Disinvestment from South Africa: they did well by doing good.
As U.S. businesses and investors weigh a return to investment and operations in South Africa, one might consider the effect the withdrawal or disinvestment from South Africa of the 1980s had on the rates of return of the disinvesting firms. Was there a measurable effect, either positive or negative, on the returns of firms that chose to leave South Africa? The effect of divestment on portfolio performance, specifically its effect on risk and return, has received much attention. Grossman and Sharpe (1986) provide an excellent analysis of this topic while Hauck (1986), Jenkins (1986), Kaempfer et al. (1987a), and Kaempfer and Moffett (1988) examine the potential and actual effects on South Africa of American firms disinvesting from that nation. Feigenbaum and Lowenberg (1988) consider the effect of divestment on the likelihood of disinvestment. However, little attention has been paid to the effect of disinvestment on the firms that actually withdrew (see Meznar et al., 1994, on this topic). This question has serious implications as to the fiduciary responsibilities of management, who due to the political and social pressures of proposed divestment withdrew from or sold off investment in their South African subsidiaries. Since the goal of firm management is to maximize shareholder value, if a firm voluntarily suffers a decrease in its rate of return due to disinvestment, management has violated its obligation to shareholders.
This paper examines the effect of disinvestment from South Africa on American firms in order to determine if any perceivable effect on the firms' rates of return occurred at the time of the announcement of disinvestment. Perhaps this analysis can tell us the effect on IBM, for example, of leaving an extremely profitable business in South Africa just six months after its president proclaimed that "(w)e are not in business to conduct moral activity, we are not in business to conduct socially responsible action. We are in business to conduct business" (Akers, 1986, p. 117). Did IBM withdraw strictly for business reasons, or did the social pressures of the divestment/disinvestment activists play a role? How exactly did the market judge such a move? Analysis of the empirical data may lead to answers as to whether the market connected the divestment/disinvestment pressures of social activists with legitimate economic reasons for a firm to withdraw from South Africa, or whether it perceived withdrawal as simple public relations or excess perquisite consumption by management. This research provides some answers to the above questions.
II. EMPIRICAL ANALYSIS OF THE EFFECTS OF DISINVESTMENT
South Africa has had problems and varying degrees of social unrest since World War II, but foreign investors found it most risky and uncertain in the late 1970s, particularly after the Soweto uprisings beginning in 1976. Student and social activist protests against American investment in South Africa began in the early 1980s. Therefore, the analysis here reviews the effect of the announcement of disinvestment on the rates of return of the disinvesting firms from 1980 to 1991, as the majority of U.S. firms left South Africa in the mid-1980s. Analysis of just this "subset" of disinvestments is justified since, according to figures compiled by the Investor Responsibility Research Center of Washington, D.C., "of the approximately 350-400 U.S. companies with direct investment in South Africa in January 1984, . . . by mid-1988 only 138 U.S. companies remain(ed) in South Africa" (Hufbauer et al., 1990, p. 243).
This research focuses on an event study using daily returns in order to analyze what effect a particular event (such as the announcement of some relevant news) might have on the returns of a particular firm. Data are from the Center for Research into Security Prices (CRSP) tapes from the University of Chicago that provide monthly and/or daily information on stocks traded on the New York and American Stock Exchanges and on NASDAQ. Using the daily CRSP tapes, the analysis reviews the announcement effect for all available disinvesting firms from 1980 to 1991. The initial performance benchmark is the firms' average rates of return over the 250-day period (approximately one year) immediately preceding the announcement of disinvestment. A t-test determines if the announcement period's return differed significantly from the average return of the past year. The analysis repeats the same tests using an ordinary least squares market regression approach. It also performs a comparison against the market-proxy benchmarks of the CRSP tape composite portfolios in order to circumvent any statistical problems that might have arisen from event-date clustering. Obviously, though, if a firm's riskiness changes after the announcement of disinvestment, its expected rate of return also will change. Thus, even though the event study looks at the portfolio of firms that disinvested rather than at individual securities, all firms would experience this change in risk so that it would not be diversified away in the portfolio composition. Therefore, a simple comparison of returns would be misleading since as risk changes, returns also must change. In order to correct for this problem, one can "standardize" or "normalize" the returns of each time period of analysis by dividing them by their respective standard deviations (Masulis, 1980, p. 155; Brown and Warner, 1985, p. 21). The event studies of interest thus are duplicated using standardized variables.
