Price and volume effects associated with changes in the Dow Jones averages.
This study is organized as follows. The next section summarizes the results of previous studies of changes in the roster of the Standard and Poor's Index of 500 Common Stocks (S&P 500) and presents Merton's attention hypothesis as an alternative to the price pressure hypothesis. The following section describes the sample and methodology used to investigate changes in the roster of the Dow Jones Averages. The empirical results of these tests are presented in the next section. A summary of the results with conclusions drawn from the analysis follow.
Previous studies examining changes in the composition of stock indices have focused on price and volume responses associated with changes in the S&P 500, see for example, Shleifer (1986), Harris and Gurel (1986), Goetzmann and Garry (1986), Lamoureux and Wansley (1987), and Jain (1987). Three findings are consistent across these studies: (1) event period abnormal returns for firms added to the S&P 500 are strongest after 1976, coinciding with significant growth in index trading of the S&P 500, and an introduction of the Stock Price Notification Service; (Shleifer 1986; Harris and Gurel 1987; Lamoureux and Wansley 1987); (2) a 3% abnormal event day return is associated with addition to the S&P 500; (Shleifer 1986; Harris and Gurel 1987; Lamoureux and Wansley 1987; Jain 1987); (3) event period trading volume is greater than base period trading volume; (Shleifer 1986; Harris and Gurel 1987; Lamoureux and Wansley 1987).
The price pressure hypothesis has been offered as an explanation for the observed price and volume effects (Shleifer 1986; Harris and Gurel 1987; Lamoureux and Wansley 1987). The price pressure hypothesis implies that traders who provide the liquidity required to eliminate excess demand will require compensation in the form of a price reversal. The price pressure hypothesis does not imply a definite post event period pattern for trading volume.
The price pressure hypothesis has not been uniformly supported in this literature. Consistent with the price pressure hypothesis, Harris and Gurel (1986), and Lamoureux and Wansley (1987), using observations from 1978-1983 and 1976-1985, found statistically significant post event period price reversals, and abnormal event period trading volume. Firms deleted from the index experienced negative event period abnormal returns, and abnormal event period trading volume. Some have argued, (Harris and Gurel 1987; Lamoureux and Wansley 1987), that the portfolio re-balancing activities of index fund managers generates excess demand. Pruitt and Wei (1989) found that net changes in institutional ownership in the quarters surrounding addition to the index are significantly positively related to event day abnormal returns.
Studies conducted by Shleifer (1986), Goetzmann and Garry (1986), and Jain (1987) over the same sample periods as Harris and Gurel (1986), and Lamoureux and Wansley (1987), do not find statistically significant price reversals in the post event period. Event period abnormal returns of 3% for firms added to the Standard and Poor's Supplementary Indexes are reported in Jain (1987). These indices are unlikely to be tracked by index traders.
Event period price and volume effects could be the result of a change in an index roster providing the market information about the future cash flows of the affected firms. However, a firm is chosen for inclusion in an Average based on widely available public information, and the purpose of the Dow Averages is to mirror the general condition of the economy, DJIA, and the transportation industry, DJTA. The personnel at Dow Jones & Company are not professional investment analysts. Thus, the market is unlikely to view the inclusion in an Avenge as a forecast of increased future cash flows.(1)
The Dow Averages and Standard and Poor Indices differ in two major ways. First an existing service notifies subscribers of impending changes in the S&P indices.(2) Changes in the roster of the Dow Averages appear in the Wall Street Journal (WSJ). Second no authorized index fund or derivative security is based on either the DJIA or the DJTA. Thus a change in the composition of either Dow Average does not force professional fund managers to reallocate their portfolios immediately.
However, "unauthorized" funds based on the Dow averages may exist. Any such funds and the "closet indexers" managing them cannot be unambiguously identified. If managed in the same fashion as authorized funds, then even though we can't identify the funds or the managers we should observe the effects of their trading in the market. The price pressure hypothesis does not distinguish between authorized or "unauthorized" transactions.
Merton (1987) has developed a model of capital market equilibrium in which an investor's optimal portfolio is incompletely diversified due to information costs. A comparative static examination of the equilibrium pricing relationship reveals that an increase in the relative size of a firm's investor base will reduce the firm's cost of capital and increase the market value of the firm. Merton writes:
. . . It is of course possible for the investor base to increase without the firm spending anything. For example, a newspaper or other mass media story about the firm or its industry that reaches a large number of investors who are not currently shareholders, could induce some of this number to incur the set-up costs and follow the firm. . . .
