The price and volume effects related to changes in the S&P 500 index

(1)

The Price and Volume Effects related to

Changes in the S&P 500 Index

ABSTRACT

This thesis studies the price and volume effects of changes to the S&P 500 Index. A sample of stocks added to and deleted from the index between September 2008 and October 2015 shows symmetric but no permanent effects. Results from empirical tests provide evidence for the price-pressure hypothesis. After an addition is announced, prices increase to almost 3 percent. When the change is implemented, the increase is fully reversed. The trading activity is at most between the announcement and the effective date.

Keywords: S&P 500; index inclusion; price; volume JEL Classification: G10; G12; G14

Bachelor’s thesis in Finance and Organization University of Amsterdam

Faculty of Economics and Business

Author: Joep Lubberdink

Student nr.: 10008586

Date: February 1, 2016

Field: Finance

Supervisor: Dr. J.J.G. Lemmen Coordinator: Dr. P.J.P.M. Versijp

(2)

2 Hierbij verklaar ik, Joep Lubberdink, dat ik deze scriptie zelf geschreven heb en dat ik de volledige verantwoordelijkheid op me neem voor de inhoud ervan.

Ik bevestig dat de tekst en het werk dat in deze scriptie gepresenteerd wordt origineel is en dat ik geen gebruik heb gemaakt van andere bronnen dan die welke in de tekst en in de referenties worden genoemd.

De Faculteit Economie en Bedrijfskunde is alleen verantwoordelijk voor de begeleiding tot het inleveren van de scriptie, niet voor de inhoud.

(3)

3

The S&P 500 Index is an American stock market index, consisting of 500 large companies, founded in 1957. It tends to be the performance benchmark for the US equity markets. Many investors track the S&P 500 Index. The composition is determined by S&P Dow Jones Indices. Changes made in the Index have a big impact on investors, as it is one of the most commonly followed stock indices. Funds that follow the S&P 500 Index should adjust their portfolio when there occur changes in the

composition. When many index funds add the stock to their portfolios, this causes a demand shock, which leads to an increase in trading volume and stock price. Standard and Poor’s state that the changes made are meant to maintain the stability of the Index, and the selections are solely based on publicly available information (S&P Dow Jones Indices, 2015). Therefore, changes in the Index should be information-free events. Nevertheless, former research of Harris and Gurel (1986) and Shleifer (1986) provides significant evidence of a 3-4% abnormal return for the added stock on the first trading day after S&P announces their addition to the Index. Chen et al. (2004) find that after an addition to the Index is announced, trading volume is more than three times higher relative to the 60 trading days preceding the announcement. Chen et al. also show a permanent increase in stock price of added stocks, but no permanent decline for stocks removed from the Index. These results are inconsistent with earlier research about the effects of index changes that imply a symmetric price effect to index inclusion a deletion.

The price effect of additions to and deletions from the S&P 500 Index is widely covered by previous research. Different reasons are given for explaining the price effect. This thesis contributes to the existing literature by testing the explanations in a more recent period. Furthermore, provides this thesis new evidence to the effect of changes in the index, by investigating the volume effect. The price response to changes in the Index after the financial crisis in 2008 will be studied, looking at abnormal returns. The change in trading activity is studied by measuring the volume ratios, adjusted for market volume.

This thesis is organized as follows: the next section discusses the existing scientific literature concerning this topic and describes the four hypotheses following from literature. Section 3 will explain the data and methodology used in this thesis. In the fourth section, the results of the empirical study will be presented and finally, section 5 concludes and gives recommendations for further research.

2. Literature review

The price and volume effects of inclusion and deletion of stocks from the S&P 500 Index are studied a lot. The existing literature studied various time periods, beginning in 1965 and continuing until 2001. The empirical evidence is quite strong that stocks added to the index experience a positive price effect.

(5)

5 With the exception of Harris and Gurel (1986) all studies report that the price effect of inclusion is permanent. Studies about the effect on abnormal returns after an exclusion have different outcomes. However, most studies focus on additions.

First, the composition of the S&P500 Index is explained why changes in the index occur. The results of prior research are shown by a literature table and the four hypotheses concerning the price effect are discussed.

2.1 Changes in the S&P 500 Index Composition

In general, changes in the S&P 500 Index are caused by deletions. Companies involved in a merger, acquisition or significant capital restructuring and companies that substantially violate one or more of the inclusion criteria will be removed from the Index (S&P Dow Jones Indices, 2015). Since the Index has to maintain the number of 500 shares, additions follow the deletions. Standard and Poor’s uses four eligibility criteria for including companies, which are not always strictly enforced: the market capitalization has to be over US$ 5.3 billion; the company must have adequate liquidity, the stock should trade for a minimum of 250,000 shares and have a public float of at least 50%; the firm must be a domestic and leading in an important U.S. industry; last year reported earnings must be positive (S&P Dow Jones Indices, 2015). The companies that meet these requirements can be picked by the U.S. Index Committee. Only firms unanimously approved by the committee are put into a candidate replacement pool at discretion. Until 1976, no public announcement of changes in the Index were made. The only way investors could obtain information about the Index was to call Standard & Poor’s. After September 1976, changes in the S&P 500 Index were announced after the market closed on Wednesday and the effective change occurred the next trading day at the market’s opening. In October 1989, Standard and Poor’s started with preannouncing changes in the index. The date of changes in the index is announced by Standard and Poor’s after the market closes on the

announcement date (AD) and the changes take place on the effective day (ED). The period between AD and ED varies from one day to more than a month. But on average, there are seven trading days between AD and ED.

