A PRIL 2004 U.S. STOCK INDEXES: IS THERE A BEST CHOICE? Research Report by Jim Quinn Email: jim.quinn@qqro.com There are numerous stock indees available to U.S. stock market investors, ranging from the Wilshire 5000, which essentially covers the entire U.S. market, to stock indees that cover very small segments or sectors of the market. This paper will compare U.S. stock indees that either address the broad U.S. market or address a segment of the U.S. market defined by market capitalization (i.e. large-, mid- or small-cap). A logical starting point for one who wants to gain eposure to the U.S. market would be to buy all the stocks in the U.S. and weight those purchases by market capitalization. There are two problems with this strategy. First, many of these stocks are illiquid, so when they are bought and sold, the transaction costs can be quite high. Second, an average-sized mutual fund would be spending only a few dollars on each of the smallest U.S. stocks. For eample, a $100 million fund attempting to replicate the Wilshire 5000 would be establishing a position of less than $10 each in many of the Wilshire stocks. It is not economically practical to establish or maintain holdings of this size. An alternative solution is to buy a subset of the U.S. market and rebalance the portfolio periodically to maintain a representative inde. Most U.S. stock indees are designed as such a subset. Some will attempt to represent the entire U.S. market, or a marketcap segment thereof; some attempt to mirror an industrial sector; and some seek to depict a style (e.g.: value or growth). If a subset of the U.S. market is selected, periodic changes will be required to the content of the subset to ensure that it continues to represent the selected segment or sector of the market.
C ONTENTS C HARTS AND TABLES 3 Summary 5 I. Inde Construction 7 II. Size and Sector Representation 11 III. Investability 15 IV. Turnover 17 V. Historical Returns 23 VI. Conclusions 7 Figure 1: Summary Statistics for Dow Jones, Russell, S&P, and Wilshire Indees 8 Figure 2: Number of Stocks in Various Market-Cap Ranges 8 Figure 3: Percentage of Inde s Total Market Capitalization in Various Market-Cap Ranges 9 Figure 4: Economic Sector Representation for the Broad-Cap Indees 10 Figure 5: Economic Sector Representation for the Large-Cap Indees 10 Figure 6: Economic Sector Representation for the Mid-Cap Indees 10 Figure 7: Economic Sector Representation for the Small-Cap Indees 13 Figure 8: Investability Profile for the Broad-Cap Indees 13 Figure 9: Investability Profile for the Broad-Cap Indees 13 Figure 10: Investability Profile for the Large-Cap Indees 13 Figure 11: Investability Profile for the Large-Cap Indees 13 Figure 12: Investability Profile for the Mid-Cap Indees 13 Figure 13: Investability Profile for the Mid-Cap Indees 13 Figure 14: Investability Profile for the Small-Cap Indees 13 Figure 15: Investability Profile for the Small-Cap Indees 16 Figure 16: Turnover Reported by Inde Providers and ETFs (%) 18 Figure 17: Cumulative Return for the Broad-Cap Indees 18 Figure 18: Cumulative Return for the Large-Cap Indees 18 Figure 19: Cumulative Return for the Mid-Cap Indees 18 Figure 20: Cumulative Return for the Small-Cap Indees 19 Figure 21: Broad-Cap Inde Volatilities for December 1994 December 2003 19 Figure 22: Annualized Daily and Monthly Volatility for All Market Segments 20 Figure 23: Sharpe Ratios for All Market Segments 1996-2003 20 Figure 24: Sharpe Ratios by Year for the Broad-Cap Indees 21 Figure 25: Sharpe Ratios by Year for the Large-Cap Indees 21 Figure 26: Sharpe Ratios by Year for the Mid-Cap Indees 21 Figure 27: Sharpe Ratios by Year for the Small-Cap Indees 22 Figure 28: Tracking Error of the Broad-Cap Indees vs. the U.S. Market 22 Figure 29: Tracking Error of the Broad-Cap Indees vs. the U.S. Market U.S. Stock Indees: Is There a Best Choice? 2
Summary Can the inde returns be achieved by a money manager? Ideally, an inde will be constructed so that it is possible for an inde-fund manager to achieve the inde returns. For this to happen, the stocks in the fund must be investable and the turnover of the fund must be reasonable. If an inde has many stocks with wide spreads between the bid price to buy shares and the asking price, it will be very difficult for the inde-fund manager to achieve the returns of the inde, because the manager will typically be buying at the ask and selling to the bid. Turnover also is an issue. The inde-fund manager must pay commissions and a portion of the bid ask spread every time a stock is added to or deleted from the inde. In this paper we compare market-cap-weighted indees provided by Dow Jones Indees, Frank Russell Company, Standard & Poor s Corp. and Wilshire Associates. Section I includes a brief discussion of how each of these inde families is constructed and reconstituted. Section II provides an evaluation of the indees with regard to their market capitalization and sector representation. Section III evaluates the investability of the constituent stocks in each inde, and Section IV assesses the historical turnover of each inde. Finally, Section V presents historical returns, volatility and Sharpe ratios for each of the indees. For the broad capitalization indees, the tracking error versus the U.S. market is presented. Throughout this paper, we refer to the specific inde names on numerous occasions, in the tet and in the charts and tables. For simplicity of presentation, we abbreviate the names of the indees. The full inde names of the indees that are included in this study and the abbreviations we use are provided below: Dow Jones U.S. Total Market Inde (Dow Jones TMI) Dow Jones U.S. Large-Cap Inde (Dow Jones Large) Dow Jones U.S. Mid-Cap Inde (Dow Jones Mid) Dow Jones U.S. Small-Cap Inde (Dow Jones Small) Russell 3000 Inde (Russell 3000) Russell 1000 Inde (Russell 1000) Russell 2000 Inde (Russell 2000) Russell Top 200 Inde (Russell Top 200) Russell Midcap Inde (Russell Midcap) U.S. Stock Indees: Is There a Best Choice? 3
