Critics of index investing often argue that the strategy is not sensitive to valuation. They feel that a simple strategy of buy, hold and rebalance is folly: there are times when the market is cheap or overvalued, and it makes sense to shift toward or away from equities accordingly.
It’s hard to argue with that principle. There are certainly periods when markets appear frothy and others that look like excellent buying opportunities. The problem is, whose valuation should you believe?
On April 2, Vanguard’s chief economist Joseph Davis went on BNN and said that valuations are right around their historical averages, so expected stock returns over the next several years should also be near their historical averages, which in the US is about 9% a year.
Three days later, The Globe and Mail ran an article about Prof. Robert Shiller’s method for valuing the market. According to Shiller, the earnings yield of the S&P 500 is currently 4.5%, compared with the historical median of 6.3%, which means the market is significantly overvalued. This translates to expected returns of 2.6% annually over the next 10 years.
What is the average investor to make of this? Of course, investment firms (even Vanguard) have a vested interest in being bullish on equities—I get that. But traditional valuation measures are easily quantified: the trailing P/E ratio of the S&P 500 today is about 16.2, compared with the historical average of about 15. So it’s indeed pretty close to average.
And while Shiller is an academic with fewer conflicts of interest, that doesn’t necessarily mean he’s right. While his predications have a good track record, his data were also saying equities were overvalued last fall, yet since October the US market is up about 25%. If you made a tactical shift away from stocks based on his valuation, you missed an enormous rally: 2012 saw the best first quarter for US stocks since 1998.
Diversify, rebalance and stop guessing
This is my concern about making tactical moves based on valuation, which is really just a form of market timing. First, you need to base your actions on data that can tell conflicting stories. I don’t know how you decide which criteria to use with any confidence, since all of them will be right during some periods and dead wrong during others. (Ken Fisher’s The Only Three Questions That Count includes a lengthy discussion about why P/E ratios don’t always have predictive value.) Then you need to execute your strategy consistently and unemotionally. I suggest that’s much easier said than done—especially after a couple of your moves backfire.
Simply owning a broadly diversified portfolio can go a long way toward making all of this less important. If you’re holding government bonds, corporate bonds, real-return bonds, stocks from around the world (with a mixture of value and growth, large and small), real estate and several currencies, chances are that there will always be both overvalued and undervalued assets in the mix, whatever yardstick you want to use. And a disciplined, rules-based rebalancing strategy is a self-correcting mechanism that ensures no asset class has too much or too little influence for very long. It’s not a valuation tool, but it does help you build in a measure of “buy low, sell high” without resorting to forecasts.
Maybe the question should be around time horizon – are stocks over valued compared to where they wil be in 25 years? Nope.
10yrs – 99% no
5yrs, likely no.
1 yr? 50/50 shot.
Dan,
Excellent post…how would you invest lump sum payments? For example, I typically contribute to my kids RESPs in the first quarter fo the year and usually buy all at once according to my asset allocation. Is it better to buy in increment (dollar cost average)?
Thanks
Karim
@Karim: The evidence suggests that you should simply invest the money when you have it. This is especially true if transaction costs are an issue (i.e. you are using ETFs as opposed to mutual funds). DCA is great for people who need to make regular contributions from their paycheque anyway, and it has some emotional value (fewer regrets), but it’s not an optimal strategy. All of the data show that investing $1,200 on January 1 is likely to outperform $100 every month.
You can always use dollar value averaging if you are concerned about becoming overweight equities when they are overvalued.
@Andrew F: A great strategy, though difficult to implement for many investors, I think. Readers who want to learn more can read Value Averaging: The Safe and Easy Strategy for Higher Investment Returns by Michael Edleson.
The way Joseph Davis phrases his projection in the BNN interview you linked to is strikingly different from the projections he had given to the Financial Times a day before: http://www.ft.com/intl/cms/s/0/f9fd175e-77e2-11e1-b237-00144feab49a.html
In the BNN interview, he suggests returns “in the range of 9 to 10%”, whereas in the Financial Times article he suggests a *50% chance* of real returns of 6% (i.e. roughly 8-8.5% nominally). That’s a pretty significant difference! It’s disappointing to see people from Vanguard making these kinds of misleading statements.
I personally don’t think either estimate is credible without at least some adjustment factor for demographics and current valuations.
As a couch potato, I don’t believe in forecasts anyway :-)
That being said…my portfolio is down as I have a very small (20%) weighting to fixed-income as I figured that as I am young…..why the hell not? It’ll bounce back sooner or later.
Anyway, I just had to mention Dan that the title of this post should instead read “Depends whoM you ask”, with an “M”, as the pronoun in question is being used as an indirect object. Keep up the excellent articles!
