Sections

Active Management: The portfolio tax

C Thomas Howard argues that active equity fund managers are superior stock pickers but destructive portfolio managers, to the extent that stockpicking skill is completely wasted

Active US equity mutual funds underperform their respective benchmarks on average. Does this imply a lack of stock-selection skill? That is the conventional wisdom  but, as is often the case, the truth is much more complicated. 

A growing body of research challenges this conventional wisdom, revealing that stock-selection skill is common among US active equity mutual fund managers, allowing them to generate returns that more than cover fund fees. 

A 2013 paper by Jonathan Berk and Jules H van Binsbergen, for instance, concludes that 80 or more display stock-picking skill and that this skill persists for up to 10 years. I, along with others, argue that the observed underperformance is due to non-performance pressures and incentives that lead to building underperforming portfolios, rather than evidence of the lack of skill. Berk and RC Green, as well as RB Cohen, C Polk, and B Silli, have recently offered profit maximisation arguments for why funds choose to turn themselves into closet indexers and thus underperform. 

I refer to the reduction in fund performance resulting from these non-performance investment decisions as an implicit ‘portfolio tax’ imposed by destructive portfolio management decisions, many driven by the desire to grow large colliding with the requirements imposed by the US fund distribution system.  

A fund’s performance (alpha) can be thought of as the fund’s selection skill (SS) minus its portfolio tax (PT). Using a large sample of active US equity mutual fund returns, I estimate that SS averages nearly 5% annually (net of fees and benchmark return), while PT averages a bit higher, resulting in an across-the-board negative alpha. The impressive stock-picking skill of buy-side analysts is completely offset by the fund’s subsequent portfolio management decisions. The implicit portfolio tax is on average four times the explicit average fund fees (1.3%) and thus the most important destroyer of performance. 

1. Annual charge in selection skill March 1997-March 2010

2. Active US equity MF performance decomposition

Because of the interplay between SS and PT, it is not possible to estimate a fund’s SS directly from its returns. In order to address this issue, I regress one month ahead fund returns (net of fees and best-fit benchmark return) on measures of AUM, R-squared, and overdiversification (based on the relative portfolio weight methodology set out by Cohen et al in their 2010 paper, Best Ideas). Each of these measures has been shown to be a negative predictor of future fund returns: the larger the AUM, the higher the R-squared, and/or the more it overdiversifies, the worse is future fund performance. The portfolio tax is imposed by growing a fund, closely tracking a benchmark, and taking an increasing number of low-conviction positions.  

 

3. Top 10 US fund complexes

4. Bottom 10 US fund colexes

My sample includes all active US equity mutual funds over the period March 1997 through March 2010 and results in nearly 183,000 fund-month observations. Each of the three regression coefficients was of the correct sign (negative) and highly significant, meaning the regression itself was highly significant. Each of the AUM, R-squared, and overdiversification variables were truncated so that the regression’s intercept provided an estimate of the average SS over this time period: the AUM variable is set equal to zero at $3m, the R-squared at 0.55, and the overdiversification variable when holding high relative weight stocks, implying zero portfolio tax when a fund’s AUM is $3m or less, its R-squared 0.55 or less, and it holds only high-conviction stocks. 

The estimated average fund SS was a highly significant 4.64%, which provides further evidence of stock-picking skill among active US equity mutual funds. 

Figure 1 shows the annual change in SS by fund age and AUM quintile. SS increased 5.4 basis points annually, implying not only that funds are superior stock pickers, they also get better over time. This is apparent in the last column, as SS grows as the fund ages. The bottom row shows, however, that SS increases decline as AUM grows larger, with the over-$1bn quintile-five funds experiencing SS declines over time. Fund age is the friend of SS, while size is the enemy, particularly when AUM exceeds $1bn. 

The regression coefficients, along with the corresponding value for each variable, were used to estimate SS and PT for each fund-month during the estimation period. The resulting SS results were used to estimate SS and PT during an ex-post period of April 2010 through March 2014, as shown in figure 2.  

On average, funds displayed selection skill of 4.81%, but imposed a portfolio tax of 5.05%, resulting in an expected alpha (SS–PT) of -0.24%. None of the value created by active equity managers was passed on to investors, with the portfolio tax large enough to destroy investor value relative to investing in the benchmark. Strikingly, over 90% of the funds displayed selection skill, a result that stands in stark contrast to conventional wisdom. But even this extraordinary level of skill was insufficient, as only 41% of expected alphas were positive. 

The top 10 and bottom 10 US equity fund complex results are shown in tables 3 and 4, respectively (based on 191 US complexes with at least two active equity mutual funds). The

top complex was T2 Partners, with an SS of 13% and an expected alpha of 10.79%. The worst complex was Dunham with a SS of 0.25 and an expected alpha of -5.12%. The largest portfolio tax was imposed by Capstone, with a PT of 8.07%.  

Large and well known complexes dominate the largest PT list in the middle of table 4. Ironically, Vanguard, the industry leader in reducing fund fees, imposes on its active equity fund investors a PT that is larger than all but eight of the 191 US active equity fund complexes. 

Here are some of the decisions that lead to the imposition of PT: 

• Hiring a portfolio manager to make decisions not directly based on analyst recommendations (Stefan Frey and Patrick Herbst, in a 2010 paper using a large data base of buy-side analyst recommendations for a large European mutual fund complex, found that investment decisions based on analyst recommendations generated alpha and significantly outperformed PM decisions that deviated from analyst recommendations); 
• Closely tracking a benchmark; 
• Staying in a style box; 
• Implementing a risk management programme; 
• Managing volatility, drawdown, and/or tail risk; 
• Overdiversifying beyond 40 stocks; 
• Paying AUM-based rather than performance-based compensation; 
• Growing large, particularly surpassing $1bn in AUM; 
• Team portfolio management rather than solo management; 
• Prescribing sector and industry constraints for portfolio construction. 

Many of these decisions are made with good intentions but the law of unintended consequences has the last word. The bottom line is that, while active equity funds are superior stock pickers they are destructive portfolio managers, and portfolio decisions, beyond building a portfolio of the top 40 or so conviction-weighted analyst picks, impose some level of portfolio tax. 

C Thomas Howard is CEO and director of research at AthenaInvest and Emeritus Professor of Finance at the Daniels College of Business, University of Denver 

Have your say

You must sign in to make a comment

IPE QUEST

Your first step in manager selection...

IPE Quest is a manager search facility that connects institutional investors and asset managers.

  • QN-2488

    Asset class: Euro Investment Grade (Enhanced) Passive ESG Credits.
    Asset region: Europe.
    Size: EUR 500 to 600 million.
    Closing date: 2019-01-10.

  • DS-2497

    Closing date: 2019-01-09.

  • QN-2498

    Asset class: Fixed Income Investment Grade.
    Asset region: Global Developed Markets.
    Size: $50m.
    Closing date: 2019-01-07.

  • DS-2499

    Closing date: 2019-01-02.

  • DS-2500

    Closing date: 2019-01-10.

Begin Your Search Here