S&P SuperComposite 1500 Inde (S&P 1500) S&P 500 Inde (S&P 500) S&P MidCap 400 Inde (S&P 400) S&P SmallCap 600 Inde (S&P 600) Wilshire 5000 Total Market Inde (Wilshire 5000) U.S. Stock Indees: Is There a Best Choice? 4
I. Inde Construction A brief discussion is provided in this section to describe the construction and reconstitution policies for each of the indees discussed in this paper. All of the indees are market-capitalization weighted. Dow Jones TMI The Dow Jones U.S. TMI includes three size-based sub-indees: Large-Cap, Mid-Cap and Small-Cap. The Dow JonesTMI currently includes approimately 1,600 stocks, and the inde is designed to capture approimately 95% of the investable U.S. market cap. Approimately 70% of the U.S. market cap is captured in the large-cap inde, 20% in the mid-cap inde and 5% in the small-cap inde. The market caps in the Dow Jones TMI are float adjusted, meaning that if another corporation or insiders own a portion of a company s stock, that portion is ecluded from the market cap when the weights are established. Buffer zones are used during the reconstitution process to reduce turnover. The buffer zones make it more difficult for a constituent not currently in the inde to be added and make it more difficult for an eisting constituent to be deleted. As a result, the first 67.5% of the float-adjusted U.S. market cap is large cap, 67.5% to 75% can be either large or mid, 75% to 85% is mid, 85% to 92.5% can be either mid or small, 92.5% to 95% is small, and 95% to 99% can either be small or not in the inde. Dow Jones reconstitutes its indees on a quarterly basis. Russell 3000 The Russell inde family includes the Russell 1000 and the Russell 2000. Russell ranks all of the U.S. echange-traded stocks by total market capitalization each year at the end of May. The top 1,000 stocks are included in the Russell 1000 and stocks with rankings of 1,001 to 3,000 are included in the Russell 2000. The largest 200 stocks in the Russell 1000 are placed in the Top 200 inde and the rest of the stocks in the Russell 1000 are placed in the Russell Midcap inde. For ranking purposes, the total market caps are used, but once the indees have been established, the weights of each stock in the inde are based on the floatadjusted market cap of the stock. Russell indees are reconstituted once per year on the last day of June, based on market caps at the end of May. S&P 1500 The S&P family of indees uses three indees to represent the U.S. market. The S&P 500, S&P 400 and S&P 600 consist of 500, 400 and 600 stocks respectively, selected to represent the large-, mid- and small-cap market segments. Together, the three indees make up the broad-cap S&P 1500 inde. S&P has a committee that selects the stocks for each inde. Market capitalization is not the only consideration. Committee members attempt to achieve a balanced representation of the various sectors, and they incorporate a profitability screen into their selection process. Stocks U.S. Stock Indees: Is There a Best Choice? 5
are deleted from the indees due to mergers and delistings and if the stock deviates significantly from the membership criteria. When a stock is deleted from one of the indees, a new stock is added to maintain a fied number of stocks in each inde. Stocks are added and deleted on an asneeded basis. Wilshire 5000 The Wilshire 5000 inde essentially consists of all of the echange-listed common stocks in the United States. The number of stocks in the inde varies; currently there are approimately 5,200 stocks in the inde. Since all common stocks are included, inde reconstitution is not an issue for this inde. Stocks of new IPOs are added to the inde at the end of each month and companies are removed when they merge with another company or are delisted from an echange. U.S. Stock Indees: Is There a Best Choice? 6