The test statistic used to determine the statistical significance of the abnormal returns arising from the announcement of disinvestment is a t-test. Unlike the formulation of the t-test advocated by Brown and Warner of the ratio of abnormal return to the portfolio's standard deviation found over the estimation or comparison period (1985, p. 7), the t-statistic in the analysis here divides the portfolio's abnormal return (as determined through the procedures previously discussed) by the portfolio's standard error over the event period, not the estimation period. According to Boehmer et al. (1991), use of this test statistic, which they refer to as the "ordinary cross-sectional" method, does not explicitly assume that the event does not affect the variance of the portfolio's returns as does use of the traditional t-test. If one assumes no change in a portfolio's variance after a particular event (in this case, announcement of disinvestment), one might accept the existence of abnormal returns when none in fact actually exist simply because a change in variance well might accompany such an event. A change in variance and hence return may correspond to an announcement of disinvestment from South Africa as the market may have perceived a firm's withdrawal from the increasingly violent and recessionary South Africa as a decrease in the variability of the firm's overall operations. An additional benefit to using this particular test statistic is evidence that the ordinary cross-sectional test statistic seems not to be affected by the existence of event-date clustering even though, as usual, it nominally requires that abnormal returns be uncorrelated across firms (Boehmer et al., 1991, p. 259 and 268). In a paper addressing abnormal performance in cases of uncertainty as to the actual or specific event-date, Ball and Torous (1988, p. 148) also support the use of this test statistic, in particular use of the event-period standard deviation rather than the estimation period standard deviation. Thus, the analysis here reformulates the conventional t-test to the "ordinary cross-sectional test."
III. DATA AND RESULTS
Due to the extensive public attention given to the divestment/disinvestment debate in the 1980s, several sources provide information on which companies left South Africa and when. For example, divestment movement publications such as the Unified List (1990) prepared by Richard Knight of the Africa Fund and those published by the Investor Responsibility Research Center (Cooper, 1989) list several hundred companies that pulled out of South Africa in the 1980s. However, careful analysis of precise announcement dates available from public sources such as The Wall Street Journal and The New York Times result in a sample of 52 companies from 1977 to 1992 that had made distinct public and published announcements of disinvestment. Out of these 52 disinvesting companies, 40 had complete returns data as identified through the use of CUSIP numbers on the available version of the CRSP tapes for the relevant period of 1980 until the end of 1991. NASDAQ-listed firms are not available on the particular version of the CRSP tapes used in this analysis, thus reducing the available set of disinvesting companies. Additional situations such as Revlon's also complicate the analysis. Revlon publicly announced disinvestment on December 5, 1986 but was taken private (off the exchange and hence off the data tapes) as of December 26, 1986. Thus, Revlon was deleted from the disinvestment sample. Some firms were lost as they turned out to be subsidiaries of other firms. For example, Kentucky Fried Chicken (KFC) announced disinvestment plans on February 18, 1987, but the company is a subsidiary of PepsiCo, which did not announce disinvestment plans until April 20, 1990. (Appendix A lists the relevant firms and their respective announcement dates.)
The analysis indicates the announcement date for each firm - that is, the date a newspaper published the announced plans for disinvestment - by a dummy variable and adjusts time for all firms to "event-time" rather than calendar time so that day 0 (when the dummy variable is set equal to 1) is the common event date (announcement date) for each firm. This procedure "diversifies" the effects of any particular time period. Additionally, the analysis combines the firms into one large portfolio of the 40 firms rather than viewing each in isolation. This procedure also helps to diversify so as to abstract from any idiosyncratic influences arising from the individual firms.
