Why Market Timing Fails
In a recent study, market timing based on Robert Shiller’s well-known Cyclically Adjusted Price-to-Earnings (CAPE) ratio failed to produce market-beating returns. Here I offer an explanation of why this doesn’t work.
Shiller’s CAPE is one way to try to measure whether stocks are currently over- or under-valued. If CAPE gives correct results, you might think it’s self-evident that getting out of the market when stocks are overvalued would be a good idea. Based on this reasoning, the study results seem to imply that CAPE is not a good valuation measure, but this isn’t necessarily correct.
Even if CAPE is completely accurate, it still isn’t necessarily useful for market timing. The problem is that it takes time for stock prices to readjust. Suppose that CAPE says prices are 10% too high. If the market reacted quickly, then next year’s returns would be 10% lower than normal. But prices don’t react this quickly.
Suppose that it takes 10 years for stock prices to adjust and bring the CAPE measure back to “normal.” This means that over those 10 years, returns were 1% lower per year than they would have been without this adjustment. So, if you normally expect to earn inflation+5%, then when CAPE says stocks are overvalued by 10% you might expect only inflation+4% for the next decade.
However, inflation+4% is still better than what you could expect from bonds, real estate, or any other common investment category. So, you can be exactly right about stocks being overvalued but still not be able to use this knowledge to make any extra money. Being out of the stock market is a bad idea more often than not.
Shiller’s CAPE is one way to try to measure whether stocks are currently over- or under-valued. If CAPE gives correct results, you might think it’s self-evident that getting out of the market when stocks are overvalued would be a good idea. Based on this reasoning, the study results seem to imply that CAPE is not a good valuation measure, but this isn’t necessarily correct.
Even if CAPE is completely accurate, it still isn’t necessarily useful for market timing. The problem is that it takes time for stock prices to readjust. Suppose that CAPE says prices are 10% too high. If the market reacted quickly, then next year’s returns would be 10% lower than normal. But prices don’t react this quickly.
Suppose that it takes 10 years for stock prices to adjust and bring the CAPE measure back to “normal.” This means that over those 10 years, returns were 1% lower per year than they would have been without this adjustment. So, if you normally expect to earn inflation+5%, then when CAPE says stocks are overvalued by 10% you might expect only inflation+4% for the next decade.
However, inflation+4% is still better than what you could expect from bonds, real estate, or any other common investment category. So, you can be exactly right about stocks being overvalued but still not be able to use this knowledge to make any extra money. Being out of the stock market is a bad idea more often than not.
I'm an indexer and don't employ any timing strategy myself. The only one I have seen that in my opinion seems to offer validated benefit is Mebane Faber's Tactical Asset Allocation based on a 200DMA. Basically it underperforms aboyt 70% of the time, however since 1900 (or whenever data starts) there have been 3-4 really disastrous periods during which executing his model would have had you out of the market and saved you a steep drawdown. Steep enough that, despite the relative rarity of these periods and the under performance the rest of the time, being rescued from those periods results in a higher overall return.
ReplyDelete@Anonymous: It's always hard to tell with such things, but I'd be worried that this method's success is just data mining and that it has no expectation of success in the future.
DeleteWell unless someone has finally developed the time machine, attempting to analyze anything is just data mining and has not expectation of success in the future. Just because passive indexing would have worked well over the last 100 years does not make its future certain.
DeleteIt's actually just math with this method. It will under perform in flat/oscillating and general up trends. And it will outperform during down trends, with the out performance proportional to the depth of the drop. So it really comes down to whether you think another 2008 or 1930s might happen. If it does, this offers unemotional, mathematical, rule-based protection. If it never does, you'll have paid for protection you never needed.
@Anonymous: It's true that just about everything amounts to data mining in a technical sense, I was reacting to your statement that this method's success came from avoiding just 3 or 4 really bad periods. If bad periods in the future look a little different making this method not work as well, presumably it will not beat the market.
DeleteAgreed. This particular method requires periodic (once every couple decades) intervals of very steep declines, of the 08/09, Great Depression caliber, in order to outperform. If those never happen again in my investing lifetime, so we never see more than a 10-20% correction, you would mathematically expect this method to under perform. That's not a bet I'd want to take!
DeleteI’ve had an on and off debate with myself about market timing. I’ve concluded there are 2 types of market timing:
ReplyDeleteThere is conscious market timing where we buy something that we have concluded (whether through intensive analysis or a gut feel) the cycle is at its lowest point or sell something because we have concluded also that it has reached the high point of the cycle. This is impossible and is at the heart of your message. This type of market timing falls within the domain of active investors who are trying to squeeze every bp of yield. As you clearly state, it can’t be done consistently so don’t go there. Agreed. Better to leave the neuroticism to the “pros”.
At the same time, I’ve also been having an internal debate that despite our awareness of the follies of conscious market timing, aren’t we all engaging in market timing, perhaps at a more subconscious level? At some point in time, you have to jump off the fence and become a participant in the market. You have to buy or sell something and even if you are consciously aware that you are not factoring in current market events into your plan/strategy, isn’t there always subconsciously an element of market timing in any type of transaction that will set the table for an enhanced positive or negative outcome? Take an example of 2 people who began investing at different periods and both oblivious to the current market climate. They were in it for the long-term and were disciplined enough to rebalance and average their costs accordingly. The difference being one invested in the 2007 before the credit crisis and another invested in 2001 right after the dot com bubble. Clearly their portfolios would perform differently even though they were not trying to consciously time the market and pick an optimal entry point. This has to have some impact. I’m not sure if there has been any research on this, so would love to hear other perspectives. How much would dollar cost averaging or some other strategy mitigate this?
Cheers
@Aman: You're right that even if you aren't engaged in active market timing strategies, you still have to choose when to put new money into the market or pull money you need to spend out of the market. For me, the new amounts I have to place in the market tend to come along in small enough chunks that I just invest them immediately into whichever ETF is below my target percentage. If you come into a very large sum of money (say more than a year or two of pay), some people advocate averaging it into the market over a period of time. As long as you don't spread it out over too long a time, this makes some sense. However, we should resist trying to wait until conditions look right because all evidence says people who wait for the right conditions are really waiting for high prices because the market feels safe after prices rise significantly.
DeleteI think if you come into a lump sum from a rational point of view it is better to invest the sum all at once. But if you had put your $1M inheritance into the market in Jan 2008, "rational point of view" doesn't feel so good. So from a behavioural point of view some people recommend investing over a period of a year or so. I've already decided that when/if that day arrives I'm investing it as a lump sum.
DeleteMichael, when you say you invest whatever comes along immediately, what kind of amounts are you talking about? Wouldn't the cost of trading eat into that significantly? (I'm envisioning you investing after every paycheque or something... maybe that's where I'm wrong..)
Delete@Anonymous: I have a threshold of usually between $3000 and $5000. So, you're right that I don't try to invest money from every pay cheque. Once I get to that threshold, I don't try to second guess the market and wait for a lower entry point.
DeleteI'm surprised that Norm Rothery used this study. The original study is an all or nothing switch between stocks and cash. Further, it would have been better to use individual stocks which have periods of under/over valuation rather than the index. Indexing is a shortcut for the groups of investors that skip doing bottom up analysis on stocks or other securities.
ReplyDelete@Anonymous: I've had my fill of doing bottom-up analyses of stocks. The competition is far too fierce. Maybe you're one of the few who can succeed at this, but the math says that the majority of those who try to beat the market this way will fail.
Delete