syouth1-blog

[syouth1-blog]

Comeback? Actually...

This is going to be an “actually” posting. Basically, I’m going to link to a WSJ story on trends in active-fund success and then I’m going to (try to) constructively quarrel with it based on data I’ll present. So, in Twitter parlance, it’s an “actually”, which some people find really annoying. If that’s you, consider yourself warned and hit the back button or close this window!

Takeaways

WSJ story argues active funds are on comeback trail, active U.S. stock funds in particular.

We’ve seen active-fund success rates improve from their 2016 lows (48% as of Q1 2017), but improvement has been driven more by active allocation (54%), foreign-stock (57%), and taxable bond (59%) funds, not U.S. stock funds (43%). 

Active U.S. stock fund success rates dipped a bit in 1Q 2017 (43%) but have pushed well off the lows of past three years (low-30%s in 2Q and 4Q 2014, 4Q 2015, and 2Q 2016). 

WSJ Story

This is the piece in question--’Active Managers Stage a Comeback’. Overall, it’s well done. I don’t have an issue with the premise as the slump in active investing is definitely newsworthy and they’ve approached the topic in a sensible way. 

(Full disclosure: The story draws heavily on data supplied by Morningstar, where I work.)

Comeback?

The gist of the WSJ story is that active funds have awoken from their long slumber. To amplify that theme, it quotes famed value investor Bill Miller and a few others.  

The thing is, active fund success rates are only slightly better lately than they were in recent years, though clearly they’ve improved off the 2016 lows. To illustrate, here’s a heat-map of quarterly active-fund success rates going back three years. (Each row is color-coded; green = highest success rate over the three-year time frame for that asset class; red = lowest success rate.)

From this, we can see that, yes, active success rates have risen a bit from early last year, when they were atrociously low. But when viewed over the sweep of the three-year period I examined, they haven’t exactly skyrocketed.

This is especially true of active U.S. equity funds. Here’s a chart that magnifies the trend in active U.S. equity fund success and compares that to the success rate of all funds. 

What we can see is that active U.S. stock fund success rates appear to have dipped a bit since last year’s second-half, not improved.

If not active U.S. equity, what’s been the difference-maker in the last few quarters? Judging from the below, international equity, where nearly 57% of unique active funds beat their benchmark in the first quarter.

If we compare active-fund success rates in the first quarter of 2017 to last year’s first quarter, here’s what it looks like:

As we can see, it’s not really an active U.S. stock fund comeback story, though we saw improvement among such funds. Rather, we’ve seen performance perk up the most among allocation funds, foreign-stock funds, and taxable bond funds, with sector equity funds (which are less numerous) also contributing.

Conclusion

The piece was well done but some of the nuance was lost. Hence this “actually” posting, which is maybe annoying but hopefully is useful to those of you who chose to read it.

Quarterly success rates are a bit of a slippery measure to begin with. Given mortality (i.e., funds dying) and a lack of persistence (i.e., funds that win in one quarter not repeating in the next), it’s more of a read on whether conditions are more conducive to active-fund success. It won’t really tell you much, though, about whether active funds have become more skillful, or at least better equipped to outperform over the long haul. Those rules haven’t changed--it’s a tough game.

Methodology

I was a little lazy in the way I calculated “success rates”. For example, the above is survivorship-biased  and uses recent category classifications (as of March 31, 2017) rather than as-of classifications (i.e., at the beginning of each quarterly period). It also doesn’t remove fund-of-funds from the mix. Sorry, I was time-constrained and doing a history-true, survivorship-bias free pull of the data would have taken too long. Net-net, I doubt that these issues are distorting the findings, but they’re worth noting.

Also worth noting: The WSJ story cut off at 2/28/17 whereas the analysis i conducted above went through 3/31/17. Not a major difference, but worth calling out.

[syouth1-blog]

Winning Funds are Losing Assets?

There’s been some chatter lately that even winning active funds–i.e., those that have beaten their benchmarks recently–are getting hit with outflows. That would be remarkable given investors’ propensity to chase performance. In general, if a fund has beaten its bogy, it’s usually been able to rake in assets. So, if we saw investors pulling money from winning funds, it would represent a pretty dramatic reversal. In this posting, we’ll take a closer look at the data to see if investors are, indeed, pulling their assets from winning funds and offer a few thoughts on trends that might be driving this behavior.

Key Findings

Investors pulled a net $99 billion from “winning” active funds and $214 billion from “losing” funds in the year ended January 31, 2017.

Much of the pain was concentrated in U.S. equity funds, where “winning” active funds saw nearly $100 billion in net assets walk out the door.

Over the three years ended January 31, 2017, “winning” active funds gathered a net $425 billion while “losing” funds got hit with a staggering $1 trillion in net outflows.

It’s not clear that the rash of selling from “winning” funds in year ended January 2017 represents a new trend (the three year numbers suggest its a relatively recent phenomenon).

Approach

I defined “winning” funds as those which generated positive excess returns versus their category-assigned benchmark over the trailing 1 and 3 years ended January 31, 2017. For example, if a Large Value fund beat the Russell 1000 Value Index–which is the index assigned to the Large Value category–over the past year, it’s a winner. And if it lags that index it’s a loser. (The data source in all cases is Morningstar, my employer.)

Then I tallied the estimated net flows to these funds over the corresponding one- and three-year periods. Morningstar makes these estimates based on each fund’s reported net assets, performance, and any distributions (estimated flows are the “plug” in the formula).

From there, I simply aggregated the flow data for all winning and losing funds.

Findings

I found that investors have indeed pulled assets from winning active funds over the past year, but not the last three years. All told, investors pulled a net $99 billion from active funds, and $214 billion from losing funds, in the twelve months ended January 31, 2017.

The outflows from winning funds are concentrated mainly in U.S. equity funds, especially Large Growth, though some international equity and allocation funds also got hit. On the other hand, winning active bond funds took in cash. 

The picture normalizes when we look over the past three years, though. As shown, winning funds took in about $425 billion in net assets over that span, though it’s worth noting that more than half of that went to bond funds. Also striking–investors pulled more than a trillion from lagging active funds in the past three years.

Ground zero for the shift from active to passive has been U.S. equity. Given that, I telescoped in on active U.S. equity funds. Here’s what those results look like:

Strikingly, it didn’t seem to matter much if a fund beat its index–it probably got redeemed anyway. In total, investors yanked a net $93 billion from winning active U.S. stock funds in the year ended January 2017. And if a fund lagged, well, then it was likely orders of magnitude worse ($165 billion in net outflows, in aggregate).

When we look back over three years, we can see an even clearer break between funds that beat and those that lagged: Funds that lagged got annihilated while those that beat more-or-less subsisted.

Conclusion

As should be evident from the above charts, the last year was unusual in that winning active funds got so little love from investors. That was especially true of U.S. stock funds, large-cap funds in particular, with Large Growth being ground zero in the stamped to passive.

But when we expand the time horizon to three years, the picture looks a bit more familiar, with winning funds hauling in assets and losing funds shedding them. This suggests that the seemingly indiscriminate selling over the year ended January 2017 doesn’t form a trend–yet.

Indeed, the recent rash of outflows from winning active funds bears monitoring for signs of continuation and even acceleration. Investors tend to extrapolate past performance into the future. Given active funds’ struggles in recent years, this could translate to investors assuming that passive will continue to outperform and, for that reason, replacing their stakes in active funds with allocations to passives. 

[syouth1-blog]

Round Peg, Square Hole

My most recent posting was pretty long so I'll g so I'll keep this one short. Essentially, it compiles some thoughts I tweeted out recently on a small fund that seems to have messed up, big time. 

The fund, Sandalwood Opportunity, appears to have invested on margin in low-quality, illiquid paper via a subadvisor it hired. The fund's travails illustrate the danger of stuffing a daily-liquidity product like an open-end mutual fund with assets that are hard to transact in. The Tweets follow. (Note: In cases in which Tweets included accompanying charts, those charts immediate precede the Tweets in the embeds below.)

1) JR riffs a bit on the Sandalwood Opportunity Fund here. It's a doozy. || What to Know About 'Explore Bonds' https://t.co/GAEO3IRw7j

— Jeffrey Ptak (@syouth1)

December 9, 2016

2) What this chart--which is ugly--doesn't say is how it happened. There seems to be a story there. https://t.co/Zn1RuI8hpb

— Jeffrey Ptak (@syouth1)

December 9, 2016

3) The fund lost ~7% in about 12 days in mid-June. Unexplained by market. Occurred right around time they canned a subadvisor, Whipoorwill.

— Jeffrey Ptak (@syouth1)

December 9, 2016

4) What happened? Looks like fund got hit w/big redeem, realized Whipoorwillcouldn't liquefy its sleeve, canned them, force-sold the assets

— Jeffrey Ptak(@syouth1)

December 9, 2016

5) Here's chart of flows in fund (I class) as % of prevday's net assets around time of drawdown. Big redeem coincide w/DD. Quiet then bang. pic.twitter.com/CGutAjdgjc

— Jeffrey Ptak(@syouth1)

December 9, 2016

6) Here's how Sandalwood-which has been paid millions to mismanage the fund-had to say. 'Middle Market Credit sleeve'..'manager'=Whipoorwillpic.twitter.com/DhGEbBgbSW

— Jeffrey Ptak(@syouth1)

December 9, 2016

7) Contrary to what they say, fund was not in heavy redeem throughout 2Q. Redeems spiked in mid-June, when it blew up. But why'd it blow up? pic.twitter.com/SX24DulwSY

— Jeffrey Ptak(@syouth1)

December 9, 2016

Sorry, chart with x-axis labeled correctly this time. pic.twitter.com/SX24DulwSY

— Jeffrey Ptak (@syouth1)

December 9, 2016

8) Prob blew up b/c Whipoorwill's assets waaayless liquid than they realized, possibly margined at fund and/or subadvisorlevel. Squeezed.