II. Size and Sector Representation A summary of some key statistics regarding the market capitalization of the inde constituents for the Dow Jones, Russell and S&P indees is provided in Figures 1 through 3. The broad-cap indees are the Dow Jones TMI, the Russell 3000, and the S&P 1500. The Dow Jones TMI and the S&P 1500 currently contain approimately the same number of stocks, with the Dow Jones TMI s stocks having a slightly higher average and median market capitalization. Unlike the S&P 1500, the number of components included in the Dow Jones TMI changes over time to ensure 95% coverage of the U.S. market. The Russell 3000 contains significantly more stocks, and the average and median market capitalization is much lower, since a greater proportion of the stocks in the Russell 3000 have a market cap of less than $1 billion. A couple of other observations regarding market capitalization follow: The Russell family of indees provides the most precise breaks in terms of market capitalization, particularly shortly after the annual inde reconstitution. The Dow Jones indees are also fairly precise, but overlap somewhat due to their buffer zones. The S&P large-, mid- and small-cap indees show the most overlap. The Dow Jones large-, mid- and small-cap indees all have a higher average market cap than the corresponding S&P indees. The Russell 2000 has the lowest average market cap of all of the small-cap indees. Figure 1: Summary Statistics for Dow Jones, Russell, S&P and Wilshire Indees (USD Mil.) Inde Number of Average Median High Low Indees Market Cap Stocks Market Cap Market Cap Market Cap Market Cap Dow Jones TMI 12,372,772 1,600 7,733 2,095 311,066 112 Dow Jones Large 9,042,604 247 36,610 18,548 311,066 5,914 Dow Jones Mid 2,337,573 513 4,557 3,951 12,049 1,121 Dow Jones Small 992,595 840 1,182 1,111 3,337 112 Russell 3000 12,698,628 2,940 4,319 793 311,066 42 Russell 1000 11,572,636 991 11,678 3,789 311,066 489 Russell 2000 1,125,992 1,949 578 469 2,064 42 Russell Top 200 8,439,984 201 41,990 21,953 311,066 8,369 Russell Midcap 3,132,652 790 3,965 3,044 17,024 489 S&P 1500 11,561,352 1,500 7,708 1,685 311,066 64 S&P 500 10,164,019 500 20,328 8,722 311,066 902 S&P 400 956,007 400 2,390 2,069 11,799 336 S&P 600 441,326 600 736 626 4,865 64 Wilshire 5000 13,236,905 5,242 2,525 253 311,066 0 Data as of December 31, 2003. All market cap values are estimates calculated with data provided by Dow Jones Indees. Data represents full market capitalization rather than float market capitalization. U.S. Stock Indees: Is There a Best Choice? 7
Figure 2: Number of Stocks in Various Market-Cap Ranges Market-Cap Ranges (USD Bil.) Indees >50 10-50 5-10 2-5 1-2 0.5-1 0.1-0.5 0.05-0.1 <0.05 Totals Dow Jones TMI 45 198 182 394 411 238 132 0 0 1,600 Dow Jones Large 45 185 17 0 0 0 0 0 0 247 Dow Jones Mid 0 13 165 295 40 0 0 0 0 513 Dow Jones Small 0 0 0 99 371 238 132 0 0 840 Russell 3000 44 190 173 396 479 623 1,019 15 1 2,940 Russell 1000 44 190 173 395 175 13 1 0 0 991 Russell 2000 0 0 0 1 304 610 1,018 15 1 1,949 Russell Top 200 44 154 3 0 0 0 0 0 0 201 Russell Midcap 0 36 170 395 175 13 1 0 0 790 S&P 1500 45 187 157 296 302 289 219 5 0 1,500 S&P 500 45 185 137 103 29 1 0 0 0 500 S&P 400 0 2 20 184 141 51 2 0 0 400 S&P 600 0 0 0 9 132 237 217 5 0 600 Wilshire 5000 46 196 180 409 486 649 1,473 525 1,278 5,242 Figure 3: Percentage of Inde s Total Market Capitalization in Various Market-Cap Ranges Market-Cap Ranges (USD Bil.) Indees >50 10-50 5-10 2-5 1-2 0.5-1 0.1-0.5 0.05-0.1 <0.05 Totals Dow Jones TMI 41.36 31.66 10.25 10.09 4.85 1.42 0.38 0.00 0.00 100.00 Dow Jones Large 56.59 41.79 1.62 0.00 0.00 0.00 0.00 0.00 0.00 100.00 Dow Jones Mid 0.00 5.94 47.96 43.22 2.88 0.00 0.00 0.00 0.00 100.00 Dow Jones Small 0.00 0.00 0.00 23.95 53.62 17.67 4.76 0.00 0.00 100.00 Russell 3000 39.88 29.66 9.51 9.75 5.38 3.50 2.32 0.01 0.00 100.00 Russell 1000 43.76 32.55 10.43 10.68 2.49 0.09 0.00 0.00 0.00 100.00 Russell 2000 0.00 0.00 0.00 0.18 35.09 38.51 26.10 0.11 0.00 100.00 Russell Top 200 60.00 39.69 0.31 0.00 0.00 0.00 0.00 0.00 0.00 100.00 Russell Midcap 0.00 13.31 37.71 39.45 9.19 0.32 0.02 0.00 0.00 100.00 S&P 1500 44.26 31.74 9.57 8.22 3.78 1.82 0.59 0.00 0.00 100.00 S&P 500 50.35 35.88 9.70 3.63 0.43 0.01 0.00 0.00 0.00 100.00 S&P 400 0.00 2.47 12.61 58.18 22.49 4.17 0.09 0.00 0.00 100.00 S&P 600 0.00 0.00 0.00 5.67 40.39 38.49 15.37 0.09 0.00 100.00 Wilshire 5000 39.43 29.48 9.46 9.62 5.23 3.49 2.82 0.29 0.18 100.00 Data as of December 31, 2003. All market cap values are estimates calculated with data provided by Dow Jones Indees. Data represents full market capitalization rather than float market capitalization. U.S. Stock Indees: Is There a Best Choice? 8
Bar charts are provided in Figures 4 through 7 to show the percentage of each inde by sector. The analysis utilizes the ten broad economic sectors of the four-tiered Dow Jones Global Classification Standard (DJGCS). As indicated in the bar charts, there is a fair amount of consistency with regards to economic sector breakdown within the broad-cap, large-cap and mid-cap segments, but there is more disparity within the small-cap segment. Figure 4: Economic Sector Representation for the Broad-Cap Indees 25% Broad-Cap Sector Breakdown, % 20% Dow Jones TMI Russell 3000 S&P 1500 Wilshire 5000 15% 10% 5% 0% Basic Materials Consumer, Cyclical Consumer, Noncyclical Energy Financial Health Care Industrial Technology Telecommunications Utilities Data as of December 31, 2003. All market cap values are estimates calculated with data provided by Dow Jones Indees. Data represents full market capitalization rather than float market capitalization. U.S. Stock Indees: Is There a Best Choice? 9