Since five S&P 500 constituents have multiple publicly listed share class lines, the index now consists of 505 shares. Discovery Communications, Google, Comcast Corp., Twenty-First Century Fox and News Corp. are represented by two share class lines each in the index (S&P Methodology Update). These double listings are not taken into account in this thesis.

2.2 Price and Volume effects of changes in the S&P 500 Index by prior studies

The effects of additions to the S&P 500 Index are widely covered by prior research. Little research has been done about the price effect of deletion, because it is hard to obtain a sufficient sample size.

(6)

6 Only Chen et al. (2004) investigated the impact of a removal, with a sample of 235 deletions. Other studies, include Harris and Gurel (1986), Beneish and Whaley (1996) and Lynch and Mendenhall (1997) analyzed less than 20 deletions. The volume effect of changes in the index is studied by Harris and Gurel (1986), Beneish and Whaley (1996) and Chen et al. (2004). In this thesis, both effects are analyzed for either inclusions and deletions of stocks.

Table 1 shows the most important results of existing literature studying the effects of changes in the S&P 500 Index with a comparable approach. Based on these studies the methodology in this thesis is determined.

(7)

7 Table 1: Results of previous studies concerning the price and volume effects of changes to the S&P 500 Index

Author(s), year Sample Size Sample Period Estimation Period Methodology Event Window Results (Addition/Deletion)

Shleifer (1986) 246 additions 1966-1983 CAR¹ 1966-75 (N=144) 1976-83 (N=102) AD (-20, -1) -2.86*** -1.49 AD -0.192 2.79*** AD (1, 10) -0.065 -0.859 AD (11, 20) 1.12* -0.154

Harris & Gurel (1986) 194 additions

1973-1983 AD (-250,40) CAR 1973-77 (N=110) 1978-80 (N=34) 1981-83 (N=50) AD (0,1) 0.21 2.97*** 3.25*** AD (0,5) 0.69* 2.74*** 2.79*** MVR² AD (0,1) 1.21 1.87*** 3.45*** AD (0,5) 1.01 1.47*** 1.79*** Jain (1987) 87 additions 22 deletions 1977-1983 AD (-60,60) CAR AD (-5,-1) AD (-1,0) AD (0,1) AD (1,5) -0.2 3.1*** 3.3*** -0.5 Dhillon & Johnson

(1991)

187 additions 1978-1988 AD (-250,-120) 1978-83 (N=86) 1984-88 (N=101)

CAR AD (-60,-2) -3.71* -2.30

AD (0,1) 2.38* 3.55***

AD (2,60) -6.48* -1.31

Lynch & Mendenhall (1996)

34 additions 15 deletions

1990-1995 AD (-872,-673) CAR (AD +1, ED-1) 3.807***/ -12.960***

(AD, ED+2) 5.164***/(AD,ED):-14.760*** (ED, ED+2) -1.801***/(ED):5.590*** (ED, ED+7) -2.106**/(ED, ED+5):4.639*** Beneish & Whaley

(1996)

103 additions 1986-1994 1986.01-89.09 (N=70) 1989.10-94.06 (N=33)

CAR (AD-1, AD) 0.277 -0.263

(AD, ED+1) 3.674*** 5.903***

(AD, ED+10) 4.638*** 4.928***

(AD, ED+20) 5.483*** 3.862***

MVR AD 0.990 1.399**

Denis et. al (2003) 236 additions 1987-1999 AD (31,211) CAR AD (0,1)

AD (0,30)

4.65*** -0.4 Chen, Noronha &

Singal, (2004)

760 additions 235 deletions

1962-2000 (AD-60, ED+90) CAR 1962.07-76.08

(N=279/145) 1976.09-89.09 (N=263/28) 1989.10-2000.12 (N=218/62) AD (0,1) -0.047 -0.407* 3.171***/-1.168 5.446***/-8.462*** (AD, ED) 8.899***/-14.436*** (AD, ED+20) -0.742/1.189* 3.123***/-1.642 6.396***/-4.170 (AD, ED+60) 0.588/2.172 3.556***/-1.715 6.189***/0.394 Turnover ratio AD 0.763***/0.691*** 3.741***/3.523*** 3.703***/3.487*** ED 12.323***/16.495***

Elliot, Ness, Walker & Warr (2006) 147 additions 1993-2001 AD (-250,-50) AR AD-1 0.10 AD 5.67*** AD+1 0.07 ED-1 1.68*** ED 2.24*** ED+1 -1.16***

(8)

8 2.3 Hypotheses for price movements around changes to the S&P 500 Index:

According to previous research, four hypotheses can be defined. The following sections describe these hypotheses. Table 2 shows the effect of the different hypotheses on stock price of the firm added to or removed from the S&P 500 Index. The discussed hypotheses lead to the hypothesis development of this thesis.

2.3.1 Long-term downward-sloping demand curve hypothesis (or imperfect substitutes hypothesis)

The efficient market hypothesis states that stocks have perfect substitutes and assumes perfectly elastic demand. So demand curves for stocks are horizontal, and any supply or demand shock doesn’t have an effect on stock price. However, since stocks don’t have perfect substitutes, their demand curves slope down. Therefore, the long-term downward sloping demand curve hypothesis is also called the imperfect substitutes hypothesis. This hypothesis is supported by the studies of Shleifer (1986), Beneish and Whaley (1996), Lynch and Mendenhall (1997) and Wurgler and Zhuravskaya (2002).