— Jeffrey Ptak(@syouth1)

December 9, 2016

9) From footnote on borrowings. Based on funds' average monthly net assets, works out to around 10% leveraged during year. Pitiful situation pic.twitter.com/XWArnnMR1L

— Jeffrey Ptak(@syouth1)

December 9, 2016

10) Overall, egregious failure. Charged huge sums, ran fund into ground, didn't have decency to transparently explain it. @sec_enforcement

— Jeffrey Ptak(@syouth1)

December 9, 2016

Hopefully the Tweets speak for themselves on this one.

[syouth1-blog]

Wakey, Wakey: Trends in Active Fund Pre-Fee Excess Returns

In a recent posting, I compared the prices of US active mutual fund to estimates of future pre-fee excess returns. In summary, I found that the annual expenses of most active funds met or exceeded a generous estimate of their potential before-fee excess returns. That is, many funds look like they’re priced to fail.

What I didn’t include in that posting, though, was detail on how I derived those estimates, which people might reasonably quarrel with, by the way. For instance, those estimates could be too low if I’m incorrectly extrapolating the past into the future and pre-fee excess returns sharply improve from here. I could definitely be wrong.

So, in this posting I’m providing some of the detail. Using this data, others can form their own judgments about whether active funds are priced competitively or not. That aside, I found the data interesting in other ways. For instance, I’m asked from time to time where active investing is most fruitful. From the data, one can gain a sense of which areas have been more, or less, target-rich based on the level of average pre-fee excess returns.

I close the piece with some thoughts on the implications of these findings on investors, fund companies, and fund boards alike. All are likely to be impacted and so I’ve tried to list the ways.

Key Takeaways

Over the last two decades, average pre-fee excess returns of active mutual funds have fallen in most categories; this suggests that the pre-fee payoff to investing in active funds has shrunk.

At the same time, the variation of average pre-fee excess returns among funds has narrowed in many categories; funds are performing more alike within their categories than before (pre-fee).

These trends are evident in the chart below, which dot-plots each category along two dimensions—pre-fee average excess returns (vertical axis) and standard deviation of excess returns (horizontal). Note the shift down and to the left, indicating smaller average pre-fee excess returns and a narrowing of performance of active funds within categories.

This shift is likely to have significant implications on investors (who must weigh the potential payoff of active management against its cost), fund companies (who set prices), and fund boards (who approve those prices).

Most investors should probably index in lieu of choosing active funds, but those who choose to do so are well-advised to place an even higher premium on cost; many fund companies should slash prices or, barring that, merge or fold funds or even their own firms; and fund boards should take a harder line when evaluating the viability of funds at current prices. 

U.S. Equity

Most U.S. equity categories have seen their average pre-fee excess return decline over the past two decades (i.e., the average for 10/31/16, shown in the second column from left, is below the average for all rolling periods over the full 20-year span, shown in the third column from left). Only the small-cap categories boast average pre-fee excess returns exceeding 1% for the 10 years ended October 2016.

(Each column is color-coded from highest (green) to lowest (red).)

What’s also striking is the way the variance around the mean shrunk, as shown in the two rightmost columns of the table above. For example, for the Large Blend category, we found there was a 1.17% standard deviation around the 0.04% average pre-fee excess return for the 10 years ended 10/31/16. In other words, around 68% of Large Blend funds generated pre-fee excess returns between -1.13% and 1.21%. By contrast, there was wider dispersion in the past—a 1.82% standard deviation around the 0.59% average pre-fee excess return for all rolling 10-year periods that ended between 1996 and 2016 (meaning 68% of Large Blend funds had pre-fee excess returns between -1.23% and 2.41%). This trend held across the board but was most pronounced in the small- and mid-cap categories.

What this data suggests is that the payoff to investing in active U.S. equity funds before fees has shrunk; what’s more, it appears there’s less differentiation among the active funds in these categories—they’re performing more and more alike.  

International Equity

Among active international equity funds, we can observe many of the same trends that we noted above—recent average 10-year pre-fee excess returns have generally declined when compared to the norm for all rolling 10-year periods between 1996 and 2016. Whereas it wasn’t uncommon to see average pre-fee excess returns over 3% in the past, there were recently only a few narrow or niche categories that delivered that kind of payoff. We can also see that the standard deviation of average excess returns has fallen over time in nearly every case. 

That said, a few caveats to go with this data. First, whereas we can more precisely categorize and benchmark U.S. equity funds, international equity funds are a less homogeneous group. As such, it’s possible that the benchmarks chosen for some of the international equity funds in this study aren’t a great fit. Second, some of these categories were home to relatively few funds recently and even fewer over the past few decades. Given this, we’d want to be careful to not put too much stock on historical comparisons. (For ease of reference, we’ve placed a heavier black border around the larger categories, by assets, in the above and other tables.)

Caveats notwithstanding, it appears that many of the same trends have rippled through active international equity funds. Namely, average pre-fee excess returns have come down while the variance of those returns among funds in the various categories has narrowed.

Fixed Income

The pre-fee payoff to investing in active bond funds is less than for equity funds. This is evidenced by the figures shown in the “Avg. Exc. Rtns.” columns of the tables below, which are meaningfully lower than the comparable figures in the equity tables above. (Fortunately, the fees for active bond funds usually lower than those of active stock funds.)

Strikingly, the average pre-fee excess return has been negative for some popular categories, including High Yield Bond. This is even truer of the average pre-fee excess returns of the municipal bond fund categories, which are shown in the table below.

That said, a word of caution on the results shown in the tables above: Because they’re not risk-adjusted, and because risk-taking can be especially impactful in the bond world, it’s best not to take these results too literally. In addition, benchmark misspecification can be a real issue among bond funds, as bond indexing remains more an art than a science given investability issues and other quirks of transacting in the bond market.

Implications

Caveats notwithstanding, the data suggests that the payoff to investing in active funds has shrunk. What does this mean to investors, fund companies, and fund boards?

Investors

Investors should pick their spots, know their limits, and re-double focus on costs.

In general, investors are well-advised to be choosier when selecting active funds. For instance, the data strongly suggests that the U.S. large-cap categories are largely picked over, making it difficult for active funds to scratch out an edge before fees. Thus, these types of investments would seem to be stronger candidates for indexing than active management. Despite that, actively-managed U.S. large-cap funds held around $2.7 trillion in assets as of November 2016.

Investors also would want to consider their ability to differentiate funds by pre-fee excess returns. Given that there’s been less variation in pre-fee performance, with pre-fee excess returns increasingly converging, this can make it more difficult to separate the wheat from the chaff. Investors would therefore perhaps be well-served to consider active funds in categories where potential pre-fee payoffs are large and there’s enough separation between funds to make selection easier.

One inescapable conclusion is that cost will become increasingly important in active-fund selection. With average pre-fee excess returns appearing to shrink, investors must place an even greater premium on low expenses, for otherwise there will be little if any net-of-fee excess returns to speak of. Moreover, with average pre-fee excess returns appearing to converge in a number of categories, cost is likely to become an even greater differentiator after fees and thus even more predictive of future performance than before.  

The world is now awash in index funds and ETFs, giving investors an easy way to invest in broad swaths of the market at very low cost. Thus, the acid test is no longer active fund A versus active fund B but rather any active fund versus the relevant index fund or ETF for the category concerned. Many active funds will flunk that test and, as such, investors who lack the patience or resolve to hunt for skilled active managers that can beat the odds are probably better off indexing.

Fund Companies

Fund companies should face reality, seek scale, cut costs, or fold; they should also

The cost of funds has come down through the years, but not as quickly as pre-fee returns. It’s likely that some cyclical factors have driven down the payoff to active investing, further depressing average pre-fee excess returns. But the downward trend in average pre-fee excess returns does not appear to be isolated—it’s widespread. Further, there’s a narrower range of pre-fee performance in many categories, implying that many managers are using similar methods to nose into the same trough. Fund companies must face this reality.

When they do, they’ll seek scale by combining with other firms, merging funds, or otherwise rationalizing their operating expenses. And they’ll slash expenses aggressively. We’ve seen several large fund mergers to this point but they’ve not yielded meaningful savings to fund investors—i.e., the funds’ expense ratios didn’t fall. These aren’t true scale deals and, thus, likely won’t stand the test of time.  

Firms appear to have throttled back on fund launches recently but there’s still a glut of sub-scale funds. As of November 2016, there were around 7,600 unique active funds in our database. Of those, about 2,700 held less than $100 million in assets and roughly 4,800 had less than half a billion under management. Many of these small funds will need to merged or liquidated away.

With less product to push and fewer distribution avenues (amid increasing consolidation and convergence of practices among institutions and platforms), fund companies must also adapt by reining in their operations. Fewer portfolio managers and analysts will certainly help firms to economize, but firms will also have to slash distribution, marketing, and back-office costs. These changes will be painful but necessary.

They’ll have to cut expense ratios much more aggressively than they have done to this point, or find other ways to offer lower-cost versions of their capabilities. As mentioned, we have seen expense ratios drift lower, but the decline hasn’t been as steep as the falloff in pre-fee performance. That’s obviously not sustainable, even allowing for the possibility that pre-fee excess returns will improve in future years. Active-fund complexes can merge, liquidate, and cost-cut to their heart’s content, but if expense ratios don’t come down then it probably doesn’t matter.

