Job Market Paper
The Policy Duet: Coordination in U.S. Monetary—Fiscal Communication
Abstract: Monetary and fiscal authorities regularly communicate their assessments of the economy, yet the macroeconomic effects of alignment between their messages remain poorly understood. To study these effects, I construct a new dataset of approximately 250,000 U.S. policy documents from the Federal Reserve, Treasury, White House, and Congress spanning 1911–2025. Using this dataset, I build a communication coordination index that measures alignment between monetary and fiscal authorities’ messages about inflation. I identify coordination shocks in a structural VAR estimated over 1951–2025, using political proximity between Fed policymakers and the congressional majority as an external instrument. A positive coordination shock raises real activity, inflation, and short-term interest rates. Counterfactual analysis suggests that a substantial part of the expansion operates through lower uncertainty about real economic conditions. A New Keynesian model with imperfect information rationalizes these findings: households observe current demand conditions but cannot distinguish their persistent and transitory components. More aligned signals from monetary and fiscal authorities improve households’ inference, reducing uncertainty and precautionary saving, thereby stimulating aggregate demand.
Working Papers
Louder Than Rates: The Systematic Nature of Central Bank Communication (with Tiziana Assenza, Fabrice Collard, Philipp Wangner)
[CEPR Discussion Paper DP21691] [TSE Working Paper N°1761][Policy & Media Coverage]
Abstract: Do central banks decide systematically how much to communicate when explaining their policy decisions? Using all U.S. Federal Open Market Committee policy statements since 1994, we measure communication effort through the change in Shannon entropy and estimate a forward-looking communication rule. We find that communication is systematic: the Federal Reserve communicates more when inflation is expected to exceed target and output is expected to fall below potential. This finding is robust across a variety of sensitivity exercises. We then develop a New Keynesian model with imperfect information showing that systematic communication acts as a second policy instrument, stabilizing expectations and complementing interest-rate policy, especially at the zero lower bound.
The Uncertainty Channel of Monetary Policy Communication [Supplemental Appendix] (Submitted)
Abstract: This paper shows that, on average, more Federal Reserve communication lowers monetary policy uncertainty and raises real activity. I construct a monthly index of Fed communication intensity from 9,298 speeches by Federal Open Market Committee (FOMC) members. I identify communication shocks through heteroskedasticity, exploiting the rise in the volatility of communication intensity as the Fed moved toward greater transparency in the 2000s. A one-standard-deviation increase in communication intensity lowers monetary policy uncertainty by about 0.24 standard deviations. In a structural vector autoregression (VAR), the same shock raises industrial production by about 0.3 percent, lowers the unemployment rate by about 0.2 percentage points, and raises durable consumption by about 0.5 percent. The policy rate, the yield curve, and consumer sentiment show no statistically significant response on impact. These findings are suggestive of an uncertainty channel of monetary policy communication.
Work in Progress
A Filtering Approach to News-Based Measures: Application to the EPU Index (with Christian Gouriéroux, Nour Meddahi, Serge Nyawa, Wenxuan Wang, Danfeng Yue)
Abstract: Text-based economic indices are widely used in empirical research, but their reliability depends on how they are built. Many of these indices are based on a dictionary of words and a set of newspapers, both chosen by the researcher. The resulting measures are sensitive to these choices and thus subject to measurement error. Drawing on the literature on volatility and realized volatility, we propose filtering methods, notably the Kalman filter, to reduce this error. We apply the approach to the Economic Policy Uncertainty (EPU) index of Baker, Bloom, and Davis (2016). In various regression analyses, the filtered index delivers more precise estimates and an improved empirical fit relative to the original measure. These findings show that filtering can make text-based economic indicators more reliable.