It should be stressed that the current shareholders may already know all the information contained in such stories. Nevertheless, if the form of the prior public releases of the information did not capture widespread attention among investors who do not follow the stock and if the new form does, then the firm's investor base will increase and the stock price will rise. Thus, our model is consistent with the observation that stock price sometimes reacts to a broad and widely circulated report about the firm, even when all the substantive information in the report has been previously announced.
In contrast to the price pressure hypothesis Merton's attention hypothesis predicts a permanent change in market value.
Examination of the population of roster changes for the DJIA and the DJTA provides an opportunity to test this attention hypothesis. By examining all the changes in the Dow Averages from 1962 to 1991 we have a set of news events which provide the market with no firm specific information about future cash flows, but do draw attention to the affected firms. Additionally, media coverage of changes in the two Averages differs substantially. Changes in the roster of the DJIA are reported on the front page of the WSJ and prominently in other media while changes in the roster of the DJTA are reported in the footnotes to a table in the market data section of the WSJ.
The firms which comprise these indices are large and widely held by investors. Merton's model implies that marginal changes in press coverage draw investors' attention to the firm resulting in an increase in the value of the firm. The Merton hypothesis implies that while professional investors may be cognizant of information which affects the value of the firms in both Dow Averages and the likely candidate replacement firms, investors who do not continuously monitor these firms are the ones who will be attracted to a firm at the margin as a result of the broad media coverage associated with an index change. Firms removed from an index should not experience a significant price change. Removal from an index will not affect the number of investors who have already incurred the information costs associated with including a firm in their optimal portfolio. As these announcements do not release any valuable firm-specific information we would not expect the value of firms dropped from an Average to decline.
Merton's model makes no predictions concerning the trading volume of affected firms surrounding generalized news releases. Announcements which attract investors to a firm may produce a transitory increase in trading volume. Whether the relative trading volume of the affected firm will permanently increase as a result of a generalized news release is unclear. Volume effects, if present, are expected to be short lived and only associated with announcements of additions to the DJIA.
DATA AND METHODOLOGY
The sample consists of all changes, from 1962 to 1991, in the roster of common stocks whose prices are used to compute the DJIA and DJTA. Except for the substitution of nine firms in the roster of the Dow Jones Rail Road Average in January of 1970, at which point the Average was renamed the Dow Jones Transportation Average, deletion from the rosters of both the DJIA and DJTA have generally been precipitated by corporate control events.(3)
The 30 year sample period of the study contained 11 changes in the DJIA, 41 changes in the DJTA, and one change in a Dow Utility Average(4,5) The majority of the changes in the DJIA occurred in the 1980's while changes in the DJTA are more evenly distributed throughout the sample period. The timing of these changes are illustrated in Table 1.
Unlike the S&P 500, for which the Stock Price Notification Service is available, no information prior to the news release announcing the roster change in the Dow Jones Average is available to the general public. The firms involved in the change and licensed information services are notified, after close of the NYSE, on the day prior to the appearance of the announcement in the WSJ. Our event date is the day on which notification of a change in a Dow Jones Average appears in the WSJ.
Table 1. CHANGES IN THE DOW JONES AVERAGES 1962-1991 DJIA DJTA 1964 1 1965 3 1967 1 1968 2 1969 2 1970 10 1971 2 1976 1 1 1979 2 1980 3 1982 1 4 1983 1 1984 1 1985 2 1 1986 2 1987 2 2 1988 2 1989 2 1991 3 1 11 41 Notes: Because four firms added to the DJTA were formed through merger or consolidation at the time they were added to the Average and publicly traded shares of Consolidated Rail Corp were issued immediately preceding the inclusion of this firm in 1987, tests for price and volume effects use a sample size of 36. The average market value for a firm in the DJIA was $18.88 billion, as of December 31,1990. The stocks included in the DJIA have generally accounted for approximately one-quarter of the market value of issues traded on the NYSE. The average market value for a firm in the DJTA was $1.84 billion, as of December 31, 1990.