~Maxwell C.
To Maxwell: actually, “Depends Who” is correct. You only use “whom” if the phrase is “Depdens on Whom”. -_-
Excellent post Dan. I love it! Mainly because I always have difficulty putting everything in all at once, especially when equities have risen.
TD e-series global potato 10-year return of around 4% is not that bad! Especially since this was the decade that followed the decade of exuberance. Who knows what future decades will bring?
While it is true that the price level of the market does affect future performance, it is very difficult, if not impossible, to predict exactly _when_ performance will be affected. Since you can’t time a correction in prices, valuation really only helps in the long term. And even though high prices reduce the equity premium, it still exists. Put another way, when the market is ‘frothy’, your investment has a lower expected return than when it isn’t. However, its expected return is still _positive_ and, in an efficient market, theoretically greater than the risk-free rate. Therefore you’re still better off staying invested than trying to time the market. You know you’re going to make less than you would in other circumstances, but you still stand to make more than you would sitting out.
Now if you’re talking about shifting your asset allocation based on valuations from an asset class with a lower long-term expectation to one with a higher long-term expectation (such as a different geographical region, or a different size or value weighting), it may theoretically be possible to gain an edge. However, this is much easier said than done, requires a great deal of discipline, can have significant tax consequences, and will still be largely dominated by luck. I strongly agree that it is better to simply follow a disciplined plan and let the market go ahead and fluctuate.
@Karim: Actually, the same logic that suggests it is preferable to stay invested rather than timing the market explains why it is better to invest in a lump sum. Basically, it is better to have funds invested than not invested.
Equities and other assets with risk components should always, over the long term, pay a premium over risk-free investments. Therefore, on average, you are always better off having your cash fully invested than not. That means not pulling it out to time the market, as well as getting it in as quickly as possible in the first place. (Of course, it is always possible that you’ll get lucky and have your cash on the sideline during a downward correction, but in the long run you’re obviously better off not relying on luck!)
Also I’m certainly not saying all one’s money should be in equities; it is certainly logical to keep a percentage of one’s portfolio in risk-free fixed income investments based on one’s personal risk tolerance. But that’s a long-term asset allocation decision.
Very interesting discussion.
The idea of market valuation can be used in so many ways. I do not use it to sell anything, based on it, unless it is extreme. At this time, I think the market is “fairly valued”, a couple months ago, I think it was overvalued.
With any new cash, that I had on hand, I chose to sit on it in February. The market did pull back a little. I would feel a little happier putting some in now, but in reality, will probably wait a little longer.
In these turbulent times, within a sideways market, it changing every few months!
My view of waiting a little, is I don’t like seeing my $10,000 investment worth $9000 next month. This should not bother me, as a long term investor, but it does a little ;) I guess I am still a believer in margin of safety, realizing that it is a very inexact science.
Again, great discussion!
This was the very issue I had trouble getting over when I swtiched to a passive strategy, it’s pretty easy to look with hindsight and say you could have predicted past market swings and I had the impression that this is why I was paying an additional fee for actively managed funds. However it’s only when I took the time to educate myself on what was actually happening that you realize how fruitless market timing is.
I understand the point you are making and could not agree more about valuations being very relative depending on the metric being used, or “expert” consulted. The examples in your post make this very clear. You also had other posts that illustrated very well the uselessness of market forecasts.
I do wonder if there are exceptions however? Canadian government bonds come to mind.
Let’s say an investor does not have any government bonds in his portfolio and wants to add government bonds. The timing right now is not very compelling based on two factors I can think of. Namely the lack of a maturity premium and probable lost of purchasing power.
The yield to maturity of the 10 year Canada government bond is around 2.1% at this moment. A few banks in Canada offer saving accounts with interest around 2.1%.
Putting your money in a savings account is of course not foolproof vs. a 10 year government bond. Who knows, maybe interest rates for saving accounts in the next 10 years will go down and the 10 year government bond with the current yield to maturity of 2.1% will prove to be the better investment?
However, right here and now there is no extra return for tying up your money for 10 years vs. a saving account with flexibility. This makes government bonds a hard sell.
There is also inflation. The bank of Canada targets a core inflation of 2%. Often “total” inflation is higher than core inflation, so a 10 year government bond will probably not even protect purchasing power. Again, for an investor who wants to add government bonds to his portfolio, the timing does not seem very favourable.
Yes there are assumptions in my reasoning with respect to government bonds: i) future rates of inflation and ii) future interest rates on saving accounts. In investing history is not always a guide, but from a historical perspective government bonds at this moment strike me as very unappealing. Holding cash in a saving account at around 2% with its flexibility seems like a better alternative.
