Rob Carrick wrote an article late last year that suggested four actively managed mutual funds which may be worth a look by open-minded DIY investors. All four funds had beaten the S&P/TSX Composite Index’s 8% return over the last 10 years (as of October 31, 2014). Three out of four of the funds had even crushed the index return since their inception.
Without more information, it would be difficult for DIYers to determine whether the outperformance was merely the result of tilting the portfolio towards known risk factors (such as the small cap or value factors), or if the managers had any skill that may resemble alpha. By running 3-factor regressions on the monthly returns for each fund since inception, we can better understand what factors are driving their returns.
Mawer Canadian Equity – A
Since July 1991, the fund has outperformed the index by +1.00%. After the fund returns are put through the wringer, we find that most of the excess returns are a result of the fund’s value tilt over the period (HML coefficient of 0.29). The true alpha of the fund drops from 1.00% to -1.17% (almost identical to the fund’s MER of 1.21%).
Performance: July 1991 to October 2014
| Fund / Index |
10-Year Annualized Return (%) |
20-Year Annualized Return (%) |
Annualized Return Since First Full Month (%) (07/1991-10/2014) |
| Mawer Canadian Equity – A |
10.60 |
10.66 |
9.87 |
| S&P/TSX Composite Index |
8.01 |
8.73 |
8.87 |
| Difference (%) |
2.59 |
1.93 |
1.00 |
Sources: Morningstar Direct, Dimensional Returns 2.0
Regression Results: July 1991 to October 2014
| Fund |
Adjusted R Square |
MKT | SMB | HML |
Alpha (annual) |
| Mawer Canadian Equity – A |
0.90 |
0.89 | -0.13 | 0.29 |
-1.17 |
Sources: Morningstar Direct, Andrea Frazzini Data Library, Dimensional Returns 2.0
Leith Wheeler Canadian Equity – B
With annual outperformance relative to the index of +1.49% per year since May 1994, this fund would appear to be adding value for investors. Unfortunately, when we stress test the returns since inception, the alpha drops from +1.49% to -1.59% (once again, very close to the annual MER of 1.49%). Similar to the Mawer fund, most of Leith Wheeler’s excess returns came from its tilt towards the value factor (HML coefficient of 0.32).
Performance: May 1994 to October 2014
| Fund / Index |
10-Year Annualized Return (%) |
20-Year Annualized Return (%) |
Annualized Return Since First Full Month (%) (05/1994-10/2014) |
| Leith Wheeler Canadian Equity – B |
8.45 |
10.28 |
10.09 |
| S&P/TSX Composite Index |
8.01 |
8.73 |
8.60 |
| Difference (%) |
0.44 |
1.55 |
1.49 |
Sources: Morningstar Direct, Dimensional Returns 2.0
Regression Results: May 1994 to October 2014
| Fund |
Adjusted R Square |
MKT | SMB | HML |
Alpha (annual) |
| Leith Wheeler Canadian Equity – B |
0.90 |
0.95 | -0.09 | 0.32 |
-1.59 |
Sources: Morningstar Direct, Andrea Frazzini Data Library, Dimensional Returns 2.0
Beutel Goodman Canadian Equity – D
This fund is already starting underwater with a negative alpha of -0.15% per year since inception. Although this doesn’t seem so bad (considering the fund has a cost of 1.38% per year), a 3-factor regression exposes the true alpha to be an even lower -2.20%. Once again, the value factor helped the fund achieve performance that was better than expected (with an HML coefficient of 0.33).
Performance: January 1991 to October 2014
| Fund / Index |
10-Year Annualized Return (%) |
20-Year Annualized Return (%) |
Annualized Return Since First Full Month (%) (01/1991-10/2014) |
| Beutel Goodman Canadian Equity – D |
8.68 |
9.10 |
8.90 |
| S&P/TSX Composite Index |
8.01 |
8.73 |
9.05 |
| Difference (%) |
0.67 |
0.37 |
-0.15 |
Sources: Morningstar Direct, Dimensional Returns 2.0
Regression Results: January 1991 to October 2014
| Fund |
Adjusted R Square |
MKT | SMB | HML |
Alpha (annual) |
| Beutel Goodman Canadian Equity – D |
0.89 |
0.88 | -0.08 | 0.33 |
-2.20 |
Sources: Morningstar Direct, Andrea Frazzini Data Library, Dimensional Returns 2.0
Fidelity True North – B
This was the only fund that had a positive alpha before and after running the 3-factor regressions. The +1.16% alpha dropped to +0.50% per year after the regression results were in (for those of you who are interested, I subsequently ran a 4-factor regression with the additional momentum factor, which dropped the alpha to -0.15% per year).