The event study results represent three basic categories based upon the method used to calculate abnormal returns: the comparison period returns approach, the market-adjusted returns approach, and the ordinary least squares approach (see Appendix B for the applicable equations). The analysis applies each method to the disinvesting portfolio over varying event periods in order to determine the magnitude and significance, if any, of the excess or abnormal returns arising from the announcement of disinvestment from South Africa.
Table 1 details the abnormal mean returns (and volume) determined using the comparison period returns approach for various event-windows. The corresponding t-statistics are indicated in parentheses below. Averaging returns for the period approximately one-year before the announcement date forms the comparison period return for the disinvesting portfolio. Except for the one-day event period consisting of the announcement day itself, trading volume is always significantly higher than the average trading volume of the preceding 248-day comparison period. For abnormal returns, the event-window of the announcement day and the days both immediately preceding and following the announcement (days -1,0,1) exhibits a positive abnormal mean return of 0.002814, significant at the 5% level. (The t-statistic of 4.0520153 seems relatively large in relation to the rule of thumb that a t-statistic of approximately 2 indicates statistical significance. However, this result is due to the use of a t-statistic calculated over the event period and the corresponding adjustment in degrees of freedom for the smaller sample size. See the previous discussion on the benefits of using the ordinary cross-sectional test statistic rather than the conventional t-statistic for this type of analysis.) The evidence indicates returns significantly greater than the mean return of the portfolio over the comparison period when announcement of disinvestment from South Africa occurs. Trading volume also is significantly higher than over the comparison period. Thus, one can conclude that announcement of disinvestment resulted in positive abnormal returns to the disinvesting firms over the three-day period immediately surrounding the announcement. The significance indicates that some pre-announcement activity takes place and that it continues at least one day after announcement. The results of the event-window consisting of the announcement day and preceding day (days -1 and 0), while not statistically significant at the 10% level, further corroborate the overall result of this study with positive abnormal mean returns of 0.00339850, significant at the 12% level. Again, this overall result seems to indicate that investors are reacting to the announcement and are rewarding the firms for their disinvestment plans.
Interestingly, the results of the market-adjusted returns approach repeats the pattern of a significant announcement effect over the event-periods of days (-1,0,1) and (-1,0) using the comparison period returns approach (table 1). Statistically significant positive abnormal returns (measured as the difference between the portfolio's mean return and the return on a market proxy) exist only for those same two event-windows. For abnormal returns as measured as the difference between the portfolio's mean return and that of the CRSP equally-weighted "market" composite portfolio, positive abnormal mean returns exist that are significantly greater than zero at the 5% level for both of the indicated event-windows. The same holds true at a 10% level of significance for abnormal returns calculated relative to the CRSP value-weighted portfolio. In determining abnormal performance using the S&P 500 as the market proxy, abnormal returns also are significantly positive at the 10% level for the three-day event period of days (-1,0,1), and significant at the 12% level for days (-1,0). Thus, over these two particular event-windows, the disinvesting portfolio outperforms all three market proxies at significance levels of 5% and 10%. Again, the market apparently immediately rewards those firms that announce plans to withdraw from operations in South Africa. Table 1 does not show other event periods with returns significantly higher than expected based on comparison with the market. For the CRSP value-weighted and the S&P 500 proxies, for example, over the event period of the two days leading up to the announcement and the announcement day itself (days -2, -1,0), [TABULAR DATA FOR TABLE 1 OMITTED] the portfolio's mean returns are significantly greater than those of the market proxies at a 5% significance level. (This additional information is available on request from the author.)
Much of the pattern found in the results of the market-adjusted models also appears in the results arising from using the OLS market-predicted returns as a performance benchmark. As before, the event-windows of days (-1,0,1) and (-1,0) each display significantly greater mean returns than an ordinary least squares regression of the returns of the disinvesting portfolio run against the returns of the three market proxies would predict. Except for the regression residuals of the S&P 500 model, which are significant at an 11% level, the abnormal returns of the disinvesting portfolio over these two particular event-periods are significantly higher at the 10% level than the analysis would predict using the regression equations. The residuals also are significant for the CRSP value-weighted and S&P 500 equations over the event-window of days (-2, -1,0). The comparison period returns, the contemporaneous market returns, and a market regression model's prediction reveal that the market apparently rewards the disinvesting firms through higher returns, especially for the two particular event-windows of days (-1,0,1) and (-1,0).
