Research Overview

My research program develops a coherent econometric framework that links multilevel factor structures, fractional integration, long-memory dynamics, and stress-scenario propagation across macroeconomic, financial, energy, and climate-related systems.

Across applications (from growth-in-stress densities to electricity markets, exchange-rate volatility, long economic histories, labour participation, pollution, and credit dynamics), most of my research papers uncover persistent common components, heterogeneous cross-sectional structures, and non-negligible tail risks that shape macroeconomic adjustment and policy design.

You can find related publications in the Publications and Working Papers sections.

Click each section to explore.

Scenario-Based Predictive Densities: I develop econometric frameworks that generate full predictive densities under both baseline and stressed configurations of the latent factors driving macro-financial systems. These methods are applicable to growth, inflation, volatility, energy markets, and other macroeconomic variables.

  • Construct severe yet plausible scenarios from the joint distribution of high-dimensional factor models
  • Produce full predictive densities instead of point forecasts or isolated quantiles
  • Generalize Growth-in-Stress (GiS) to other macro-financial variables
  • Quantify how shocks—e.g., COVID-19—alter the tails of predictive distributions

Technical contribution: a multidimensional scenario-consistent methodology that extends traditional risk tools such as GaR by exploiting multilevel dynamic factor structures and tail-sensitive predictive environments.

FI-ML-DFM for High-Frequency Volatility

  • Defines global-in-time factors + intermittent hourly factors
  • COVID-19 induces permanent shifts in the global factor
  • Intermittent factors show 1-year mean reversion
  • CDS comovement driven by global + local + semi-pervasive factors

Dynamic Multi-Level Factor Models with Stochastic Trends

  • Allows long memory + short memory + stochastic trends
  • Prewhitening via cross-sectional and local averages
  • Factor estimators are asymptotically normal
  • Nord Pool application shows fractional cointegration

Technical contribution: a unified asymptotic framework combining long memory and multilevel factor structures.

  • Wavelet-based GLS estimation for time-varying loadings
  • Handles stationary + nonstationary factors
  • Explains cross-hour dependence in Nord Pool electricity markets

This links time-varying loadings with long-memory innovations.

Fractionally Integrated Panel Models

Proposes FI panels with multilevel dependence: top-level and block-level factors.

EPC–GDP Causality under Structural Breaks

Shows heterogeneous causal relationships across the Americas, depending on crisis episodes.

Periodic FI Models for Power Markets

Electricity prices exhibit fractional cointegration and strong diurnal periodicity.