightly less in magnitude. The total effect over the interval [-50,0] is about 3.5 percent, and the upward drift from -50 to -2 is not significant. The CAR for the interval [-1,0] for the 32-firm sub-sample is also highly significant, with a test-statistic of The results from the evenly spaced six-day intervals from -50 to +2 are similar. The only interval for which the CAR is significantly different from zero is the interval [-5,0], which includes the tracking stock announcement. The empirical results thus far are strongly suggestive of a positive share-price reaction to tracking announcements, but it is possible that since we have examined only aggregations across firms that they could be influenced by a relatively small subset of firms. We examine this issue in tables 4 through 6 and figures 2 and 3. Table 4 presents abnormal returns by firm for an 11-day period centered around the announce-

11 46 JOURNALOF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 FIGURE 2. T-STATISTIC OF ABNORMAL RETURN BY FIRM (FOR SAMPLES OF 32 AND 35 FIRMS) USX1 SGO PZB USX2 TCM CC RAL CMS TC12 RN AHE TC13 PCO UW GP KM MCIC TDS

12 The Reaction of Security Prices to Tracking Stock Announcements 47 FIGURE 2. T-STATISTIC OF ABNORMAL RETURN BY FIRM (CONTINUED) (FOR SAMPLES OF 32 AND 35 FIRMS) AGPH SNC1 KRI PKN JCP ATHM HEB GENZ TT QNTM SVRN ADRX DD SPLS CVC DLJ CD T-stat of Mean C

13 48 JOURNALOF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 FIGURE 3. CUMULATIVE ABNORMAL RETURN (CAR) BY FIRM MIN =.35 MAX =.35 USX1 SGO PZB USX2 TCI1 CC RAL CMS TC12 RN AHE TC13 PCO UW GP KM MCIC TDS

14 The Reaction of Security Prices to Tracking Stock Announcements 49 FIGURE 3. CUMULATIVE ABNORMAL RETURN (CAR) BY FIRM (CONTINUED) MIN =.35 MAX =.35 AGPH SNC1 KR1 PKN JCP ATHM HEB GENZ TT QNTM SVRN ADRX DD SPLS CVC DLJ CD Mean CAR

15 50 JOURNALOF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 TABLE 4. ABNORMAL RETURNS AND CUMULATIVE ABNORMAL RETURNS BY FIRM CAR Firm [-1,0] t-stat USX USX RAL b RN a PCO KM a SGO a TCI CMS AHP USW MCIC a PZB CC TCI TCI GP TDS a AGPH b PKN HB a QNTM DD DLJ SNC JCP b GENZ SVRN SPLS CD KRI ATHM TT ADRX CVC Mean Median Mean Excluding Three Firms with Simultaneous Announcements Notes: This table reports the daily abnormal returns for the 11-day period surrounding the announcement and the cumulative abnormal return over the interval [-1,0] for each firm in the sample. The reported test statistic is for the null hypothesis that CAR i [-1,0]=0. The test statistic is distributed Student-t with 228 degrees of freedom, or approximately Gaussian with unit variance. The test statistic for the mea CAR[-1,0] is asymptotically distributed Gaussian with unit variance. The mean CAR is reported for both the full sample of 35 firms and the sub-sample which excludes three firms that made simultaneous announcements. a Indicates a firm that subsequently canceled the proposed tracking stock. b Indicates a firm that made a simultaneous announcement.

16 The Reaction of Security Prices to Tracking Stock Announcements 51 TABLE 5. NON-PARAMETRIC TESTS FOR PERCENTILES Null Hypothesis Number Number of Percent Test-statistic of Firms CAR i [-1,0] > 0 Positive Number of positive CAR i [-1,0] % 4.23 equals half sample; Full Sample Number of positive CAR i [-1,0] % 3.89 equals half sample; Excluding Firms with Simultaneous Announcements Notes: This table reports a non-parametric test for percentiles for the full sample of 35 firms. It is a test for the null hypothesis that the number of positive CAR i [-1,0] equals one-half of the sample. In particular, the test-statistic is [ ] N* Np = 2 Np(1 p N(0,1), where N is the total number of trials, p is the probability of a particular outcome under the null, and N* is the actual observed number of particular outcomes. For the full sample, N = 35, N* = 32, and p = 5. a TABLE 6. MEAN CUMULATIVE ABNORMAL RETURNS FOR SELECTED SUB-SAMPLE OF FIRMS Interval Mean CAR[t 1,t 2 ] Test-statistic Excluding 6 Firms that Subsequently Canceled; N=29-50 to to to First 18 Firms in Sample; 1/91-5/97; N=18-50 to to to Last 17 Firms in Sample; 6/97-12/99; N=17-50 to to to Including 5 Firms with Simultaneous Announcements; N=40-50 to to to Notes: This table reports mean CARs over the intervals specified for the full sample of 35 firms and the sub-sample, which excludes the 3 firms that made simultaneous announcements. The reported test-statistic is a test for the null hypothesis that the mean cumulative abnormal return equals zero and is asymptotically distributed Gaussian with unit variance.