IV. STANDARDIZATION OF VARIABLES
Because one might question the behavior of the disinvesting portfolio's variance subsequent to the announcement of disinvestment, an additional test standardizes the returns by dividing them by their standard deviations over the comparison or estimation period and over the announcement period. Again, this procedure corrects or controls for the effects on return of a change in variance arising from the announcement of disinvestment. Table 2 demonstrates that, with the normalization or standardization of the variables, they are now all statistically greater than their comparison period means. However, the most interesting result is for the return variable where, as in the "regular" comparison period returns approach as shown in table 1, the pattern of a statistically significant abnormal mean return (i.e., significantly greater than the return experienced over the comparison period) is still evident for the event-windows of days (-1,0,1) and (-1,0).
TABLE 2 Normalized Abnormal Returns of the Disinvesting Portfolio Over Relevant Event Dates Event Period: days (-1,0,1) (-1,0) (0,1) Comparison Period Returns Method (abnormal return = normalized return-comparison period normalized mean return) Abnormal Return 1.75197 1.99333 1.73136 (5.207)(**) (6.468)(*) (3.036) Volume 13.5802 13.4573 13.3982 (44.27)(***) (27.64)(*) (31.32)(**) Market Return Method (abnormal return = normalized return-market proxy return) CRSP Equally-Weighted 0.61238 0.96139 0.32039 (1.635) (4.091) (0.789) CRSP Value-Weighted 1.90172 1.46932 0.76017 (2.649) (5.146) (1.794) S&P 500 1.15196 1.51197 0.81257 (2.849) (4.743) (2.134) OLS Method (abnormal return = normalized return-OLS predicted return using market proxy) CRSP Equally-Weighted 1.61845 1.90863 1.57883 (5.142)(**) (9.047)(*) (2.919) CRSP Value-Weighted 1.67941 1.97594 1.62845 (5.296)(**) (10.15)(*) (3.003) S&P 500 1.67278 1.96564 1.62163 (5.346)(**) (10.31)(*) (3.033) Notes: T-statistics are in parentheses. Statistical significance as follows: * significant at 10%; ** significant at 5%; *** at 1%. To be "normalized," the variables are divided by their respective standard deviations in order to abstract from the effects, if any, of a change in variance due to the announcement of disinvestment.
Table 2 also displays the results for the normalized variables using the market-adjusted returns approach where the abnormal mean portfolio return is found by subtracting one of the normalized market proxies from the normalized mean portfolio return. Unlike the normalized comparison period returns approach, which reinforces the results of table 1, the market-adjusted standardized returns do not support the results of table 1. Evidence for statistically significant abnormal returns over the event-periods of days (-1,0,1) and (-1,0) basically no longer exists when the variables are normalized. However, for the S&P 500 market proxy, the normalized abnormal return is significant at the 10% level and that using the CRSP value-weighted index is significant at 11% over days (-1,0,1). For days (-1,0), using the CRSP value-weighted proxy gives abnormal returns significant at 12%.
However, the OLS approach reveals a pattern of statistically significant abnormal returns for the event-windows of days (-1,0,1) and (-1,0). For days (-1,0,1), using OLS regressions of the disinvesting portfolio's total returns against the returns on the three market proxies, again all normalized to account for any change in variance, finds the abnormal mean returns all significantly different from zero at the 5% level. For the CRSP equally-weighted market index over days (-1,0), abnormal mean returns are significantly greater than zero also at the 5% level, with the abnormal returns corresponding to the CRSP value-weighted and the S&P 500 indexes both significant at the 10% level.