Given what appears to be increasing competition for alpha, fund companies will also have to grapple with how they define their capabilities, i.e., what they truly excel at, and re-double their focus in those areas. The competition is too stiff to make “dabbling” in non-core areas successful. So, as firms think about which funds to keep and which to ditch, they’ll of course need to ask themselves what they’re likely to succeed at. (For some firms, the honest answer will be “nothing”; those firms are well-advised to sell themselves or shut down altogether.)

Firms will also need to think in new ways about how they define their “capabilities”. After all, it’s possible that the erosion of pre-fee excess returns reflects discovery and increasing adoption of factors like size and value. If so, then tried and true methods of the past won’t offer the same pop going forward.  As such, firms would need to consider whether it makes more sense to pursue go-anywhere approaches that afford portfolio managers greater latitude, versus hemming them into a narrower style or segment. In this way, they could more freely exploit premia or mis-pricings. It goes without saying that this will be a challenge for many fund firms, akin to developing a new muscle group. Many won’t be able to pull it off.

Fund companies will need to be more fastidious about preserving the performance of funds. In a world where excess returns are more precious and investors are choosier about investing in active funds, fund companies will no longer have the luxury of letting successful funds outgrow their capacity, become bloated, and mean revert. Instead, they’ll clearly articulate capacity limits and close funds before they become too large.

They’ll also devise new ways to generate revenue that’s less dependent on ramping assets. The current model confers revenue growth only as funds grow AUM. This is a workable model insofar the strategy doesn’t lose efficacy as assets pile in, but most do and fund companies have not tended to manage capacity well. Innovations in performance-based fees would help to mitigate this risk insofar as it would encourage fund companies to shut strategies before bloat sets in. It would also better align with shareholder interests, linking the investors’ success with the fund company’s.

Fund Boards

Fund boards should take a harder line, challenging fund companies about the viability of funds that appear priced to fail.

Fund boards are too often content to evaluate funds versus a peer group, which might have been customized in various ways, sometimes to the point of absurdity. And they’re prone to rationalization about performance and the value of active investing in general. For evidence, choose an underperforming active fund, open its annual or semi-annual report, and locate the section in which the fund board explains the rationale for approving the manager’s contract and fees. It’s often generic boilerplate, devoid of substance, with any edges sanded-off so as to not embarrass the fund’s advisor. It’s time for a mindset change.

In evaluating the viability of an active fund and the fees it levies, boards need to be asking “how do we define success for investors in this fund when a low-cost index fund is readily available to them?” If they did so, then the math would resolve to comparing an active fund’s pre-fee value add to its price. Many funds would fail this more stringent, but needed, test.  

As that realization dawns, boards should challenge fund companies about the future of overpriced active funds they’re offering and encourage them to take the steps needed to cut expense ratios or, if that’s not possible, wind these funds down in an orderly way.

Conclusion

It appears that average pre-fee excess returns have declined and the variation among funds has narrowed. This will present new challenges to investors who select active funds, the fund companies that offer the funds, and the boards charged with overseeing them. With low-cost index funds and ETFs now readily available, each of these stakeholders must adapt. In all likelihood, this will mean lower usage of active funds among investors and consequently a shake-out of the active fund industry, which will have to slash costs and shrink. 

[syouth1-blog]

The Price Is Wrong (Updated)

In this piece, we compare U.S. equity mutual funds’ annual expenses to our estimate of their potential future pre-fee excess returns. We demonstrate that many funds are priced to fail—their fees approach or exceed their potential future pre-fee excess returns. Whereas investors might have tolerated overpriced funds like these in the past, they’re unlikely to do so in the future. Given this, fund companies face a stark choice—slash fees or eliminate costly funds altogether. 

(Update as of 12/16/16: I’ve tacked on a section to the end of this post that does the same thing as what follows immediately below, but limits the analysis to funds that don’t levy 12b-1 fees. I got some heat from a few readers that by comparing funds’ all-in expense ratio to estimates of future pre-fee excess returns, I was in effect conflating the cost of management and the cost of distribution/marketing/advice. I happen to strongly disagree--from the investor’s standpoint, a fee is a fee. But I hope the new section helps those who disagreed with the original approach.)

Key Takeaways

Average pre-fee excess returns of most U.S. stock categories have pinched lower in recent years and now sit well below the norm of the past few decades.

Fees haven’t fallen as steeply and, thus, more than two-thirds of U.S. equity funds levy annual expenses that would wipe out their estimated future pre-fee excess returns. Those funds recently held nearly $2 trillion in assets.

Of the funds that charge less than our estimates of future pre-fee excess returns, many have little margin for error (i.e., their fees are no more than 0.30% below potential future excess returns before fees).

Fund over-pricing is most rampant among U.S. large-cap funds and least widespread among U.S. small-cap funds.

Faced with the prospect of reaping smaller pre-fee excess returns in the future, fund companies will likely have to slash expenses or mothball costly funds that are unlikely to succeed

Background

In the past, investors have tended to judge funds’ performance based on returns vs. peers or a relevant benchmark. If funds put up good numbers, investors stuck with them (for a while at least). This isn’t the most refined or sophisticated approach; but it’s straightforward and, so, largely explains why investors have embraced it.

We’ve seen that practice change in recent years, however, given the proliferation of low-cost ETFs and index funds. Indeed, whereas investors might formerly have been willing to look past an active fund’s inability to beat a relevant index after fees (provided it beat its average peer, perhaps), they’re showing far less tolerance for underperformance now. Rather, an active fund is only worth its salt if it can beat its benchmark after fees.

By this standard, investors seeking to assess whether an active fund is worth the bother really need just two pieces of data:

The fund’s annual cost

The fund’s potential future pre-fee excess returns versus the benchmark concerned

The larger 2 is compared to 1, the more worthwhile an investor would consider the fund, and vice versa.

Study

We already possess the first piece of data--the fund’s annual expense ratio. That leaves us to estimate potential future pre-fee excess returns. There are many ways to do so but this study makes history its guide: We compiled the rolling annualized pre-fee excess returns of every U.S. open-end active equity fund in our database for the 10-year periods ended October 2006 through October 2016. We derived those excess returns by calculating the difference between a fund’s gross annualized returns and the return of the benchmark index assigned to its category. (We used each fund’s end-of-rolling-period category classification to determine which index to compare it against.)

From there, we calculated category averages (again based on each fund’s end-of-rolling-period classification) and formed a time series of those category averages. Those time series, in turn, gave us the ability to assess the recent level, historical trend, and range of pre-fee excess returns for each category.

For example, here is the time series of the average rolling 10-year pre-fee excess returns of the U.S. large-cap categories:

From the above we can observe recent level (as of October 31, 2016, 0.76% for Large Value; 0.04% for Large Blend; 0% for Large Growth), trend (150 to 300 basis points lower since the global financial crisis), and range (-0.85% to 1.62% for Large Value; -0.04% to 1.68% for Large Blend; -0.34% to 3.51% for Large Growth).

Findings

We can repeat this exercise for all other categories. For example, here’s a table that summarizes the most recent average pre-fee excess return and historical average pre-fee excess returns for the major U.S. stock categories:

From this, we can take away a few things. First, average pre-fee excess returns have generally declined and now sit below the historical averages. Second, the standard deviation of those excess returns—that is, the variance around the mean—has fallen across the board. What that suggests is the pre-fee payoff to active management has shrunk while our ability to “bound” the range of outcomes has grown.

That said, we opted to adjust the latest figures. Why? Active investing has been in a deep funk and while some of it reflects structural change (i.e., “paradox of skill”, etc.), there’s probably also a cyclical dimension (i.e., stylistic headwinds, returns dispersion, market trend, etc.). If we met the longer-term average half way, then here are the pre-fee excess returns we’d project:

Having assembled this data, we turned our attention to assessing whether funds are priced “right” (i.e., annual expenses < potential future pre-fee excess returns) or “wrong” (the reverse). Here’s how that picture looks based on the most recent expense ratio data in our database:

To summarize the above, 4,635 of 6,708 active U.S. equity funds in our database were recently levying expense ratios that would wipe out their potential future pre-fee excess returns (i.e., the estimates in the table above). At their current price, and assuming those estimates approximate actual pre-fee excess returns in the future, these funds would have no prayer of beating their benchmarks after fees.

How much money is sitting in these potentially overpriced active funds? $1.9 trillion, as shown below:

It’s worth noting that, of the funds whose expenses do not exceed our estimates of future pre-fee excess returns, the margin often isn’t very wide. To illustrate, the chart below breaks down these 2,073 active funds based on the difference between their expense ratio and the pre-fee excess return estimate for their category.

As is evident from the table, most active large-cap funds have less than 30 basis points of breathing room (i.e., their annual expenses are usually not more than 0.30% lower than our estimate of future pre-fee excess returns). By contrast, given their larger potential future pre-fee excess returns, it appears that most active small-cap funds have greater maneuverability and therefore would seem to represent a better bet.

Conclusions

Many active U.S. stock funds are too expensive to succeed. This is especially true of U.S. large-cap funds, the clear majority of which charge well more than our estimates of potential future pre-fee excess returns. All told, it appears that overpriced active U.S. equity funds (i.e., those whose fees exceed likely future excess returns) still hold nearly $2 trillion in assets, and there’s little margin for error among funds that are less expensive. The exceptions are small-cap funds, where it appears fees are still below estimated future pre-fee excess returns.

To succeed, expensive funds must buck the trend toward lower pre-fee excess returns and generate significant outperformance in the future. This is unlikely. Therefore, fund companies must decide between slashing fees to levels that give these funds a chance to outperform their benchmarks, or mothball them altogether.

Caveats

Our estimates could be unduly conservative, in which case many more funds could outperform their benchmarks and by wider margins than shown above. But it is worth remembering that in forming the estimates of future pre-fee excess returns, we grossed-up recent excess returns to allow for the possibility that those recent figures were depressed by fleeting factors.