I would be very interested in your thoughts on this topic.
P.S. sorry for the long comment…
Good article that covers the basics.
But when I look at any long term chart of equities the use of a 10 month simple moving average to add or subtract risk seems to work.
Compare a 50/50 portfolio of XBB and XIU using this rule in addition to rebalancing annually and taking 25% capital gains taxes – to a straight 50/50 buy and hold with rebalancing.
In the last 12 years this would have added just several trades and would have preserved capital with a higher risk adjusted return (higher return relative to risk taken).
@Andrew: I would argue that 12 years is too short a sample to draw conclusions. Just because something worked over the last 12 doesn’t mean it will over the next 12 or 22 or 32. And if it does work, the very act of people exploiting it will eventually eliminate the effect. (The beauty of a (mostly) efficient market!)
Nathan,
I recognize the short time period but I mentioned it because it is an easy portfolio to test and compare results.
However Mebane Faber has analyzed this approach using data since 1900.
You can read his paper “A quantitative approach to tactical asset allocation” to see how simple and effective using the 10 month moving average is to give much better risk adjusted returns:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=962461
Standard and Poor’s also recently did a similar analysis of moving average crosses from 1969 to 2011 and found similar things – using moving averages reduces risk and over time therefore gives better portfolio performance. A 50x200SMA cross beat the SP500 index 36% of the time since 1969 and had a compound average rate of growth (CAGR) of 7.4% vs 6.6% for the index. What is interesting is that it reduced risk considerably in the sense that the worst drawdown in any given year was minus 15.2% vs minus 38.5% for the index and that there was an average of only 1 signal to trade per year. Since 2000 the 50x200SMA cross beat the index 70% of the time with an average return of 5.7% CAGR vs a lousy minus 0.6% for the index.
There are reasons this phenomenon is not arbitraged away and I would not want to get into them all because the post would be too long (it already is!) but in a basic sense what the long term moving average does is provide signals about trend and momentum. One can clearly see topping processes in bull markets over long time frames that are broken when the trend is broken, indicated by the moving average.
There are other complementary oscillators and indicators that one can learn to read quite effectively to gauge momentum. As far as I can tell there are two ways to “beat the market” that consistently seem to work: value approaches that are usually based on fundamental analysis and momentum systems based sometimes on changes in fundamentals and sometimes just technicals. Faber’s paper proves the latter works. There is much literature on why value works as well and why it is not arbitraged away.
Morningstar recently introduced two ETFs that use these approaches FXM (value) and WXM (momentum). Dan has written posts about them.
Hmm, interesting. Certainly the momentum and value effects are real and there are reasons why they have not been arbitraged away. And the value effect can certainly be captured in the long term. The evidence I have seen to date though suggests that timing the value effect is much the same as trying to time the market in general, and that while negative momentum can be avoided, trying to capture positive momentum after costs is problematic. Still, sounds interesting; I’ll give that paper a closer read and do a bit more research when I have a chance!
The logic in that paper appears to be primarily based on behavioral explanations. While that might explain why the effect has persisted over a long sample period, it does also suggest that it may evaporate as it becomes more well known (think January Effect). IMO the only factors that can be trusted long term are those with associated risk stories – market, size, value, or those for which there is a real barrier to arbitrage (arguably the momentum effect), but in the latter case, by definition, they should be difficult to exploit.
I’ll keep my eye on this though, along with the over-performance of low-beta stocks, which also has a purely behavioural explanation, but has also persisted quite some time. Cheers.
Why fight the tape? (say when rebalancing)
I have long seen a sweet spot in mid to lower beta, dividend growth stocks which is now being arbitraged away somewhat. It is for instance only recently been captured by ETFs and some bigger mutual funds. My strategy has always been to divide my equity risk between the index and this type of stock but I fear that everyone is on the bandwagon because of the ZIRP policy of central banks. I just hope that emotions start to take sway once any real recovery starts and the growth mantra starts up again.
Aside: I think there is a higher probability that bonds are overvalued on a long term basis but it would take a long post to explain why. For other reasons I also assign a higher probability to stocks being overvalued than not being overvalued. I see through the lenses of scenarios not forecasts – thats another long post about risk management!
Lots of great discussion here!
@Philippe V: The expected returns of bonds are easier to estimate than those of equities, in the sense that there is a narrower range. However, in the last few years people have argued with supreme confidence that bonds were overvalued and were going to get clobbered because interest rates were sure to rise. They were spectacularly wrong. There’s no possibility in the medium term that 10-year bonds will deliver returns similar to those of the last 30 years or so: that’s simple math, not a forecast. But government bonds still have a place in a long-term portfolio, with lowered expectations.