Ignoring the market-adjusted returns approach, one can conclude that the results in table 2 for the return variables, normalized by their standard deviations, back-up and reinforce the results of table 1 - that a significantly positive announcement effect appears to exist. This is particularly true for abnormal performance as measured as the difference between the event-period's mean portfolio return and mean return over the comparison or estimation period. In other words, the performance of the disinvesting portfolio over various event-windows does significantly differ from its past performance. Table 1 also shows that the disinvesting portfolio significantly outperforms the returns on the market proxies. However, the results of table 2 weaken that conclusion. For the market returns approach, once the return variables are standardized by their respective standard deviations, the abnormal performance of the disinvesting portfolio is weaker and barely exists for the interesting event-periods of days (-1,0,1) and (-1,0). One may argue that since the test statistic used is the ordinary cross-sectional test statistic that uses the standard error as calculated over the event window and not over the estimation period as with the traditional t-test, the effect of any change in variance already has been accounted for so that one may discount to some extent the results of the market-adjusted return approach of table 2. Thus, one no longer can say with as much conviction that the disinvesting portfolio has significantly positive abnormal mean returns, at least over the event-periods of days (-1,0,1) and (-1,0). However, its returns do significantly differ from past returns, and some evidence exists that the portfolio does better than the market.
V. POSSIBLE EXPLANATIONS FOR THE RESULTS
General economic theory predicts that disinvesting firms lose or give up valuable opportunities and subsequently should perform worse. Thus, announcement of disinvestment should lead to a decrease in return, given profitable operations in South Africa. This corresponds to predictions of empirical studies such as the multinational effect described by Doukas and Travlos (1988) regarding international diversification. They find that as firms increase their international exposure, returns increase. Disinvestment or withdrawal from South Africa is the opposite of such diversification, so returns should decrease at announcement. However, Kaempfer et al. (1987b) find no statistically significant difference in share price over the period 1984 to 1986 for firms invested in South Africa versus firms that were South Africa "free." This conclusion in some sense counters the prediction of the multinational effect. Agency theory also predicts a negative effect at the announcement of disinvestment if management is consuming too much social responsibility - that is, more than what is optimal for the firm. However, these explanations do nothing to explain the evidence presented here indicating that the disinvesting portfolio outperforms the market and its own past at the announcement of disinvestment.
On the other hand, these explanations would support the results in Meznar et al. (1994), which also looks at withdrawal from South Africa but finds a negative announcement effect. These authors use a different variable of interest (cumulative abnormal return), and longer announcement or event-periods (up to 40 days) and calculate abnormal return as the difference between return and what would be predicted using a regression of own past returns. Event studies typically use short, one- or two-day event periods for analysis, assuming that relevant information is quickly accounted for in stock market returns. This, coupled with an analysis period focussing on 1985 to 1991 (a period that includes the severe market downturns of both 1987 and 1991), would indicate that influences other than the announcement of disinvestment might account for the negative results.
Specifically, with own past return as the performance benchmark, those two severe market downturns may have influenced the results such that the announcement period's returns had to be less than those of the past. This probably is most clearly indicated by the fact that their abnormal return measure continues to decrease even 10 days after the announcement, rather than moving back to zero, as one would expect in an efficient market with information rapidly accounted for in returns.
However, several different measures of abnormal return that correspond more closely to the event study literature (e.g., the comparison period and market-adjusted returns approaches, as well as OLS) varify the present study's results. And examining their figures 1 and 2 shows up-ticks around day zero even when using their cumulative abnormal return measure. Thus, the present study's findings do indicate that a positive announcement effect is still valid (and valuable).
A positive and significant increase in returns appears to exist for the two-to-three days surrounding the published announcement of plans to withdraw from South Africa. Such a positive announcement effect would correspond with or match the predictions of the voluntary sell-off literature where announcement of the sale of a corporate subsidiary leads to an increase in returns (Alexander et al., 1984; Jain, 1985; Klein, 1986). However, for the positive announcement effect to arise from a voluntary sell-off, the transaction must be of positive net present value to the selling firm. This generally was not the case for the South African disinvestments, especially after 1984 and 1985. Specifically, numerous examples exist of disinvesting firms selling their South African assets not at market value but at "fire-sale" prices. For example, Mobil withdrew in 1989 by selling its reportedly $400 million of assets for approximately $150 million, and Bell & Howell apparently "was so anxious to get out that in April (of 1986) it sold the entire subsidiary - with an estimated book value of $5 million - to a South African conglomerate for $1" (Hufbauer et al., 1990, pg. 230 and 243). However, the removal of purchasing restrictions such as those imposed by municipal authorities on firms doing business in South Africa may have increased domestic sales of the disinvesting firms by more than any loss taken from disinvestment so that returns after the announcement of disinvestment are significantly higher than before.