The study could also be incomplete insofar as it considers returns but not risk. Indeed, if we considered measures of risk-adjusted “excess return”, such as alpha or Sharpe Ratio differentials, it is possible that the picture presented above would brighten. Given other research we’ve conducted, which shows a similar erosion of risk-adjusted excess returns across time, we are doubtful this would be the case. But a simplistic test like this study’s should be caveated, hence this mention.

Readers should likewise note that in calculating excess returns we considered a fund’s end-of-period category classification. Given that this classification might not precisely reflect the fund’s style exposures throughout the 10-year rolling period, it’s possible that there’s a mismatch. We do not believe this imprecision would call the study’s overarching conclusions into question, but it’s an issue nonetheless.

Update (as of 12/16/16)

A few readers strongly disagreed with the analysis and conclusions above. Specifically, they said it was “worthless” because it included funds whose expense ratios bundle both management fees and fees for marketing and distribution (i.e., “12b-1 fees). In conducting the research originally, the question I was asking was “how many funds are likely to survive in current form?’. If you believe, as I do, that investors will increasingly judge funds based on performance versus a relevant index, then it’s the bottom-line return that matters. A fund’s “bottom-line return” is its net-of-fee return--net of all fees.

That said, there was a lot of back and forth on this topic overnight, with those disagreeing variously suggesting the analysis betrays a failure to appreciate the difference between “management” and “advice” or of a clear vendetta against active management. Rather than engage further in that, I figured we’d update the analysis and let it speak for itself. That updated analysis, which is identical to the above but limited to funds that levy no 12b-1 fees, follows.

The key takeaways are largely the same: 

More than half of U.S. equity funds levy annual expenses that would wipe out their estimated future pre-fee excess returns. Those funds recently held more than $1 trillion in assets.

Of the funds that charge less than our estimates of future pre-fee excess returns, many have little margin for error (i.e., their fees are no more than 0.30% below potential future excess returns before fees).

Here are updated versions of some of the charts above:

To summarize the above, 1,773 of 3,246 active U.S. equity funds in our database were recently levying expense ratios (but not 12b-1) that would wipe out their potential future pre-fee excess returns. 

How much money is sitting in these active funds? $1.1 trillion, as shown below:

Of the funds whose expenses do not exceed our estimates of future pre-fee excess returns, the margin often isn’t very wide. To illustrate, the chart below breaks down these 1,473 active funds based on the difference between their expense ratio and the pre-fee excess return estimate for their category.

As is evident from the table, most active large-cap funds have less than 30 basis points of breathing room (i.e., their annual expenses are usually not more than 0.30% lower than our estimate of future pre-fee excess returns). All told, those funds held around $850 billion in assets as of October 31, 2016, as shown in the final chart below.

I hope this update is useful.

[syouth1-blog]

The Look of a Winner (Is a Loser)

Investors tend to have some pretty engrained misconceptions of what "winning" funds look like. For instance, winning funds lay waste to the index and category peers; they do so over the short- and long-term; they corner really well, deftly avoiding big drawdowns and rocking during rallies; they don't rattle around much; they succeed like clockwork. They're Tom Brady. 

For those who have gotten to know markets, randomness, and the resultant unpredictability of short and even intermediate-term performance, we know this is nuts. Winning funds do not succeed anywhere near linearly. Performance is jagged; success and failure arrive abruptly; it often takes years to grind out an advantage; and so forth. This is pure torture for many investors, who bail (and that pattern reveals itself in the form of hideous dollar-weighted returns; if there's any consistency in markets, it's that, but I digress).

Study

However, this concept is often too abstract so I thought I'd try to semi-simply illustrate it through an example. Here's what I did (which will win no points for elegance or precision but last time I checked this blog was free):

Grouped together all diversified U.S. open-end equity mutual funds (i.e., the nine style-box categories; active and index funds; no ETFs)

Limited to unique funds (i.e., oldest shareclass)

Calculated the twenty year annual excess returns of the unique funds I grouped (excess returns = fund's total return minus return of benchmark index assigned to the category that fund was assigned to)

Sorted the funds into deciles by excess returns (top=group with highest excess returns; bottom=group with lowest excess returns)

There were around 680 unique funds that had twenty-year excess returns, so we're talking about 68 per decile grouping.

Findings

Here's the predictable stairstep pattern from the top to bottom decile when sorted by excess return:

The left-hand side of the above chart is Tom Brady. This is what investors idealize and, too often, expect. But reality is Jay Cutler--maddeningly inconsistent, disappointing, only occasionally thrilling.

To illustrate, here are the average maximum drawdowns of those same decile groupings over the twenty-year period we examined: 

As you can see, the best-performing funds did indeed have smaller drawdowns than the worst-performing. But it's not like it was bloodless--the top-decile, Tom Brady'est funds of the bunch got more than cut in half at some point, on average.

Now you might think 'ok, this is the cost of investing in the equity market, but I bet the winners lag a whole lot less than the losers'. And you'd be right: As shown below the more-successful funds did indeed lag less often (measured as number of rolling 36-month periods during the twenty year span where the decile grouping had negative average excess returns) than the less-successful funds. 

But it's not like they were strangers to underperformance. In fact, the best-performing funds lagged their indexes in more than one of every three rolling three-year periods. So, investors in these funds spent roughly a third of the past two decades looking up, not down, at the index (when measured over rolling three-year periods). What's more, the shortfall was significant--on average, the top-decile funds underperformed by 274 basis points annually during these lagging three-year periods, as shown below.

So then maybe you reason 'sure they fail and yeah it's not immaterial when they do, but it probably doesn't get as bad as it does for the worst funds'. You'd be right again, but only to a point: At their worst, winning funds tended to underperform their indexes by a smaller margin, on average, than the losing funds did. But it's not like they didn't throw pick-sixes (into heavy coverage, off the backfoot) like Jay Cutler.

To illustrate, the chart below plots the average worst three-year underperformance the funds in these decile groupings suffered during the two-decade period I examined:

At their very worst, the very best funds underperformed by more than nine percentage points per annum, on average. That’s not a good look.

Conclusion

To summarize, in fund investing it's often the case that the look of a winner is a loser.

Caveats and Technical Notes

This study is survivorship-biased, as dead funds by definition didn't have a twenty-year track record as of October 31, 2016 (because they're, well, dead). What this means is that the excess returns, drawdowns, number of lagging periods, average performance in lagging periods, and worst performance in lagging periods look better than was reality.

That said, mortality is lower in the upper deciles given that outperformers survive/thrive. Therefore, survivorship bias affects the results of these top deciles less. Given this, the performance stats I've reviewed for these funds, in particular the frequency and magnitude with which they lag their benchmark indexes, shouldn't be much affected. And that's the point I wanted to underscore anyway -- that winners often don't look the part.

A few other notes: This is a very simple illustration in that it only considers reward, not risk. When we account for risk, it's likely that it somewhat scrambles our decile groupings. But this was a simple illustration meant to simulate the, let's face it, fairly simplistic way performance tends to be assessed in practice (this lowest-common denominator even dictates the way "professional" investors decide to hire and fire managers, much as they might sniff at that notion).

Finally, because it uses end-of-period category classification, there's the real possibility that the excess return calculations don't fully capture the stylistic traits of these funds over the two-decade timespan we studied. (I did it this way for a mundane, somewhat lame reason -- the 'excess return' calculation in Morningstar Direct pulls in the most recent category classification.) Does this render the study invalid? Probably not. I think it could distort some of the results, but not in a widespread, indicting way.

[syouth1-blog]

Consistency Is A Mirage

I ran across a recent study that examined how consistently Australian mutual funds outperformed their peers. To do so, the study tallied-up the number of top-quartile Australian funds based on their trailing one-year returns as of June 2012. Then it tracked how many of those top-quartile funds remained top-quartile in the subsequent years. For instance, the number of funds that were still top-quartile based on their one-year returns for the period ended June 30, 2013, and so forth for 2014, 2015, and 2016. The study found that only 2% of the “winning” funds as of June 2012 were still left standing as of June 2016. (Note: At least I think that’s how the study worked; it’s pretty vaguely explained and labeled in the chart.)

I’ve seen “persistence” studies like these before and I think they are unhelpful in that they set an unattainably high standard and then use funds’ inability to meet it as a blanket indictment of active investing. I’m someone who has gone on record numerous times to say that most people should use passive, not active, funds. But I don’t feel persistence studies like these are an especially rigorous, of even intellectually honest, way to make the case for indexing.

Here’s why: When you look through the other end of the looking glass at “winning” funds, you find that they don’t meet this standard either. That is, they’re inconsistent over the short-term or, put another way, lack persistence. Success comes in clumps that arrive on their own timetable, not predictably, and alternate with periods of underperformance. That’s reality.

Study

To illustrate, I evaluated the consistency of 439 unique U.S. equity funds that had placed in the top-quartile of their Morningstar category over the ten-year period ended September 30, 2016. For each of these 439 “winning” funds, I compiled data on their category percentile ranks during each of the rolling one-year periods that fell within the ten years ended September 30, 2016. There were 109 such rolling periods and thus 47,851 observations (i.e., 439 winning funds x 109 rolling one-year percentile ranks per fund).

From there, I attempted to replicate the study with one notable difference: Whereas the study looked at just five non-overlapping 12-month periods–the years ended June 2012, 2013, 2014, 2015, and 2016–I considered every non-overlapping 12-month period from October 2006 to September 2016. But I followed the same procedure–identify the funds which ranked in the top-quartile of a given one-year period, then see how many of those repeated in the next non-overlapping year, did so again in the third consecutive year, four-peated, and five-peated.

