Thu, 31 Oct 2013
Research by author Rick Ferri took the active-passive debate beyond individual investments and found passive-oriented portfolios will likely outperform active portfolios across time spans, number of holdings, and asset types.
Christine Benz: Hi. I'm Christine Benz for Morningstar.com. I recently attended the annual Bogleheads Conference, where I had the opportunity to sit down with investment advisor and author, Rick Ferri. We discussed his recent research on indexed versus actively managed fund portfolios.
Rick, there's been a lot of research comparing how actively managed funds have done versus their passively managed counterparts. You recently issued a white paper, where you looked at actively managed portfolios versus index fund portfolios. I'd like to start by discussing briefly your methodology, how you assemble those active portfolios and also what you looked at in assembling the index fund portfolios.
Rick Ferri: There's a lot to that answer because there is a lot to the study. But what we were interested in is moving the debate between active and passive to a different level. It has always been on the individual asset-class level: How many large-cap U.S. active funds outperform the S&P 500? How many bond funds outperform the Barclays Aggregate Bond Market Index? We thought that we need to move this to a different level, which is we need to be looking at indexing versus active management on a portfolio level because people just don't own one large-cap U.S.-equity fund and they're done, or just a bond fund and they're done. They combine U.S. equities, international equities, bonds, perhaps Treasury Inflation-Protected Securities, different asset classes, real estate.
We wanted to look at how at a diversified portfolio of index funds would perform relative to a diversified portfolio of actively managed funds from the same categories. We wanted to see whether the index funds at a portfolio level had better results than at the individual level or worse results, and that was the purpose of the study.
Benz: When assembling that portfolio of active funds, the natural question is how did you decide how to assemble those portfolios? I know you used the CRSP database; that's a survivorship bias-free database, meaning that it incorporates dead funds, funds that have gone away oftentimes because they were lousy. But how did you come up with those portfolios, and how did you attempt to account for the fact that investors oftentimes do use some screening when they put together their active portfolios? Maybe they look for strong-performing funds or funds with low expense ratios or whatever it might be?
Ferri: Again, that's a long answer. The database itself needed to be cleaned because within each category--there were categories in the CRSP database. But let's say the large-cap U.S. category also contained some more mid-cap funds and balanced funds, so we had to take those out. They also contained all different classes; A shares, B shares, C shares, institutional shares. So, B shares, of course, are back-end load and they have high fees, so we took out B shares. C shares are a level load, with a high fee all the way through. So we took those out. Institutional shares you have to have $1 million to buy, which we took those out. So we were left with the A shares, which a lot of times have front-end loads, but we didn't count the loads and all of this.
We had to cull out the database based on the costs and the different share classes, and also clean the database to make sure that there were no funds in there that really shouldn't have been in that category. For each of the different categories--there were 10 different categories that we came up with--we did that. Now, as far as constructing the portfolios, we constructed an index fund portfolio, which we used a very simple one. It was a three-fund portfolio of total U.S. stock market, total international, and total bond index funds. We used the Vanguard funds initially in this three-fund portfolio design because they have the most history. We didn't do any hypotheticals. We used actual fund performance that investors could have gotten these returns in these funds. It was all actual. So we used investor share class, which is the highest-cost Vanguard fund, and we always used the index fund because that was the oldest for each of the categories. They weren't all Vanguard funds. There was a few iShares funds in there as well in a couple of categories because they were the oldest index funds.
But then we just simply put a portfolio together very simply of 60% stock, 40% bonds. Of the stock portion, it was 40% in U.S. total stock market, 20% international total market. Then the 40% was in the Vanguard Total Bond Market Index fund, and we carried that forward. Over here on the active side, from each of those three categories, we randomly selected a mutual fund that was in the database. Now, as you said, this is a survivorship-bias-free database. If we randomly selected a U.S. equity fund and then started running forward and that fund merged or liquidated, then we stopped at that point, picked another U.S.-equity fund randomly from what was available at that particular time, and then we kept going forward.
So we had a consistency of the portfolio of active funds [to mimic] what an investor would do. We tried to mimic as close as we could the investor experience, at least randomly selecting funds. Then we compared the two together to see what the performance was, and we did it 5,000 times. That was the breakpoint that if you did it more than that, it didn't really make any statistical significantly different outcomes. If you did it less than that, there was some statistical difference in the outcome. This is what we did for all 32 of the studies that we did with this.
