Wavelet Estimation for Factor Models with Time-Varying Loadings

TL;DR

This paper develops a novel high-dimensional factor model with time-varying loadings, employing wavelet-based estimation to handle both stationary and nonstationary factors, demonstrated through simulations and application to electricity market data.

Contribution

It introduces a two-stage estimation method combining principal components and wavelet-based generalized least squares for time-varying loadings in factor models.

Findings

Effective estimation of time-varying loadings demonstrated via Monte Carlo simulations.

Successful application to Nord Pool electricity market data.

Model accommodates both stationary and nonstationary factors.

Abstract

We introduce a high-dimensional factor model with time-varying loadings. We cover both stationary and nonstationary factors to increase the possibilities of applications. We propose an estimation procedure based on two stages. First, we estimate common factors by principal components. In the second step, considering the estimated factors as observed, the time-varying loadings are estimated by an iterative generalized least squares procedure using wavelet functions. We investigate the finite sample features by some Monte Carlo simulations. Finally, we apply the model to study the Nord Pool power market's electricity prices and loads.