The positive announcement effect found in this analysis gives some credence to various claims of market inefficiencies, here in the form of market segmentation arising from divestment policies or investment restrictions for some institutional investors. The announcement of disinvestment for the most part causes those prohibitions to disappear, and one can conjecture that the actual or potential renewal of investment in these firms is enough to give their returns a boost. However, it seems rather difficult to believe that such a market segmentation could consistently survive since investors not so limited would arbitrage such an anomaly away.
Yet, if market segmentation actually exists such that the sale or abandonment of investments and operations in South Africa leads to increased, positive trading activity in the shares of the disinvesting firm, then a rationale for the existence of a positive effect at the announcement of disinvestment is that announcement is seen as a "trigger" leading to increased future investment in the shares of these firms, particularly by large institutional investors such as public pension funds that had been restricted by divestment rules. Alternatively, investors, as well as managers, may wish to "consume" social responsibility and therefore are more favorably inclined towards firms that announce plans to leave South Africa, countering the agency theory argument against positive returns. This also may explain why divestment policies were implemented with relative ease for most public pension plans if plan managers were really representing participants' socially conscious wishes.
Probably the most popular explanation for a positive effect at the announcement of disinvestment, particularly by anti-apartheid activists, would be the beneficial effect on a firm arising from the removal of what has been called the "hassle" factor. Once the firm publicly announces that it no longer plans on doing business in South Africa, it is dropped from protestors' "hit-lists" of public denunciations and proposed boycotts. In terms of actual economic benefits, however, one must wonder at the market's rewarding these firms for what appears noneconomic and/or nonproductive reasons. Remember, however, that one of the most forceful methods of "hassling" firms involved imposing the municipal purchasing restrictions that had direct effects on current and future sales and income.
In the case of South African disinvestment by U.S. firms, the rather serendipitous combination (for firm management) of bad public relations and actually bad economic prospects for continued investment in South Africa came together at the same time. In other words, firm management realized that future general economic activity in South Africa did not look promising as increasing civil strife, continuing general economic malaise exacerbated by a falling world price of gold over this period, and an unprecedented subcontinental drought all combined to drastically lower estimates of economic growth and, at the same time, increase the risks of continuing to do business in South Africa. The profitability and stock market increases of the mid- to late-1980s also may have prompted withdrawal from South Africa at this particular time. Feigenbaum and Lowenberg (1987, p. 115) argue that the firms most likely to disinvest are those that can most easily afford to do so. Thus, inside the firm, coincident with disinvestment pressures by activists outside the firm, management was coming to the conclusion that investment in South Africa did not make sense for economic as well as social reasons. This apparently corresponded with the view of the market in general which, according to the results presented here, positively rewarded those firms that announced plans for disinvestment.
U.S. firms that made a public announcement of plans to withdraw from South Africa experienced a positive and significant increase in returns in the period immediately surrounding the announcement of disinvestment. Trading volume for the disinvesting portfolio also was significantly higher than historical trends would predict. One can explain these effects through the predictions of general economic theory and the results of various empirical studies on related or corresponding issues. Thus, one can explain the positive announcement effect as a "public relations" answer with economic backing due to the resolution of uncertainty concerning continued involvement or investment in the deteriorating South African economy. Thus in the 1980s, social and economic forces arguing for withdrawal from South Africa found a receptive audience in corporate management and the market.
The implications of these findings are interesting, especially if one ties together the announcements of disinvestment with the timing of various government and private investment policies. A glance over the dates of announcement in appendix A shows that the majority came in 1986 and 1987 for this data set (14 and 15, respectively, out of 40 firms in total). The autumn of 1986 is particularly important. IBM, perhaps the most significant firm to publicly announce disinvestment plans, did so in October. Eight firms followed in the two remaining months of 1986, some within days of IBM's announcement. Coke's announcement came just a month earlier in September. Policies that may have prompted these announcements include California's public funds divestment announcement of September 1986 and the U.S. government's imposition of sanctions in October of 1986 restricting any further investment in South Africa. Thus, it appears that policies implemented to influence firm investment, both directly (such as the government prohibition) and indirectly (the divestment pressures), played a critical role in prompting firms to withdraw from or disinvest from South Africa.
