Active Vs. Passive - JOI Articles

January / February 2009
Markets in Crisis

IN THIS ISSUE

EDITORIAL BOARD

VIEW ISSUE AS PDF

Articles

Active Vs. Passive

Written by David Blanchett and Craig Israelsen

Thursday, 01 January 2009 00:00

Page 1 of 5

In a previous article (see Blanchett and Israelsen [2007]), the authors conducted an analysis of the benefits of active management to correct for a variety of common shortcomings in previous studies and to address the findings of a previously published study by Holmes [2007]. In this article, the authors update their analysis to reflect the impact of mutual fund expense ratios, or fees, by analyzing the returns of active managers on a gross basis (i.e., by adding back the expense ratio).

Determining the net benefit of active management on a gross fee basis is important since different investors pay different levels of management fees, which are typically inversely related to investable assets. Most previous studies, commonly based on mutual funds, use returns for active managers that are reduced by management fees. Since institutional investors tend to pay less than smaller, retail investors, the decision of whether to invest actively or passively involves different considerations at different asset levels. The purpose of this study is to investigate the benefits of active management before fees, in order to determine at what breakpoints active management does in fact lead to greater returns than passive investing.

Previous Research

Similar to past studies, the previous article published by the authors (Blanchett and Israelsen [2007]) found that active management tended to under-perform a respective index net of fees. However, the authors are surely not the first, or the last, to weigh in on this great debate. There have been numerous studies completed attempting to determine the benefit of active management (see, for example, Davis [2001]; Arnott, Berkin and Ye [2000]; Sorensen, Miller and Samak [1998]; Carhart, [1997]; Gruber [1996]; Malkiel [1995]; or Brinson, Hood and Beebower [1995], Carosa [2005], Holmes [2007]).

When viewing the markets as a whole (i.e., a zero-sum game), it only makes sense that in the aggregate, all active money managers should under-perform the market by the total fees associated with active investing, both explicit and implicit. However, there has been sound empirical and/or "behavioral" argument in favor of actively managed funds (Timbers [1997]).

While the aforementioned articles have each provided insight as to the benefit of active management, the authors have found four common problems in active and passive studies, each affecting prior research in varying degrees.

1. Potential Survivorship Bias: A common method to create a sample population for analysis is to look back at all funds as of a current date (e.g., Dec. 31, 2004). This is a "reach-back" technique and is common in active versus passive studies due to its simplicity. While simple, this strategy exposes the data set to survivorship bias since only those funds in existence as of the "look-back date" are included in the analysis.

2. Head Count vs. Invested Assets: The majority of active versus passive studies have determined the aggregate benefit of active management as the average performance of the active managers in the study population. Such a strategy ignores the actual assets invested in each fund and consequently the net returns to investors. For example, an analysis based entirely on head count would equally consider a fund with assets of $10 billion and a fund with $1 million, despite the fact one fund has 10,000 times more assets than the other. Measuring active versus passive methodologies using total invested assets, rather than just a raw head count of funds, is a more correct approach, and has been previously employed (see Israelsen [2006]).

3. Culling The Data Set To Distinct Funds: Including multiple share classes of a given fund in active versus passive analysis is common since investors actively invest in different share classes. However, the primary difference in expenses related to varying share classes are the costs associated with distribution, which is not directly related to the question of whether or not active management adds value after fees. 12b-1s and other forms of revenue share represent fees for distribution, not active management. Also, a fund with more share classes would have a larger impact on the results in a pure head count analysis. A fund company with 14 share classes of the same fund would have 14 times the weight of a fund with only one share class. This is not a trivial issue. For instance, as of year-end 2007, there were 15,136 U.S. equity mutual funds in the Morningstar Principia database. This total includes all share classes. When employing the "distinct portfolios only" filter in Morningstar's Principia database (which selects only one share class of a multiple share class fund), the number of funds was reduced to 4,576, a reduction of nearly 70 percent.

4. Index Dependency: The index selected to represent passive management can have a substantial impact on whether or not active management successfully adds value, since the performance of different indexes can vary considerably. While it is the case that many index funds use an S&P index as their benchmark, using only S&P indexes to represent the performance of "passive management" is inaccurate and misleading. There are many other index providers (MSCI, Russell, etc.) and there is wide variance among the performance of various prominent equity indexes, as demonstrated previously by Carlson [1970] and Israelsen [2005]. As the reader will see from the conclusions, the benchmark index selected for any analysis can have an impact—sometimes considerable—on the results.

In summary, there are a variety of complicating factors that should be considered when attempting to determine the actual benefit realized by those who invest in active management. This paper attempts to address each of these issues.

1 | 2 | 3 | 4 | 5 |