Factor-augmented tree ensembles

Abstract

This article proposes an extension for standard time-series regression tree modelling to handle predictors that show irregularities such as missing observations, periodic patterns in the form of seasonality and cycles, and non-stationary trends. In doing so, this approach permits also to enrich the information set used in tree-based autoregressions via unobserved components. Furthermore, this manuscript also illustrates a relevant approach to control over-fitting based on ensemble learning and recent developments in the jackknife literature. This is strongly beneficial when the number of observed time periods is small and advantageous compared to benchmark resampling methods.Empirical results show the benefits of predicting equity squared returns as a function of their own past and a set of macroeconomic data via factor-augmented tree ensembles, with respect to simpler benchmarks. As a by-product, this approach allows to study the real-time importance of economic news on equity volatility.

Multidimensional dynamic factor models

“…Having shaped the data in the form prescribed in section 2.1 we are now ready to specialise the MDFM for this household problem. Similarly to recent work on semistructural models including Hasenzagl et al (2022a,b) and the empirical application in Pellegrino (2023b), we identify the model via economics-informed restrictions in order to extract interpretable unobserved components.…”

“…The dynamics for the latent factors and the estimation method proposed for this model are illustrated in section 6. The estimation is based on penalised quasi maximum likelihood estimation (PQMLE) and built on an ECM algorithm similar to the one employed in Pellegrino (2023b).…”

“…(ii) When the penalty is not active, the CM-step is computed from equation 8 since We structure the CM-step by following analogous steps to those in Pellegrino (2023b).…”