Is the case for active management stronger in retirement?

The company posits that the top retirement managers will be those who boast the qualities identified in its 2014 paper on accumulation--low expenses and being sponsored by an organization wherein the investment managers tend to own a high amount of their own funds. These are clear benefits that meet with the approval of none other than Vanguard's Jack Bogle. Low costs and high manager ownership are good things.

So, too, is the third attribute added for low downside capture. That name smacks of jargon, but the concept is straightforward and the desirability obvious. Downside capture is measured by comparing a fund's return with that of the index, for months when the market declines. Funds with low downside capture are those that did not "capture" the market's full losses--ideal for a fund that must meet investor withdrawals.

After a few introductory pages, the paper begins in earnest by showing how the three screens improve the quality of actively managed funds. American Funds uses a 4% annual withdrawal rate, increased by 3% per year. Its first test is to compare the results for the unscreened group of all active managers in the United States to the much smaller screened group of funds that met the company's three filters. The paper measures the percentage of funds that beat a 50/50 mix of the relevant two indexes (S&P 500 and MSCI EAFE All Country World ex-USA) over various rolling 10-year periods from 1995 to 2014.

The results are spectacular.

One hundred percent is a figure that rarely fails to attract attention. I did not think it possible that 100% of the screened funds, in 100% of time periods, could have beaten the indexes. And indeed they did not. That figure represents the percentage of times that the pool of screened funds succeeded. That is, the monthly returns for all the screened funds were averaged, so that a single return stream was created. These monthly returns were connected to form the long-term return, which was then compared against that of the index blends.

That is fine, in a sense. Academic research frequently uses such a pooled approach. On the other hand, investors certainly will not be buying 52 U.S. large-cap funds! Thus, the test cannot be said to mimic an investor's experience. Also, blending the monthly returns mutes some of the individual funds' volatility, which is directly relevant for a study about fund behaviour through withdrawals. I do not think that pooling accounts for most of the screen list's superiority, but in the absence of additional data I cannot say for certain.

Another issue: Is "Key Steps" the value effect in disguise? In sorting on downside capture, the paper tilts toward lower-volatility "value" stocks that tend to carry relatively low price multiples and have relatively high yields. Perhaps the screened active funds outperformed merely because they were value funds; a blend of value indexes would have done just as well (or better).

So I checked on that. One chart compares the 20-year performance for an index portfolio that was 50% S&P 500/50% MSCI EAFE All Country World ex-USA against an active screened portfolio composed of 50% U.S. stock funds and 50% foreign stock funds. I re-ran the numbers while using value indexes--Wilshire 5000 Value for domestic stocks and the MSCI EAFE Value benchmark internationally. The margin did shrink, but only modestly. The value effect exists, but it does not dominate.

What's the catch, you ask? You know there is one, if for no other reason than the wording of this column's deck.

Well, here it is ... and it's a mighty big one. The study does not identify the screened active list of funds before the test period of 1995 through 2014. Instead, it uses current data on manager ownership and in-period data for the two screens of expense ratios and downside capture. American Funds says that it conducted the study that way so as to avoid survivorship bias. For the 20-year figures, which require that funds be in existence for the entire time period, American Funds would have been required to toss out the funds that merged or liquidated during the study period, which would have biased the figures upwards.

Hmmm. For me, the cure was worse than the disease.

At any rate, this is a list of active managers that one would have loved to have owned in hindsight. So "Key Steps" is not a proof. It is instead a theory paper--a theory of how a portfolio that uses strong active managers, if they could somehow be identified before the fact, could handily outdo indexed strategies in meeting retirement needs.

I still think the thesis could work. There's no way to get manager ownership data from 1994, because U.S. authorities did not start collecting that until a few years ago (it is still not collected in Canada). But it appears to be a fairly stable data point, particularly as American Funds calculates it for companies overall, as opposed to each fund individually. The second screen of expense ratios rarely changes. So if the third factor of relative downside capture rates proves to be relatively stable (which American Funds says is the case, and which strikes me as correct), then one could do what American Funds did not--use the screens in advance to identify funds that retain (most of) their positive features of high manager ownership, low costs, and better bear-market performance.

I will have my assistant run some numbers to see.