U.S. Firms with Public Announcements of Disinvestment from South Africa
A. Firms included in the Disinvesting Portfolio and Date of Announcement:
American Brands, Inc. (5/1/87) American Home Products Corp. (9/4/89) AT&T (4/7/86) Bell & Howell Co. (2/7/86) Bundy Corp. (12/9/86) CPC International Inc. (4/3/87) Chrysler Corp. (1/27/83) Citicorp (6/17/87) Coca Cola Co. (9/18/86) Dow Chemical Co. (2/9/87) Eastman Kodak (11/20/86) Emery Air Freight Corp. (7/2/87) Emhart Corp. (1/28/87) Exxon (12/31/86) Federal-Mogul Corp. (9/16/88) Firestone Tire & Rubber (5/19/87) Fluor Corp. (12/8/86) Ford Motor Corp. (6/15/87) Foster Wheeler Corp. (2/23/87) General Electric Co. (4/21/86) General Motors Corp. (10/21/86) Goodyear Tire & Rubber Co. (6/8/89) Hewlett-Packard Co. (3/22/89) Honeywell, Inc. (10/23/86) ITT Corp. (5/12/87) IBM (10/22/86) Johnson Controls Inc. (11/24/86) McGraw Hill Inc. (2/27/87) Merck & Co. (12/1/87) Mobil Corp. (4/27/89) NCR Corp. (5/5/89) Newmont Mining Co. (4/4/88) Norton Co. (3/4/87) Raychem Corp. (8/29/89) Sara Lee Corp. (10/31/86) Tambrands Inc. (1/28/87) Unisys Corp. (8/20/88) Upjohn (6/7/90) Warner Communications (10/23/86) Xerox Corp. (3/20/87)
B. Firms Deleted from Portfolio due to Subsidiary Status:
KFC (Kentucky Fried Chicken) (a subsidiary of PepsiCo)
Sterling Drug Inc. (a subsidiary of Kodak)
C. Firms for Which No Identifying CUSIP Number Could be Found:
Allegis (Hertz) International Playtex Westin Hotels & Resorts
D. Firms Not Publically Traded During Relevant Period:
Revlon Group Inc. RJR Nabisco (KKR)
E. Firms Which Did Not Appear on the CRSP Tapes (Unknown Reasons):
Burroughs Phibro-Salomon Inc. Diamond Shamrock Corp. USX Corp. PepsiCo
Abnormal Return Measures
Comparison Period or Mean-Adjusted Returns Approach:
[Mathematical Expression Omitted]
[Mathematical Expression Omitted]
Market-Adjusted Returns Approach:
[A.sub.p,t] = [R.sub.p,t] - [R.sub.m,t]
(where [R.sub.m,t] = return on the various market proxies for day t)
Ordinary Least Squares Market Predicted Returns Approach:
[A.sub.p,t] = [R.sub.p,t] - ([a.sub.i] + [b.sub.i]*[R.sub.m,t])
(where [a.sub.i] and [b.sub.i] are the OLS predicted coefficients estimated from the preceeding 248 day comparison period)
Note: The specification of these equations is from Brown and Warner (1985), pages 6 and 7.
This is a revised version of a paper presented at the Western Economic Association International 69th Annual Conference, San Diego, Calif., July 8, 1995 in a session entitled "International Securities Prices and Investment." The author appreciates the valuable comments provided by the session discussant and audience, as well as by two anonymous reviewers. The paper discusses a portion of the results in the author's dissertation research at U.C. Riverside, and thus additional thanks are due to R. Robert Russell, Robert A. Haugen, Herb Johnson, Keith Knapp, Brian Archibald, James Berens and Patricia Watters.
Akers, John, Business and Society Review, Spring 1986, 117.