A market model methodology utilizing standardized prediction errors developed by Dodd and Warner (1983) based on methodology developed by Patell (1976) and Dodd (1980) was used to measure price effects. For each security, j, added to or dropped from an index, we estimate the market model and compute an excess return, or prediction error, for period t as:
[PE.sub.jt] = [R.sub.jt] - ([a.sub.j] + [b.sub.j][R.sub.mt]),
where [R.sub.jt] is the return on security j for period t, [R.sub.mt] is the return on the market for period t, and [a.sub.j] and [b.sub.j] are the ordinary least squares estimates of the coefficients of the market model regression for firm j. Daily returns from the Center for Research in Security Price equally weighted index were used as a proxy for the return on the market. The market model parameters are estimated from day -150 to day -25 relative to the announcement date.(6)
A standardized prediction error is computed as:
[SPE.sub.jt] = [PE.sub.jt]/[s.sub.jt],
where: [s.sub.jt] is the standard error of the forecast for security j for period t. We assume the standardized prediction errors are serially and cross-sectionally independent, normally distributed with a mean of zero, so that the standardized prediction errors are assumed to be t-distributed.
To measure abnormal returns over a specific interval for firm j we sum the prediction errors to give the cumulative prediction error,
[CPE.sub.j] = [summation of] [PE.sub.jt] where t = [T.sub.1j] to [T.sub.2j]
where: [T.sub.1j] and [T.sub.2j] are the relevant event time intervals. For a sample of N securities the mean cumulative prediction error is given by
[Mathematical Expression Omitted]
An interval test statistic is formed by standardizing the individual [SPE.sub.j] for the number of days in the interval for firm j as
[SCPE.sub.j] = [summation of] [SPE.sub.jt] / [([T.sub.2j] - [T.sub.1j] + 1).sup.1/2] where t = [T.sub.1j] to [T.sub.2j]
The interval test statistic is given by:
Z = [summation of] [SPE.sub.j] / [(N).sup.1/2] where j = 1 to N
The individual [SPE.sub.jt]s are assumed to be unit-normal and independent under the null hypothesis of no abnormal returns, therefore both [SCPE.sub.j] and Z will be approximately unit-normal.
Because daily volume distributions have been shown to be right skewed and leptokurtotic, Ajinkya and Jain (1989), the natural logarithm of daily volume is used to construct tests of abnormal event period trading volume.(7) The null hypothesis of no abnormal event period relative trading volume was tested using the t-test described in Lamoureux and Wansley (1987), and is described below.
[V.sub.it] = ([VOL.sub.it] / [VOL.sub.mt]), relative volume for firm i,
[VOL.sub.it] = Logarithm of the daily trading volume for firm (i) on day t.
[VOL.sub.mt] = Logarithm of the daily trading volume on the New York Stock Exchange on day t.
[V.sub.in] = Average non-event period relative volume for firm (i)
[V.sub.in] = [(126).sup.-1] * [summation over t] [V.sub.it] for t = (-63, -62, . . ., -1; ;1, 2, . . . ., 63)
for p = the number of firms in the portfolio
EV = [p.sup.-1] * [summation over p] [V.sub.i0]
NEV = [p.sup.-1] * [summation over p] [V.sub.in]
t(p-1): [p.sup.1/2*] (EV - NEV) / [Sigma]([V.sub.in])
[Sigma]([V.sub.in]) = Sample standard deviation, [V.sub.in], for firms in the portfolio.
Three event periods, day (0), days (-1,1) and days (-1,15), were chosen to investigate the existence of volume effects. For multiple day event periods, (-1,1) and (-1,15), relative volume is averaged through the event period for each firm and compared with the base period average of relative volume in the calendar quarters before and after the event period.
We do not expect the managers at Dow Jones & Company to have better estimates than the market of the future cash flows associated with the firms comprising the Industrial and Transportation Averages. Hence the announcement of a change in an index should not release valuable firm-specific information to the market. Our null hypotheses are that announcements of changes in the composition of the Dow Jones Averages will produce neither a price nor volume effect. To the extent that these announcements draw attention to the firms which compose the Dow Averages, shares of the affected firms may experience increased price and trading volume responses in the event period.
