Permanent and Transitory Components of GNP and Stock Prices
Quarterly Journal of Economics CIX (February 1994) 241-266. This is my favorite solution to the permanent/transitory decomposition issue for GNP and stock prices. I use bivariate autoregressions of consumption and GNP, and of dividends and stock prices. Consumption and dividend growth are unpredictable, so act as stochastic trends for GNP and stock prices. A movement in stock prices with no current change in dividends is completely transitory, so can be labeled an “expected return” shock. A movement in stock prices with a change in dividends is permanent and so is a “permanent earnings” shock. Note the QJE switched Figure II and III.