Figure 5: Economic Sector Representation for the Large-Cap Indees 25% 20% Large-Cap Sector Breakdown, % Dow Jones Large Russell Top 200 S&P 500 15% 10% 5% 0% Basic Materials Consumer, Cyclical Consumer, Noncyclical Energy Financial Health Care Industrial Technology Telecommunications Utilities Figure 6: Economic Sector Representation for the Mid-Cap Indees 25% 20% Mid-Cap Sector Breakdown, % Dow Jones Mid Russell Midcap S&P 400 15% 10% 5% 0% Basic Materials Consumer, Cyclical Consumer, Noncyclical Energy Financial Health Care Industrial Technology Telecommunications Utilities Figure 7: Economic Sector Representation for the Small-Cap Indees 30% 25% 20% Small-Cap Sector Breakdown, % Dow Jones Small Russell 2000 S&P 600 15% 10% 5% 0% Basic Materials Consumer, Cyclical Consumer, Noncyclical Energy Financial Health Care Industrial Technology Telecommunications Utilities Data as of December 31, 2003. All market cap values are estimates calculated with data provided by Dow Jones Indees. Data represents full market capitalization rather than float market capitalization. U.S. Stock Indees: Is There a Best Choice? 10
III. Investability As discussed in the introduction, the investability of an inde is important to investors attempting to replicate the inde or to use the inde as a universe. Broadly speaking, investability influences an investor s ability to achieve the returns of an inde. Investability bears directly on transaction costs. In an investable inde, the investor is able to establish positions in the inde constituents at a reasonable transaction cost. In a less investable inde, many of the inde constituents are costly to buy or sell. A portfolio manager will need to incur transaction costs upon initiating the positions in the fund, whenever there is a net inflow or outflow of money from the fund, and whenever stocks are added to or deleted from the inde. Perhaps the best measure of investability would use the effective bid-ask spread for the inde constituents. However, the true bid-ask spread is not directly observable, and is also variable. We will use the percentage of the average daily volume as a proy for transaction costs. Past research has indicated that temporary price impact is a function of the trade size 1. The first step in the process of evaluating investability was to decide how much money would be invested in the inde. We decided to make the amount substantial, because a small investment spread across the large number of inde constituents will always be investable. Only a large investment would stress the system in a way that would give us meaningful results. We also decided to vary the amount invested by market segment, with broadcap indees receiving the largest investment. Finally, we made the investments in the large, mid and small-cap segments add up to the investment in the broad-cap inde, with the portions approimating the percentage of the market that the segments represent. The investment amounts were: Broad-Cap Segment: $1 Billion Large-Cap Segment: $700 Million Mid-Cap Segment: $200 Million Small-Cap Segment: $100 Million To conduct the investability comparison, we distributed the fied investment in the inde to each inde constituent based on its weight in the inde. This dollar amount was then divided by the constituent s closing price on December 31, 2003, to determine the number of shares that would need to be purchased. Then we compared the number of shares being purchased to the average daily volume for the prior si-month period, to see what percentage of the average daily volume of each stock was being purchased. If the percentage of the average daily volume was low, the shares would be considered easy to purchase, while if the percentage was high the shares would 1 For eample in Keim & Madhavan s 1996 paper in The Review of Financial Studies they estimated a regression of temporary price impacts on traded size and powers of trade size, with trade size defined as the number of shares traded divided by shares outstanding. They found the temporary impact was significantly related to the trade size. U.S. Stock Indees: Is There a Best Choice? 11
be difficult to purchase. Note that all of these percentages can be scaled up or down by choosing an alternative amount to invest, and that we are making no judgment regarding what percentage of average daily volume (ADV) will create a market impact. This analysis is simply a way to evaluate relative investability between indees. For each inde we calculated a percent of ADV for each inde constituent. We ranked these percentages and calculated the cumulative weights for each constituent (the weight of that constituent plus the weights of all easier-topurchase constituents). For Figures 8, 10, 12 and 14 we plotted the cumulative weights as a function of %ADV, where the investment in each inde is as shown above. For clarity we plotted only 13 points for each inde. This analysis may tend to favor an inde with a larger number of stocks, so we conducted an additional investability analysis. In the second analysis, the amount invested in the inde with the smallest number of stocks is the same as before, but the amount invested in the other two indees in each segment is scaled up such that the average invested in each stock is the same for all indees in a segment. The results of the second analysis are shown in Figures 9, 11, 13 and 15. In the first analysis we were measuring the investability of the indees and in the second analysis we were measuring the investability of the stocks in the indees. These plots could be thought of as cumulative probability distributions, where the y-ais value is the probability that $1 