Tables 2 and 3 present results for additions to the DJIA and DJTA.(8) Firms added to the DJIA experience a statistically significant 2.14% cumulative abnormal return in the event period (-1,0).(9,10) All eleven firms added to the Average experience positive abnormal returns. The Z statistic for the Wilcoxon signed rank test is 2.93 which is statistically significant at the 5% level.(11) In the two week period following the change, (1,15), the cumulative abnormal return is negative, but not statistically significant. During the two week period prior to the announcement, (-15,-2), the cumulative abnormal return is positive, but not statistically significant. The t-test for differences in relative volume indicates that on the event date trading volume for firms added to the DJIA is significantly greater, at the five percent confidence level, than trading volume in the quarters immediately surrounding the event date. Average daily volume in the event periods (-1,1) and (-1,15) is not statistically different from the average daily trading volume in the respective base periods.
Firms added to the DJTA experienced a negative two day, (-1,0), return of .26%, which is not statistically significant. Twelve of the thirty six firms added to the DJTA had positive abnormal returns on the event date. In the two week period following the change, the cumulative abnormal return is negative but not statistically significant. During the two week period prior to the announcement the cumulative abnormal return is positive, but not statistically significant. Average daily volume in the event periods (0), (-1,1) and (-1,15) is not statistically different from the average daily trading volume in the respective base periods.
These results for firms added to the DJIA and DJTA are consistent with Merton's attention hypothesis. A one-tailed difference in means test for the two day abnormal returns of those firms added to the DJIA and DJTA was conducted. A t-statistic of 1.973 indicates the abnormal returns for firms added to the DJIA are significantly greater than the abnormal returns of the firms added to the DJTA at the 5% confidence level.
We also examined the impact of institutional ownership. During the period one month prior to through one month following the announcement of a Dow index change the proportion of shares held by institutional investors in the firms added to DJIA decreased on average, although the decrease was not statistically significant. In addition, a regression of the two day abnormal return (-1,0) on the change in institutional ownership does not indicate a statistically significant relationship between these variables. This result is not consistent with a change in institutional ownership producing the observed change in value.
SUMMARY AND CONCLUSIONS
Firms added to the DJIA experienced significant positive abnormal returns and larger relative trading volume on the event date. Firms added to the roster of the DJTA experienced neither abnormal returns nor greater trading volume on the event date. Firms dropped from either the DJIA or the DJTA did not experience abnormal returns or increased trading volume in the event period. No statistical relationship was found between abnormal event period returns and institutional ownership.
Dow Jones & Company has not authorized any index trading based on either the DJIA or DJTA. Still the systematic price and volume responses documented for firms added to the DJIA roster may be due to the portfolio rebalancing of unauthorized index traders. However, the price and volume effects produced in this manner are inconsistent with the price pressure hypothesis.
Table 2. ABNORMAL RETURNS AND VOLUME RESPONSE FOR FIRMS ADDED TO THE DOW JONES AVERAGES 1962-1991 Industrial Transportation n=11 n=36 # Positive a. (-1,0) 11(*) 12 Event Period MAR% MAR% (-1) 1.193(*) -0.068 (2.462) (-0.170) (0) 0.945(*) -0.195 (1.950) (-0.491) CAR%. CAR%. (-15, -2) 0.489 1.483 (-0.180) (1.691) (-1,0) 2.139(*) -0.263 (3.594) (-0.957) (1,15) -0.356 -2.210 (-0.011) (-0.682) T-Test Statistics for the Difference in Relative Volume Event Period Industrial Transportation (0) 1.862(*) -0.488 (-1,1) 0.601 0.139 (-1,15) -0.290 0.211 Notes: MAR = Mean Abnormal Return, t-statistics reported in parentheses. CAR = Cumulative Abnormal Return, Z-statistics reported in parentheses. * Significant at the 5% confidence level. ** Significant at the 10% confidence level. a. The null hypothesis is that the proportion of positive prediction errors equals .5. The test statistic is a Wilcoxon signed rank statistic.
The observed abnormal returns associated with additions to the DJIA may also be explained by a belief on the part of market participants that the managers of the index have superior information which is revealed through changes in the index. However, given that both the DJIA and DJTA are maintained by Dow Jones & Company, and no price or volume effects were found for changes in the DJTA, market participant's reaction to perceived superior information on the part of the Dow Jones & Company is an unlikely source of the event day price and volume responses for firms added to the DJIA.