Performance: October 1996 to October 2014
| Fund / Index |
10-Year Annualized Return (%) |
20-Year Annualized Return (%) |
Annualized Return Since First Full Month (%) (10/1996-10/2014) |
| Fidelity True North – B |
8.22 |
N/A |
9.32 |
| S&P/TSX Composite Index |
8.01 |
8.73 |
8.16 |
| Difference (%) |
0.21 |
N/A |
1.16 |
Sources: Morningstar Direct, Dimensional Returns 2.0
Regression Results: October 1996 to October 2014
| Fund |
Adjusted R Square |
MKT | SMB | HML |
Alpha (annual) |
| Fidelity True North – B |
0.94 |
0.92 | -0.05 | 0.04 |
0.50 |
Sources: Morningstar Direct, Andrea Frazzini Data Library, Dimensional Returns 2.0
Once the returns are put through a 3-factor analysis, most of the funds’ alpha disappears and even turns negative. Even though this analysis may seem somewhat sophisticated, it serves no useful purpose in determining which funds or risk factors will outperform in the future. It does suggest that if you are looking to add a value tilt to your portfolio (in order to increase your expected returns), there are cheaper ways to do this using low-cost ETFs, rather than paying for traditional active management.
[…] that they use the best managers who have the best chance to outperform the market. They can even manipulate the numbers to look better over certain periods of time. Only a very small percentage of individual managers […]
Great article! I’m curious as to why the DFA core and vector Canadian and US funds have not had great returns though they are more heavily tilted. I understand value has been doing poorly for the US but I thought it was doing well for CAD and Int. Thanks!
@Jon: The DFA Canadian Vector Equity Fund Class F (DFA600) returned 3.15% annualized between August 1, 2011 and August 31, 2018. During this same time period, the Canadian stock market returned 6.38%, while Canadian small cap value stocks returned 1.72% (so tilting towards these small cap value stocks hurt the performance of DFA600).
The DFA US Vector Equity Fund Class F (DFA223) returned 8.64% annualized between November 1, 2003 and August 31, 2018. During this same time period, the US stock market returned 9.65%, while US small cap value stocks returned 9.14% (so the underperformance is a combination of tilting towards these small cap value stocks, foreign withholding taxes, and management fees of the fund).
Could you do the same analysis for the Tangerine Investment Funds that you recommend in your white paper “The One-Fund Solution”?
@George – broad market index funds would generally be expected to have zero factor tilts (with a market beta of 1.00). If the index fund excluded small caps (as some of them do), they would be expected to have a slightly negative SMB factor loading.
Tristan,
I’ll only say one more thing. If it’s evidence you want you will have to wait quite a while before you get that with any fund tracking the DJ Select Value Index. Index fund XCV should be embarrassed that an actively managed fund is consistently doubling their returns. The money being left on the table is from investors choosing XCV over a consistent outperformer. If I had to choose between these two it would be a no brainer decision. The Canadian value index carried no value to anyone prior to 2006 as it wasn’t tracked by any investment. Thanks for your input but I am performance driven. I have all the evidence I need. I have consistently beaten the markets since I went DIY and when I no longer want to or cannot manage my own portfolio I’ll go with a low fee managed company like Mawer that has provided outstanding service and performance for it’s customers.
Bernie, I agree that one should be comfortable with whatever investing style that one chooses. But I prefer to look at the evidence before making that decision, otherwise you are leaving money on the table.
Regarding your comment about the Mawer Fund, on the contrary, Justin showed that had you invested in a fund or ETF that tracked the DJ Select Value Index (had such a security been available) from July 1997 to October 2014, you would have done better than the Mawer Fund. This 14 year time period carries more weight than the last 5 years, and is consistent with overwhelming evidence in the literature. Going forward there is no reason to expect a different result. Of course, there always will be a few active managers that will beat the risk adjusted index, but the evidence also shows that it is not possible to pick, in advance, who these managers will be, so it is prudent not to try. This is not my opinion, this is just the evidence.
