We show that the stock market index price reaction to monetary policy announcements by the Federal Open Market Committee (FOMC) is explained mostly by changes in the default-free term structure of yields, not by changes in the equity premium. We reach this conclusion based on a new model-free method that uses dividend futures prices to obtain the counterfactual stock market index price change that results purely from the change in the default-free yield curve induced by the monetary policy surprise. The yield curve change in turn partly reflects a change in expected future short-term interest rates, as measured by changes in professional forecasts, and partly a change in the term premium. We further find that the even/odd week FOMC cycle in stock index returns is also largely due to an FOMC cycle in the yield curve rather than the equity premium.

Predictive regressions of market returns on option-implied moments measured before pre-scheduled FOMC meetings show that tail risks play an important role in understanding the market risk premium around FOMC announcement days. Skewness and kurtosis, which capture investors' expectation of the tails of the return distribution, robustly predict post-FOMC returns both in-sample and out-of-sample. The predictability lasts up to one week and is stronger for expansionary monetary policy shocks. The signs of the corresponding risk premiums are consistent with economic intuition, illustrating the role of periods with high risk premiums to confirm theoretical predictions.

We consider identification of monetary shocks and their causal impacts in quasi- linear environments where (i) agents may possess subjective beliefs and (ii) monetary authorities manage current and future interest rates (e.g., forward guidance). Assuming rational expectations or risk-neutrality trivially enables identification. Without those assumptions, identification of monetary shocks from asset prices hinges on a Long-Run Neutrality condition, roughly meaning policy does not affect the compensation for permanent risks. We construct a non-parametric test of the Long-Run Neutrality condition, related to the literature on FOMC announcement effects, and argue that it is violated in the data. Finally, we present some example models in which the Long-Run Neutrality condition is violated, illustrating how this condition is generally distinct from conventional notions of monetary neutrality.