Given the prominent media coverage of the DJIA, the observed responses are likely due to the increased attention market participants directed towards firms affected by the announced changes. Firms in the DJTA have smaller market values and account for a smaller proportion of daily trading volume on the NYSE. The DJTA receives far less general media coverage than does the DJIA and, the announcements of changes in the DJTA are displayed much less prominently in the WSJ. As a result less attention may be drawn to the firms which are affected by the changes in the roster of the DJTA.
Table 3. INDIVIDUAL DAILY ABNORMAL RETURNS FOR FIRMS ADDED TO THE DOW JONES AVERAGES 1962-1991 Industrial Transportation Mean Mean Abnormal Abnormal Return (%) t-statistic Return (%) t-statistic -10 -.745 -1.54 -.388 -0.98 -9 .589 1.21 .324 0.82 -8 -.286 -0.59 .691 1.74 -7 -.005 -0.01 .027 0.07 -6 -.383 -0.79 .253 0.64 -5 -.182 -0.38 -.004 -0.01 -4 -.109 -0.22 -.038 -0.09 -3 -.183 -0.38 .158 0.40 -2 -.057 -0.12 .497 1.25 -1 1.193 2.46(*) -.068 -0.17 0 .945 1.95(*) -.195 -0.49 1 .086 0.18 .365 0.92 2 -.291 -0.60 .069 0.17 3 .237 0.49 -.286 -0.72 4 .281 0.58 -.308 -0.77 5 .235 0.49 -.085 -0.21 6 .344 0.71 .171 0.43 7 -.303 -0.63 .294 0.74 8 -.088 -0.18 -.554 -1.40 9 .263 0.54 -.042 -0.11 10 .020 0.04 .198 0.50 Notes: * Significant at the 5% confidence level.
Direct all correspondence to: John Polonchek, Oklahoma State University, Department of Finance, College of Business Administration, Stillwater, OK 74078.
1. Conversations with Dow Jones personnel support the non-information (future cash flows) content of the changes in the roster of the DJIA and DJTA. Objective criteria, such as the number of shares outstanding, the proportion of shares held by insiders, earnings growth, and industry ranking are used to screen the list of potential candidates. This information is readily available to the public prior to the event. In addition, the Wall Street Journal announcements of changes in the Avenges emphasize the fact that the substitutions do not reflect any opinion regarding the price prospects for the firms added to or dropped from the indices.
2. See Lamoureux and Wansley (1987) for a discussion of the implementation of the Stock Price Index Notification Service.
3. Five of the changes to the roster of the DJIA were precipitated by; one merger, two acquisitions, one bankruptcy, and one equity restructuring. Twenty six of the changes to the roster of the DJTA were precipitated by; fifteen mergers, ten buyouts/acquisitions, and one bankruptcy. These events generate the difference in our sample between the number of firms added to an Average and the number of firms dropped from an Average.
4. To be included in the sample a candidate firm must have had daily returns available on the CRSP tape for a period of time spanning the period starting one year prior to the event date and one month following the event date. Firms with contemporaneous announcements were omitted from the sample.
5. One change in the roster of the Dow Jones Utilities Average occurred during this time period. In April of 1986 Cleveland Electric Illuminating Company, the firm in the index, was merged with Toledo Edison Company, the resulting firm Centerior Energy Corporation was added to the index.
6. We also estimated the market model parameters over the corresponding period following the announcement date (+25 to +150). These results were not statistically different from those reported using the pre-event estimation period.
7. Firms comprising the DJIA account for a larger portion of NYSE trading volume than firms in the DJTA. In the two quarter window surrounding the event date the average daily trading volume for a firm from the DJIA sample, [V.sub.in], was 0.25% of the average daily trading volume of the New York Stock Exchange. During the same two quarter window the average daily trading volume for a firm from the DJTA sample was 0.09% of the average daily trading volume of the New York Stock Exchange.
8. We also investigated deletions from both Dow Averages using the same methodologies and tests for the firms added to the rosters. These results are consistent with the Merton hypothesis. No price or volume effects were observed during the event period for deletions from either index.
