Umberto Collodel

with Michela Baretti (Universidad de Alicante) and Jacopo Zacché (Central Bank of Malta)

We provide the first long-run systematic evidence on whether financial markets price political pressure on the Federal Reserve as a signal of future monetary policy. Using approximately four decades of Reuters newswire articles from 1988 to 2025, classified by an open-weight LLM ensemble into dovish and hawkish pressure, we estimate the daily response of interest rate expectations across the term structure, the US dollar, inflation compensation, and market volatility. Dovish pressure lowers near-term rate expectations by roughly one basis point per event — equivalent to a four-percent increase in the implied probability of a rate cut at the next FOMC meeting — with the effect decaying monotonically and vanishing beyond the three-month horizon. Hawkish pressure has no detectable effect. The response is driven entirely by presidential commentary, accompanied by dollar depreciation, with stable inflation compensation and market volatility. Excluding the Trump presidencies does not weaken the estimate, ruling out the possibility that the result is confined to an idiosyncratic recent episode. We interpret the pattern as markets pricing anticipated monetary accommodation, rather than a credibility tax on the nominal anchor.

Traditional high-frequency identification of monetary policy communication effects operates ex-post, precluding evaluation of alternative strategies before publication. We introduce an ex-ante framework that simulates heterogeneous market reactions to central bank communication before release. Our methodology employs LLMs to construct an agent-based simulation of 30 synthetic traders with heterogeneous risk preferences, cognitive biases, and interpretive styles. These agents process ECB press conference transcripts and forecast Euro interest rate swap rates across 3-month, 2-year, and 10-year maturities. Cross-sectional forecast dispersion provides a model-based measure of market disagreement, validated against realized OIS volatility. Analysing 283 ECB press conferences (June 1998–April 2025), we document Spearman correlations of approximately 0.5 between simulated and realized disagreement, rising to 0.6 under iterative prompt optimization — robust across prompting strategies, temporally stable, and significantly outperforming simple language complexity scores.

Paris School of Economics, 2022 · Supervisor: Prof. Agnès Bénassy-Quéré

Three chapters on macroeconomic forecasting and financial crises. Chapter 3 shows that early warning systems for external crises fare relatively well in terms of forecasting performance, comparing global and domestic indicators and analysing output losses from missed versus predicted crises. Chapter 4 evaluates short-term forecasts of economic activity by major institutions and the private sector, examining how forecast errors correlate to the business cycle and touching on the political dimension of forecasting. Chapter 5 introduces a novel dataset of IMF archival documents — compiled via a manual dictionary and term-frequency approach — to measure how crisis discourse has evolved across the Fund's membership over decades.