An interesting discussion! There are brilliant/lucky active managers to be sure with WB an icon who continue to outperform the index. According to the research they are in the minority. The difficulty is going forward because managers/fund companies develop a celebrity status and investors flock there. What happens when these managers leave or the company is bought out like PH&N was? Will the outperformance last? Will you stick with them?
The other difficulty with active managers is that their outperformance is based on what? Is it luck? No thanks. Is it skill in researching companies that will outperform? Good luck! What remains is factor investing by these managers combined with luck. Justin has exposed Mawer for their value tilt. But factor investing is often limited with different factors outperforming at different time periods. Knowing what factor will continue to outperform in current economic conditions is fraught with not being a science. So index investing buying the whole market is better: no luck necessary, no skill req’d., low fees, and lower risk with better diversification. A win win!
Regarding DGI, this is becoming too common amongst the retired bunch. What is frightening there is the complete reversal! Investing in bonds for capital gains and in equities for income!! I think I’ve lived too long to see such an upside down world :).
Tristan,
You’re entitled to your opinion. I have always said one should go with the investing style that works for them and what they’re comfortable with. In my case I prefer to go with dividend growth stocks. It’s not the passive style that index investors prefer but I love the feeling of directing my own course. It’s been very successful for me since I released my long time underperforming active manager. I do study many different investing strategies and do a lot of back testing with different scenarios. I give credit where it’s due and criticise when I feel when it’s warranted.
I don’t personally invest with Mawer but I have studied their well run funds for quite a while. You and Justin have not shown any evidence that any index funds beat the Mawer Canadian Equity Fund over the short, intermediate or long term. The evidence clearly shows the contrary! As you say the vast majority of actively managed funds do underperform index funds but there will always be a few that excel. If I wasn’t into DGI I would definitely go with Mawer’s flagship fund “Mawer Balanced Fund”, I wish I had before I decided to go it alone in 2008. I may suggest this fund for my survivors in my will but in the interim I’m going with what has done well for me.
Good day & happy investing.
Bernie, active managers can often outperform over short time periods such as 1,3 and 5 years, but it is the long term that matters. And the evidence is very clear that over the long term (20 years) the vast majority of active managers underperform the appropriate risk adjusted index. I prefer to base my investing decisions on the evidence, rather than on an active managers opinion. I’d rather get the certainty of market returns rather than the remote hope of outperformance coupled with the almost certainty of underperformance.
Tristan,
Lets not get carried away. The “DJ Select Value Index” may show it outperformed Mawer Canadian Equity Fund in one time frame, that being June 30, 1997 (inception date of index) to current. Interestingly, this index failed miserably, by a very wide margin, in every other time frame when compared to the Mawer fund. That said, you can not buy an index. I can only find one ETF that tracks this index, that being “iShares Canadian Value Index ETF (XCV). XCV has only been around since Nov 6, 2006 (8 years) as opposed to Mawer Canadian Equity Fund which originated on June 3, 1991 (23.5 years).
Because of the relatively short life of XCV realistic performance comparisons can only be done with less than 10 year time frames. These are the current 1-Yr, 3-Yr & 5-Yr total returns to Jan 22, 2015 courtesy of Morningstar.ca:
iShares Canadian Value Index ETF (XCV):
1-Year = 3.50%
3-Years = 9.08%
5-Years = 8.89%
Mawer Canadian Equity Fund-A:
1-Year = 14.94%
3-Years = 16.74%
5-Years = 14.46%
Based on these performance numbers I ask you, where would you put your money?
Bernie, the point is that when you want to see if an active manager has outperformed, after fees, an index fund or ETF, obviously you must make an apples to apples comparison. When Justin performed a regression analysis on the active funds he found there was a significant value tilt. Therefore it only makes sense to compare the active funds with a value index, in this case the DJ Select Value Index. Putting it the other way round, if you were investing money today, and were considering, say, the Mawer Fund, given that this fund had underperformed the DJ Select Value Index over the last 10 years by almost 1%, would you put your money with Mawer, or an ETF that tracked the DJ Select Value Index?