Deletions from Dow Jones Averages 1962-1991 CAR(-1,0) Day(0) Difference in n (Z-score) Relative Volume T-statistic DJIA 6 -1.072 0.618 (-1.025) DJTA 10 0.600 0.413 (0.193)
9. We report the results for the two day event window (-1,0) to allow for the possibility of leakage of information during the trading day prior to the announcement appearing in the WSJ.
10. To account for the possibility of serial correlation of the abnormal returns we re-estimated our test statistics using the methodology developed by Mikkelson and Partch (1988). This correlation correction has been used extensively in the literature, see for example, Cowan, Nayar, and Singh (1990), Lee (1992), Mais, Moore and Rogers (1989). The adjusted z-statistics are reported below.
Industrials Transportations (-1,0) (-1,0) Additions 2.139% -0.263% (3.54) (-0.94) Deletion -1.072% 0.600% (-1.01) (0.16)
11. The abnormal returns range in value from .6% to 4.57%, five of the eleven range in value from 2.16% to 2.76%. Thus, it is not likely that the result is driven by an outlier.
 Ajinkya, Bipin and Prem Jain. (1989). "The Behavior of Daily Stock Market Trading Volume." Journal of Accounting and Economics, 11: 331-359.
 Corrado, Charles and Terry Zivney. (1992). "The Specification and Power of the Sign Test in Event Study Hypothesis Tests Using Dally Stock Returns." Journal of Financial and Quantitative Analysis, 27(3): 465-478.
 Cowen, Arnold, Nandkumar Nayar and Ajai Singh. (1990). "Stock Returns Before and After Calls of Convertible Bonds." Journal of Financial and Quantitative Analysis, 25(4): 549-554.
 Dodd, Peter. (1980). "Merger Proposals, Management Discretion, and Shareholder Wealth." Journal of Financial Economics, 8: 105-138.
 Dodd, Peter and Jerold Warner. (1983). "On Corporate Governance: A Study of Proxy Contests." Journal of Financial Economics, 11: 401-438.
 Goetzmann, William and Mark Garry. (1986). "Does Delisting from the S&P 500 Affect Stock Price?" Financial Analysts Journal,: 64-69.
 Harris, Lawrence and Eitan Gurel. (1986). "Price and Volume Effects Associated with Changes in the S&P 500 List: New Evidence for the Existence of price Pressures." Journal of Finance, 41(4): 815-829.
 Jain, Prem. (1987). "The Effect on Stock Price of Inclusion in or Exclusion from the S&P 500." Financial Analysts Journal: 58-65.
 Lamoureux, Christopher and James Wansley. (1987). "Market Effects of Changes in the Standard and Poor's 500 Index." The Financial Review, 22(1): 53-69.
 Lee, Scott. (1992). "Management Buyout Proposals and Inside Information." Journal of Finance, 47(3): 1061-1079.
 Mais, Eric, William Moore and Ronald Rogers. (1989). "A Re-Examination of Shareholder Wealth Effects of Calls of Convertible Preferred Stock." Journal of Finance, 44(5): 1401-1410.
 Merton, Robert. (1987). "A Simple Model of Capital Market Equilibrium with Incomplete Information." Journal of Finance, 42(3): 483-510.
 Mikkelson, Wayne and Megan Partch. (1988). "Withdrawn Security Offerings and the Issuance Process." Journal of Financial and Quantitative Analysis, 23(2): 119-134; Errata, 23(4); 487.
 Patell, James. (1976). "Corporate Forecasts of Earnings Per Share and Stock Price Behavior: Empirical Tests." Journal of Accounting Research, 14: 246-276.
 Pruitt, Stephen and K. C. John Wei. (1989). "Institutional Ownership and Changes in the S&P 500." Journal of Finance, 44(2): 509-513.
 Shleifer, Andrei. (1986). "Do Demand Curves for Stocks Slope Down?" Journal of Finance, 41(3): 579-590.
|Printer friendly Cite/link Email Feedback|
|Author:||Polonchek, John; Krehbiel, Tim|
|Publication:||Quarterly Review of Economics and Finance|
|Date:||Dec 22, 1994|
|Previous Article:||A re-examination of the operating leverage-financial leverage tradeoff hypothesis.|
|Next Article:||Predictability of stock returns and real output.|