Think Twice Before You Ditch That Laggard Fund in Your Portfolio
Our research finds that “hired” funds underperformed “fired” funds in future periods, on average.
Institutional Investors: Hiring Versus Firing
Are investors skilled in hiring and firing investment managers? The literature suggests not. For instance, an influential 2008 study of approximately 3,700 large institutions found that these well-resourced and informed investors tended to hire managers that had outperformed their benchmarks in recent years and fire those that underperformed. Yet, the study found that the hired managers underperformed in the years following hiring, on average, while the fired managers outperformed after getting canned.
Does this trend hold for mutual fund managers as well? It’s harder to assess given that--unlike the study of institutions, which were allocating discrete, identifiable, large sums--mutual fund flows are diffused across millions of investors, reflecting a multitude of hiring and firing decisions. Nevertheless, one can use funds’ organic growth rates as a proxy for hiring and firing decisions in the aggregate.
Mutual Fund Investors: Hiring Versus Firing
Employing this method (see Methodology section at end), we found that active U.S. stock funds “hired” by investors (that is, those with sharply positive organic growth in a given calendar year) fared better against their benchmarks than “fired” funds (those with sharply negative organic growth) before hiring and firing, respectively. However, the performance of hired funds eroded post-hiring, while fired funds saw their performance improve after they’d gotten the hook, as shown below.
Exhibit 1: Active U.S. Equity Funds: Average Annual Excess Return Before Versus After Hiring/Firing (1/1/96-12/31/18; Versus Category Benchmark Index)
Source: Morningstar Data.
These findings also held when we narrowed the scope of the study to a single representative share class of each fund (for example, the oldest share class) instead of all share classes (the approach taken in Exhibit 1). The oldest share class tends to be cheaper, on average, than other share classes, likely explaining why the average annual excess returns of hired and fired funds shown below exceed those in Exhibit 1.
Exhibit 2: Active U.S. Equity Funds: Average Annual Excess Return Before Versus After Hiring/Firing (1/1/96-12/31/18; Versus Category Benchmark Index; Oldest Share Class)
Source: Morningstar Data.
We also evaluated each fund’s returns versus its peer group average. Not surprisingly, the average annual excess returns are highest of all--for hired and fired funds alike--under this approach, as the peer-group average returns are net of all fund expenses, whereas the category benchmark returns exclude fees (as indexes are costless). The findings are directionally the same, though, with hired funds outperforming their peer-group averages by a larger margin than fired funds before hiring/firing and then by a smaller margin afterward, as shown below.
Exhibit 3: Active U.S. Equity Funds: Average Annual Excess Return Before Versus After Hiring/Firing (1/1/96-12/31/18; Versus Category Average; Oldest Share Class)
Source: Morningstar Data.
While these findings suggest that mutual fund investors lack skill in hiring and firing managers, an important caveat applies: The average excess returns of “fired” funds are survivorship-biased upward, reflecting the impact of excluding the poor results of funds that were merged or liquidated away and thus didn’t finish the periods in question. These dead funds were more prevalent in the “fired” cohort because funds that underperform and suffer outflows of assets were likelier to be mothballed.
This was evident when we changed our performance measure from average excess return to success rate—that is, the number of funds that outperformed a benchmark or peer group divided by the total number of funds that began the period. When we measured success this way, treating mergers and liquidations as failures, we found that “hired” funds were still a bit likelier to outperform after hiring than “fired” funds after firing, as shown below.
Exhibit 4: Active U.S. Equity Funds: Average Success Rate Before Versus After Hiring/Firing (1/96-12/18; Versus Category Benchmark Index; Oldest Share Class)
Source: Morningstar Data.
The clearest takeaway is that while investors appear to have put a lot of weight on past performance in making hiring and firing decisions, it’s insufficient on its own. Past outperformance tends to beget future underperformance, and vice versa for funds that have gone through a slump. Returns are only a starting point, which is why we’ve long encouraged investors to consider performance together with other factors that our research has found to be more predictive, like fees.
Ironically, the strongest argument for considering past performance is that, as a practical matter, other investors are likely to continuing paying attention to past performance. Consequently, it’s likely to have a bearing on a fund’s likelihood of surviving in the future which, in turn, influences an investor’s odds of success. Put differently, underperforming funds are likelier to die and there’s no way to succeed in a dead fund, explaining why the success rates of “hired” funds slightly exceeded that of “fired” funds.
But investors are likely to be better served putting past performance in a broader context, so as to ensure that it informs rather than drives the hiring or firing decision.
Our study encompassed nearly 15,200 share classes of active U.S. equity funds, including dead funds, that existed from Jan. 1, 1996 through Dec. 31, 2018. These funds spanned the following Morningstar Categories: large value, large blend, large growth, mid-cap value, mid-cap blend, mid-cap growth, small value, small blend, and small growth.
We compiled each fund share class’s rolling 36-month annual net returns and compared those returns to those of each share class’s assigned Morningstar Category index and Morningstar Category average each December to derive its annual excess returns. The first 36-month rolling period ended Dec. 31, 1998, and the last ended Dec. 31, 2018.
We also compiled each fund share class’s calendar-year organic growth rate (calculated as its net asset flows for that year divided by its net assets at the beginning of the year). To be eligible for inclusion in the analysis, a share class must have at least $50 million in net assets at the beginning of the calendar-year in question. (This control sought to prevent outliers--that is, asset movements at tiny funds--from distorting the analysis.)
We defined “hired” funds as eligible share classes whose organic growth rate was among the 25% highest of all eligible share classes that experienced positive organic growth in a calendar year. We defined “fired” funds as eligible share classes whose organic growth rate was among the 25% lowest of all eligible share classes that experienced negative organic growth in a calendar year.
We defined the hiring/firing period (over which we measured organic growth) as the calendar year which immediately followed a 36-month rolling period. For instance, the hiring/firing period for the 36-month rolling period ended Dec. 31, 2003, was calendar-year 2004. Thus, to determine the pre-hiring/firing performance of “hired” or “fired” funds, the analysis looked back at the performance of such funds in the 36-month period that ended at the beginning of the calendar year in question.
To determine the post-hiring/firing performance of “hired” or “fired” funds, the analysis measured the excess returns of such funds over the 36-month period beginning on the first day of the hiring/firing period. Continuing with the example, the post-hiring/firing period for the calendar-year 2004 was the 36 months beginning Jan. 1, 2004, and ending Dec. 31, 2006.
We calculated average annual excess returns of each cohort (“hired” versus “fired”) for the “before” and “after” hiring/firing periods by averaging the average annual excess returns of all eligible funds for each of the 36-month periods. There were 19 “before” periods (that is, the 19 36-month periods ended Dec. 31, 1997, through Dec. 31, 2015) and 19 “after” periods (the 19 36-month periods ended Dec. 31, 2000, through Dec. 31, 2018).
We assigned each fund share class to a Morningstar Category benchmark and Morningstar Category based on its category classification at the beginning of each rolling 36-month period. If a fund was classified in one of the nine eligible Morningstar Categories at the beginning of the period but didn’t survive to the end of the period (denoted by the absence of an excess return for that period), we considered it “dead.” We treated dead funds as “failures” in the success rates we calculated.