¹ If recessions of standard magnitude occurred at random intervals of mean length Ṯ instead of in predictable cycles, the stock market would rise in every period in which a recession was not signaled and decline only when a recession could be forecast, producing a “leading indicator” type of result. In that case, however, the depth of the stock market reaction should be limited to the effect of the recession occurring now rather than Ṯ periods from now—a rather small order of magnitude.

The only business cycle model consistent with an efficient stock market that is a leading indicator is one in which business activity is a diffusion. While there is evidence consistent with such a view of economic activity, it is not the model most people have in mind.

² Indeed, we can already see how this relates to our earlier question. Why would we hold any real capital into a foreseeable recession unless its current price already discounts the recession? But if we hold wheat into the harvest at a price above the expected harvest price, why shouldn't we do the same for business inventories or capital equipment in the face of a recession?

³ The key to this result is that the elasticity of wealth with respect to oil inventories (above and below ground) is unity regardless of the price elasticity of consumption.

⁴ It is tempting to refer to this situation as nonoptimal, but clearly if there are some money or subjective costs that keep individuals from holding the portfolio which is predicted in the absence of costs, it is the actual behavior of the market that is optimal, rather than my overly simple model. Nevertheless, the discrepancy between the two models is quite illuminating.