invested in the inde will be invested in a stock where the purchase represents less than the %ADV shown on the -ais. The plot with the steepest slope is the most investable, because for the same level on the y-ais, say 0.9, the steeper curve will result in a lower %ADV on the -ais. The most investable inde also appears furthest to the left and above the others. Based on how this first analysis is structured, all else equal, an inde with a larger number of stocks will be more investable than an inde with a smaller number of stocks. For eample the S&P 500 is the most investable inde in Figure 10, but when the investment amount is scaled up in Figure 11 to account for the larger number of stocks in the S&P 500, it becomes the least investable. The same phenomenon is apparent in Figures 12 and 13, where the Russell Midcap goes from being the most investable to the least investable. However, with the small-cap indees a different pattern is apparent. The Dow Jones Small-Cap inde appears as the most investable inde in Figure 14 and 15. As more dollars are added to the Dow Jones Small-cap inde in Figure 15 the graphs for the Dow Jones and S&P indees move closer U.S. Stock Indees: Is There a Best Choice? 12
together but the Dow Jones Small-Cap maintains its advantage. The same pattern holds when comparing the Dow Jones Mid-Cap and the S&P 400 in Figures 12 and 13 (Dow Jones Mid-Cap is more investable in both). Figure 1 of this report indicates that the Dow Jones Indees have higher average and median market caps than the S&P indees. The primary reason why the Dow Jones indees are more investable is likely that a higher range of market capitalization is being targeted with Dow Jones versus S&P. It is fairly well accepted that larger capitalization stocks are typically more liquid than small capitalization stocks. Figure 8: Investability Profile for the Broad-Cap Indees Figure 9: Investability Profile for the Broad-Cap Indees Using Same Dollar Allocation for Each Inde Dollar Allocation Scaled to the Number of Stocks in the Inde 1.2 1.0 Dow Jones TMI Russell 3000 S&P 1500 1.2 1.0 Dow Jones TMI Russell 3000 S&P 1500 Cumulative Probability 0.8 0.6 0.4 Cumulative Probability 0.8 0.6 0.4 0.2 0.2 0.0 0% 1% 2% 3% 4% 5% 6% % ADV 0.0 0% 2% 4% 6% 8% 10% 12% % ADV Figure 10: Investability Profile for the Large-Cap Indees Figure 11: Investability Profile for the Large-Cap Indees Using Same Dollar Allocation for Each Inde Dollar Allocation Scaled to the Number of Stocks in the Inde 1.2 1.0 Dow Jones Large Russell Top 200 S&P 500 1.2 1.0 Dow Jones Large Russell Top 200 S&P 500 Cumulative Probability 0.8 0.6 0.4 Cumulative Probability 0.8 0.6 0.4 0.2 0.2 0.0 0% 2% 4% 6% 8% 10% 12% % ADV 0.0 0% 2% 4% 6% 8% 10% 12% % ADV Data as of December 31, 2003. All market cap values are estimates calculated with close, shares and float factor data provided by Dow Jones Indees. Data calculated using average daily volume figures from July 1, 2003 to December 31, 2003. U.S. Stock Indees: Is There a Best Choice? 13
Figure 12: Investability Profile for Mid-Cap Indees Figure 13: Investability Profile for Mid-Cap Indees Using Same Dollar Allocation for Each Inde Dollar Allocation Scaled to the Number of Stocks in the Inde 1.2 1.2 1.0 1.0 Cumulative Probability 0.8 0.6 0.4 Dow Jones Mid Russell Mid S&P 400 Cumulative Probability 0.8 0.6 0.4 Dow Jones Mid Russell 2000 S&P 600 0.2 0.2 0.0 0% 2% 4% 6% 8% 10% 12% 14% % ADV 0.0 0% 2% 4% 6% 8% 10% 12% 14% % ADV Figure 14: Investability Profile for Small-Cap Indees Figure 15: Investability Profile for Small-Cap Indees Using Same Dollar Allocation for Each Inde Dollar Allocation Scaled to the Number of Stocks in the Inde 1.2 1.2 1.0 1.0 Cumulative Probability 0.8 0.6 0.4 Dow Jones Small Russell 2000 S&P 600 Cumulative Probability 0.8 0.6 0.4 Dow Jones Small Russell 2000 S&P 600 0.2 0.2 0.0 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% % ADV 0.0 0% 5% 10% 15% 20% 25% 30% 35% 40% % ADV Data as of December 31, 2003. All market cap values are estimates calculated with close, shares and float factor data provided by Dow Jones Indees. Data calculated using average daily volume figures from July 1, 2003 to December 31, 2003. U.S. Stock Indees: Is There a Best Choice? 14
IV. Turnover Inde investors typically desire long-term eposure to a market or sector. Turnover is not desirable for inde investors or their portfolio managers, primarily because turnover results in costs (brokerage fees, bid-ask spreads, administrative costs, etc.) and ultimately reduces returns. Turnover is difficult to compare among the major inde families, because each inde provider uses a different method of calculation. If the three methods of calculation were applied to any of the indees considered in this study, the S&P method would yield the highest turnover, the Russell the second highest and the Dow Jones method the lowest. Figure 16 provides a tabular summary of the turnover that has been reported by the providers of the Dow Jones, Russell and S&P indees. We have also included turnover data reported by the echange-traded funds tracking the Dow Jones, Russell, and S&P indees. Definitive statements regarding comparative turnover cannot be made due to the different methods of calculation. Interestingly, the reported turnover of the S&P indees more closely