Skeptical? Me too.
The rational is that a) deflation had price levels reduced by 18% from '29 to '32, and b) high dividend payment levels, in the range of 14%, contributed to a recovery contribution that is not factored in. These things are true, and not irrelevant, but they still tell a very misleading story. Hulbert only hints at the math, but it can be reproduced this way.
The JDI top (eliminating decimals) in 1929 was 381. Less 18% for deflation, this would have only been 312 and change at the 1932 bottom.
At the bottom, the DJI was at 41, a number which does not enter into the calculations.
From the Yahoo finance historical data, the highest DJI value in 1936 was 185, a number Hulbert does not cite. Take 185, and multiply it by 1.14, 4 times to get 4 years of compounding, and the answer is (voila!) 312 and change.*
OK. With a little detective work and some relatively simple math we can get to the desired answer.* But does anything smell maybe a bit fishy?
First off, Hulbert corrects the top value for deflation, but does not correct the gains from the bottom for inflation between 1932 and 1936. Trust me, there was some. Not a lot perhaps, but enough to to require some adjustments in the math.
The 14% dividend yield number from 1932 would have fallen dramatically as stock prices rose from 41 to 185 during the years from '32 to '36. This is huge, but totally left out of the reckoning.
Even beyond the specious math, the underlying assumptions are questionable. An investor would have had to stand pat through the entire 1929 to 1932 collapse, and then be in a position to reinvest dividends during the recovery. Possible, I suppose, but highly unlikely.
I've always respected Hulbert, but this article uses cherry picked data, flawed logic, and even some invalid numbers, to reach a very questionable conclusion.
Update: Something else troubles me about the whole approach. I think there is an apple and an orange here, maybe even an onion.** Comparing the the level of the stock market index to itself over time is one thing. Comparing the level at one point in time to any measure of total return at a different time is something else again. The total return idea has its own validity, but in this context strikes me as being a totally erroneous comparator.***
Update 2: Suppose Hulbert is 100% right, and I'm 100% wrong. From 1929 to 1936 you still have a compounded annual return of ZERO!
Update 3 (5/17) The ex post facto math in multiplying 185 by 1.14^4 to get a derived theoretical number from the previous bottom, based on presumed dividend returns reinvested strikes me as being bogus.
_________________________________
Update 3 (5/17) The ex post facto math in multiplying 185 by 1.14^4 to get a derived theoretical number from the previous bottom, based on presumed dividend returns reinvested strikes me as being bogus.
_________________________________
* I can't be sure that this is how it was done, but if not, it's a heck of a coincidence.
** Perhaps what I was smelling.
*** But I'm just the trombone player
No comments:
Post a Comment