Tristan,
I think I finally understand what this entire post was about after reading your last response to Bernie. I think many of us beginners are first, confused what this term “value tilt” means. A tilt be definition, might be a bias. But what is “value” then? (many of us don’t understand this as a financial term.) And ultimately, what’s the big deal with a “value tilt”?
If what you say is true, I think many of us do not understand there are also many “indices” around. Including one called a “value index”. I think many of us think that a Canadian equity index ETF or fund should simply track THE Canadian equity index (let’s say the S&P/TSX Composite). Many of us didn’t realize there are also such a thing as a “DJ Select Value Index”.
Ken, a value tilt is owning an ETF containing value stocks (stocks screened for various metrics such as low price to book ratio). Historically, value stocks have outperformed growth stocks over the long term, but with greater risk. However there can be long periods when they underperform. Having some of your equity allocation in a value ETF is a value tilt. However, this makes your portfolio more complicated to manage. There is a huge value in keeping it simple, so especially for someone starting out, and maybe for always, I think it’s better to go with just total market ETFs as described at Canadian Couch Potato.
Interesting results, whets appetite for more. Looks like all these funds had a tiny large cap tilt (the negative SMB value) and a modest value tilt (according to Bernstein’s cut-off that 0.3 or more would make it a definite value fund). I presume the T-stats and p-values indicate statistically significant results?
Could you perform a regression that includes low volatility and momentum, two other commonly cited factors?
Overall, it seems another confirmation that value-investing, especially, works. I’m surprised that this wouldn’t motivate you to include value ETFs in your model portfolios … maybe worth a post to explain why not?
Finally, how did the funds do in terms of return vs volatility risk compared to a benchmark? Thanks for your efforts.
@CanadianInvestor – the T-stats were significant for the MKT, SMB, and HML factors for the first 3 funds (the Fidelity fund only showed significance for the MKT factor). None of the alphas were significant.
The model portfolios for our PWL clients do include DFA funds (which tilt towards the small, value and profitability factors, and also use momentum screens): https://www.pwlcapital.com/en/Advisor/Toronto/Toronto-Team/PWL-Portfolio-Model
I prefer to keep my model portfolios for DIY clients very simple and plain-vanilla (this decision was a direct result of working with many DIY investors in the past).
Volatility risk for Canadian value stocks has been lower than the S&P/TSX Composite Index over the measured periods, so it was no surprise to see that these funds also had lower volatility than the index since their inception.
I wouldn’t agree with the statement that those funds don’t add value. They have a 20 year record and have beaten index by a good margin. Yes, they may be value tilted. But 20 years ago, there were no consensus that showed value outperformed passive index. The managers had to choose to be value tilted.
The problem with using retrospective analysis is that you cannot simply assume that the outcome will persist into the future. Now the managers know that value and small cap outperform, it is more likely than not the value of these stocks will be pushed up. So the managers who can outperform will likely be analyzed years later and their success be attributed to some other tilt.
William,
I disagree! Performance numbers are ALWAYS post fees. The data clearly shows these funds outperformed the index when held over the time periods noted. This article is clearly an exercise in torturing the numbers until they confess to the author’s agenda. I don’t take this article seriously as it is biased and likely a conflict of interest considering the author’s occupation.
A better way to think about this analysis is that is shows (for the first 3 funds at least) that the S&P/TSX Composite Index may be an inappropriate benchmark. If the managers were compared to a more appropriate large cap value-tilted benchmark (such as the DJ Canada Select Value Index), the alpha would turn negative.
07/1997-10/2014
DJ Canada Select Value Index = 10.84%
Mawer Canadian Equity – A = 9.96%
Leith Wheeler Canadian Equity – B = 8.84%
Beutel Goodman Canadian Equity – D = 8.47%
Precisely, when compared to an index that has similar factor weightings, the active funds lag by an amount that is very similar to their costs (MER plus TER). Comparing to a broad market index is not appropriate as factor weightings are quite different
” If the managers were compared to a more appropriate large cap value-tilted benchmark (such as the DJ Canada Select Value Index), the alpha would turn negative.”