Findings

The chart below cross-sections these winning funds over time. For instance, in the first rolling one-year period (12-months ended 9/30/07), 142 of the 439 long-term winners landed in their category’s top-quartile. Of those 142 funds, 48 repeated the following year ended 9/30/08 (34%), 19 three-peated the year ended 9/30/09 (13%), 4 four-peated the year ended 9/30/10 (3%) and 2 five-peated in the year ended 9/30/11 (1%). In the second rolling one-year period (ended 10/31/07), 139 of the 439 long-term winners were top-quartile. Of those, 44 repeated (32%), 11 three-peated (8%), 1 four-peated (1%), and none five-peated (0%). And so forth for all of the other rolling one-year periods through 9/30/16.

When we average out the repeat, three-peat, four-peat, and five-peat rates across the rolling periods, here’s how it looks:

What’s striking is how consistently inconsistent these long-term winners were. Roughly half were one-and-done–they scored a top-quartile ranking in the one-year period concerned but then didn’t repeat in the next year. About one-third managed to repeat, but only 10% three-peated, roughly 4% four-peated, and only 1% of these winners five-peated. While they’re totally different universes (U.S. equity funds vs. Australian funds), these figures aren’t all that different from the study’s. And these are funds that have been long-term winners!

(Keep in mind that this is the percentage of these winning funds that landed in the top-quartile in a given one-year period. But not all winning funds do this. In fact, only about one-third of these winning funds landed in the top-quartile of a rolling one-year period on average. So, when we account for this, we find that there’s a puny 0.4% chance that one of these long-term winning funds, chosen at random, would five-peat.)

Silver Linings and Caveats?

Is there a silver lining here? Well, these average repeat, three-peat, four-peat, and five-peat rates are better than mere chance (25.0%, 6.3%, 1.6%, 0.4%, 0.1%, respectively).  Also, they’re perhaps depressed a bit by survivorship bias, as these percentile rankings don’t include dead funds. If we added those funds–most of which are duds–back into the distribution it’s possible that some of these winning funds might have exhibited a bit more consistency than we’re observing. But in all likelihood, it wouldn’t turn out materially different than what’s shown. 

What if we extended the rolling periods from one-year to three-years? It stands to reason that we’d see more consistency–after all, we’d be starting with long-term winning funds (over the ten years ended 9/30/16) and then subdividing that decade into fewer segments. So there’s not too much mystery about the direction. To address that issue, as well as the practical limitation a ten-year period imposes (i.e., one can track three-year repeats and a few three-peats, but not four-peats or five-peats), I extended the time-frame to twenty years from ten years.

Over the twenty-year period ended September 30, 2016, there were 193 unique U.S. equity funds that landed in the top-quartile. Here’s some data on the consistency of their rolling three-year category rankings:

The results weren’t dramatically different. These winning funds had repeat top-quartile showings (in successive three-year non-overlapping periods) about 40% of the time, on average, as shown below. But it fell off sharply from there, with these funds three-peating only about 15% of the time, on average, with four-peats (4%) and five-peats (0.6%) seldom observed.

I would note that there’s some hair on this data. For one, survivorship bias becomes more pronounced the longer you extend the time-frame and twenty years is pretty long indeed. Therefore, it is possible that these winning funds were more consistently successful at the relevant times than portrayed here. What’s more, since we’re using end-of-period category classification for purposes of ranking these funds historically, we’re essentially assuming that their Morningstar category assignment as of 9/30/16 held going all the way to 10/1/06. A more-robust study than I’ve conducted would control for these issues (if time permitted I’d do it myself but alas), but I suspect not markedly change the outcome and key takeaways.

Conclusion

While opponents of active fund investing cite inconsistency as another reason to index, that argument is a red herring. Even the most durably successful funds exhibit little consistency when measured over rolling one- or three-year periods, as evidenced by the results shared above. Success for these winning funds came unpredictably, in clumps, not metronomically.

Arguing against active funds because they’re inconsistent is harmful. It reinforces an unattainable standard that leads investors to make poor decisions, such as chucking a worthy fund away when it doesn’t succeed like clockwork. There are plenty of reasons to index but inconsistent performance among active funds isn’t one of them.

[syouth1-blog]

When Winners Fail

Quick summary: While it’s generally assumed that the best funds don’t underperform for long, that’s not the case. Winning funds—which I defined as those that have beaten their Morningstar category index since-inception (over a period not less than ten years)—lagged their index in about three of every ten rolling ten-year periods during their lifetime, on average. And it’s not a mean reversion story per se—these performance setbacks appear to have occurred at various stages of the winning funds lives’, not just in the last few years as returns began to roll over. What it means is that even in the best-case scenario--picking a fund that proceeds to win over its lifetime--there will be plenty of doubt-inducing moments along the way.

Sequoia

A few weeks back, a friend and I were corresponding about the Sequoia Fund. If you’re not familiar, Sequoia is a pretty storied fund that, through its namesake, the late Bill Ruane, had a long and profitable association with Warren Buffett.

As it happens, Sequoia has been very successful in its own right during its nearly four-decade run. Indeed, from Sequoia’s July 15, 1970 inception through Jan. 31, 2016, it had notched a 13.8% annualized gain, which bested the S&P 500 by nearly 3% per annum.

Very few funds have put up those kinds of numbers. But lately Sequoia has had a comedown due to its very large stake in troubled drug-maker Valeant Pharmaceuticals.

Everyone is probably familiar with the Valeant story so no need to belabor it. But Sequoia’s history with Valeant is kind of interesting. The firm initiated its position in Valeant in the second quarter of 2010 and quickly built up an 11-million-plus share stake by the end of that year. Then, for the next five years, they sat on it, neither adding nor trimming. Because Valeant soared over that span (a cumulative $1,088% gain from December 2010 to July 2015), far outpacing Sequoia’s other holdings, it became a larger and larger position, topping out at more than 30% in the summer of 2015.

When the bottom dropped out of Valeant shares last summer, Sequoia fell with it: The fund had lost more than 24% in the six months ended January 31, 2016 vs. only 7% for the S&P 500 Index. That deep slump has put Sequoia in an uncomfortable spot, with some questioning whether management has lost its mojo. (Heightening the scrutiny, Sequoia disclosed that several board members resigned their positions, reportedly because they disagreed with management’s handling of the fund’s Valeant position.)

What is Failure?

All of which brings us back to this chat I was having with my friend. He seemed ready to write-off Sequoia because, thanks to the recent performance falloff, its 10-year numbers looked ho-hum vs. the indexes. He has a point – a decade is, practically speaking, a pretty long time to underperform. After all, most investors judge a fund on its three-year returns, so by that standard ten years is an eternity!

That got me wondering – how often do winning funds suffer prolonged (i.e., ten years) bouts of underperformance? How frequent an occurrence is it? To delve into that, I compiled the since-inception excess returns of all unique U.S. equity funds in Morningstar, Inc.’s database, measuring each fund’s performance against its assigned Morningstar category index. Then I scoped-out funds that had incepted after February 1, 2006 (since they lacked a ten-year record) as well as funds that had negative since-inception returns (since we’re interested in winning funds).

Once I’d narrowed the universe in this way, I got to work calculating rolling ten-year excess returns for each fund. Then I tallied-up the number of losing ten-year periods each winning fund had suffered since-inception, and summed up the number of losing ten-year periods across all funds. Here’s what I found:

“Winning” funds (i.e., those that are at least ten years-old and have beaten their Morningstar category index since inception) lagged their Morningstar category index in three of every ten rolling ten-year periods, on average. Put another way, the average winning fund spend about a third of its days with a losing ten-year number.

The math being what it is, the least old of these winning funds will tend to have fewer losing ten-year rolling periods, which makes sense—if we’re limiting the universe to funds that have beaten their category index since inception but are at least ten years-old, it stands to reason that an eleven year-old fund isn’t going to have many losing ten-year rolling periods (otherwise it wouldn’t be a “winner” to begin with). With that in mind, I took another cut at the data, examining losing rates by fund age.

When we look at it this way, we can see that the losing rate of winning funds can be even higher than the 30% headline figure mentioned above.

Finally, I examined one more cut of the data, this time assessing how losing rates varied with the magnitude of since-inception return. To do this, I sorted the winning funds from largest to smallest by magnitude of since-inception excess returns and then grouped into quartiles. 

As one would expect, the winningest funds—the top quartile by since-inception excess return—had the lowest losing rate. But it’s important not to lose sight of the big picture: Very few funds beat their benchmarks, let alone beat by more than 1%, over the long haul. So the losing rates of the third and fourth quartiles are instructive – these funds were underwater in about four or every ten rolling ten-year periods, on average.

So to bring it full-circle, while I get what my friend was saying about Sequoia, the data underscores the point that there’s nothing especially indicting about a losing ten-year number. Rather, what it argues is that long slumps happen to the best of ‘em, and somewhat routinely at that. (By the same token, it’s not a get-out-of-jail card for winning funds that have lagged for a decade or more; that’s where additional due-diligence comes into play.)

Counter-argument

I could see a counter-argument being made that all we’re really detecting here is the downslope of mean reversion. That is, old “winning” funds with high losing rates aren’t really winners—they’re has-beens that had been good once but have since entered an irreversible slide, which makes the high losing rates a prelude to, well, being a loser. In that sense, as this counter-argument goes, a losing ten-year number really would be damning, as it would signal the end was nigh.

With that in mind, I took a different pass through the data to determine when these rolling ten-year “fails” had occurred. For the counter-argument to hold, then those failing periods would have come recently--enough to dent the funds’ returns as mean reversion took hold, but not so deep as to push the funds’ since-inception returns below the benchmarks (that is, to make them losers).