As many investment funds follow the index to replicate its performance, they have to adjust their portfolio’s after a change in the index occurs. Therefore, they have to buy the added stock and sell the removed stock. Because the value of all the S&P 500 index funds is about ten percent of the index portfolio value (Beneish & Whaley, 1996), the impact of their trading activity is enormous. The addition to and deletions from the index will cause a demand and supply shock, respectively, and thus a change in price. An important implication of the downward-sloping demand hypothesis is that stock price increases almost 3% on the announcement date (Shleifer, 1986).

The downward-sloping demand curve hypothesis implies that when a stock is removed from the index and contemporaneous selling of stock by index funds result in stock price decrease. Dhillon and Johnson (1991) provide evidence consistent with the imperfect-substitutes hypothesis only if stocks, bonds, puts and calls for the same firm are close substitutes.

2.3.2 Price-pressure hypothesis (or short-term downward-sloping demand curve hypothesis)

Harris and Gurel (1986) show a 3.13% abnormal return after a stock is added to the index, but find a reversal of this initial price increase. However, they only focus on additions. They also find an increase in trading activity on the day the addition is announced. On average, they find trading volume on the first day after inclusion is announced is 1.89 times as large as the daily mean volume over the eight weeks prior to the announcement. Positive abnormal trading volume is consistent with the price pressure hypothesis. Index funds have to rebalance their portfolio after a change occurs in the S&P 500 Index. This rebalancing causes a significant increase in price after the announcement date, but returns to normal level after three weeks.

(9)

9 Lynch and Mendenhall (1997) show a symmetric price effect after a change in the S&P 500 Index is announced. This increase (decrease) in price stops after the firm is effectively added to (deleted from) the index. Lynch and Mendenhall also find a significant increase in trading volume on an announcement date and a day before the effective change. These results are consistent with the price pressure hypothesis and inconsistent with the efficient market hypothesis. Chen et al. (2004) only observe a partial price reversal in the 1989 to 2000 period. They find an increase in abnormal return from 5.4% on the announcement date to almost 9% by the effective inclusion date. But 60 trading days following the effective date there is 6.2% left. In the case of deletions, the abnormal return completely reverses itself after 60 days.

Beneish and Whaley (1996) also show that trading volume increases around the

announcement date. They state that the increase in either trading volume and price is permanent. Because index funds buy the added stocks and hold them. Therefore, the supply of these stocks will decrease and price rises. Index funds adjust their portfolio after the effective change date, to minimize their tracking error. Arbitrageurs use the opportunity of what Beneish and Whaley call the ‘S&P game’. Arbitrageurs trade as soon as the change in the index is announced, since they know that indexers wait until the effective day to rebalance their portfolio. After Standard & Poor’s

implemented the new announcement policy, this effect increased in the transitory period because of the presence of risk arbitrageurs (Beneish & Whaley, 1996). The price pressure causes transitory abnormal returns. The higher the number of trading days between the announcement date and the effective change date is, the greater the demand is by risk arbitrageurs. This is an important difference with the long-downward-sloping demand curve hypothesis, in which the price increase is permanent. Because the price pressure is also caused by a downward-sloping demand curve, but there is no permanent effect, the price-pressure hypothesis is also called the short-term downward-sloping demand curve.

2.3.3 Liquidity hypothesis

Harris and Gurel (1986) argue that the announcement of an addition to the index leads to an increase in trading volume of the added stock. They show that this volume effect has grown over time. From 1973 until 1977 the average volume on the first day after the announcement is 1.21 times as large as the daily mean volume over the eight weeks preceding the announcement date, versus 2.81 in 1978-1973. Beneish and Whaley (1996) also find a large increase in the trading volume after an addition is announced. On the first trading day after the announcement, the stocks trading volume, adjusted for market volume, is 3.484 times as large as the 60 trading days prior to the announcement. Before September 1989, the increase in trading activity after an announcement was even larger. Because back then, Standard & Poor’s announced the change to index on the same day as the change was made effective. In this period, they find an overall average abnormal trading volume of 7.311 times normal.

(10)

10 The liquidity hypothesis not only states an increase in volume but also lowers transaction costs and, therefore, lowers investors expected rate of return. Contrary to the other hypotheses, the liquidity hypothesis expects the increase in trading volume to remain after the effective inclusion day (Hegde and McDermott, 2003)

The asymmetric price response to index changes by Chen et al. (2004) and Lynch and Mendenhall (1997) is inconsistent with the liquidity hypothesis. They show price reversal after the effective inclusion date.

2.3.4 Information hypothesis

Jain (1987) and Dhillon and Johnson (1991) provide evidence for the information hypothesis. Dhillon and Johnson show an increase in the price of bonds after the firm’s stock is included in the index. They suggest that new information is added to the stock after inclusion because of expected increase in trading volume and liquidity. Jain (1987) states that because Standard & Poor’s prefers to include stable firms into the index, the inclusion of a firm signals a reduction in the riskiness of its stocks.

Denis et al. (2003) show a change in earnings expectations after a stock is added to the index. They give two possible explanations for increasing stock prices after inclusion in the index. First, index inclusion may lead to improved operating performance, because of closer scrutiny of management. Jain (1987) and Elliot et al. (2006) also provide evidence for improved quality of

management. Therefore, they argue that stock inclusion is not an information-free event. Alternatively, investors may have recognized superior analytical abilities of Standard and Poor’s in choosing stocks for the index. So they increase their earnings expectations of a firm after it is added to the index. Another implication of the information hypothesis is given by Chen et al. (2004). They say that the addition of a stock to the index leads to an increase in the investor awareness. This increase causes higher demand for this stock and, therefore, an increase in price and trading volume. However, after a stock is removed from the index, investor don’t become ‘unaware’ of the stock. So there should be no decrease in price after deletion. Chen et al. provide evidence of an asymmetric price response for changes in the S&P 500 index.