approimates the reported turnover of their corresponding echange traded funds (ETFs). It appears that S&P s turnover calculation may more closely reflect the turnover of the ETFs linked to them. All of the broad-cap and large-cap indees have turnover percentages in the single digits, while all of the mid-cap and small-cap indees have average turnover percentages in the double digits. The broad-cap and large-cap indees have lower turnover for a couple of reasons. First, a stock can only enter or leave the inde from one direction. A growing stock will enter the inde and a shrinking stock will leave the inde. With the mid- and smallcap indees, stocks can leave the inde from the upper border or the lower border. A stock can grow into the inde or out of the inde. Likewise, a stock can shrink into the inde or out of the inde. Second, smaller stocks tend to be more volatile than larger stocks. Each of the inde families uses different methods to control turnover in its indees. While their individual approaches to minimizing turnover vary, the actual turnover results are fairly close between these three inde families. Russell reconstitutes its indees once per year. Frank Russell Company produced a paper entitled Russell Indees, Eamining the Frequency of U.S. Reconstitution (Gardner, et al., 2001) which evaluated the projected differences in turnover if its indees were reconstituted quarterly or semiannually, instead of annually. The author found that the turnover would be more than 30% higher for the Russell 1000 and more than 60% higher for the Russell 2000 with quarterly reconstitution. U.S. Stock Indees: Is There a Best Choice? 15
Figure 16: Turnover Reported by Inde Providers and ETFs (%) Dow Dow Dow Dow Jones S&P Russell Jones S&P Russell Jones S&P Jones S&P Russell Year TMI 1500 3000 Large 500 1000 Mid 400 Small 600 2000 1992 1.13 NA NA 0.66 1.18 NA 9.69 5.84 15.41 NA NA 1993 1.05 NA NA 0.84 2.64 NA 18.32 10.32 23.20 NA NA 1994 2.16 NA NA 1.99 3.78 NA 17.68 9.89 27.00 NA NA 1995 2.46 5.97 NA 2.43 5.00 NA 10.73 15.57 22.32 13.73 NA 1996 2.72 5.34 NA 1.77 4.58 NA 14.59 14.36 24.86 16.37 NA 1997 1.95 5.36 NA 2.18 4.92 NA 11.81 17.91 14.55 21.84 NA 1998 2.76 8.81 NA 2.82 9.46 NA 20.02 31.38 23.23 24.38 NA 1999 2.97 5.47 1.60 3.41 6.16 2.60 28.85 28.87 29.88 24.39 24.10 2000 7.68 7.67 3.30 12.30 8.91 4.00 55.74 36.41 55.40 36.41 32.50 2001 1.60 4.25 1.20 4.05 4.43 2.40 12.65 16.98 17.30 15.63 20.60 2002 1.16 3.69 1.30 1.61 3.82 2.60 11.71 10.72 9.47 10.99 16.80 2003 0.92 1.79 2.30 1.36 1.45 2.70 10.27 8.60 9.60 10.98 20.50 Average 2.51 5.37 1.94 3.10 4.69 2.86 18.51 17.24 23.87 19.41 22.90 ETF Avg 5.00 NA 5.03 NA 5.15 7.33 NA 22.39 NA 20.33 29.67 Turnover for the Dow Jones indees was higher than normal in 2000 because float-adjusted weights were implemented in the indees that year. Dow Jones Indees reconstitutes its indees on a quarterly basis, but uses a buffer zone to reduce turnover (See Section I for details). The buffer zone makes it more difficult for a stock that is not currently in the inde to be added to the inde and also makes it more difficult for a stock that is currently in the inde to be deleted from the inde. The buffer zones have historically been an effective way of reducing turnover, while still allowing for a more frequent reconstitution process. Standard and Poor s relies on a committee to make decisions to add or delete stocks from its indees. Stocks are added and deleted from the indees throughout the year, not on any specific schedule. The S&P inde committee controls the turnover by requiring that a stock significantly deviate from the membership criteria before it is deleted. U.S. Stock Indees: Is There a Best Choice? 16
V. Historical Returns Another way to compare the indees is simply to look at the historical returns. In this section, we compare historical returns, volatilities, and Sharpe ratios. For the broad-cap indees we also evaluate how well each of the indees tracks the U.S. market as a whole, using the Wilshire 5000 as a proy for the U.S. market. Returns: Cumulative returns for the broad-cap, large-cap, mid-cap and small-cap indees are plotted on Figures 17, 18, 19, and 20 respectively for the period December 31, 1994, through December 31, 2003. All of the indees were indeed to a value of 100 on December 31, 1994 for comparison purposes. As seen in Figures 17 and 18, the returns for the broad-cap and large-cap indees have been very close throughout the period studied. There is more of a separation in returns in the mid- and small-cap indees than in the broad- and large-cap indees. Higher mid-cap returns for the S&P 400 and lower small-cap returns for the Russell 2000 were observed in the 1994-2003 period. Two papers authored by Peter Jankovskis have addressed this phenomenon. Another paper, authored by Quinn and Frank Wang, also sheds some light on the under-performance of the Russell 2000. Jankovskis s paper on small-cap indees contends that the Russell 2000 inde under-performs the S&P 600 inde, due to the annual Russell reconstitution. The Russell 2000 historically under-performs in July and August, which is when the ecess returns for the additions to the inde are typically negative. Quinn and Wang s paper uses event studies to quantify the ecess returns following the reconstitution for stocks