What I found is that while a fair number of failing ten-year periods had occurred within three years of January 31, 2016, they hardly dominated. Indeed, nearly 60% of the failing ten-year periods happened at least three years prior, almost 40% five-plus years before. This would seem to rebut the counter-argument that high fail rates of winning funds are simply a result of recent performance tailing off. Rather, it appears that many of these winning funds successfully weathered performance setbacks they’d suffered years before. The exhibits below present the results two ways—by age of fund and by excess return quartile.

Technical Notes and Arcana

A few notes:

The study used end-of-period (January 31, 2016) Morningstar category classification to calculate “excess return”. Therefore, it is arguably vulnerable to distortion given that the funds might not have spent the whole since-inception period in that same category which, in turn, introduces the risk that the index isn’t a good fit. I could have corrected for this using other methods, such as examining the funds’ historical category classifications and then calculating since-inception excess returns based on the monthly “history-true” excess return stream. The truth is that this is a ton of work and hugely data-intensive and I didn’t have time. :( Could it materially distort the results? If anything, it would probably shrink the pool of winners and increase the losing rate of the winners that remained. So, think of the results presented above as the best-case—in reality, losing rates are probably a bit higher than shown.

The study is survivorship biased, as I was focused on “winners” as of January 31, 2016 and, well, you can’t win if you’re dead. L On one hand, that shouldn’t present an issue because we’re interested in assessing how often winners lose, not on how often all funds lose. But one could argue that if today’s mortality rate is lower than it was in the past, then perhaps the true losing rate over time is higher than we’ve presented here. (Suppose that some of those merged/liquidated funds were at one time winning funds but then performance eroded and, as that happened, their losing rates would have leapt.) If anything, that would mean the results presented above understate losing rates.

For the heck of it, I also calculated the cumulative excess returns of each winning fund since inception at one-month intervals. For instance, if fund A incepted on March 31, 2000, then I would have calculated its since-inception excess return versus its category benchmark on April 30, 2000, May 31, 2000, June 30, 2000, and so forth all the way to January 31, 2016. What I was interested in knowing was how much time, in aggregate, these winning funds spent lagging their benchmarks, on average, during their life. The answer? Winning funds finished with a lower since-inception return than their benchmark about 13% of the time.

[syouth1-blog]

U.S. Equity Fund Investor Success Rate

One of my favorite datapoints is “investor return”. If you’re unfamiliar, “investor return” estimates the performance of the average dollar invested in a fund. It’s calculated in essentially the same was as an internal rate of return, ie the return that would make the net present value of an investment (i.e., its future cash flows less the initial outlay) zero.

Calculating investor return is pretty simple. To derive it, we only need a fund’s beginning net assets, ending net assets, monthly investor asset flows to/from the fund during the period, and the fund’s monthly return during the period.

Once we’ve got a fund’s investor return, we can compare it to the fund’s published return that you’d see quoted on Morningstar.com. This offers a sense of how well investors timed their purchases and sales. For instance, if the fund’s published return exceeded its investor return for the same period, it suggests investors mistimed their purchases or sales.

Here’s an example of a fund that’s seen hideously large investor-return gaps. And, by contrast, here’s a fund that been used more prudently by investors.

Study

In general, researchers have found that investors do a pretty lousy job of timing their investments. Given their propensity to chase performance, this doesn’t come as a huge surprise. What investor return does is offer a way to quantify the damage investors do to themselves. (For more on this topic, check out my colleague Russ Kinnel’s annual “Mind the Gap” study here.)

But I thought I’d take a slightly different tack. What I’m interested in is how often the return on the average dollar invested in funds beats their benchmarks. This is a variation of other studies I’ve conducted examining success rates, pre- and post-tax. But here I'm trying to get a sense of how often investors--as represented by the average dollar invested--succeeded.

To that end, I compiled the published and investor returns of all unique U.S. equity funds that existed as of February 1, 2006. Then I made the following comparisons:

The funds’ published returns vs. the returns of their category benchmark indexes

This will tell us how many funds survived the ten-year period and beat their benchmark

The funds’ investor returns vs. the returns of their category benchmark indexes

This will tell us how many funds survived the ten-year period, beat their benchmark, and beat on an investor-returns basis (i.e., the return on the average dollar in the fund exceeded the benchmark’s return)

The funds’ published returns vs. their investor returns

This will tell us how many funds’ had investor-return “gaps” where the average dollar invested in the fund gained less than the fund itself

Findings

There were 2,366 unique equity funds at the beginning of the period (February 1, 2006). Of that number, 876 funds died at some point (through merger or liquidation) during the ensuing 10-year period ended Jan. 31, 2016, leaving 1,490 funds. 464 of the 1,490 surviving funds generated returns that beat their category benchmarks. And, finally, 197 of those 464 outperforming funds had investor returns that exceeded the returns of their benchmarks.

Here’s what it looks like in chart form:

So, roughly 37% of funds that existed at the beginning of the period didn’t survive to the end. And since only about one-third of those survivors beat their benchmarks, it means that only 20% of U.S. equity funds that began the ten-year period beat their benchmark by the end of it. But because investors mistimed their investments in more than half of those outperforming funds, what you find is that only 197 of the original 2,366 U.S. equity funds—about 8%--boasted 10-year

investor

returns that surpassed their benchmark’s.That’s a pretty low success rate, about one in twelve. And it’s even more sobering when you consider that we didn’t take risk or taxes into account. It looks bad.

Caveats

That said, a few caveats worth mentioning:We have investor return data for 300 of the 464 funds that outperformed (197 had investor returns that beat the benchmark over the ten-year period; 103 had investor returns that lagged the benchmark). What about the other 164? We don’t calculate investor returns for these funds because at some point they absorbed another fund through merger and the influx of assets from the merger would short-circuit the calculation. If the investor returns of all 164 of those funds would have beaten their benchmarks, then it would have pushed the investor-returns success rate to around 15%, best-case. Still low, but not as dismal as before. While not commonplace, there are underperforming funds (i.e., published returns less than benchmark returns) whose

investor returns

beat the benchmarks. Indeed, there were 73 such funds in our dataset. When taken together with the funds who’s published and investor returns beat the benchmark—the aforementioned 197 funds—the success rate rises to 11%. (Note: Of the 1,026 funds that survived and underperformed, 377 lacked investor returns for the same reason mentioned above.)Investor returns can be a quirky datapoint, sensitive to a number of nuanced, inter-related factors. This is something I’ve

written about

in the past. In addition, it’s important to keep in mind that a dollar-weighted return is just that—

dollar

-weighted. What that means is, hypothetically, you could have 99 prudent but small (in dollar terms) investors in a fund and 1 impulsive but large investor and that large investor’s impulsivity would drown out the small investors’ laudable behavior, at least where the investor returns calculation is concerned.Along  those lines, it's incorrect to conclude, based on the above findings, that around 90% of monies invested in U.S. equity funds have lagged the benchmarks. In calculating investor success rates, we're assuming funds are equal-weighted. In real-life, funds are all different sizes by AUM.  

Still…

All that said, it’s not a pretty picture. Of the 949 surviving funds that had a ten-year investor return, only 248 recorded a positive investor-return “gap” (i.e., investor returns > published returns). The remainder saw shortfalls (i.e., investor returns < published returns), indicating that investors had misused the clear majority of funds. What’s more, the gaps were pretty sizable—the average shortfall was 2.24% per annum. The following chart shows the distribution of investor-return gaps for the 701 funds that saw a shortfall over the ten-year period ended Jan. 31, 2016:

When you couple the lack of success—as evidenced by low investor-return success rates—with the large shortfalls shown above, it looks kind of grim.

Conclusion

Active investing isn’t hopeless but it’s more difficult than many investors realize. By estimating the odds of success, as I’ve done here and in other studies, I’m hoping that I can offer a more clear-eyed sense of the odds. In that way, investors exploring active investing can do so in a more informed way.

I believe that for the vast majority of investors, indexing is the better choice.

[syouth1-blog]

U.S. Equity Funds: Approximate 2015 Beat Rates

Now that 2015 is in the books, I had a look at U.S. equity fund returns and tallied up beat rates by category and for the asset-class as a whole. This yields a sense of how successful active investing was last year, in sheer numbers terms.

(For purposes of this exercise, I defined a 'beat' as a fund whose returns beat its Morningstar Category Benchmark Index over the full calendar. For Large Value category, the Russell 1000 Value Index would be the category benchmark, and so forth for the other eight categories. I counted each dead U.S. equity fund--which I defined as a fund which began the year in one of the nine categories but died during the year--as a fail. The study encompasses funds only, not ETFs, and I used oldest shareclass to limit to unique funds.)

Beat Rates

All told, around 40% of U.S. equity funds beat their category index in 2015. That's not exactly a banner year for active funds but worth mentioning that this represents a marked improvement over dismal 2014 campaign, as shown below.

Long story short, value funds had a much better time of it in 2015 than they did in 2014, though it's also worth noting that only two categories posted beat rates higher than 50% (LV and SV).

Why'd the year go better? Off the top of my head it probably reflects style 'messiness' that worked to active funds' advantage. For instance, it paid to have more exposure to core and growth in 2015. That probably benefited value managers, whose portfolios skew a little in that direction.

Beat Margins

I also calculated average 'beat margins', which is the average excess return of funds that beat their category index. I did so for 2015 and 2014 to allow comparison. What you'll see below is that beat margins widened a bit in 2015, so the payoff to active investing was--on average--a little better.

All told, the average U.S. equity fund that beat its index in 2015 did so by a margin of around 2.40%, vs. 1.80% in 2014. As one would expect, beat margins tended to be wider as you moved down-cap to mid- and small-.

