Forecasting using sparse cointegration

Ines Wilms*, Christophe Croux

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

15 Citations (Web of Science)

Abstract

This paper proposes a sparse cointegration method. Cointegration analysis is used to estimate the long-run equilibrium relationships between several time series, with the coefficients of these long-run equilibrium relationships being the cointegrating vectors. We provide a sparse estimator of the cointegrating vectors, where sparse estimation means that some elements of the cointegrating vectors are estimated to be exactly zero. The sparse estimator is applicable in high-dimensional settings, where the time series is short compared to the number of time series. Our method achieves better estimation and forecast accuracy than the traditional johansen method in sparse and/or high-dimensional settings. We use the sparse method for interest rate growth forecasting and consumption growth forecasting. The sparse cointegration method leads to important forecast accuracy gains relative to the johansen method.
Original languageEnglish
Pages (from-to)1256-1267
Number of pages12
JournalInternational Journal of Forecasting
Volume32
Issue number4
DOIs
Publication statusPublished - 2016
Externally publishedYes

Keywords

  • Lasso
  • Reduced rank regression
  • Sparse estimation
  • Time series forecasting
  • Vector error correction model
  • VECTOR AUTOREGRESSIVE MODELS
  • COVARIANCE ESTIMATION
  • VARIABLE SELECTION
  • ERROR-CORRECTION
  • RANK
  • REGRESSION
  • LASSO

Cite this