Department of Economics and Business Economics

Statistical tests for equal predictive ability across multiple forecasting methods

Research output: Working paper/Preprint Working paperResearch

Standard

Statistical tests for equal predictive ability across multiple forecasting methods. / Borup, Daniel; Thyrsgaard, Martin.

Aarhus : Institut for Økonomi, Aarhus Universitet, 2017.

Research output: Working paper/Preprint Working paperResearch

Harvard

Borup, D & Thyrsgaard, M 2017 'Statistical tests for equal predictive ability across multiple forecasting methods' CREATES Research Paper, no. 2017-19, Institut for Økonomi, Aarhus Universitet, Aarhus.

APA

Borup, D., & Thyrsgaard, M. (2017). Statistical tests for equal predictive ability across multiple forecasting methods. Institut for Økonomi, Aarhus Universitet. CREATES Research Paper No. 2017-19

CBE

MLA

Borup, Daniel and Martin Thyrsgaard Statistical tests for equal predictive ability across multiple forecasting methods. Aarhus: Institut for Økonomi, Aarhus Universitet. (CREATES Research Paper; Journal number 2017-19). 2017., 67 p.

Vancouver

Author

Borup, Daniel ; Thyrsgaard, Martin. / Statistical tests for equal predictive ability across multiple forecasting methods. Aarhus : Institut for Økonomi, Aarhus Universitet, 2017. (CREATES Research Paper; No. 2017-19).

Bibtex

@techreport{cbe915ed93a7443d8fa6d16d370f8648,
title = "Statistical tests for equal predictive ability across multiple forecasting methods",
abstract = "We develop a multivariate generalization of the Giacomini-White tests for equal conditional predictive ability. The tests are applicable to a mixture of nested and non-nested models, incorporate estimation uncertainty explicitly, and allow for misspecification of the forecasting model as well as non-stationarity of the data. We introduce two finite-sample corrections, leading to good size and power properties. We also provide a two-step Model Confidence Set-type decision rule for ranking the forecasting methods into sets of indistinguishable conditional predictive ability, particularly suitable in dynamic forecast selection. In the empirical application we consider forecasting of the conditional variance of the S&P500 Index.",
keywords = "forecast comparison, multivariate tests of equal predictive ability, Giacomini-White test, Diebold-Mariano test, conditional forecast combination",
author = "Daniel Borup and Martin Thyrsgaard",
year = "2017",
month = may,
day = "17",
language = "English",
series = "CREATES Research Paper",
publisher = "Institut for {\O}konomi, Aarhus Universitet",
number = "2017-19",
type = "WorkingPaper",
institution = "Institut for {\O}konomi, Aarhus Universitet",

}

RIS

TY - UNPB

T1 - Statistical tests for equal predictive ability across multiple forecasting methods

AU - Borup, Daniel

AU - Thyrsgaard, Martin

PY - 2017/5/17

Y1 - 2017/5/17

N2 - We develop a multivariate generalization of the Giacomini-White tests for equal conditional predictive ability. The tests are applicable to a mixture of nested and non-nested models, incorporate estimation uncertainty explicitly, and allow for misspecification of the forecasting model as well as non-stationarity of the data. We introduce two finite-sample corrections, leading to good size and power properties. We also provide a two-step Model Confidence Set-type decision rule for ranking the forecasting methods into sets of indistinguishable conditional predictive ability, particularly suitable in dynamic forecast selection. In the empirical application we consider forecasting of the conditional variance of the S&P500 Index.

AB - We develop a multivariate generalization of the Giacomini-White tests for equal conditional predictive ability. The tests are applicable to a mixture of nested and non-nested models, incorporate estimation uncertainty explicitly, and allow for misspecification of the forecasting model as well as non-stationarity of the data. We introduce two finite-sample corrections, leading to good size and power properties. We also provide a two-step Model Confidence Set-type decision rule for ranking the forecasting methods into sets of indistinguishable conditional predictive ability, particularly suitable in dynamic forecast selection. In the empirical application we consider forecasting of the conditional variance of the S&P500 Index.

KW - forecast comparison, multivariate tests of equal predictive ability, Giacomini-White test, Diebold-Mariano test, conditional forecast combination

M3 - Working paper

T3 - CREATES Research Paper

BT - Statistical tests for equal predictive ability across multiple forecasting methods

PB - Institut for Økonomi, Aarhus Universitet

CY - Aarhus

ER -