Monetary policy works largely through expectations: core transmission channels, including long-term yields, mortgage rates, and investment, depend on expected future policy. We build a model in which markets interpret policy decisions as informative about the central bank's real-time readings of the economy—readings that persist and guide future policy. At each meeting, the central bank anticipates how markets will interpret its decision. That interpretation, in turn, depends on how the central bank systematically responds to its readings of the economy. We identify a time-inconsistency problem: meeting-by-meeting rate choices can make market expectations excessively sensitive to policy surprises, leaving the economy inefficiently exposed to mistakes in the central bank's readings. We quantify this mechanism's importance in a New Keynesian model with imperfect and private central bank information, estimated using macroeconomic data, Federal Reserve forecasts, uncertainty estimates for productive capacity, and inflation-swap responses to policy surprises. The optimal ex ante rule responds much less aggressively to estimates of the economy's productive capacity than does meeting-by-meeting policy, limiting the amplification of assessment errors through market expectations and achieving substantially better macroeconomic stabilization.

This paper develops a theory of forward guidance that captures two key features of central bank communication: announcements are not fully state-contingent, and the private sector is wary of policy mistakes. The central bank tailors announcements along two central dimensions: (i) vagueness—how tightly the announcement constrains future policy; and (ii) data dependence—how aggressively future policy responds to a public signal of the economy. Three main lessons emerge. First, uncertainty affects communication differently depending on its source: sufficiently high uncertainty about either demand or cost-push shocks calls for vaguer guidance, but only demand-side uncertainty increases the optimal degree of data dependence. Second, strict rules are costly: fully precise guidance entails sizable stabilization losses, especially when the Phillips curve is steep. Third, because of the stabilizing role of announcements, optimal forward guidance differs sharply from simply communicating the best forecast of the future optimal policy rate (Delphic guidance).

Monetary policy decisions affect economic activity largely by shaping private-sector expectations. The difficulty of anticipating how agents will interpret a decision is thus an important source of uncertainty about how strongly policy affects the economy (transmission uncertainty). I study a model in which agents infer the central bank's persistent real-time economic assessments from its decisions in a way that the central bank cannot perfectly predict. Systematic policy shapes this inference, making the extent of transmission uncertainty endogenous. I show that, under discretion—when the central bank chooses its preferred rate after observing its assessments—transmission uncertainty attenuates the policy response, in line with the familiar cautionary principle. However, this caution is inefficient. First, the discretionary rule is too cautious in its overall scale: a stronger policy response improves stabilization without increasing exposure to interpretation risk. Second, a self-reinforcing feedback loop between uncertainty and caution can generate arbitrarily large, avoidable losses.

I study how monetary policy shapes financial intermediaries' incentives to acquire information about credit risk, and how this affects credit supply and macroeconomic stabilization. The analysis highlights a trade-off between limiting amplification through financial attention and pursuing the central bank’s preferred stabilization policy. Next, I plan to study how monetary policy affects the allocation of attention between aggregate and firm-specific risks, and to quantify financial attention's role in shock propagation.

Uncertainty about monetary transmission is a classic rationale for gradualism in monetary policy. We ask when that logic actually goes through. Our results distinguish attenuation from persistence: uncertainty can dampen policy movements without making interest rates more persistent. Whether it does so depends on its source and the economy's structure. Ongoing work studies how transmission and forecast uncertainty interact and what share of observed monetary policy gradualism can be explained by transmission uncertainty.