Working papers
Working papers
R&R Journal of Empirical Finance
Abstract: This paper introduces a general linear multifactor asset pricing methodology that integrates systematic risk measured at different frequencies into a single pricing equation. Our setup allows for horizon-dependent risk exposures, consistent with the idea that investors with different investment horizons may respond differently to systematic factor variation. Empirical results show that frequency-specific ICAPM specifications outperform traditional models and match benchmark performance, with significant prices of risk concentrated at horizons beyond three years. The approach reveals low-frequency pricing information in ICAPM factors, with risk prices broadly consistent with ICAPM theory.
Updated version coming soon
Abstract: This paper introduces multiple-shock impulse response functions, which consider the cumulative effects of simultaneous shocks occurring within one period, together reflecting an underlying (unobserved) primitive shock. The concept generalizes individual shock impulse response functions, accounts for the dependence between shocks, and is applicable to various multivariate time series models. Simulation studies highlight its necessity for accurately interpreting the total effect of shocks and mitigating potential temporal aggregation issues. The multiple-shock approach is applied within a global vector autoregression framework, examining the European debt crisis as a primitive shock through monetary policy and uncertainty shocks in lower credit rating countries, and using nation-wide shocks as proxies for an area-wide equity shock. The results demonstrate that the interaction between monetary policy and uncertainty can significantly amplify the effects on economic and financial variables, and that using multiple nation-wide shocks aligns more closely with theoretical expectations. Secondly, we use the multiple-shock approach in a multivariate GARCH model to analyze the effects of the Swiss Franc-Euro peg removal and Brexit on currency volatility, observing varied currency covariance responses that could impact asset allocation strategies.
Winner PhD competition of the 12th European Central Bank Conference on Forecasting Techniques
Work in progress
Overhyped? Can ML Models Reliably Predict Stock Returns?
with Yanki Kalfa and Allan Timmermann.
Abstract: Hyperparameters determine the architecture of machine learning (ML) models and can greatly affect their forecasting performance, yet there is little consensus on how to choose the range and grid of hyperparameters to search over. We provide an extensive examination of which hyperparameters are most important for popular ML models' out-of-sample forecasting performance using a large U.S. dataset on individual stock returns and firm characteristics. We find that some choices of hyperparameters virtually guarantee good out-of-sample return forecasts while others lock in poor forecasts. This poses a challenge because many empirical studies fail to provide details on how they set their hyperparameters. We also find that time-series validation methods do not offer a definitive solution to the dependence of out-of-sample return forecasting performance on the underlying range of hyperparameters.
Draft coming soon
Publications
Journal of International Money and Finance, Volume 143, 103073
Open access: https://doi.org/10.1016/j.jimonfin.2024.103073
Abstract: Like other central banks, the ECB resorted to asset purchase programs (APPs) to replace conventional policy measures. We examine their impact on the Euro area with a focus on the heterogeneity among its constituents and across financial markets. Our analysis combines a Bayesian structural VAR with an identification scheme based on market surprises at the announcement time, effectively capturing structural dynamics. At the Euro area level, APPs stimulate the economy, lower government bond yields, elevate stock prices, and reduce corporate and sovereign stress. The impact shows heterogeneity in the stock market with a widened value-growth spread in stocks and varying sector impacts, particularly favoring financial stocks, and across countries with stronger effects on southern Euro area countries. Our results show strong spillover effects between countries, indicating challenges in the precise targeting of APPs.
Banner photo: Cottonbro