I noticed Martin Wolf of the FT has a review of the recent book by Andrew Smithers (http://www.ft.com/cms/s/0/38164e12-c330-11de-8eca-00144feab49a.html).
Reading from the review there are two things I agree with:
1: There is always a correct price for any market that some people call “Fair Value”.
That is a bit confusing since Fair Value means about ten different things to different people.
I am assuming that what Smithers is talking about, is what International Valuation Standards (IVS) calls “Other Than Market Value” (OMV) which is what the price ought to be (in the opinion of the person doing the valuation) if the market was not in what George Soros and IVS calls “disequilibrium”.
That’s the same thing Warren Buffet talks about when he says “I can do a valuation”, which evidently is something he excels at (the trick is to buy when the price is lower than the “value” and sell when it’s higher).
2: Markets rotate around what Smithers calls “Fair Value”.
They do by definition (according to International Valuation Standards), the premise there is that over a long period of time (a hundred years) market participants over-price things as often as they under-price things.
Where I disagree:
First how you calculate Fair Value or OMV whatever you want to call it.
Smithers uses the “q” factor which he defines on his website (http://www.smithers.co.uk/faqs.php ) as:
q is the ratio between the value of companies according to the stock market and their net worth measured at replacement cost.
It can be defined to include or exclude corporate debt. When debt is included, we refer to the ratio as Tobin’s q, as it was in this form that Nobel Laureate James Tobin introduced the concept. For stock market purposes, however, it is easier to exclude debt and we refer to it in this form as “equity q”.
The data from which q is calculated are published in the “Flow of Funds Accounts of the United States Z1”, which is published quarterly by the Federal Reserve. This data source is available from 1952 onwards.
q is one of the two valid methods of measuring the value of the stock market. The other is the cyclically adjusted P/E. As they are both valid measures they both give the same answer, subject to small variations arising from the differences in data sources.
I presume that’s the explanation that economists give.
Valuation is nothing to do with economics..
To do a valuation you don’t need to be a Nobel prize-winning economist, a simple peasant can do a valuation (twenty chickens equals one cow); typically economists think about the reasons things were priced a certain way in the past, a valuation is by definition a prediction about something that might happen in the future, I have never met an economist that is prepared to write a price down on a piece of paper for an event that will occur in the future.
Specifically how much you can reasonably expect to sell (or buy) something for, in the future.
For example if you are a banker and you loan someone some money, then (if you are a smart banker) you get a valuation done to estimate the minimum price (he doesn't need to know the maximum) he/she might be able to sell the collateral that the borrower puts up, if at some time in the future, the borrower decides not to give you the money back.
In “valuation” speak “q” as it is defined there (as far as I can figure) is nominally a measure of Depreciated Replacement Cost (NYSE:DRC) which is one of three common methods of calculating value (the others are a sales comparison (basically mark to market), and an income capitalization (of future revenues)).
By definition if you did the valuations right and the market is not in disequilibrium each approach should give the same answer. That provides a convenient way of checking.
From a valuation perspective the way “q” is calculated is potentially unreliable because it does not account for intangible value (i.e. the value of a brand, proprietary technology or market presence).
Using P/E ratios suffers from two drawbacks, first those are by definition historical (what matters for valuation is the future), and second the discount rate is nominally considered constant in that valuation approach (which is not true).
One way to do a valuation of the US stock market to determine OMV is outlined in previous articles (seekingalpha.com/article/149741-the-seve...), basically from a valuation perspective all you need to do is (a) make sure that the result is dimensionally correct (P/E isn’t), (b) get sufficient data and get it to work (c) validate it.
There are other ways to work out “Fair Value” for example the accountants who worked out “Fair Value” of the assets of Lehman; Fannie May, Freddie Mac, Bear Sterns, AIG etc (www.marketoracle.co.uk/Article8177.html).
Sadly it would appear that the methodology that they used was not very reliable, which was how come over a period of about three months those companies were magically transformed from being worth hundreds of billions of dollars to about zero; remember Hank Paulson’s immortal one-liner in July 2008 “The US Banking System Is A Safe and Sound One” – (either he was lying or he got his valuations wrong).
Personally I use nominal GDP divided by the 30-Year Treasury, there is a logic there (slightly complicated), but regardless it seems to work.
The proof of any valuation is in the pudding, like for example if you do a valuation of something, did it turn out (in the future) that the person managed to sell (or buy) the something for more or less what you figured?
A good example of where valuations were not done properly was the housing market in USA where bankers did or commissioned valuations, but unfortunately when they came to sell their collateral they found that the price they could get was less than the valuations they had done said the price would be (that's if they bothered to do a valuation at all - many didn't).
The reason that I think the approach that I use is valid is simply because it works:
For example using that approach I could predict (a) that the S&P would bottom at 675 (exactly) in January (which it did in March +/- 1% if you are talking intra-day), (b) rally strongly after (it did), and (c) not look back until the Dow hit 10,000 (it did not look back – it might be now but that’s another story).
I have never seen the theories of “q” or P/E” or counting ducks, or whatever systems there are around, used to make “valuations” of that accuracy, (i.e. predictions of the price stocks could reasonably be expected to sell for in the future).
Until someone comes up with a better methodology that's the one I propose to use for valuations of stock markets.
Comparing “q” to OMV
I have not posted the chart from the FT article because it is copyright.... I redrawn the essential component to compare:
Some comments, in no particular order:
1: I don’t understand how “q” divided by price minus one (presumably) can be less than minus one as it apparently was in 1920? That implies that the estimated “value” at that point in time was less than zero, or did I miss something?
2: According to Smithers (not shown) the line that you get from “q” is more or less exactly the same as the line you get from P/E. That sounds fishy. When markets are in disequilibrium you would expect them to be different.
3: There is a trough in 1920 that looks worse than the Great Depression; I can’t find any reference to that on Wikipedia.
4: The mispricing in the 80’s looks as bad as the Great Depression, which is odd.
5: According to Smithers the mispricing in 2000 (Dot.com) was twice as bad as 1929, I find that hard to believe.
6: Also according to Smithers the market is still overvalued, and was at S&P 675.
My analysis suggests otherwise, and that the market is under-priced (mispriced down), although that doesn’t mean it will go up or that it won’t reverse 20% suddenly).