Luis Winter

We propose a function-on-function linear regression model for time-dependent curve data that is consistently estimated by imposing factor structures on the regressors. An integral operator based on cross-covariances identifies two components for each functional regressor: a predictive low-dimensional component, along with associated factors that are guaranteed to be correlated with the dependent variable, and an infinite-dimensional component that has no predictive power. In order to consistently estimate the correct number of factors for each regressor, we introduce a functional eigenvalue difference test. While conventional estimators for functional linear models fail to converge in distribution, we establish asymptotic normality, making it possible to construct confidence bands and conduct statistical inference. The model is applied to forecast electricity price curves in three different energy markets. Its prediction accuracy is found to be comparable to popular machine learning approaches, while providing statistically valid inference and interpretable insights into the conditional correlation structures of electricity prices.

We develop a suite of real-time nowcasting tools that are compatible with the precision sampler for missing data in linear state-space models. We introduce an alternative approach to measure the impact of data releases and revisions on the GDP nowcast in the absence of the Kalman filter, and describe a data-driven scenario analysis framework that exploits the sampler's unique ability to condition on additional information. Leveraging the computational efficiency of the precision-based method, we conduct a large-scale model comparison for nowcasting German GDP. Our application demonstrates that scenario nowcasts conditional on expert GDP projections improve prediction accuracy and post-COVID uncertainty assessment compared to solely model-driven approaches.