Spades wrote:Lazyday, here's some of my ideas for an over valuation of the market.
1. The current Shiller P/E multiple is over 24. The average from 1900-2010 is 16. That is 50% above the average Shiller P/E multiple.
It might be true that the market is expensive.
But this isn't enough. Are stocks so expensive, that they will return similar or less than bonds, going forward from today?
What evidence do you see for a strong market to invest in?
I don't know if the market will be strong or weak, but I think it will do better than bonds in the long run and I'm better off with stocks.
Even if the market may be expensive, it is destructive to wealth to buy and sell, instead of buy and hold and rebalance. Possible exception IMO (not all Bogleheads) is that we might conservatively adjust our AA when stocks are so expensive that expected returns are similar to those of bonds. I don't think this has happened many times other than around 1999/2000.
We could roughly estimate future real bond returns simply using the yield of long term TIPS, which is 1.71%. You could get a little more complicated, and also look at nominal Treasuries and adjust for estimated inflation.
For stocks, we can use Gordon Equation and Earnings yield.
I'm not an expert on the Gordon Equation, but we start with the dividend yield, which is 2.36% for the Russell Global index, 1.75% in US, 2.81% Ex-US. http://www.russell.com/Indexes/data/fac ... _Index.asp
Add to this the real earnings growth rate, or the real dividend growth rate. Wiki says this has been 2.6% since 1980. That's high historically, which might be because of the effect of net share buybacks. The dividend yield above ignores share buybacks companies do. (Though of course we should exclude buybacks related to options compensation.) Perhaps the earnings growth rate was increased through net buybacks. Or still has yet to? There might be pent up growth which the Gordon Equation has missed so far. Over time, companies have shifted much earnings from paying dividends to buying back shares. How this shows up in the Gordon Equation over time is confusing to me, and makes me trust it less.
Our rough guess is 1.75% US dividends or 2.81 Ex-US, plus growth.
In US, we could add 1.75% dividends and 2.6% growth, for 4.35% estimated real returns.
Ex-US has higher dividend yield by 1.06, if growth is the same, we would use 5.41% estimated real returns.
Another rough estimate is Earnings Yield, which is simply 1/PE. This is both simplistic and optimistic, and I think relatively few academics use it. Jeremy Siegel of Wahrton is one. It has the fault of using recent earnings which move a lot. Might be improved by smoothing using 7 or 10 years of earnings, but adjusting upwards for growth. Also growth may be a little slower than earnings; if all earnings is reinvested, some excess growth happens, but some investment is needed for depreciation and other loss, probably larger. So the model might be improved by subtracting a small amount. However, there are complaints about the Gordon Equation too, and it is sometimes thought to underestimate returns.
The site above shows PE in US 16.51 (6%), Ex US 13.5 (7.4%), Global 14.7 (6.8%).
These numbers are all very rough estimates, and the second set is optimistic. Some might think the first set is slightly optimistic if the growth rate is inappropriate, but a lower growth rate might need to be adjusted upward for ignoring increasing net share buybacks over time. Even if we adjust the first set downward in a realistic way, it will still be significantly higher than the real expected return from bonds, in my opinion. And the second is much higher, though admittedly imperfect.
In contrast, had this exercise been done around early 2000, I believe both methods would have shown simalar expected broad stock return to expected bond returns. Not half or one quarter as much as bonds, but similar to bonds. So, these very rough tools might have pointed one towards reducing holdings of largecap stocks, though other equity asset classes would have passed the test. One might have either conservatively and carefully shifted towards them or to TIPS. A few years later, the broad stock market would have looked ok again. This round trip would probably be a once in a lifetime event, as I don't expect these tools to show expected stock returns to equal expected bond returns in my lifetime again.
You can make your own go through using the two methods above with your own inputs, and will come up with different results, but I'd be surprised if you come up with expected stock returns close to bond returns if you are honest. Also look at others work, but be careful about their bias, and make the distinction between expected returns in the future forever/unspecified, and expected returns in the short run or medium term. I'm talking about forever. Also valid for any term, it's a guess for any future period starting today. Goal is avoiding major bubbles, not market timing, so short or medium term forecasts aren't as useful.
Some credit for some of these ideas goes to Larry Swedroe, though not the blame, his thoughts were surely superior and possibly even more conservative.