Shleifer (1986), Harris and Gurel (1986) and Wurgler and Zhurasvkaya (2002) present evidence inconsistent with this hypothesis and state that inclusion is an information-free event. They argue that the stock price should experience a permanent increase as an effect of information provided by the announcement of inclusion. Their results don’t show a permanent effect.

(11)

11 Table 2: Hypotheses and expected effect on stock price

Panel A: Inclusions

Hypothesis pre-AD (-10,0) AD (-1,1) ED (-1,1) permanent ED(0,)

Downward-sloping demand curve inconclusive positive negative positive Price pressure inconclusive positive negative negative Liquidity inconclusive positive negative neutral Information inconclusive positive negative neutral

Panel B: Deletions

Hypothesis pre-AD (-10,0) AD (-1,1) ED (-1,1) permanent ED(0,)

Downward-sloping demand curve inconclusive negative positive negative Price pressure inconclusive negative positive positive Liquidity inconclusive negative positive neutral Information inconclusive negative positive neutral

Note: this table shows the different hypotheses and their expected effect on abnormal return for stocks added to or removed from the S&P 500 Index.

3. Data and Methodology

To examine the price effect of additions to and deletions from the S&P 500 Index, an event study in the spirit of MacKinlay (1997) is used. This methodology is used to analyze the abnormal returns around the event dates. In this thesis, the event dates are the dates that Standard and Poor’s announces the changes in the index (AD) and the dates that the changes are made effective (ED).

The effective dates of changes are obtained from the Center for Research in Security Prices (CRSP), which obtains these dates from Standard and Poor’s itself. The announcement dates had to be searched for each change apart. Back until May 2013, the announcement dates of changes are found by the official press releases at the Standard and Poor’s website (us.spindices.com). But there are no press releases available before May 2013. So these announcement dates are obtained by the websites of Wall Street Journal and PR Newswire. Daily stock returns are obtained by Datastream, as the database of CRSP is updated yearly, and returns don’t go further than December 2014.

3.1 Event Windows

In this study different event windows surrounding the event dates are analyzed. The sizes of the event windows are determined by earlier research. The event windows around the announcement date are comparable to those of Shleifer (1986). The first event window starts ten days prior to the

announcement date and runs through the announcement date: AD (-10,0). This event window shows the development of the stock price prior to the announcement date. The second event window runs

(12)

12 from the day before the announcement date through the next trading day: AD (-1,1) and the final window covers the period ten days after the announcement: AD (0,10).

The event windows around the effective change date are chosen in the spirit of Beneish and Whaley (1996) and Chen et al. (2004). First the abnormal returns are examined around the effective date: ED (-1,1). To analyze the permanent impact, the effects after ten and twenty days are observed for the windows ED (0,10) and ED (0,20), respectively.

Figure 1 shows that for 113 stocks added to the index the event window AD (0,10) includes the effective inclusion day. The same event window for deletions includes the effective change date for all removed stocks. The number of trading days has grown over time. Until October 1989, when Standard and Poor’s introduced the preannouncing of changes to their Index, there was no difference between the announcement date and effective change date. During the period October 1989 through June 1994 the mean number of days between the announcement of an addition and the effective inclusion day was 4.15 (Beneish & Whaley, 1996). As Figure 1 shows, from September 2008 until October 2015, the mean number of days between AD and ED is 7.14 days for additions and 5.63 for deletions. The difference in days between the additions and deletions is because the last day the stock is included in the S&P 500 Index is counted as deletion day. The next trading day is the effective inclusion day.

3.2 Estimation Window

An estimation window is required to calculate the normal return. According to prior studies from Harris and Gurel (1986) and MacKinlay (1997), the parameters of the market model are estimated by one year in trading days (252 days). The event period is not included in the estimation period, to prevent for influence by the event when measuring the normal return. So the event period and

estimation period don’t overlap. Therefore, the estimation period starts 272 days before the event date and ends 20 days prior to the event date.

(13)

(14)

14 3.3 Sample

In this thesis, the firms will be analyzed that are added to and removed from the S&P 500 Index during the period September 2008 through October 2015. In September 2008, Lehman Brothers went bankrupt which caused the start of the global financial crisis. Although this period contains two big financial crises, it is not covered by existing literature yet.

The original sample consists of 160 additions and 150 deletions.1 The surplus of additions is because of the listing of five double share class lines. As mentioned before, these are not taken into account in this thesis. Stocks are filtered from the sample if the addition to or deletion from the index was influenced by a significant simultaneous event, like a merger or acquisition. Their stock prices will be disturbed too much by this event to show an accurate effect of the index inclusion or deletion. From the additions, 33 firms are deleted from the sample because their addition to the Index was caused by a spin-off (18), acquisition (11), merger (3) or company split (1). Another 18 firms are excluded from the sample because there was no data available for the given period. For calculating the volume ratios, only five firms are excluded because of missing data. As a result, the final addition sample for analyzing the abnormal returns consists of 104 firms. The final sample for the volume ratios consists of 117 stock inclusions.

