J. Bradford De Long, Lawrence H. Summers; Equipment Investment and Economic Growth: Reply, The Quarterly Journal of Economics, Volume 109, Issue 3, 1 August 199

Major long-run swings in the U. S. stock market over the past century are broadly consistent with a model driven by changes in current and expected future dividends in which investors must estimate the time-varying long-run dividend growth rate. Such an estimated long-run growth rate resembles a long distributed lag on past dividend growth, and is highly correlated with the level of dividends. Prices therefore respond more than proportionately to long-run movements in dividends. The time-varying component of dividend growth need not be detectable in the dividend data for it to have large effects on stock prices.

We examine interest and inflation rates from 1879 to 1913. Deflation prior to 1896 was followed by inflation. Average U. S. inflation was 3.1 percentage points higher in the years after 1896, yet nominal interest rates were no higher after 1896. This nonadjustment of nominal rates would be consistent with rational expectations if inflation was not forecastable, and indeed univariate tests show little sign of serial correlation. But gold production does forecast inflation. The relationship between mining and inflation was such that expected inflation should have risen 300 basis points between 1890 and 1910. We consider explanations of this failure to foresee the shift in inflation after 1896 and conclude that it is not persuasive evidence that investors ignored relevant information, but does suggest great uncertainty about the appropriate model for analyzing the economy.