Question: Apples n oranges? Should we use a 30-year "look back weighted average" on GDP?
Discussion:
I think one of the problems is that Doug is assuming an "ergodic" relationship between stock prices and GNP growth (and total returns traditionally include prices with the assumption of dividends reinvested in the same stock index, but before taxes and not counting transaction fees). Doug may roll his eyes. But really, asset valuation is not necessarily an inherently logical process.
In particular, we might break a stock "price" down in the following way. To make it concrete let's assume a $100 stock. Assume a car factory.
$25 might be the actual physical cost of setting up the factory. $15 might be the "brand effect" -- why you with your factory that is long established have a putative advantage over someone with an identical factory but just starting out. $25 might be a "bid to normal rate of return effect." By this I mean that if the 30 year govt bond rate of return is 5%, you might demand a 7% return on a stock with a higher risk premium. If the car can be made and marketed at a nominal cost of $40 a share but pays dividends of, say, 50% at that price, then it is clearly undervalued. The price will be bid up to the "risk adjusted rate of return." But we're not done yet. $10 might be the "anticipation effect" a reasonable assessment of this particular firm's prospects of increasing its business given such factors as how "hot" its cars are, how the economy is faring, and so on. $25 might be Keynes' "pretty face effect." That is, everyone's estimate of what everone else thinks the stock is worth.
Now, you can play with these numbers all you want. (On NASDAQ issues the last category would be 99% of total share price. )
But here's the point: in the example above, 40% (and even more, if you consider the subjective element of "brand") of the total price is "subjective anticipation." It is possible, then that at any given moment in its entire history, a component of the stock market's total return is not based on "fundamentals"--such as might be deduced ex-post from GNP growth, but from the psychological condition which determines asset valuation. Because on any given day D1 there is *always* an anticipation about future prospects for D2, D3....Dx, there is *always* a valuation component of the stock index which is, as it were, disjoined from the objectively measured economy (after the fact). Ergo your stock index will be only loosely connected to GDP.
There are other factors which could be at work. For one, even the riskless rate of return on a govt bond demands a calculation about inflation which translates into an inflation premium. We charge a higher rate of return, anticipating inflation today, then might be justified after the fact. This forces stocks to offer an even higher "risk adjusted" premium. The net effect of these attitude-driven adjustments is to make capital more dear and to slow economic growth. But paradoxically the valuation process for stocks, and perhaps even bonds, could outpace the growth in real productivity, because on average "anticipation" exaggerates and affects performance.
It's a thought, anyhow. If we posit instead a mechanistic connection between real GDP and stock valuations than it ought to be possible to come up with relatively good S&P 500 predictor just by tracking quarterly GDP.
-- Gregory P. Nowell Associate Professor Department of Political Science, Milne 100 State University of New York 135 Western Ave. Albany, New York 12222
Fax 518-442-5298