From the 150 firms that are deleted from the S&P 500 Index, 76 of them were acquired by another company. Nine companies are dropped from the sample because they were involved with a spin-off, merger or company split. Another 8 firms are excluded because of redomestication and two went bankrupt. Data was missing for 10 stocks for calculating the abnormal returns and 6 for

measuring the trading activity. The final sample of deletions consists of 50 firms for abnormal returns and 54 firms for trading volume.

3.4 Abnormal Return

The price effect as a result of changes to the S&P 500 index is based on abnormal and cumulative abnormal returns. The abnormal returns are calculated using the daily returns of a security relative to the S&P 500 index. The abnormal return on stock i on day t is given by the equation:

Where represents the actual stock return and is the normal return. The normal return is predicted by the relationship between the stock and the S&P 500 index (expressed by the intercept

and slope ) and the actual return on the index. The average abnormal return is defined as follows:

(15)

15

Where N is the total number of sample stocks. To estimate the total impact of the addition or deletion over the event window, the individual abnormal returns are added up. This gives the

cumulative abnormal return:

The mean value of the identical events is represented by the cumulative average abnormal return:

To determine whether the cumulative average abnormal returns are statistical significant, a statistical test is required. A (student’s) t-test does not have sufficient power because it is sensible for event-induced volatility. Therefore, the standardized cross-sectional test of Boehmer, Masumeci and Poulsen (1991) is used. They provide evidence that their method is robust to variance induced by the event and immune to the distribution of abnormal returns over the event window. First, the all abnormal returns are standardized by the forecast-error corrected standard deviation:

Where is the standard deviation of security i’s abnormal returns during the estimation period and is the average market return during the estimation period. To test the average abnormal return, the standardized average abnormal return is divided by the square root of the number of

securities and its standard deviation. The test statistic on day t, , is given by the equation:

The test statistic for the cumulative average abnormal return is measured in the same way. To test the cumulative average abnormal return, , the following equation is used:

(16)

16

Where is the standardized cumulative average abnormal return over N firms and is the sample standard deviation of over the event period.

To provide certainty in testing significance, the t-test of Brown and Warner (1980) also is used. This is specifically developed for event studies and reads as:

3.5 Volume Ratio

In order to decide whether a change in the S&P 500 Index has an effect on trading activity, trading volumes, adjusted for market volume, are analyzed around announcement date and effective date. This is calculated by the Volume Ratio in line with Harris & Gurel (1986):

and are the actual trading volumes of security i and the total S&P 500 Index at day t, respectively. and represent the average trading volume of the stock and the total S&P 500 Index in the 60 trading days prior to the announcement day. The volume ratio, is a standardized measure of period t trading volume in security i, adjusted for market variation. If there is no change in volume during event-period t relative to the preceding 60 days, its expected value is 1. is the mean volume ratio of the total sample of securities.

The trading activity is tested for significance using a t-statistic. This t-statistic tests whether the mean of the volume ratios is different from 1 by the following equation:

The mean volume ratios are calculated in a three-day interval around the announcement dates and effective dates of changes to the index. An adjustment had to be made by the effective dates of the deletions. CRSP counted the last day a firm was part of the index as the effective deletion day.

However, in many cases this was a Sunday. In this thesis, the first trading day after deletion is determined as the effective deletion day.

(17)

17 4. Results of empirical study

In this section, the results of the event study are discussed. Table 3 shows the price effect of stocks added to or removed from the S&P 500 Index. The results are based on cumulative average abnormal returns relative to the S&P 500 Index’s total return. Table 4 presents the change in trading activity around stock inclusion and deletion. Trading activity is measured by trading volume, adjusted for market variation.

The results of Table 3 refer to the implications of Table 2. The price effect on the announcement of a stock’s addition to the index is similar to previous studies conerning index inclusion. The event window AD (-1,1) presents a cumulative average abnormal return of 2.5%. When tested by the method of Boehmer et al. (1991) the result is significant at 1% level. Testing with Brown and Warner ‘s (1980) t-test results in significance at the 10% level. Figure 2 shows the effect clearly. This effect is comparable to the price increase of around 3% other researchers found. The positive return around the announcement date is consistent with all four hypotheses.

The returns after the effective inclusion date show a significant price reversal. First, there is a small decrease in abnormal return, just after the effective date. This is consistent with the four

hypotheses and suggests the selling of the included firms by arbitrageurs (Beneish and Whaley, 1996). Index funds adjust their portfolios for changes in the index after the effective inclusion date. Since arbitrageurs know this, they buy the added stock at the announcement date and close their position around the effective date. In this way the arbitrageurs profit from the price pressure. This is called the ‘S&P Game’.

The permanent fall in price makes the initial abnormal returns fully disappear. This reaction is consistent with the price pressure hypothesis and inconsistent with the downward-sloping demand curve, liquidity and information hypotheses. One explanation for the permanent strong decrease in abnormal returns is the financial crisis. It is possible that the added stocks are not stable enough for a crisis and underperform relative to more experienced S&P500 index constituents. Another reason is that the market capitalization of added firms is smaller than that of existing constituents.

Just looking at the signs, table 3 shows a symmetric price effect. However, the sample of the deleted firms is relatively small, and this results in less significant outcomes. Also for deletions, the price effect on the announcement date is around 3%, consistent with prior studies. Stock prices of stocks effectively removed from the index exceed the preannouncement prices. Recovering the initial fall in stock price is consistent with the price pressure hypothesis, but these high abnormal returns are remarkable. An explanation for this could be a selection bias. Most firms are removed from the S&P 500 Index because they are acquired by another company. But these stocks are filtered from the sample. Only the stocks remain that are removed from the index because of market capitalization. This suggests a decrease in stock price further before the announcement date than investigated in this

(18)

18 event study. The positive abnormal returns after the effective deletion suggest a recover of this price fall. Another possible explanation is the small sample size.