added to the Russell 2000, and found that the return of the Russell 2000 is adversely impacted by the reconstitution. This significant impact is attributed to the large number of illiquid stocks in the Russell 2000. Jankovskis s paper on mid-cap indees seeks to eplain why the S&P 400 outperformed mid-cap indees from Russell and Wilshire for the period 1997-2001. Jankovskis does an attribution analysis for sector representation, size, and value/growth, and finds that these factors eplain only a small portion of the S&P 400 s superior returns. He finds that the positive ecess returns that S&P 400 stocks eperience immediately prior to leaving the inde and moving to the S&P 500 inde eplain most of the difference in performance between the S&P 400 and the Russell and Wilshire mid-cap indees. U.S. Stock Indees: Is There a Best Choice? 17
Figure 17: Cumulative Return for the Broad-Cap Indees Figure 18: Cumulative Return for the Large-Cap Indees 350 December 1994 December 2003 400 December 1994 December 2003 300 250 Dow Jones TMl Russell 3000 S&P 1500 Wilshire 5000 350 300 Dow Jones Large Russell Top 200 S&P 500 250 200 150 100 400 350 300 12/94 06/95 12/95 06/96 12/96 06/97 12/97 06/98 12/98 06/99 12/99 06/00 12/00 06/01 12/01 06/02 12/02 06/03 12/03 Figure 19: Cumulative Return for the Mid-Cap Indees Dow Jones Mid Russell Midcap S&P 400 December 1994 December 2003 200 150 100 300 260 220 12/94 06/95 12/95 06/96 12/96 06/97 12/97 06/98 12/98 06/99 12/99 06/00 12/00 06/01 12/01 06/02 12/02 06/03 12/03 Figure 20: Cumulative Return for the Small-Cap Indees Dow Jones Small Russell 2000 S&P 600 December 1994 December 2003 250 200 150 100 12/94 06/95 12/95 06/96 12/96 06/97 12/97 06/98 12/98 06/99 12/99 06/00 12/00 06/01 12/01 06/02 12/02 06/03 12/03 Data based on price-return inde values from December 31, 1994 to December 31, 2003. 180 140 100 12/94 06/95 12/95 06/96 12/96 06/97 12/97 06/98 12/98 06/99 12/99 06/00 12/00 06/01 12/01 06/02 12/02 06/03 12/03 U.S. Stock Indees: Is There a Best Choice? 18
Volatility: Average volatilities are etremely close between indees within each of the market segments. The volatilities of the indees change quite a bit over time, but the changes are synchronized. Figure 21 depicts volatility over time for the broad-cap indees. The individual lines in this graph are almost indistinguishable. Figure Figure 21: Broad Cap Inde Volatilities for Dec 1994 Dec 2003 22 is a table that summarizes volatility over the period Dec-ember 1996 to December 2003. This table shows that some of the volatility may be Dow Jones Large Russell Top 200 lost when monthly returns are used to calculate S&P 1500 volatility instead of daily returns. 45% 40% 35% 30% 25% 0% 12/94 06/95 Willshire 5000 12/95 06/96 12/96 06/97 12/97 20% 15% 10% 5% 06/98 12/98 06/99 12/99 06/00 12/00 06/01 12/01 Monthly volatility calculated on a daily basis using price-return data. 06/02 12/02 06/03 12/03 Figure 22: Annualized Daily and Monthly Volatility for All Market Segments Volatility (%) Cap Range Indees Daily Monthly Broad-Cap S&P 1500 18.51 16.17 Wilshire 5000 18.39 16.51 Dow Jones TMI 18.83 16.61 Russell 3000 18.46 16.29 Large-Cap S&P 500 18.80 16.29 Russell 1000 18.68 16.35 Dow Jones Large 19.56 16.92 Russell Top 200 19.46 16.85 Mid-Cap Russell Midcap 18.10 16.80 S&P 400 19.23 18.01 Dow Jones Mid 17.93 17.06 Small-Cap S&P 600 18.66 19.19 Russell 2000 19.30 20.46 Dow Jones Small 19.37 20.65 Daily volatility values calculated using daily price-return inde values from December 31, 1994 to December 31, 2003. Monthly volatility values calculated using monthly price-return inde values from December 31, 1994 to December 31, 2003. U.S. Stock Indees: Is There a Best Choice? 19
Sharpe Ratio: The Sharpe ratio of a stock inde is the return of the inde minus the risk-free rate of return, divided by the volatility of the inde. The Sharpe ratio is often used to compare investments with different levels of risk (for which volatility is a proy). Investors seek investments with high Sharpe ratios. Figure 23 presents the Sharpe Ratios for all of the indees for the period December 1996 through December 2003. All of the Sharpe ratios within each segment are very close with the eception of the S&P 400 s higher ratio and the Russell 2000 s lower ratio. This is consistent with the observations made in the comparison of returns. Figures 24 through 27 present the Sharpe ratios for the period from 1996 to present. Sharpe ratios of less than 0 are not meaningful, so if the inde return was negative for a period, the Sharpe ratio is not reported on the chart. None of the indees consistently showed higher Sharpe ratios than the others over the period. Figure 23: Sharpe Ratios for All Market Segments 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 December 1996 December 2003 Dow Jones Russell S&P 0.00 Broad-Cap Large-Cap Mid-Cap Small-Cap Figure 24: Sharpe Ratios by Year for Broad-Cap Indees 2.00 1.80 1.60 1.40 December 1996 December 2003 Dow Jones TMI Russell 3000 S&P 1500 Wilshire 5000 1.20 1.00 0.80 0.60 0.40 0.20 0.00 1996 1997 1998 1999 2000 2001 2002 2003 All data based on price-return inde values. Sharpe Ratios calculated using the 3-month Treasury Bill. U.S. Stock Indees: Is There a Best Choice? 20