Benz: I want to get into the conclusions. But before we get into that, first, I'd like to just follow up on that randomized idea in terms of your active sample because our research at Morningstar shows that investors don't select funds randomly. In fact, when we look at where flows are going, generally speaking, investors choose lower-cost funds. They appear to be shopping on track record some of the time. So you did incorporate a basic sort of expense cut at least to account for that fact.
Ferri: Well, initially we did all funds, regardless of what expense they were because we wanted to look at the raw numbers versus the whole database. But we know that a lot of investors will use factors to determine which funds to get into and that cost is a big factor. So we ran everything again looking at only the lowest-cost funds, basically the 50 percentile down. We only used those funds. In every category lopped off the higher half, so we didn't use those. And we ran all the numbers again and all the different tests that we did. So that was the one factor that we did look at. We did many, many tests. It took many months to compile all this data.
Our goal was to show that this is a different way of looking at active versus passive, and that the next iteration of this, which someone else is welcome to take the algorithms and continue to work on--maybe Morningstar would look at that we're not just randomly selecting funds or not just randomly selecting low-cost funds. We have different ways in which people select funds. We realize that, and that's the next iteration of all of this. We just stopped with this one test of the lowest-cost and that's where we ended the study.
Benz: That actively managed data set, you mentioned front-load funds, but also no-load funds were in the mix, as well.
Ferri: That's right. There were front-loaded funds, but we didn't count the load. And then there were no-load funds.
Benz: In terms of findings in the interest of full disclosure, your firm does manage index fund portfolios for your clients. So you definitely are a believer in index products. But in terms of your takeaways from all of this research, what were the general conclusions?
Ferri: The general conclusion was there is what we call the passive portfolio multiplier. As you add more index funds to a portfolio, the probability of the portfolio outperforming a portfolio of active funds moves up incrementally. So if you're looking at just one category, large-cap stocks or bonds, you're on the 75% or so level. In other words about 75% of the time an index fund in each of the categories will outperform the majority of the funds, including the ones that didn't survive. But when you put it together in a portfolio, U.S. stocks, international stocks, and total bond, you put those all together in a portfolio, the probability of the portfolio outperforming a portfolio of active funds actually moves up into the mid-80s. So you get this bump up in performance relative to active funds.
The reason for that is because when you look at the actual data of active funds outperforming index funds in each of the categories, the ones that outperform, outperform by [a little], but the ones that underperform, underperform by [more than that]. So even though 75% underperform, they're underperforming by a lot, and the 25% that outperform are only outperforming by a little. So you could have a portfolio where six out of 10 of the funds outperformed their benchmarks, but the portfolio itself underperformed indexes because the outperformers were only outperforming by a little and the underperformers were underperforming by more. So it is interesting.
Benz: You had a related conclusion, which is that the more active finds one layered on within a given asset class--and we know that investors sometimes do this, they own say multiple large-cap funds, for example--you found that the more funds that were in the mix within a given asset class, the more the advantage accrued to the index portfolio.
Ferri: That's correct. In fact, my co-author, Alex Benke from Betterment, who did all the statistical work, it was his suggestion that we do this. He said that what they found, and I agree with him, is that a lot of investors will buy two or three or four U.S. equity funds or two or three international funds or two or three bond funds. So what we did was we ran the study, instead of just looking at one fund in each category, we ran the study looking at two funds in each category and three funds in each category versus the index fund.
We found that when we ran those numbers, the probability of the index fund portfolio outperforming started going up considerably. So a term that Peter Lynch used to use is "diworsification." It's actually better to just pick one actively managed fund and hope for the best than it is to diversify among actively managed funds in each category because your probability of outperforming index funds actually goes down when you do that.
Benz: You also found that time horizon matters here, and that here again the longer the holding period, based on your research at least, the more the advantage accrues to the index portfolio.
Ferri: That's correct. We looked at a lot of things. We looked at time advantage; we looked at adding more asset classes to a portfolio. Everything we looked at, when we increased the time from five years to 10 years to 15 years, the probability of the all-index fund portfolio outperforming incrementally grew. When we looked at going from three asset classes to five to eight to 10, the probability of the all-index fund portfolio outperforming grew. Then as investors started putting more actively managed funds in the portfolio, two or three per asset class, the probability of the all-index fund portfolio outperforming grew. In the long term, an all-index fund portfolio all the time really has the highest probability of getting people to their financial goals.
Benz: Rick, thank you so much for being here to share this research.
Ferri: Thank you very much.