Research

Firm heterogeneity shapes the distribution of attention across firms and, in turn, monetary transmission. We develop a heterogeneous-firm rational-inattention framework that endogenizes the joint size–attention distribution. A natural prior is that attention rises with firm size because mispricing affects more sales. Yet larger firms face lower demand elasticities and desired pass-through, which reduce the profit loss from mispricing and desired-price sensitivity to nominal shocks. Demand curvature determines which force dominates, generating increasing, decreasing, or hump-shaped attention. Empirically, attention is hump-shaped in size. This distribution shapes monetary policy’s heterogeneous effects across firms and whether nominal shocks raise or lower measured TFP.

We document that exchange-rate cyclical components repeatedly align with their eight-year lags, consistent with limit-cycle theory: without shocks, the economy returns to the same state after a fixed interval. This motivates a limit-cycle benchmark that forecasts exchange-rate changes using the same-horizon change at the corresponding phase of the preceding cycle. Like the random walk, a benchmark long considered difficult to outperform, it uses only exchange-rate history, without macroeconomic or financial predictors. It lowers out-of-sample mean squared forecast errors relative to the random walk for most of the twenty currencies in our sample at twelve-, sixteen-, and twenty-quarter horizons.

We study how business cycle fluctuations shape the pace and direction of innovation. Non-green innovation is procyclical, while green innovation is countercyclical, both over the cycle and in response to economic shocks. We explain this divergence in a dynamic general equilibrium model with endogenous green and non-green innovation. Green patents derive more value from distant profits, making green innovation less sensitive to short-run fluctuations. Endogenous innovation costs reinforce this mechanism, making green and non-green innovation effective substitutes. Evidence from market-implied patent valuations and scientist earnings supports the mechanism. The model implies larger and greener effects of innovation subsidies during downturns.

Firm-level R&D changes are usually small, with occasional large increases that are procyclical and predict patent quantity and quality. We develop a medium-term business-cycle model with heterogeneous innovators who accumulate raw ideas and face fixed project-launch costs. Past shocks reshape their raw-idea distribution, making subsequent shock propagation history dependent. Low volatility leaves more innovators close to launching, so subsequent contractions postpone more projects and cause larger losses. In the calibrated model, identical Great Recession shocks generate long-run cumulative TFP losses of 4.7% starting from the pre-recession distribution, versus 3.3% in a counterfactual using the average distribution. The model traces this vulnerability to the preceding Great Moderation, generating the Great Twins hypothesis. Cross-country evidence supports this prediction; firm-level R&D data support the mechanism involving the pre-recession concentration of innovators near the launch threshold. The central bank therefore faces a trade-off between reducing businesscycle volatility and limiting the severity of subsequent recessions.

Market power has risen substantially and unevenly across U.S. sectors. We study how this change affects monetary non-neutrality. In a menu-cost model, higher desired markups flatten firms’ profit losses, widen inaction regions, and reduce the frequency of price adjustment, thereby strengthening the real effects of monetary shocks. Empirically, contractionary monetary shocks have larger output effects in high-markup states, and higher markups are associated with lower frequencies of both price increases and decreases, even after controlling for inflation. A quantitative model disciplined by micro-price evidence implies that monetary non-neutrality rose by 42% between 1980 and 2016, with changing markups accounting for 75.9% of the increase. About 90% of the markup effect operates through the endogenous decline in price-adjustment frequency.

Excess sensitivity, the significant response of long-term interest rates to monetary policy, is a well-known puzzle. This paper shows that excess sensitivity is more pronounced following monetary policy easing than tightening, a phenomenon we term the asymmetric excess-sensitivity puzzle. By contrast, no comparable asymmetry is observed around the release of other macroeconomic data. A model with an ambiguous signaling channel of monetary policy can account for these findings. The model predicts that monetary easing has stronger and excessive effects on long-term interest rates but weaker effects on real GDP than monetary tightening. We provide evidence that supports the proposed mechanism.

Government consumption stimulus is typically deployed alongside other fiscal measures that may affect the fiscal multiplier. This paper investigates the interaction of the fiscal multiplier with a prime component of such policy mix: unemployment insurance (UI) extensions. We document empirically that longer pre-existing UI duration significantly reduces government spending multipliers, using state-level military spending shocks and variation in UI duration across U.S. states. To interpret the evidence, we build a small-open-economy model with search-and-matching frictions and heterogeneous households. The model separates two mechanisms: (i) a level effect, whereby longer UI duration reduces households’ precautionary savings and thus lowers the fiscal multiplier; and (ii) an endogenous UI duration effect, whereby higher government spending lowers unemployment, leading to an earlier expiration of UI extensions. Quantitatively, the calibrated model indicates that the second channel dominates.

Canonical macroeconomic and financial models require credit to be equal to its fundamental component, i.e., the net present value of the net flows to creditors. Per this conventional view, credit booms are expected to precede increased flows to creditors. However, data suggests otherwise. To rationalize the novel empirical findings, we develop a model with financial frictions and heterogeneous firms, allowing for the existence of credit bubbles. We show that a positive credit bubble shock raises the aggregate credit and output while depressing the credit’s fundamental component through firms’ precautionary savings.