Figure 25: Sharpe Ratios by Year for Large-Cap Indees 1.60 1.40 1.20 December 1996 December 2003 Dow Jones Large Russell Top 200 S&P 500 1.00 0.80 0.60 0.40 0.20 0.00 1996 1997 1998 1999 2000 2001 2002 2003 Figure 26: Sharpe Ratios by Year for Mid-Cap Indees 2.50 December 1996 December 2003 2.00 Dow Jones Mid Russell Mid S&P 400 1.50 1.00 0.50 0.00 1996 1997 1998 1999 2000 2001 2002 2003 Figure 27: Sharpe Ratios by Year for Small-Cap Indees 3.00 2.50 2.00 December 1996 December 2003 Dow Jones Small Russell 2000 S&P 600 1.50 1.00 0.50 0.00 1996 1997 1998 1999 2000 2001 2002 2003 All data based on price-return inde values. Sharpe Ratios calculated using the 3-month Treasury Bill. U.S. Stock Indees: Is There a Best Choice? 21
Tracking Error: Tracking error is a measure of how closely a portfolio tracks an inde. A low tracking error indicates that the portfolio tracks the inde very closely. We used tracking error to determine how well the three broadcap indees tracked the U.S. market. We used the Wilshire 5000 returns as a proy for the U.S. market returns, and compared the Wilshire 5000 returns with the Dow Jones TMI, the Russell 3000, and the S&P 1500. Figure 28 depicts the annual and period tracking errors for these three indees versus the Wilshire 5000, and Figure 29 displays the underlying data. The tracking errors for all three of these indees against the U.S. market are low. Figure 28: Tracking Error of the Broad-Cap Indees vs. the U.S. market 7.00% Tracking Error vs. the Wilshire 5000 6.00% 5.00% Dow Jones TMI Russell 3000 S&P 1500 4.00% 3.00% 2.00% 1.00% 0.00% 1996 1997 1998 1999 2000 2001 2002 2003 1996-2003 Figure 29: Tracking Error of the Broad-Cap Indees vs. the U.S. Market (%) Period Dow Jones TMI Russell 3000 S&P 1500 1996-2003 1.96 3.22 2.61 1998-2003 1.95 3.56 2.70 2000-2003 1.68 4.09 2.74 1996 1.37 1.10 1.67 1997 2.47 2.40 2.78 1998 2.67 2.58 3.00 1999 2.11 1.58 2.17 2000 2.95 4.95 5.04 2001 0.89 6.43 1.57 2002 1.15 0.90 1.25 2003 0.69 0.56 0.91 All tracking error values calculated against the Wilshire 5000. All data calculated using daily price-return inde values. U.S. Stock Indees: Is There a Best Choice? 22
VI. Conclusions U.S. stock indees calculated by Dow Jones Indees, Frank Russell Company, Standard and Poor s, and Wilshire Associates were eamined from a number of perspectives. There are very few differences among the indees competing within a given market segment. This is especially true for the broad-cap and large-cap indees. A few additional conclusions emerge from the analysis: In terms of inde composition, the Dow Jones indees tended to contain higher market cap stocks than the Russell and S&P indees. This difference probably contributed to Dow Jones s better results on the investability measure. The Dow Jones, Russell, and Standard and Poor s inde families have significantly different composition requirements and reconstitution policies; however, the turnover appears to be somewhat similar within a given market size segment. The S&P 400 had the best returns in the mid-cap segment, and the Russell 2000 had the lowest returns in the small-cap segment. Otherwise, all of the indees within a given segment had very close returns, especially considering the high volatility of returns. The volatilities for all of the indees were similar. References Gardner, Grant, A. Kondra, and M. Pritamani. Eamining the Frequency of U.S. Constitution. Working Paper. Frank Russell Co. 2001. Jankovskis, Peter. The Impact of Russell 2000 Rebalancing on Small-Cap Performance. Journal of Investment Consulting, vol. 5, no. 2 (2002): 23-29 Jankovskis, Peter. The Returns of the S&P 400: Implications for Active Mid-Cap Managers. Working Paper. OakBrook Investments. Keim, Donald, and A. Madhavan. The Upstairs Market for Large-Block Transactions: Analysis and Measurement of Price Effects. The Review of Financial Studies, vol. 9, no. 1 (1996): 1-36. Quinn, Jim, and F. Wang. How is Your Reconstitution, Has Rebalancing Become a Drag? Journal of Indees, 4th Quarter 2003: 34-38 U.S. Stock Indees: Is There a Best Choice? 23
About the Author Jim Quinn initiated his research on inde investing in September 2002. He conducted a substantial research effort regarding U.S. indees while completing his Masters degree in Financial Engineering at Haas Business School (UC Berkeley), which he completed in March 2003. He published a portion of that research with a colleague, Frank Wang, in the November 2003 issue of the Journal of Indees. Mr. Quinn conducted the research that comprises this paper initially during the summer of 2003, and he presented those results at the Institutional Capital Investing Forum in Washington, D.C. in October 2003. The results were updated through the end of 2003 and are presented in this paper. Jim recently joined Gifford Fong Associates, a financial modeling firm in Lafayette, CA and is managing several research projects there. He also recently launched the San Francisco chapter of QWAFAFEW (Quantitative Work Alliance for Applied Finance Education and Wisdom) and serves as the chair for that group. The author acknowledges Dow Jones Indees, who provided the data for this project. Thanks are particularly due to Bernadette O Sullivan, John Prestbo, and Richard Ciuba at Dow Jones Indees for their help and support throughout the project. U.S. Stock Indees: Is There a Best Choice? 24