Interpretation and use of sensitivity in econometrics, illustrated with forecast combinations

J.R. Magnus, A.L. Vasnev

Research output: Contribution to JournalArticleAcademicpeer-review

Abstract

Sensitivity analysis is important both for its own sake and in combination with diagnostic testing. We consider the question of how to use sensitivity statistics in practice, and in particular, how to judge whether the sensitivity is large or small. For this purpose, we distinguish between absolute and relative sensitivity, and highlight the context-dependent nature of sensitivity analysis. The relative sensitivity is then applied to forecast combinations, and sensitivity-based weights are introduced. All of the concepts are illustrated using the European yield curve. In this context, it is natural to consider the sensitivity to autocorrelation and normality assumptions. Different forecasting models are combined using equal, fit-based and sensitivity-based weights, and compared with the multivariate and random walk benchmarks. We show that the fit-based and sensitivity-based weights are complementary, but that the sensitivity-based weights perform better than other weights for long-term maturities. Crown Copyright © 2013.
Original languageEnglish
Pages (from-to)769-781
JournalInternational Journal of Forecasting
Volume31
Issue number3
DOIs
Publication statusPublished - 2015

Fingerprint

Forecast combination
Econometrics
Sensitivity analysis
Random walk
Autocorrelation
Diagnostics
Testing
Normality
Benchmark
Yield curve
Maturity
Statistics

Bibliographical note

PT: J; NR: 24; TC: 0; J9: INT J FORECASTING; PG: 13; GA: CM6YB; UT: WOS:000357836500014

Cite this

@article{f0d33bd3ebd0448abc4a340cf85192dd,
title = "Interpretation and use of sensitivity in econometrics, illustrated with forecast combinations",
abstract = "Sensitivity analysis is important both for its own sake and in combination with diagnostic testing. We consider the question of how to use sensitivity statistics in practice, and in particular, how to judge whether the sensitivity is large or small. For this purpose, we distinguish between absolute and relative sensitivity, and highlight the context-dependent nature of sensitivity analysis. The relative sensitivity is then applied to forecast combinations, and sensitivity-based weights are introduced. All of the concepts are illustrated using the European yield curve. In this context, it is natural to consider the sensitivity to autocorrelation and normality assumptions. Different forecasting models are combined using equal, fit-based and sensitivity-based weights, and compared with the multivariate and random walk benchmarks. We show that the fit-based and sensitivity-based weights are complementary, but that the sensitivity-based weights perform better than other weights for long-term maturities. Crown Copyright {\circledC} 2013.",
author = "J.R. Magnus and A.L. Vasnev",
note = "PT: J; NR: 24; TC: 0; J9: INT J FORECASTING; PG: 13; GA: CM6YB; UT: WOS:000357836500014",
year = "2015",
doi = "10.1016/j.ijforecast.2013.08.001",
language = "English",
volume = "31",
pages = "769--781",
journal = "International Journal of Forecasting",
issn = "0169-2070",
publisher = "Elsevier",
number = "3",

}

Interpretation and use of sensitivity in econometrics, illustrated with forecast combinations. / Magnus, J.R.; Vasnev, A.L.

In: International Journal of Forecasting, Vol. 31, No. 3, 2015, p. 769-781.

Research output: Contribution to JournalArticleAcademicpeer-review

TY - JOUR

T1 - Interpretation and use of sensitivity in econometrics, illustrated with forecast combinations

AU - Magnus, J.R.

AU - Vasnev, A.L.

N1 - PT: J; NR: 24; TC: 0; J9: INT J FORECASTING; PG: 13; GA: CM6YB; UT: WOS:000357836500014

PY - 2015

Y1 - 2015

N2 - Sensitivity analysis is important both for its own sake and in combination with diagnostic testing. We consider the question of how to use sensitivity statistics in practice, and in particular, how to judge whether the sensitivity is large or small. For this purpose, we distinguish between absolute and relative sensitivity, and highlight the context-dependent nature of sensitivity analysis. The relative sensitivity is then applied to forecast combinations, and sensitivity-based weights are introduced. All of the concepts are illustrated using the European yield curve. In this context, it is natural to consider the sensitivity to autocorrelation and normality assumptions. Different forecasting models are combined using equal, fit-based and sensitivity-based weights, and compared with the multivariate and random walk benchmarks. We show that the fit-based and sensitivity-based weights are complementary, but that the sensitivity-based weights perform better than other weights for long-term maturities. Crown Copyright © 2013.

AB - Sensitivity analysis is important both for its own sake and in combination with diagnostic testing. We consider the question of how to use sensitivity statistics in practice, and in particular, how to judge whether the sensitivity is large or small. For this purpose, we distinguish between absolute and relative sensitivity, and highlight the context-dependent nature of sensitivity analysis. The relative sensitivity is then applied to forecast combinations, and sensitivity-based weights are introduced. All of the concepts are illustrated using the European yield curve. In this context, it is natural to consider the sensitivity to autocorrelation and normality assumptions. Different forecasting models are combined using equal, fit-based and sensitivity-based weights, and compared with the multivariate and random walk benchmarks. We show that the fit-based and sensitivity-based weights are complementary, but that the sensitivity-based weights perform better than other weights for long-term maturities. Crown Copyright © 2013.

U2 - 10.1016/j.ijforecast.2013.08.001

DO - 10.1016/j.ijforecast.2013.08.001

M3 - Article

VL - 31

SP - 769

EP - 781

JO - International Journal of Forecasting

JF - International Journal of Forecasting

SN - 0169-2070

IS - 3

ER -