These managers aren’t in it to stay precisely within the confines of a particular index. Their primary focus is giving value and attractive returns to their customers. If you choose to now focus on the “DJ Canada Select Value Index” I imagine there are several more funds in this category that beat this index in performance over the long term.
@Bernie – I would almost guarantee that there are a small percentage of funds within any asset class category that have beaten an appropriate index over the long term.
@Bob – Fama and French’s paper “The Cross-Section of Expected Stock Returns” was published in 1992 (over 20 years ago). The existence of the value premium was known well before that (Fama and French just put it into a workable model).
Why would you dismiss these actively managed mutual funds for “adding value”? Instead of simply buying the index these quality companies spent countless hours of research and technical analysis to get superior returns for their customers. These funds are no “flash in the pan”. They’ve been outperforming for many years and should be commended for their grades not put down!
Bernie, the point is that on a risk adjusted basis they don’t add value. After fees, they provide a worse risk adjusted return than a passive index with the same value characteristics
@Bernie – running a 3-factor or 4-factor regression is commonly used throughout the academic investment community to determine whether a particular active strategy has exhibited alpha that can’t be explained through the known risk factors (market, small, value, momentum). For the exercise, I simply collected the data, ran the regression, and posted the results (I can assure you that no numbers were “tortured” in this particular regression analysis ;)
Thanks. I was just looking at the last 2 years.
Can you point out an etf with a value tilt that has beaten mawer canadian equity then? The one i found (XCV) performed pretty poorly.
@ChristinaD – unfortunately there are no Canadian value tilted ETFs that have been around as long as the Mawer Canadian Equity – A. For what it’s worth, the Dow Jones Canada Select Value Index (the index that XCV tracks) had a return from inception (July 1997) to October 2014 of 10.84%. The Mawer Canadian Equity – A had a return of 9.86% over the same period. I’ve included the 3-factor tilts below from July 1997 to October 2014…
Mawer Canadian Equity – A:
MKT = 0.89
SMB = -0.14
HML = 0.31
Dow Jones Canada Select Value Index:
MKT = 0.98
SMB = -0.16
HML = 0.42
Yikes, you’re right! CCP has revised his model portfolios since I last visited to make them appear simpler than yours. But, I’m not sure VXC is ready for prime time. Your Vanguards have more diversification and I like that :)!
The issue with value investing is that it is subjective and that it only works with concentrated portfolios. This increases risk as diversification is lost. To mitigate risk the quest becomes to find truly undervalued stocks to hopefully create a margin of safety to protect the down side and reduce risk. It isn’t easy to estimate intrinsic value of a stock. There are metrics but it is more an art than science and value investing is more for the long term and that is it’s achilles heel as few investors or investment companies can wait! This is why value investors hold significant cash in their portfolios waiting to find this truly undervalued stock. An example is the Steadyhand Founders fund which is akin to the Mawer balanced fund. The former underperformed the index significantly because the manager held too much cash in his quest I believe to optimize the asset mix for value. Btw, your model portfolios are better than Dan’s ;)!
@harveym – thank you for your comments. There’s certainly a difference between a value investor who is constrained to being fully invested at all times and a value investor who is free to sit on cash and wait for opportunities (this investor could arguably be referred to as a market timer as well as a stock picker).
Thank you for the compliment about the model portfolios, but they’re a bit deceiving. My Vanguard model portfolios are almost identical to Dan’s Vanguard model portfolios (same asset classes, same asset mix), except that his is superior to mine in terms of simplicity :)
Hi Justin,
I really enjoy the information I obtain from your’s and Dan Bortolotti’s website. I am trying to understand the point or implication of your analysis. You wrote, “Without more information, it would be difficult for DIYers to determine whether the outperformance was merely the result of tilting the portfolio towards known risk factors (such as the small cap or value factors), or if the managers had any skill that may resemble alpha.”
I also tried to read Dan B’s explanation of this type of analysis but got lost. Could you put it in “layman’s terms”? Isn’t a better performance a better performance? If the fund manager outperforms an index, does it really matter if there is a “tilt” towards anything? I hope you understand my question.
Hi Ken,
I’m not Justin, but I’ll take a stab at this one. There are two main reasons that I see why it matters. First, it is possible to get these same tilts very cheaply with tilted index funds (ETFs). So if all the manager’s performance comes from tilt, they aren’t earning their higher fee for active management. (Also you’re losing out on diversification compared to the index fund.)