Figure 2 shows the inclusion effect on stock price. The average abnormal return is at most the day after the addition is announced. Before the announcement, the figure already shows some effect. This may occur because investors speculate on Index inclusion of a firm. For example when there are rumors that a S&P 500 Index constituent will be acquired and removed from the Index. Speculators anticipate on these rumors by buying the stock which is expected to replace the deleted firm. After (AD +2) the average abnormal return is closer to the average normal return of all firms and the return on the S&P 500 Index. A decrease in stock price can be seen at the seventh day after the addition announcement. This occurs because the effective inclusion date is, as shown in Figure 1, on average, seven days after the announcement.

Table 3: Abnormal returns around changes in the S&P 500 Index

The initial sample consists of all stocks added to or removed from the S&P 500 Index from September 2008 to October 2015. Stocks are excluded if their addition or deletion is caused by a simultaneous event, such as mergers, acquisition or a default. The cumulative average abnormal returns (%) are calculated relative to the S&P 500 Index’s total return. The CAARs are analyzed around the day that Standard & Poor’s announces the changes to the index (AD) and the day that the changes are made effective (ED). In the parentheses the interval around this event date is shown. The significance of the CAAR is tested with a standardized cross-sectional z-test of Boehmer, Masumeci and Poulsen (1991). The final column shows the percentage of firms that have a positive cumulative abnormal return over the event window.

Inclusion (N=104) Deletion (N=50)

Interval CAAR % positive CAAR % positive

AD (-10,0) 0.6059 1.7889 0.6166 57.69 -1.1951 0.0310 -0.2298 52.00 AD (-1,1) 2.5057 6.1616 1.4126 77.88 -2.6730 -1.2470 -0.5439 40.00 AD (0,10) -0.3904 -0.9294 -0.1545 46.15 7.8593 2.2426 1.2976 70.00 ED (-1,1) -0.6111 -1.7279 -2.5743 43.27 1.5324 0.8180 0.3642 60.00 ED (0,10) -3.4679 -6.8882 -4.5548 26.92 9.7086 2.5533 1.7271 60.00 ED (0,20) -5.2319 -5.5532 -4.6043 26.92 13.0474 2.8166 2.0792 66.00

Note: a list of all average abnormal returns can be found in Appendix 1. The estimation window used for these calculations is one year in trading days (252 days) ending 20 days prior to the event date. Calculations using a different estimation window (of 150 days) give comparable results. Please see Appendix 2.

(19)

19 Figure 3 shows the stock price response around the announcement of a deletion. The positive average abnormal return at the announcement date is surprising, although it is fully wiped out by the negative average abnormal returns the days preceding and following the announcement date. This results in a negative cumulative average abnormal return for this interval, as shown in Table 3.

-1,00% -0,50% 0,00% 0,50% 1,00% 1,50% 2,00% 2,50% -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 AD 1 2 3 4 5 6 7 8 9 10 AAR Normal Return Average Return Index

Figure 2: Stock price response around additions

Note: this figure shows the average abnormal return of 104 firms added to the S&P 500 Index in the period September 2008 until October 2015. The vertical axis shows the return. The horizontal axis shows the event days around the day the addition is announced. -4,00% -3,00% -2,00% -1,00% 0,00% 1,00% 2,00% 3,00% 4,00% -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 AD 1 2 3 4 5 6 7 8 9 10 AAR Normal Return Average Return Index

Figure 3: Stock price response around deletions

Note: this figure shows the average abnormal return of 50 firms removed from the S&P 500 Index in the period September 2008 until October 2015. The vertical axis shows the return. The horizontal axis shows the event days around the day the addition is announced.

(20)

20 The effect of changes in the S&P500 Index to trading activity are shown in Table 4. The trading volumes of the stock added to or deleted from the index are adjusted for market volume. After the announcement of an addition to the index, trading volume increases as predicted. On average, the trading volume of the first day after the addition is announced is 3.194 times higher as the daily mean 60 days trading prior to the announcement. This is consistent with earlier research of Harris and Gurel (1986) and Beneish and Whaley (1996). The large mean volume ratio is caused by almost all firms; 92.31% of the stock show an individual volume ratio greater than 1. The increase in mean volume ratio remains stable the first day after the effective inclusion date. This is consistent with the liquidity hypothesis. The price pressure hypothesis states that in the long term the positive volume effects reverts to the pre-inclusion level. The trading activity has to be investigated over a larger period to provide evidence for this hypothesis. Figure 4 shows a larger timeframe.

One day before the effective deletion of a stock from the index, the average trading volume is more than four times higher caused by almost 89% of the firms. A possible explanation for this effect is that arbitrageurs know that index funds adjust their portfolio after the effective date. This will lead to a supply shock and thus a decrease in stock price. Arbitrageurs anticipate on this by selling the stocks before the index funds will do.

