|
 Large- and small-cap managers existed long before the creation of size indexes and the confirmation of the size effect. Goals for an index provider such as the Russell Investment Group (formerly the Frank Russell Company) include creating an investment style benchmark, an asset allocation tool and an implementation methodology for passive investing purposes.1 By not recognizing the transition from large- to small-cap and small- to micro-cap, and by backing into a default definition of small-cap, the Russell 2000 Index has missed the mark for each of these goals.
Investment Style Benchmark
One benchmark goal articulated by the Chartered Financial Analyst (CFA) Institute is that the index benchmark be appropriate given a manager's investment style. Appropriateness is more than matching the style name. Ideally, it is matching the neutral position or passive return of the measured investment strategy. The CFA Institute Subcommittee on Benchmarks and Performance Attribution emphasized the point that "Choosing a bad or inappropriate benchmark can undermine the effectiveness of an investment strategy and lead to dissatisfaction between client and manager."2
In the mid-1980s, Russell research of manager data showed that, "Investment managers select not from hundreds of stocks but between 3,000 and 3,200 issues."3 Further surveys showed that managers who viewed themselves as large-cap managers purchased stocks down to a market-cap rank of about 1,000.
This led Russell to create the Russell Indexes that were "as simple as 1, 2, 3" with the top 1,000 stocks labeled as large-cap, the next 2,000 stocks labeled as small-cap and the combined 3,000 defined as the investable universe (see Figure 1).
Conceptually reasonable but logically flawed, Russell treated the 2,000 small-cap as the difference between its large-cap 1,000 and its investable universe 3,000. What Russell missed was that while large-cap managers stopped buying near the 1,000th market-cap-ranked company, going the other direction, small-cap managers stopped buying near the 500th market- cap-ranked company. Micro-cap managers stopped buying near the 2,000th market-cap-ranked company. This is a critical evaluation issue for how small-cap managers actually invest.
In the early 1980s, Wilshire Associates' related studies with variance analysis and performance attribution led to the observation that the performance differences between large-cap and small-cap shifted between the 500th and 1,000th market-cap-ranked stock. Years of style metric testing, along with affirmation from institutional holdings and money managers, confirmed these attribution-driven results. These results led to Wilshire's 1986 release of style indexes that divided the Wilshire 5000 (now the Dow Jones Wilshire 5000) into large-, small- and micro-cap indexes divided at the 750th and 2,500th market-cap-ranked stocks, respectively.4
Years later, Russell research showed "that many small-cap growth managers tend to let their winners run."5 Active money managers are reluctant to sell their "winners." This led Russell to create the Russell 2500 Index in 1993 to cover the stocks ranked 501 to 3,000. The release of the Russell 2500 supported the argument that the Russell 2000 Index started too low in the market-cap range for a true representation of small-cap managers.
In June 2005, Russell launched the Russell Microcap Index. This index covered the market-cap stocks ranked from 2,001 to 4,000, extending Russell's coverage, but it did not extend the coverage of their core indexes. The Russell Microcap Index's overlap with the Russell 2000 and Russell 2500 Indexes raised questions about Russell's definition of small-cap. Russell admits to receiving a fair amount of academic criticism because of the overlap with its existing small-cap indexes (see Figure 2). Russell defends this overlap by saying that it covers the micro-cap opportunity set and that small-cap managers invest down to 3,000.6
According to Russell, its analysts noticed that, "the Russell 2000 Index did not fit all of the small-cap investment products available in the marketplace. Many managers drew on somewhat large-cap stocks, while a new (and growing) class of managers focused almost exclusively on the smallest end of the spectrum."7 Russell further points out that, while use of the Russell 2500 and style indexes continue to grow, use of the "Russell 2000 has remained mostly flat."8
Russell's staff seems comfortable with index overlaps as shown by its style index methodology. Russell's error is that, "[t]he Russell indexes do not attempt to refine the market by excluding companies in the opportunity set."9
Wilshire believes that the DJ Wilshire 5000 Index10 includes all companies in the opportunity set. By definition, dividing the market into subsets requires excluding companies. Wilshire's equity research recognized that large-cap companies ranked beyond 500 began the transition to small-cfap and stopped having large-cap attributes near 1,000, whereas the reverse was true for small-caps. (Wilshire defines this transition-overlap range as mid-cap stocks.) The same type of size transition was recognized between stocks ranked 2,000 and 3,000 for small-cap and micro-cap. Russell did not acknowledge this transition until the release of their 2,500 small-cap and micro-cap indexes. For more than 20 years, Wilshire research has continued to support the 750 and 2,500 break points.11 Figure 3 shows Russell's combined index representation of the U.S. equity market supports the size break points at 750 and 2,500.
Note that independent of each other and years before Russell's 2500 and Micro-cap indexes, Wilshire size indexes perfectly bisect the Russell defined overlap space.
Holdings coverage research has continued to show results similar to those of Figure 4. Small-cap style money manager investments not included in the 750–2,500 small-cap market-cap range are balanced above and below these size breaks. With smaller security counts (1,750 versus 2,000), higher portfolio percentage covered, and balanced outliers, market- cap breaks at 750 and 2,500 hit the bull's eye for small-cfap and small-cap style benchmarks. The evidence indicates that Russell's break points are too low.
Asset Allocation Tool
Another benchmark goal articulated by the CFA Institute is that the index be "representative of the asset class or mandate."12 This is important for risk budgeting, asset allocation, and attribution. According to William Sharpe:
The usefulness of an asset class factor model depends on the asset classes chosen for its implementation. While not strictly necessary, it is desirable that such asset classes be 1) mutually exclusive, 2) exhaustive and 3) have returns that 'differ.' Pragmatically, each should represent a market-capitalization weighted portfolio of securities; no security should be included in more than one asset class; as many securities as possible should be included in the chosen asset class…13
The Russell 2500 and Russell Microcap Indexes are eliminated as tools for asset allocation because they are not mutually exclusive. This leads to a dilemma for Russell index users. On the one hand, they can restrict their asset allocation research to only the top 3,000 stocks at the cost of not being exhaustive. (This would allow users to differentiate large-cap from small-cap by ignoring the additional 1,000 Russell and 2,000 additional DJ Wilshire stocks.) On the other hand, users could select the nearly 4,000-stock Russell 3000E Index14 but would have to use it as a single asset class.
Method For Passive Investment
Passive investments compared with active money management are supposed to represent the neutral investment style at little or no cost. Cost is comprised of not just management fees but transaction and implementation costs as well. It has been estimated that due to arbitrage around index changes, "investors in Russell 2000-linked funds [annually] lose between 1.30% and 1.84%."15 This arbitrage cost amounts to more than the fees of most active managers.
This Russell 2000 performance lag is supported by the 10-year performance shown in Figure 5 as well as the significant number of negative one-standard-deviation events that occur after Russell's annual rebalance versus those of other small-cap indexes shown in Figure 6.
Conclusion
The Russell 2000 Index was created with a fundamental misinterpretation of how small-cap money managers invest. Russell did not recognize the overlap or transition between the sizes. If Russell did, it perhaps would have chosen a value in the transition area when it created the large- and small-cap indexes. Russell should have corrected this error by moving the market-cap cutoff from 1,000 and 3,000 to numbers closer to 750 and 2,500. Instead Russell created new indexes extending the reach with an overlap of its existing indexes. This has led the Russell 2000 Index to be a flawed small-cap index as a performance standard for active managers, a questionable tool for asset allocation purposes and an expensive implementation method for passively capturing small-cap performance. [Editor's Note: For a response from Russell, please turn to page 57.]
|
