Federico D'Amario
Researcher at the Bank of England. Ph.D. in Economics, Sapienza University of Rome. Research interests: high dimensional econometrics, causal machine learning, macroeconometrics, forecasting.
Recent Work
Journal Article · 2026
Double Machine Learning for Time Series
with M. Ciganovic & M. Tancioni (Sapienza University of Rome) · The Econometrics Journal
Modifies the Double Machine Learning estimator for macroeconomic time-series settings using Reverse Cross-Fitting, improving efficiency and robustness to model misspecification. We propose a calibration rule targeting a "Goldilocks zone", a region of tuning parameters that delivers stable, partialled-out signals and reduced small-sample bias.
BoE Staff Working Paper · 2026
The Economic Effects of Shocks to Bank Capital Regulation: Evidence from the United Kingdom
with W.B. Francis & S. De-Ramon (Bank of England)
Structural VAR with sign and narrative restrictions based on 2014–2016 stress test events. Banks comply primarily by reducing risk-weighted assets; a 100bp Tier 1 capital ratio increase causes a 0.18pp GDP contraction. Tighter capital requirements temporarily reduce banking sector competition.
Journal Article · 2023
Forecasting Cryptocurrencies Log-Returns: a LASSO-VAR and Sentiment Approach
with M. Ciganovic · Applied Economics
Twitter, Reddit sentiment and Google Trends within a LASSO-VAR framework for ten cryptocurrencies. Sentiment improves directional accuracy for smaller, less capitalized currencies. Post-double-LASSO Granger-causality shows sentiment does not Granger-cause returns.
Journal Article · 2026
Carbon-Penalised Portfolio Insurance Strategies in a Stochastic Factor Model with Partial Information
with K. Colaneri & D. Mancinelli (Tor Vergata) · Scandinavian Actuarial Journal
Optimal proportional portfolio insurance strategies that hedge downside risk while penalizing carbon risk. Stochastic filtering turns the partial-information problem into a full-information one for a CRRA investor. Numerical analysis shows lower carbon emission intensity without sacrificing returns.