Alexander, Gordon J., P. George Benson, and Joan M. Kampmeyer, "Investigating the Valuation Effects of Announcements of Voluntary Corporate Sell-offs," Journal of Finance, June 1984, 503-517.
Ball, Clifford A., and Walter N. Torous, "Investigating Security-Price Performance in the Presence of Event-Date Uncertainty," Journal of Financial Economics, 22:1, 1988, 123-154.
Boehmer, Ekkehart, Jim Musumeci, and Annette B. Poulsen, "Event-Study Methodology Under Conditions of Event-Induced Variance," Journal of Financial Economics, 30:2, 1991, 253-272.
Brown, Stephen J., and Jerold B. Warner, "Using Daily Stock Returns: The Case of Event Studies," Journal of Financial Economics, 14:1, 1985, 3-32.
Cooper, Alison, U.S. and Canadian Business in South Africa 1989: A Directory of U.S. and Canadian Corporations with Business Links to South Africa and Namibia, Investor Responsibility Research Center, Washington D.C., 1989.
Doukas, John, and Nickolaos G. Travlos, "The Effect of Corporate Multinationalism on Shareholders' Wealth: Evidence from International Acquisitions," Journal of Finance, December 1988, 1161-1175.
Feigenbaum, Bernard, and Anton D. Lowenberg, "South Africa Disinvestment: Causes and Effects," Contemporary Policy Issues, October 1988, 105-117.
Grossman, Blake R., and William F. Sharpe, "Financial Implications of South African Divestment," Financial Analysts Journal, July/August 1986, 15-29.
Hauck, David, U.S. Corporate Withdrawal from South Africa: The Likely Impact on Political Change, Investor Responsibility Research Center, Inc., South Africa Review Service, Washington D.C., August 1986.
Hufbauer, Gary Clyde, Jeffrey J. Schott, and Kimberly Ann Elliott, Economic Sanctions Reconsidered: History and Current Policy, second edition, Institute for International Economics, Washington, D.C., 1990.
Jain, Prem C., "The Effect of Voluntary Sell-Off Announcements on Shareholder Wealth," Journal of Finance, March 1985, 209-224.
Jenkins, Carolyn M., The Economic Implications of Disinvestment for South Africa, Special Study, The South African Institute of International Affairs, University of Natal, Durban, South Africa, 1986.
Kaempfer, William H., and Michael H. Moffett, "The Impact of Anti-Apartheid Sanctions on South Africa: Some Trade and Financial Evidence," Contemporary Policy Issues, October 1988, 118-129.
Kaempfer, William H., James A. Lehman, and Anton D. Lowenberg, "The Economics of the Call for Anti-Apartheid Investment Sanctions," Social Science Quarterly, September 1987a, 528-538.
Kaempfer, William H., James A. Lehman, and Anton D. Lowenberg, "Divestment, Investment Sanctions, and Disinvestment: An Evaluation of Anti-Apartheid Policy Instruments," International Organization, Summer 1987b, 457-473.
Klein, April, "The Tuning and Substance of Divestiture Announcements: Individual, Simultaneous and Cumulative Effects," Journal of Finance, July 1986, 685-697.
Knight, Richard, Unified List of United States Companies Doing Business in South Africa, third edition, The Africa Fund, New York, 1990.
Masulis, Ronald W., "The Effects of Capital Structure Change on Security Prices: A Study of Exchange Offers," Journal of Financial Economics, 8:2, 1980, 139-178.
Meznar, Martin B., Douglas Nigh, and Chuck C. Y. Kwok, "Effect of Announcements of Withdrawal from South Africa on Stockholder Wealth," Academy of Management Journal, 37:6, 1994, 1633-1648.
The New York Times, various issues.
The Wall Street Journal, various issues.
|Printer friendly Cite/link Email Feedback|
|Author:||Posnikoff, Judith F.|
|Publication:||Contemporary Economic Policy|
|Date:||Jan 1, 1997|
|Previous Article:||Public and private investment and economic growth in Mexico.|
|Next Article:||Scale economies and police department consolidation evidence from Los Angeles.|