17 52 JOURNALOF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 ment, and Figure 2 presents the associated firm level t-statistics over a longer interval. Although there is considerable noise in the firm level returns, there is a clear tendency for positive and significant abnormal returns during the event window. From Table 4, eleven firms earned positive abnormal returns on both day -1 and 0. Only one firm, K-Mart, earned negative abnormal returns on both days. 13 The CAR over the interval [-1,0] is positive for 30 of the 35 firms. It is positive and statistically significant at the.10 level for 11 of the 35 firms, while it is negative and significant for none of the firms. Under the null hypothesis of zero abnormal returns, half of the firms should have a CAR over [-1,0] that is negative. A non-parametric percentiles test, reported in Table 5, for this null is strongly rejected, with a test statistic of In particular, the test-statistic is [ N * Np ] a 2 = N(0,1), Np(1 p) where N is the total number of trials, p is the probability of a particular outcome under the null, and N * is the actual observed number of particular outcomes. For the full sample, N=35, N * =32 and p=.5. Figure 3 presents a plot of the CAR for each firm and, in the lower right-hand corner, the mean CAR aggregated across firms (this is the same mean CAR depicted in Figure 1). For comparison, the vertical axes are constrained to -.35 and There is again evidence of substantial idiosyncratic noise in the individual firm data, although the tendency for the CARs to increase around the announcement date is still evident. Comparing the firm CARs with the averaged CAR in the lower right-hand panel illustrates the dramatic smoothing effect of averaging the CARs across firms. Constrained to the same scale, virtually the only detectable effects are slow drift upward prior to the announcement and the spike in the two-day announcement period. The reported results are not sensitive to the specification of the estimation window, the event window, or the methodology. We tried several different estimation windows 14 and event windows with essentially no change in the statistical inference and virtually no change in the relevant point estimates. We also used, in a previous version of the paper, the methodology outlined in Schipper and Smith (1986), which ignores sampling error and estimates the variances of the disturbances over separate window, and found very similar results. The reported results are also consistent across various sub-samples of firms, as reported in Table 6. Excluding the six firms that subsequently canceled their proposed tracking stocks, the mean CAR over [-1,0] is 3.3 percent, which is slightly larger than over the full sample and suggests that the canceled tracking stocks were, on average, slightly better received by investors on the announcement day. To guard against the possibility that the results are dominated by a few firms early or late in the sample, we break down the summary statistics over the first 18 tracking stock announcements and the last 17. The first 18 announcements, which occurred prior to December 1998, had a positive CAR over [-1,0] of 3.5 percent, while the most recent 17 tracking stock announcements had a positive CAR over [-1,0] that was somewhat smaller, about 2.7 percent. Both are highly significant. Finally, including the five tracking stock announcements that are contaminated by simultaneous announcements, for a sample size of 40, also yields a positive CAR over [-1,0] of 2.7 percent. Conclusion This paper provides some empirical evidence on reaction of share prices to announcements proposing the adoption of an equity structure based on tracking stock. By comparing returns rela- 13 Incidentally, K-Mart s tracking stock proposal was subsequently rejected by shareholders. 14 We tried many estimation windows, including [-281, -161], [-281, -101], [-281, -61], [-281, -21] in addition to the reported window of [-281, -51].