@Andrew: I have just finished Mabene Faber’s The Ivy Portfolio, which is an excellent book. I’m going to save my detailed comments on this subject for a later post, but I’ll address your comment that Faber’s paper proves that his strategy works. I would argue that his paper proves that it would have worked in the past if executed with robotic precision and zero costs. There’s a big difference.
@Nathan: You’ve touched on an interesting idea. There is a long academic debate on where the value premium comes from. Most people accept that small caps have a premium because they are riskier than the market (they have a higher cost of capital and are more volatile). But value stocks, in many cases, are less volatile and not necessarily riskier, so why would they outperform? And if that’s an inefficiency in the market, why wouldn’t it get arbitraged away? One explanation is that it is a behavioural bias: investors just don’t like stodgy, low-growth companies. That would be hard to arbitrage away: the majority of investors would have to become contrarions—which, of course, is logically impossible. :) Although as you suggest, we may be seeing a little of that in the trend toward dividend stocks, which have value characteristics.
It’s good to hear that you’ve read Faber’s book. I always wondered what a buy-and-hold disciple would think about it. I look forward to your thoughts on the subject.
Faber’s approach to market-timing is more about reducing risk than increasing returns.
Dan
As far as costs – Rereading Faber I understood the portfolio had less than one round trip trade per asset class per year which is hardly more than rebalancing annually. On average Faber states there are only 3 to 4 roundtrip trades for the whole portfolio per year. This would cost 20-40 dollars in a discount broker account. Otherwise the cost is the same as owning the ETFs as usual.
Taxes could be ignored in a tax free account.
As far as precision – it is easy today to set up technical alerts from your discount broker account that are sent to your e-mail. With the alert you just do as the model mechanically instructs to buy or sell. One doesn’t even have to know how to read a moving average chart – although I would recommend it.
With a large enough portfolio and the new ultra low cost ETFs total costs would be quite low.
This portfolio seems to be only a very little bit more difficult to maintain than a 4-5 asset class CP portfolio and arguably as easy as the Uber-Tuber.
Also if you analyze the updated version of Fabers paper you can see that the financial crisis drawdown was completely avoided on the charts and this will make a huge difference to the long run performance of this portfolio relative to the market buy and hold.
@Andrew: You’re ignoring cost of the spread, but the paper you referenced does show that to be low as well. As I see it, the real risk is that the pattern will cease to hold as it becomes better-known, and will in fact under-perform for an extended period as traders continue to attempt to use it. It will be difficult to determine whether this has indeed occurred for quite some time, as it is of course expected that the strategy will under-perform over extended periods even if it is still ‘working’.
@CCP: Indeed, that’s something I’ve been quite interested in lately. As I said, I’m wary of anything without an attached systematic risk, for the reasons I’ve brought up regarding the trading system Andrew mentioned. In an efficient market, if it is logical for some investors to weight their portfolios toward value, it must be logical for others to do the opposite. One would expect that these ‘others’ might be shorter-term investors, for example, who are more concerned about avoidance of short-term losses. The historical data hasn’t shown this however. It is possible that the market is rationally discounting against the risk that some sort of catastrophic event will impact all value companies together, to a magnitude we have not yet seen. Or indeed, as you say, there could just be a strong behavioral explanation. Or some of both. I have more reading to do. :)
I would like to create some of my own charts. Where can one download the information from the S&P 500 into Excel? Appreciate any tips. Thanks :)
@John: Monthly returns going back to 2000 can be downloaded here:
http://us.ishares.com/product_info/fund/downloads/IVV.htm
This paper gives some interesting evidence for a risk story explanation of the value premium: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=330180 Basically they’re saying that during economic downturns, aka times when the expected market risk premium is high, the the value factor tends to co-vary with the market premium, making value risker in ‘bad’ times (and less risky in good times). Since investors are more risk-averse in bad times, there must be a value premium.
Also interesting is this Fama-French paper that actually looks at how a behavioral explanation would work. While FF generally don’t take a side on risk story vs behavioural explanation, and are generally seen to favour the risk explanation, they do explain how a behavioural basis would play out, and why it would not necessarily be arbitraged away: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=502605 (Essentially if there are rational and irrational investors, the actions of the irrational investors will create a premium, but because arbitraging it away is not entirely risk-free, the counteractions of rational investors will not eliminate the premium. (Rather, they will end up rationally taking the opposite tilts to those taken irrationally by the irrational investors, if that makes sense. If it doesn’t, the paper explains it better. :) ))