This section provides information on the procedures used to select sets of large, seasoned
U.S. mutual funds over the period from 1985 through 1994.
All fund information was taken from an extensive historical database provided by Micropal, an international firm based in the U.K. that maintains a comprehensive database of collective investments throughout the world. The data utilized for this study include virtually all U.S. mutual (open-end) funds available for purchase by the general public during the period from 1980 through 1994.
Of particular importance: both "dead" and "alive" funds are included in the database. For example, a fund that was merged into another in 1987 and thus "disappeared" at that time is included for all months prior to the merger, although it does not exist (in its initial form) today. This characteristic of the database makes it possible to avoid survivor bias, which plagues many studies of fund performance. A surprising number of funds disappear each year (approximately 3% by some estimates), and such funds are typically those with poor performance. Hence measurement of the historic performance of samples of funds that have existed unchanged for several years will inevitably give results subject to an upward bias. Simply put, the average performance of funds that survive is almost certainly greater than that of all funds -- those that survive and those that do not. This study is not subject to this error.
Initially, the database was filtered to remove all funds classified by Micropal as falling into the following categories: money market funds, funds specializing in tax-favored securities (e.g. municipal bond funds), funds specializing in securities of specific industries (e.g. natural resource funds, utility funds), and funds specializing in the use of options. The remaining Micropal categories are:
From the remaining database, 100 funds were selected for inclusion in each of the ten years. Consider, for example, the funds chosen for the calendar year 1986. The selection process utilized data available in December, 1985 and could, in fact, have been performed during that month. First, the availability of historic returns was checked. Only funds with at least 60 months of such returns were retained. In this case, returns from December, 1980 through November, 1985 (inclusive) were required. We define funds meeting this criterion as being seasoned at the time.
The seasoned funds obtained in this manner were ranked, based on total net assets as of the end of November, 1985. The 100 largest funds were then selected to form the "LS100" set of funds for 1986.
This process was utilized for every year from 1985 through 1995, with one minor
exception. For the set of funds used in 1985, only 59 monthly returns were required, since
only returns from January 1980 onward were used in the study.
Since each set of LS100 funds was chosen independently, there were some changes in composition from year to year. On average, less than 14 of the 100 funds were replaced in the set from one year to the next. Not surprisingly, the funds that changed tended to be smaller than those that did not. On average, the value of new funds added in a year was 9.5% percent of the total value of all funds in the set.
While the LS100 funds are only a subset of the funds in the included categories (e.g. bond, stock, balanced, global and international), they represent a substantial portion of the dollars invested in such funds. The figure below shows the total number of such funds at each year-end. Due to the rapid increase in the number of U.S. mutual funds, our 100 funds represent a diminishing percentage of the total.
At year-end 1994, the LS100 funds represented less than 3% percent of the total number of bond, stock, balanced and international U.S. mutual funds. However, approximately 44% percent of the dollars invested in such funds were invested in these 100 funds. The figure below shows the ratio of total net assets of each year's LS100 funds to that of all funds in these categories.
Since the LS100 funds invest a significant proportion of the money in their sector of the U.S. mutual fund industry, their aggregate performance is of considerable interest. This is not to say that their performance is necessarily representative of that of the remainder of the industry.
The LS100 funds are clearly larger than the remaining funds. Among other things, this suggests that their expenses will be smaller, when expressed as a percentage of the assets under management. If performance before expenses is unrelated to size, then one would expect that the average performance net of expenses of the LS100 funds would be better than that of the remaining funds.
It is important to note that the funds selected for inclusion in the LS100 set in a given year are likely to have had better performance in prior years than those not selected, since investors tend to invest new assets in funds with better historic performance. If there is any tendency for performance to be consistent through time, this would suggest that the aggregate performance of the LS100 funds might be better than that of the remaining funds.
In sum, the performance of the LS100 is likely to be at least as good as that of the remaining funds, and may well be superior. If the latter were the case, such funds would be of particular interest, since an astute investor could identify (and invest in) such funds each year.
For all these reasons, the aggregate performance of the LS100 should be of substantial
interest.
To measure the performance of the average dollar invested in large, seasoned U.S. mutual funds, we utilize a value-weighted index of the LS100 funds. More specifically, we assume that at the end of each December, a portfolio of all 100 funds selected at the time is formed, with the dollars invested in proportion to the funds' total net assets at the end of the prior month (November). The resulting shares of each fund are held throughout the subsequent year. Whenever a fund provides a dividend or capital gains distribution, the money is reinvested in the fund in question. In effect, a truly buy-and-hold strategy is followed throughout each year. At the end of the year some rebalancing is required, but only to accommodate changes in the funds included in the LS100.
The total value of a given LS100 portfolio is computed at the end of each month from the prior December through the December of the year in question. The percentage change in this value for any given month constitutes the total return for the LS100 Index in that month.
This process is repeated for each year. The resulting set of monthly returns from January 1985 through December 1994 constitutes the set of returns for the LS100 fund index over the total period.
The same procedure is utilized for subsets of the LS100 -- each such portfolio is
formed at the end of each December, using relative net asset values as of the end of
November, with each position held through the following year, with the changes in the
value of the total portfolio used as measures of performance. Data Source Micropal, Inc. © For more information, please call
617-451-1585