18 The Reaction of Security Prices to Tracking Stock Announcements 53 tive to those predicted by a market model, we find a mean abnormal return of over 3 percent in the two-day period surrounding the announcement for a sample of 35 firms. Individually, 30 of the 35 firms earned positive abnormal returns. Statistically, these results are significant and robust to alternative estimation windows and across various subsets of firms. These estimates are comparable to previous estimates of abnormal returns associated with announcements of two relatively similar equity restructurings, carve-outs and spin-offs. The empirical finding that announcements of all three forms of equity restructurings earn positive abnormal returns is consistent with the hypothesis that investors expect gains from more focused and transparent business units possibly through improved incentive structures or the ability to more closely monitor the management of the individual business groups. Important areas for future research include providing empirical evidence on the source of the observed abnormal returns, such as whether they are in fact due to greater focus, as well as providing empirical evidence on why some firms choose a tracking stock equity structure instead of another form of restructuring, such as a carve-out or spin-off. References Bagwell, Laurie Simon, and Josef Zechner Influence Costs and Capital Structure. Journal of Finance 48: Berger, Philip, and Eli Ofek Diversification s Effect on Firm Value. Journal of Financial Economics 37: Billett, Matthew T., and David C. Mauer Diversification and the Value of Internal Capital Markets: The Case of Tracking Stock. Journal of Banking and Finance. Forthcoming. Campbell, John, Andrew Lo, and Craig MacKinlay The Econometrics of Financial Markets. Princeton, NJ: Princeton University Press. Comment, Robert, and Gregg A. Jarrell Corporate Focus and Stock Returns. Journal of Financial Economics 37: Desai, Hemang, and Prem C. Jain Firm Performance and Focus: Long-run Stock Market Performance Following Spinoffs. Journal of Financial Economics. Forthcoming. D Souza, Julia, and John Jacob Why Firms Issue Targeted Stock. Journal of Financial Economics. Forthcoming. Fama, Eugene F., and Kenneth R. French Common Risk Factors in the Returns on Stocks and Bonds. Journal of Financial Economics 33: Harris, M., and A. Raviv The Capital Budgeting Process: Incentives and Information. Journal of Finance 51: Hass, Jeffrey Directional Fiduciary Duties in a Tracking Stock Equity Structure: The Need for a Duty of Fairness. Michigan Law Review 94: Hite, Gailen L., and James E. Owers Security Price Reactions Around Corporate Spin-Off Announcements. Journal of Financial Economics 12: Hite, Gailen L., James E. Owers, and Ronald C. Rogers The Market for Interfirm Asset Sales: Partial Sell-offs and Total Liquidations. Journal of Financial Economics 18: Logue, Dennis E., James Seward, and James Walsh Rearranging Residual Claims: A Case for Targeted Stock. Financial Management 25: MacKinlay, A. Craig Event Studies in Economics and Finance. Journal of Economic Literature 35:

19 54 JOURNAL OF ECONOMICS AND FINANCE Volume 24 Number 1 Spring 2000 Milgrom, Paul Employment Contracts, Influence Activities, and Efficient Organization Design. Journal of Political Economy 96: Miles, J., and J. Rosenfeld An Empirical Analysis of the Effects of Spin-Off Announcements on Shareholder Wealth. Journal of Finance 38: Nanda, Vikram On the Good News in Equity Carve-Outs. Journal of Finance 46: Schipper, Katherine, and Abbie Smith Effects of Recontracting on Shareholder Wealth: The Case of Voluntary Spin-offs. Journal of Financial Economics 12: Schipper, Katherine, and Abbie Smith A Comparison of Equity Carve-outs and Seasoned Equity Offerings. Journal of Financial Economics 15: Schipper, Katherine, and Abbie Smith Equity Carve-Outs. In Corporate Restructuring and Executive Compensation, edited by Joel M. Stern, G. Bennett Stewart, and Donald H. Chew. Cambridge, MA: Ballinger. Seward, J. K., and J. P. Walsh The Governance and Control of Voluntary Corporate Spin- Offs. Strategic Management Journal 25: Slovin, Myon B., Marie E. Sushka, and Steven R. Ferraro AComparison of the Information Conveyed by Equity Carve-outs, Spin-offs, and Asset Sell-offs. Journal of Financial Economics 37: Spiess, D. Katherine, and John Affleck-Graves The Long-Run Performance of Stock Returns Following Debt Offerings. Journal of Financial Economics 54: 47:73. Vijh, Anand Long-term Returns from Equity Carveouts. Journal of Financial Economics 51: Zuta, Shlomith Diversification Discount and Targeted Stock: Theory and Empirical Evidence. Working paper, University of Maryland. Appendix Below we provide a sketch of the event study methodology as it applies to this study, using the notation of Campbell, Lo, and MacKinlay (1997). Additional details can be found in MacKinlay (1997) and Campbell et al. The market model can be represented as R i,t = i + i R m,t + i,t with i,t N (0, i2 ) for i = 1,, N (1) where R i,t is the return on security i at time t, and R m,t is the return on a proxy for the market at time t. Since the true parameters i, i are unknown, they are estimated using a pre-event sample period. We can rewrite (1) in matrix notation as R i =X i i + i (2) where X i = [ i R m ] is an (L 1 x2) vector of ones and the market return over the estimation window; i=[ i i ] denotes the (2x1) parameter vector; and L 1 is the number of days in the estimation window. Applying ordinary least squares (OLS) yields consistent estimates of the conditional mean parameters i, even if the true errors are heteroskedastic. Abnormal returns during the event window, conditional on the contemporaneous market return and the OLS estimates i from the estimation period, are therefore