Lag Margins

Lastly, I calculated average 'lag margins', which is the average excess shortfall of funds that failed to beat their category index. As for beat margins, I did so for 2015 and 2014 to allow comparison. The chart below shows that average lag margins narrowed some in 2015, meaning that the price of failure wasn't quiet as dear.

The average U.S. equity fund that lagged in 2015 did so by a margin of around 400 basis points, which is a tad narrower than the roughly 460 basis points by which it lagged in 2014. Most categories saw this sort of narrowing, though most pronounced among mid-cap funds.

(Note: Because funds that died in a given calendar by definition lack a full-year's return, I couldn't calculate their lag margins.)

Conclusion

One year of anything is usually noise and so I'd discourage anyone from drawing sweeping conclusions from the above. But I do find it interesting to examine annual beat rates, and how they cycle higher and lower over time, because it helps to frame expectations about the potential efficacy, and consistency, of active investing.

2015 appears to have been a better year for active fund investors, with more frequent successes and bigger payoffs on average than the prior year. But it was still a bit of a slog, with the overall beat rate stuck below 50%.   

[syouth1-blog]

No, Active Management Isn't Intrinsically 'Bad'.

Keith Akre of Kraken Capital Watch blog wrote a rebuttal to one of Ben Carlson’s blog postings which drew on data I’d pulled together on the after-tax efficacy of active management. The data showed that very few active funds beat their benchmarks post-tax over the 10-year period ended 10/31/15.

For starters, thanks Keith for reading Ben’s posting and taking the time to respond to it with a posting of your own. Much appreciated.

Straight off I’ll say that the data/chart was never meant to be a blanket condemnation of active management, which I think is how Keith received it. I feel pretty strongly that active/passive isn’t a decision that investors should approach with an either/or, black/white, binary mindset. There’s a spectrum of choices, bracketed on one end by cap-weighted indexing and on the other by, call it, ‘unconstrained active’ investing. (I don’t want to speak for Ben, but having read a lot of his stuff, I’d like to say that he has a similar m.o./orientation.)

That said, I feel even more strongly that investors should approach that decision–choosing among the options that line that active/passive spectrum–with eyes wide open. And because investors are often tripped up by emotion and various other behavioral tics, it’s even more crucial to try to frame the decision for them–as they’re not so good at doing it themselves. (I’ve written about this at some length, most recently here.) So, that’s why I think it's worth taking a step back and trying to set realistic expectations by drawing on the copious data available to us.

Which brings us to the question of whether active management is likely to succeed in a given context/set or set of circumstances. I’ve looked at this in many ways, shapes, and forms over time, riffing on topics like persistence of performance, predictiveness of certain factors, etc. ad nauseum. And it’s really in that spirit that I’ve offered recent tidbits/findings on things like active-management “beat rates” as well as the piece Keith referenced which was on after-tax efficacy.

(I should probably mention here that while the views I express here are my own, I am fortunate to work at Morningstar as part of the firm’s manager research team. We analyze thousands of strategies, active and passive, the world over. Somewhat ironically, we are often accused of being softies for active managers. What’s inarguable is that we cover many actively-managed funds and recommend–i.e., have assigned positive ratings to–many of them.)

With that preface out of the way, probably makes sense to turn to some of the specifics in Keith’s blog posting, beginning with this portion I’ve excerpted below:

Instead the conclusion should be “If you randomly select a mutual fund from the entire universe of available mutual funds, you have a high likelihood of choosing a fund which underperforms its category’s index fund.”  Those are two very different arguments to make.  Just because a lot of funds lag their respective indexes does that mean that investing in any actively managed mutual fund is a bad thing?  I could take this same data and write a story that says “Data shows that multiple actively managed mutual funds outperform their respective indexes after taxes and fees over a long-term basis.”  If I had the ability to identify the outperformers, would I care how many laggards there are?

I don’t think that we’re too far apart here, but it’s a question of degree. I think many investors would be shocked by the sheer number of active funds that fail to beat their benchmark over time. Thus, your restatement of the conclusion–that many active funds outperformed their indexes—is correct but kind of incomplete without context, for the multitudes of outperformers are far outnumbered by scores and scores of underperformers.

Here’s another key section of his posting:

So here are my main objections to the study mentioned above:

1. I do not randomly select mutual funds.

2. Some funds are classified into categories by Morningstar that may not be the same objective outlined by the fund managers, making the benchmark of the study inappropriate.

3. Funds may be invested in for reasons other than beating a generic index (desire for income, completion of a portfolio strategy, exposure to a specific risk factor, etc…)

4. The assumptions of the data are that mutual funds are purchased (oldest surviving share class, which also happens to most often be the most expensive), held with no regard to tax appropriateness at the highest marginal rate, and then sold at the end of the period without regard to tax consequences).  It also assumes that any discontinued, or non-surviving funds are automatically categorized as laggards.  I know that in order to have some kind of academic rigor basic assumptions need to be made, however, this does not make it reflective of the real life experience.

On 1, no one chooses randomly. The point of the study was to illustrate how very little margin for error there is even for those who are choosing studiously. Getting it right pre-tax is really hard. The aperture narrows even further after-tax.

On 2, this is valid to a point but it’s also worth considering the way I conducted the study. Namely, I aimed to recreate the opportunity-set an investor had about ten years ago and thus used beginning-of-period category classifications (not end-of-period, as is more common). Practically speaking, I don’t think using 'prospectus benchmark index’ (which is the manager’s handpicked index) would have made a huge difference. I would have tested excess after-tax returns vs. prospectus BM index but I can’t access historical prospectus benchmark data from our DB.

On 3, that’s true. And 10-15 years ago I agree that active funds would have been pretty much the only game in town to goose one’s yield, complete a portfolio, or lever up to a given factor. But the world has changed–there’s been a proliferation of products that allow investors to pretty surgically do the same at a fraction of the cost of a trad open-end active fund. So, I can’t say I’m too sympathetic. Notwithstanding that, Keith is talking about edge cases, as there are far more middle-of-the-fairway funds in this universe I studied than there are these specialized, bolt-on types of vehicles he describes.

On 4, I disagree with Keith’s assertion that 'oldest shareclass’ funds are most often the expensive shareclass. To put this in perspective, when I run a screen today for all US equity funds and limit to oldest shareclass, the average expense ratio is 1.07%. When I don’t limit to oldest shareclass, the average expense ratio is 1.18%. Granted, this is based on today’s expense ratios, not those that prevailed ten years ago. But if anything I would expect the oldest-shareclass method to have been more friendly to active funds if I’d run it as of 10/31/05 given that it would have screened out many expensive load shareclasses (which were far more abundant at that time vs. today).

I also disagree with Keith's suggestion that assuming dead funds were 'fails’ isn’t real life. To put this in perspective, had I used end-of-period category classifications instead of beginning-of-period classifications, it would have been far less realistic in the sense that it would have, by definition, excluded all of the funds that existed at the beginning of the ten-year period but died before reaching the end of the period. Having studied the data set, I also don’t think it’s unfair to treat these dead funds as 'fails’–most are mothballed because performance has been poor, after all, and while perhaps if they’d lived they would have resurged, we’re talking real world here and the reality is that had an investor chosen such a fund at the beginning of the period, well, they’d have been out-of-luck.

A little deeper into the post Keith wonders aloud whether the quirks of the oldest-shareclass method might be enough to explain the difference between beating and lagging. The difference he mentions is 0.25%.

For instance, this study uses “oldest surviving share class”.  The fund screener on Morningstar states that the oldest surviving share class “is often (but not always) the A share class” which is almost always the most expensive share class available.  This can make for a meaningful difference in fund performance.  Most of the differences in expense ratios between A shares and the cheapest available option I found in the small cap growth category were around 0.25% per year.

As mentioned above, I find the opposite–using oldest shareclasses guides one towards cheaper funds, on average, but not pricier funds. But putting that aside, I thought it might be useful to provide some data on the average excess returns of all funds in these categories ('All’), the many that lagged ('Lagged’), and the precious few that beat ('Beat’). It follows below.

The takeaway from the above is that we’re not typically talking about 25bp shortfalls. On average, funds in these categories were lagging by 1% - 1.50% p.a., which means a supposed 25bp handicap isn’t going to close the gap much.

(To answer a question Keith poses in his posting – yes, there have been that many fund failures/liquidations in the small-growth category, which is basically a boneyard. I would also note that the test he runs for SG funds–in which he finds that about a fifth of funds did better than a relevant Vanguard SG index fund after-tax–is materially different than the study I ran, which was based on beginning-of-period classification. It’s therefore riddled with survivorship bias–the hundreds of SG funds that underperformed and were killed off since 2005.)

While I feel good about the robustness of the study, I think it’s open to a number of criticisms:

Keith rightly notes that the assumptions made in tax-adjusting the returns could be too punitive. This is fair. I don’t agree with him that the assumptions imply that all of these investors would have been well-heeled, well-advised types. (It self-contradicts in asserting that the tax rates are exceedingly high on one hand–because the investors concerned aren’t that well off–and on the other hand that they’re loaded and therefore would engage an advisor and have access to low-cost shareclasses.) However, I would note that the reason I compared the active funds to Vanguard index MFs is so that those index MFs would be subject to the very same tax-adjustments.

Though Keith doesn’t note it in his post, it’s worth noting that tax-adjusted returns are also load-adjusted. This isn’t a Morningstar thing. It’s an SEC thing, as they stipulate the calculation method. The upshot is that it could be too punitive toward load funds. However, because oldest shareclass method mostly resolves to front-load shareclasses (w/re to load funds) and because we used a ten-year measurement period (over which the load could be worked off), I don’t believe this significantly understates load-fund after-tax returns. But it’s there and worth noting.

