Earth observation
Using satellite imagery to observe economic activity at a spatial and temporal resolution that traditional statistics cannot always capture.
I’m Marwan Menaa, a PhD Candidate at Université Paris Dauphine-PSL and Banque de France. My research sits where economic measurement meets satellite imagery, textual analysis, generative AI, and large-scale data systems.

Alternative Data as New Indicators for Economic Analysis
01 / Research lens
Official statistics give us the map. Alternative data can reveal the movement in between—faster, more granular, and from entirely new angles.
Using satellite imagery to observe economic activity at a spatial and temporal resolution that traditional statistics cannot always capture.
Transforming large collections of news, narratives, and unstructured documents into interpretable measures of perceptions and expectations.
Combining machine learning and generative AI with economic validation to build indicators that are timely, robust, and useful for decisions.
From messy, high-volume traces to signals that can survive economic scrutiny.
02 / Selected work
LLM CLASSIFICATION × BERT FINE-TUNING
A Multilingual LLM-Based News Indicator of Euro Area Inflationary Pressures
A 27B-parameter LLM labels multilingual news, then a fine-tuned EuroBERT classifier scales the signal to 17 million articles in French, German, Italian, and Spanish. The resulting index tracks euro-area inflation expectations out of sample.
Open access

Search economics research across working papers, journal articles, books, and chapters from the RePEc network—by title, author, affiliation, keyword, and JEL code.
Search OpenRePEcA tool built with Jean-Charles Bricongne, Diyar Can, and Guillaume de Rouville to make economics conference and special-issue discovery more efficient and transparent.
Visit ConfWatcherNatural language processing applied to a near-universe of French newspaper articles from 2004-2022 to measure inflation attention and perceptions among households and firms.
Read Working Paper 921Doctoral research supervised by Sylvain Benoit (Université Paris Dauphine-PSL) and Thomas Renault (Université Paris-Saclay), with Jean-Charles Bricongne as advisor, at the intersection of academic research and central banking.
Ask me about the research03 / Toolkit
$ inspect toolkit
● ready to turn noise into signal
I enjoy the whole chain: framing the economic question, engineering the data, building the model, stress-testing the result, and explaining what it means.
04 / Journey
A path through central banking, capital markets, engineering, and research—each adding a different lens on how data becomes a decision.
Université Paris Dauphine-PSL × Banque de France
CIFRE thesis · LEDa · Alternative Data as New Indicators for Economic Analysis
Royal Bank of Canada Capital Markets
Rates, FX & commodities derivatives · Pricing · XVA automation
European Central Bank
Market-data infrastructure · Economic projections · Monetary analysis
Banque de France
News-based inflation indicators · NLP · Country-risk sentiment analysis
05 / Teaching
Courses that connect rigorous foundations with the tools, data, and decisions students will encounter in practice.
Université Paris Dauphine-PSL
Master 1 IEF — Ingénierie économique et financière
Université Paris-Saclay
Core concepts, instruments, and the economic logic behind how financial markets operate.
06 / Open channel
I’m always happy to discuss research, data, economics, or the odd dataset that might tell us something new.
The address is assembled only when you click, which makes bulk scraping less useful.