Macroeconomic forecasting using penalized regression methods

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

We study the suitability of applying lasso-type penalized regression techniques to macroe-conomic forecasting with high-dimensional datasets. We consider the performances of lasso-type methods when the true dgp is a factor model, contradicting the sparsity assumptionthat underlies penalized regression methods. We also investigate how the methods handle unit roots and cointegration in the data. In an extensive simulation study we find that penalized regression methods are more robust to mis-specification than factor models, even if the underlying dgp possesses a factor structure. Furthermore, the penalized regression methods can be demonstrated to deliver forecast improvements over traditional approaches when applied to non-stationary data that contain cointegrated variables, despite a deterioration in their selective capabilities. Finally, we also consider an empirical applicationto a large macroeconomic u.s. Dataset and demonstrate the competitive performance of penalized regression methods.
Original languageEnglish
Pages (from-to)408-430
Number of pages23
JournalInternational Journal of Forecasting
Volume34
Issue number3
DOIs
Publication statusPublished - 1 Jul 2018

Keywords

  • Forecasting
  • Lasso
  • Factor models
  • High-dimensional data
  • Cointegration
  • DYNAMIC FACTOR MODELS
  • PRINCIPAL COMPONENT ANALYSIS
  • APPROXIMATE FACTOR MODELS
  • TIME-SERIES
  • ADAPTIVE LASSO
  • LARGE NUMBER
  • SELECTION
  • PREDICTORS
  • SHRINKAGE
  • AUTOREGRESSIONS

Cite this

@article{2c197d0d0f3641998c4d4be1c9e86fc9,
title = "Macroeconomic forecasting using penalized regression methods",
abstract = "We study the suitability of applying lasso-type penalized regression techniques to macroe-conomic forecasting with high-dimensional datasets. We consider the performances of lasso-type methods when the true dgp is a factor model, contradicting the sparsity assumptionthat underlies penalized regression methods. We also investigate how the methods handle unit roots and cointegration in the data. In an extensive simulation study we find that penalized regression methods are more robust to mis-specification than factor models, even if the underlying dgp possesses a factor structure. Furthermore, the penalized regression methods can be demonstrated to deliver forecast improvements over traditional approaches when applied to non-stationary data that contain cointegrated variables, despite a deterioration in their selective capabilities. Finally, we also consider an empirical applicationto a large macroeconomic u.s. Dataset and demonstrate the competitive performance of penalized regression methods.",
keywords = "Forecasting, Lasso, Factor models, High-dimensional data, Cointegration, DYNAMIC FACTOR MODELS, PRINCIPAL COMPONENT ANALYSIS, APPROXIMATE FACTOR MODELS, TIME-SERIES, ADAPTIVE LASSO, LARGE NUMBER, SELECTION, PREDICTORS, SHRINKAGE, AUTOREGRESSIONS",
author = "Stephan Smeekes and Etienne Wijler",
note = "FRED-MD Database (https://research.stlouisfed.org/econ/mccracken/fred-databases/)",
year = "2018",
month = "7",
day = "1",
doi = "10.1016/j.ijforecast.2018.01.001",
language = "English",
volume = "34",
pages = "408--430",
journal = "International Journal of Forecasting",
issn = "0169-2070",
publisher = "Elsevier Science",
number = "3",

}

Macroeconomic forecasting using penalized regression methods. / Smeekes, Stephan; Wijler, Etienne.

In: International Journal of Forecasting, Vol. 34, No. 3, 01.07.2018, p. 408-430.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Macroeconomic forecasting using penalized regression methods

AU - Smeekes, Stephan

AU - Wijler, Etienne

N1 - FRED-MD Database (https://research.stlouisfed.org/econ/mccracken/fred-databases/)

PY - 2018/7/1

Y1 - 2018/7/1

N2 - We study the suitability of applying lasso-type penalized regression techniques to macroe-conomic forecasting with high-dimensional datasets. We consider the performances of lasso-type methods when the true dgp is a factor model, contradicting the sparsity assumptionthat underlies penalized regression methods. We also investigate how the methods handle unit roots and cointegration in the data. In an extensive simulation study we find that penalized regression methods are more robust to mis-specification than factor models, even if the underlying dgp possesses a factor structure. Furthermore, the penalized regression methods can be demonstrated to deliver forecast improvements over traditional approaches when applied to non-stationary data that contain cointegrated variables, despite a deterioration in their selective capabilities. Finally, we also consider an empirical applicationto a large macroeconomic u.s. Dataset and demonstrate the competitive performance of penalized regression methods.

AB - We study the suitability of applying lasso-type penalized regression techniques to macroe-conomic forecasting with high-dimensional datasets. We consider the performances of lasso-type methods when the true dgp is a factor model, contradicting the sparsity assumptionthat underlies penalized regression methods. We also investigate how the methods handle unit roots and cointegration in the data. In an extensive simulation study we find that penalized regression methods are more robust to mis-specification than factor models, even if the underlying dgp possesses a factor structure. Furthermore, the penalized regression methods can be demonstrated to deliver forecast improvements over traditional approaches when applied to non-stationary data that contain cointegrated variables, despite a deterioration in their selective capabilities. Finally, we also consider an empirical applicationto a large macroeconomic u.s. Dataset and demonstrate the competitive performance of penalized regression methods.

KW - Forecasting

KW - Lasso

KW - Factor models

KW - High-dimensional data

KW - Cointegration

KW - DYNAMIC FACTOR MODELS

KW - PRINCIPAL COMPONENT ANALYSIS

KW - APPROXIMATE FACTOR MODELS

KW - TIME-SERIES

KW - ADAPTIVE LASSO

KW - LARGE NUMBER

KW - SELECTION

KW - PREDICTORS

KW - SHRINKAGE

KW - AUTOREGRESSIONS

U2 - 10.1016/j.ijforecast.2018.01.001

DO - 10.1016/j.ijforecast.2018.01.001

M3 - Article

VL - 34

SP - 408

EP - 430

JO - International Journal of Forecasting

JF - International Journal of Forecasting

SN - 0169-2070

IS - 3

ER -