Second, many believe that the added performance of small and value comes with proportional added risk. So you don’t get the full picture if you just compare the performance to an untilted index.
@Ken C – excellent question! The analysis showed that in most of the cases, the outperformance was due to the tilt towards the value factor (and not necessarily any stock picking skill of the active managers). Tilting towards the value factor has been known for decades to increase the expected return of a portfolio. Better performance is relative – tilting your portfolio towards value companies using an ETF at a fraction of the cost of the active fund (knowing that the active fund will gain the majority of its expected outperformance from its tilt towards the value factor), seems like a reasonable alternative.
Thanks for the great work Justin.
I have seen suggested several times in the past that actively-managed bond funds have a particular advantage over the long term.
Would you consider running the same analysis on the Mawer Bond Fund compared to an equivalent passive index such as tracked by XBB?
@Westercoaster – actively-managed bond funds can arguably have some advantages over the long term relative to a passive bond index fund (as they are free to diversify their issuer risk to a greater extent and to manage the fund more tax-efficiently). The “better returns” argument does not appear to be the case for the Mawer Canadian Bond – A over the long term (which has lagged the FTSE TMX Canada Universe Bond Index by about 1.17% per year since inception).
07/1991-12/2014:
Mawer Canadian Bond – A = 6.45%
FTSE TMX Canada Universe Bond Index = 7.62%
Sources: Morningstar Direct, Dimensional Returns 2.0
Hi Justin,
Thanks for doing this! Any chance you want to run the Mawer US and International funds as well? (That would give a better picture of their balanced fund’s performance.)
Hi Nathan – you’re very welcome! Here are the results for the Mawer US and International funds…
Mawer U.S. Equity Fund – A
Adjusted R Square: 0.87
Alpha (annual): -1.23
MKT: 0.88
SMB: -0.05
HML: 0.10
Mawer International Fund – A
Adjusted R Square: 0.77
Alpha (annual): +2.89
MKT: 0.85
SMB: -0.23
HML: 0.07
Nice analysis. One concern I have about factor analysis, though, is that it denies the existence of certain types of skill by definition. For example, if someone had the skill to know in which periods small cap or value stocks would outperform the general market, factor analysis “erases” this skill. That reservation aside, I think factor analyses are a useful way to examine supposed outperformance.
@Michael James – thanks, Michael! I completely agree that factor analysis has its flaws like any other measurement tool. You just gave me an idea for another blog post with your comment :)
I’m looking forward to your next blog post on this subject. Hopefully, I’ll learn something new (or unlearn something old and wrong).
Justin, excellent post as always. This site and CCP are hands down the best info publically available. A couple of questions:
1) regarding 3F regressions: I can struggle through doing them manually, but is there an equivalent to http://www.fundfactors.com for Canadian funds? It would be way easier
2) I notice your public portfolios don’t have REITs or any factor tilting – what is your rationale?
Again, great work!
William
@William – thank you so much for your feedback – I’ll be sure to pass your praise onto Dan as well :)
1) There is no 3F Canadian regression resource that I am aware of. I agree that it is quite tedious to run these regressions – Frazzini’s factor data is in USD, so the Canadian monthly fund data must first be converted into USD before running the regressions. If there are any programmers interested in putting something together, I’m all ears :)
2) My recommendation for DIY passive investors is to keep their portfolio simple and plain vanilla. Most investors have a hard enough time staying the course – they don’t need to be second guessing even more decisions.
All the best,
Justin
Excellent article as usual, Justin! Could you point me to any sources where I could find more detailed information about the type of analyses you conducted here? I do quite a lot of data analyses as well as computer programming, and might be interested in following up on this.
All the best, ~ Rod
@Rod Martin – I think the first article I ever read on running your own factor analysis was from William Bernstein’s blog: http://www.efficientfrontier.com/ef/101/roll101.htm
I also wrote my own instructions for the process here: https://www.pwlcapital.com/en/Advisor/Toronto/Toronto-Team/Blog/Justin-Bender/November-2013/How-to-Run-a-3-Factor-Regression-Analysis
Good luck!