In economic and business data, the correlation matrix is a stochastic process that fluctuates over time and exhibits seasonality. The most widely-used approaches for estimating and forecasting the correlation matrix (e.g., multivariate GARCH) often are hindered by computational difficulties and require strong assumptions. In this paper we propose a method for modeling and forecasting correlation matrices that allows the correlation to be driven nonlinearly by common factors. Our nonlinear common factor (NCF) method simplifies estimation and provides more flexibility than previous factor-based methods. We illustrate its use on energy prices in Boston.
Bibliographical notePublisher Copyright:
© 2020 Elsevier Inc.
- Covariance matrix estimation
- Dynamic correlation
- Energy pricing
- Multivariate adaptive regression splines
- Nonlinear factor analysis
- Positive definiteness