I didn’t risk- or style-adjust the after-tax returns. It’s possible–not probable, but possible–that some of these active funds were a lot less volatile than the index funds concerned and so when you risk-adjust you find the beat rate is higher. I wouldn’t hold my breath on that one, but it’s not something I studied. Also, it’s possible–as Keith alludes to in his posting–that the beginning of period category classification isn’t representative of the funds’ style exposures throughout the ten-year period concerned. Again, I was trying to simulate the investor’s opportunity set at that T-10 time and so any style drift thereafter was peripheral to that research objective. But it’s another issue to note.

I’m sure there are other potential problems but off the top of my head those are probably among the most-valid criticisms one could levy.

Long post. Sorry about that. Hope some of it is useful.

Happy new year.

Jeff

[syouth1-blog]

MLP Funds: The Hottest Mess?

Earlier today Josh Brown reran a good post he'd written earlier this year about MLPs. His message--they aren't bonds. It's well worth reading.

It got me to wondering how investors in MLP funds (which my employer, Morningstar, classifies as "energy limited-partnership" funds) have fared amid a boom and bust that have occurred in pretty close succession, something that can spell trouble.

To that end, I pulled a little data to examine the relationship between recent asset flows (ETFs and MFs) and returns in the energy limited-partnership category (ETFs and MFs). 'Catastrophe' might be too strong a word to describe what appears to have unfolded, but 'ugly' is not. Here's exhibit A:

So what you're looking at above is a chart that juxtaposes the energy limited-partnership category's returns--plotted as a simple growth of $10k (the red line)--and the monthly net flows into the category (the blue'ish bars). And what you're probably noticing is the torrent of money that flowed into these funds in the year or two before the 2014 peak. All told, investors poured about $48 billion into MLP funds over this span. Uh oh. 

But there's another way we can look at the data: By breaking down the category's returns into shorter time segments, comparing performance from one segment to the next, and observing the amount of money that flowed into the category over that span. What we're trying to determine is when returns were trending higher, when they were trending lower, and how much money was flowing in (or out) at those different times.

At the risk of oversimplifying, investors are buying opportunely when they pump money into an asset whose returns are trending higher, and they're selling opportunely when they pull assets as returns are deteriorating. The inverse also holds: They're buying inopportunely when they invest at a time returns are trending lower, and they're selling inopportunely when returns are trending higher. As you can see from the chart above, MLP funds have seldom been in outflow in recent years so we're really talking about how opportunely investors timed their purchases.

To evaluate that, I broke the time period above into rolling six-month segments at one-month steps. So there'd be a segment that ran from 4/1/09 - 9/30/09, another from 5/1/09 -10/31/09, and so forth. I then compared the category returns of non-overlapping segments, i.e. the 4/1/09 - 9/30/09 six-month period vs. the 10/1/09 - 3/31/10 six-month period, to determine if performance was trending higher or lower over the full 12-month rolling period in question, and repeated for all other rolling one-year periods. Finally, I compared that return differential (ie, difference between the non-overlapping six-month return segments mentioned) against flows over that 12-month period and then plotted the results, as shown in the chart below.

The vertical axis of the chart is flows. As mentioned, this category has been in inflow for virtually the entirety of the period examined, explaining why all of these numbers (which are rolling 12-month figures) are positive. The horizontal axis is the 6-month return differential during the rolling 12-month periods concerned. An example will help to explain: In the 12-month period ended January 2015, investors shoveled around $15.7B into MLP funds. In the first six months of that 12-month period (2/1/14 - 7/30/14), the average MLP fund gained 12.7%; in the last six months (8/1/14 - 1/31/15), the average fund lost 8.3%. Thus, the return differential was a grim -21.0%, meaning performance was trending sharply lower at the time investors were piling into MLP funds.

How to interpret the chart? Long story short, anything to the left of the line that bisects the chart is "bad"--it's investors contributing assets to the average MLP fund at a time its returns were trending lower. The higher to the left you move, the more inopportune the purchase. By contrast, anything to the right of the line is "good"--it's investors contributing assets at a time the average MLP fund's returns were trending upward.

In the final equation, the chart doesn't look too good. There are far more 12-month periods in which performance was trending lower and yet investors were dumping money into the funds. When this happens, dollar-weighted returns (aka, IRR) are well below time-weighted (aka, stated) returns, resulting in a returns "gap". Put another way, bad as MLP funds' stated returns have been lately, they probably pale in comparison to the returns investors have actually earned. It looks like a hot mess.

And here's the thing: I concluded the study in October 2015 because I only have flows data up to that date (Morningstar will publish November 2015 flows data in another day or two). The average MLP fund's performance was awful in November (-8%) and has been downright atrocious thus far in December (-15.1% through Monday). Had I captured that performance, the results almost certainly would have turned out worse.

-     -     -     -     -

Technical notes:

- I used the US OE Energy Limited Partnerships category average for purposes of calculating growth of $10K and the six-month return differentials. There was a gap at the beginning of the period that predated the category's first month return. During that stretch, I used the returns of the Morningstar Composite MLP TR USD Index in lieu of the category average. 

- Flows include US OE MF and ETFs classified as 'energy limited partnership'. They do not include FoFs.

- Given the popularity of ETFs in this category and the fact that a lot can happen intra-month (I used monthly, not daily, flow figures), it's possible that the study is imprecise. That said, it does not seem like a good idea to assume that investors trading these funds intra-month were more adept at timing their purchases and sales.

[syouth1-blog]

Are Active Fund Managers Too Skilled for Their Own Good?

Figured it was about time I got a Tumblr for sake of those who'd otherwise have to deal with 30+ tweet storms.

Building on some of the points that Ben Carson made in a recent blog post (Ben riffed a bit on manager skill and the possibility that the active investing field might be becoming too skilled for its own good; it's well worth reading, like all of his stuff), I decided to mess around a bit by examining trends in gross, i.e., "pre-fee", excess returns of U.S. stock funds. My thinking was that if highly-skilled active managers were cancelling each other out, we should see signs of that in the gross excess returns they generate (i.e., they'd be smaller), so let's examine those over time.

To that end, I took a look at the 157 rolling seven-year periods that spanned the 20-years ended November 30, 2015. For each rolling seven-year period, I tallied up the annualized excess returns of all US MFs (cats included LV, LB, LG, MV, MB, MG, SV, SB, SG). I then calculated the arithmetic average of those excess returns for each category and period. Those averages formed a time series, shown below. (For further technical details on the way I conducted the study, see the bottom of this post.)

So, there's a lot going on in the chart above but a few takeaways:

Recent rolling 7-year excess gross returns are the lowest we’ve seen in two decades. The average fund generated a measly 12bp gross excess return in the seven-year period ended 11/30/15. What’s more, the average was negative in five of the nine categories (LB, LG, MV, MB, MG).

Excess gross returns appear to be pinching lower across nearly every category, though seems to be more pronounced in smid and growth categories. This would seem to imply that it has become more difficult for skillful managers to add value before fees.

There are some caveats to this. For one, it is not risk-adjusted so it’s possible—though I would not say probable—that things could look a little better once risk is factored in. Second, this uses end-of-period classification, which could misrepresent a fund’s positioning/style throughout the seven-year period concerned. (Fwiw, I also looked at this based on fund’s beginning-of-period category classification and it looked even worse.)

From this, one can reasonably infer some things about ‘optimal pricing’ of active funds. Namely, if they cost more than 100bps, all in, then there’s not going to be much after-fee excess returns to speak of. (For anyone wonder, the ‘excess returns > 1%’ beat rate (i.e., funds generating excess returns >1% as % of all funds) is also about the lowest it’s been in two decades, at ~33%.)

I would caution against literally interpreting the results above as 'look! skilled active managers are tearing each other to shreds, so they're doomed...doomed!'. For one thing, the 'skill' shown earlier in the time lapse (coming out of the tech/telecom/internet boom especially) is probably greatly exaggerated. By the same token, there are likely some factors unique to recent periods that have depressed active manager performance and as those conditions ease the efficacy of active investing will improve. So, as with most things, the 'truth' is probably somewhere in the middle--the payoff to active investing is less than before though not as dismal as implied by the downward slope of the chart above.

That said, the results are sobering for the US active investing complex, which is still charging a good deal too much. Tough game which should be priced as such.

*     *     *     *     *

Technical notes for my fellow nerds out there:

To build the list, I searched for all non-index funds in the M* "U.S. Equity" broad category group, including dead funds. This returns the funds that are, or were, categorized in the nine style-box categories mentioned above.

I then narrowed the list to unique funds by oldest shareclass or, in the case of the dead funds, chose a shareclass at random--sorry, that's how I had to do it because we don't tag dead funds with an 'oldest shareclass' designation.

I used end-of-period category classification for each fund. I could have used beginning of period classification, to approximate the choice-set that investors had at each of those 157 times but since we were calculating excess returns vs. a M* category index ex-post, seemed fairer to use end of period classification.

To calculate gross excess returns, I compared each fund's gross returns vs. the returns of its M* category index. Long story short, every category has an index assigned to it...the "M* Category Index". So, for LV it's Russell 1000 Value, for SG it's Russell 2000 Growth, and so forth for the other seven.

I chose a rolling seven-year period because it struck an acceptable balance between the kind of sheer noise you'd get using a shorter rolling period (which boasts the advantage of greater inclusiveness and thus data richness) and the practical challenge of using a ten-year rolling period (so long a period that you effectively screen-out most of the dead funds, few of which live more than ten years).

For those who are going to jump all over me for merely suggesting that the average gross excess returns for active funds could be anything other than zero in any of these seven-year periods, consider that this only includes US funds, only includes those with a seven-year return, etc. These funds don't represent the 'market portfolio' in aggregate so let's not assume zero-sum.)