Lack of co-movement between consumption growth differentials and real exchange rates is a traditional indicator of a disconnect of foreign exchange markets from economic fundamentals (Backus-Smith 1993 anomaly). We present novel evidence for the (dis)connect between the volatilities, as opposed to the levels, of these variables. The volatility correlations are below one, but they are larger than the level correlations. In the cross-section of countries, the volatility disconnect weakens for countries with low amount of expected growth risk and high amount of volatility risk. We provide an explanation of our empirical findings based on international risk-sharing of both expected growth and volatility news shocks.

We develop a theory of forecast coherence in a firm production setting, which yields a normative ex ante benchmark of first-best coherent forecasts and statistical tests to detect incoherence ex post. Under the null, the forecasts—and the forecast errors—of output and inputs are “close” to one another. Using the Duke Survey of top executives of large US corporations, we reject the null of coherence for 55% of CFOs in our sample. In a positive version of our model, incoherence reflects intrapersonal frictions in coordinating multiple forecasts, implying that some of the rules of thumb proposed by the managerial education literature to make contemporaneous forecasts may emerge as second-best optimal. Consistent with our model, we find that corporate performance correlates negatively with incoherence, being lowest for firms whose CFOs provide “narrow bracketing” forecasts—projecting past capital growth into the future while ignoring output and labor. We also find that the use of incoherent rules of thumb correlates negatively with corporate investment spending and positively with corporate leverage.