Table 4: Trading activity around changes in the S&P 500 Index

Index Inclusion (N=117) Index Deletion (N=54)

MVRa. STDb. % > 1 MVR STD % > 1 AD –1 0.993 0.427 -1.007 38.46% 1.151 1.556 0.711 29.63% AD 1.301 1.076 3.029 47.01% 1.326 1.455 1.645 40.74% AD +1 3.194 3.139 7.561 92.31% 2.310 2.675 3.598 79.63% ED –1 1.521 2.221 2.535 42.74% 4.395 4.586 5.440 88.89% ED 1.203 0.968 2.273 43.59% 1.764 3.573 1.572 51.85% ED +1 1.365 1.986 1.988 40.17% 1.942 3.787 1.829 42.59%

a. Mean volume ratio. The cross-security mean of the volume ratio in security i to the average volume in that security in the 60 days prior to the announcement date (AD) or effective change date (ED). If there is no effect on trading volume, the expected value of this ratio will be equal to 1.

b. The sample standard deviation of the volume ratios

c. t-statistic for testing whether the mean of the volume ratio is different from 1.

Figure 4 presents the trading activity around the announcement date of stock inclusions and deletions over time. The mean volume ratio increases after an addition is announced. Five trading days after an addition is announced, the trading volume is 6.257 times higher as the daily mean 60 days preceding the announcement. The individual volume ratios are greater than 1 in 83% of the stocks at (AD+5). After seven trading days the mean volume ratio is coming closer to the expected value of 1, the pre-inclusion level. This is consistent with the price pressure hypothesis. As shown in Figure 1, the average number of days between the announcement date and the effective change date is 7.1 for inclusions and 5.6 for deletions. Figure 4 shows the most trading activity prior to the effective

(21)

21 change date. This suggests that the volume effects of arbitrage are higher than the actual

portfolioadjustments relating to changes in the S&P 500 Index. Further research is needed to provide evidence for this suggestion. Therefore, researchers should be able to distinguish the trades of arbitrageurs from those of indexers.

5. Conclusion & Discussion

This thesis studies the effects of changes in the S&P 500 Index to the price and volume of a stock. Previous research provides four hypotheses explaining these effects: the long-term downward-sloping demand curve hypothesis, the price-pressure hypothesis, the liquidity hypothesis and the information hypothesis. These hypotheses are tested by an event study during from September 2008 through October 2015, which is an extraordinary period. The global financial crisis and the Euro crisis are included in this event study. Still, the results of this thesis show a significant increase in stock price of almost 3% after addition to the index is announced. This price effect is comparable to previous studies and consistent with the four hypotheses. The reversal of stock price after the effective inclusion date provides evidence for the pressure hypothesis. In contrast to the other hypotheses, the price-pressure hypothesis states there is no permanent price effect.

Also, the symmetric price effect on the announcement date is similar to most former research. The results in this thesis show negative abnormal returns around the announcement of a stock

0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 AD -10 AD -9 AD -8 AD -7 AD -6 AD -5 AD -4 AD -3 AD -2 AD -1 _AD AD+ 1 AD+ 2 AD+ 3 AD+ 4 AD+ 5 AD+ 6 AD+ 7 AD+ 8 AD+ 9 AD+ 10 Stock Inclusions Stock Deletions

Figure 4: Trading activity around changes in the S&P 500 Index

Note: this figure shows the mean volume ratios around the announcement date. The figure presents 117 stock added to and 54 stock deleted from the S&P 500 Index.

(22)

22 exclusion. The large abnormal returns after the stock is effectively removed from the index are

remarkable. A large reversal of the price effect between the announcement date and the effective date applies for both inclusions and deletions. Stock prices appear to make a correction for the initial announcement effect, after the change to the index is made effective. This suggests the effects of arbitrage.

The presence of arbitrageurs is shown by the trading activity around changes to the S&P 500 Index. The trading volume after the announcement date is around three times higher than the 60 days preceding the announcement. Arbitrageurs know that index funds adjust their portfolios after the effective date, since the funds attempt to minimize their tracking error. The highest trading volume is recognized between the announcement and effective change date. The arbitrageurs profit from the price pressure during this period. After seven trading days, the mean volume ratio is coming closer to the pre-inclusion level, which is consistent with the price-pressure hypothesis.

This thesis contains some limitations. The abnormal return is calculated by subtracting the normal return from the realized return. As an alternative for measuring the normal return through the market model, a multiple factors model could have been used. Well known approaches are the three-factor model of Fama and French (1993) and Carhart’s (1997) four-three-factor model, which includes momentum. There are no large differences in results expected. Another limitation derives from analyzing the trading activity. In this thesis the trading volume is not adjusted for shares outstanding. Therefore, the mean volume ratio can be affected by a few large stocks with unusual high trading volume.

The small sample size of deleted firms can make their results less accurate. The symmetric price effect found in this thesis is consistent with previous research, except from Chen et al. (2004). However, only Chen et al. studied the effect of deletion with a sufficient sample size. Future researchers investigating this subject are suggested to choose a larger period, so more deletions are included in the sample.

The results in this thesis show a transitory price and volume effect, which provides evidence for the price pressure hypothesis. The absence of a permanent effect is inconsistent with the long-term downward-sloping demand curve, the liquidity and the information hypothesis. Further research is needed to find out what drives the announcement effect and the strong price reversal, and whether these effects are caused by arbitrageurs or indexers.

(23)

23 References

Beneish, M. D., & Whaley, R. E. (1996). An anatomy of the “S&P Game”: The effects of changing the rules. The Journal of Finance, 51(5), 1909-1930.

Boehmer, E., Masumeci, J., & Poulsen, A. B. (1991). Event-study methodology under conditions of event-induced variance. Journal of Financial Economics, 30(2), 253-272.

Brown, S. J., & Warner, J. B. (1980). Measuring security price performance. Journal of Financial Economics, 8(3), 205-258.

Chen, H., Noronha, G., & Singal, V. (2004). The price response to S&P 500 index additions and deletions: Evidence of asymmetry and a new explanation. The Journal of Finance, 59(4), 1901-1930. Denis, D. K., McConnell, J. J., Ovtchinnikov, A. V., & Yu, Y. (2003). S&P 500 index additions and earnings expectations. The Journal of Finance, 58(5), 1821-1840.

Dhillon, U., & Johnson, H. (1991). Changes in the Standard and Poor's 500 List. Journal of Business, 64(1), 75-85.

Elliott, W. B., Ness, B. F., Walker, M. D., & Warr, R. S. (2006). What drives the S&P 500 inclusion effect? An analytical survey. Financial Management, 35(4), 31-48.

Harris, L., & Gurel, E. (1986). Price and volume effects associated with changes in the S&P 500 list: New evidence for the existence of price pressures. The Journal of Finance, 41(4), 815-829.

Hegde, S. P., & McDermott, J. B. (2003). The liquidity effects of revisions to the S&P 500 index: An empirical analysis. Journal of Financial Markets, 6(3), 413-459.

Jain, P. C. (1987). The effect on stock price of inclusion in or exclusion from the S&P 500. Financial Analysts Journal, 43(1), 58-65.

Lynch, A. W., & Mendenhall, R. R. (1997). New Evidence on Stock Price Effects Associated with Changes in the S&P 500 Index. The Journal of Business, 70(3), 351-83..

MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of Economic Literature, 13-39.

Shleifer, A. (1986). Do demand curves for stocks slope down?. The Journal of Finance, 41(3), 579-590.

Wurgler, J., & Zhuravskaya, E. (2002). Does Arbitrage Flatten Demand Curves for Stocks?. The Journal of Business, 75(4), 583-608.

S&P Dow Jones Indices. (2015). Press Release - S&P U.S. Indices Methodology Update. New York: McGraw Hill Financial.

(24)

24 Appendices

Appendix 1: Average abnormal returns around changes in the S&P 500 Index

Inclusions (N=104) Deletions (N=50)

Event day Event Day Event Day Event Day

ED-1 -0.001177 ED-1 0.0250512

AD-10 -0.003412 ED -0.003617 AD-10 0.0064145 ED -0.0219498

AD-9 _-0.002645 _ED+1 _-0.001316 AD-9 _0.0186025 _ED+1 _0.0122228

AD-8 _0.0021321 _ED+2 _-0.001011 AD-8 _0.0080609 _ED+2 _-0.005037

AD-7 _0.0002909 _ED+3 _-0.006392 AD-7 _-0.0083238 _ED+3 _-0.0007861

AD-6 _0.005642 _ED+4 _-0.00576 AD-6 _-0.0067647 _ED+4 _-0.0069564

AD-5 _-0.003725 _ED+5 _-0.002312 AD-5 _-0.004373 _ED+5 _0.0065895

AD-4 _0.001782 _ED+6 _-0.002184 AD-4 _0.0073606 _ED+6 _0.0084857

AD-3 _0.0027416 _ED+7 _0.0006755 AD-3 _-0.0113349 _ED+7 _0.0261459

AD-2 _-0.0005 _ED+8 _-0.00304 AD-2 _-0.0115678 _ED+8 _0.0360912

AD-1 _0.0004324 _ED+9 _-0.006455 AD-1 _-0.0325693 _ED+9 _0.0249098

AD _0.003319 _ED+10 _-0.003268 AD _0.0225441 _ED+10 _0.0173705

AD+1 _0.0213056 _ED+11 _-0.001405 AD+1 _-0.0167049 _ED+11 _-0.0050104

AD+2 _-0.00174 _ED+12 _-0.002375 AD+2 _-0.0091649 _ED+12 _0.0010486

AD+3 _-0.004238 _ED+13 _-0.001328 AD+3 _0.0114793 _ED+13 _0.0130227

AD+4 _-0.001493 _ED+14 _-0.005695 AD+4 _-0.0045999 _ED+14 _-0.0006337

AD+5 _-0.001413 _ED+15 _-4.65E-05 AD+5 _-0.0195167 _ED+15 _0.023911

AD+6 _-0.004238 _ED+16 _-0.001688 AD+6 _0.0063434 _ED+16 _0.0045937

AD+7 _-0.006759 _ED+17 _-0.005106 AD+7 _0.0202184 _ED+17 _0.0001931

AD+8 _-0.004047 _ED+18 _0.0035185 AD+8 _0.0052765 _ED+18 _-0.0010614

AD+9 _-0.001423 _ED+19 _-0.000756 AD+9 _0.0264791 _ED+19 _0.0030106

AD+10 _-0.003178 _ED+20 _-0.002758 AD+10 _0.0362382 _ED+20 _-0.0056861

Appendix 2: Cumulative abnormal returns with different estimation windows

Inclusion (N=104) Deletion (N=50) Interval (%) (%) (%) (%) AD (-10,0) 0.6059 0.4999 -1.1951 -1.4168 AD (-1,1) 2.5057 2.5136 -2.6730 -1.8333 AD (0,10) -0.3904 -0.4871 7.8593 9.1312 ED (-1,1) -0.6111 -0.6145 1.5324 1.8214 ED (0,10) -3.4679 -3.6589 9.7086 10.3707 ED (0,20) -5.2319 -5.5146 13.0474 14.7311

1. Estimation window for measuring the normal return: (-272, -20) 2. Estimation window: (-170, 0)

The price and volume effects related to changes in the S&P 500 index

The Price and Volume Effects related to

Changes in the S&P 500 Index

Contents