Forecasting substantial data revisions in the presence of model uncertainty

G.M. Koop, Anthony Garratt, Shaun Vahey

Research output: Contribution to journalArticle

15 Citations (Scopus)
7 Downloads (Pure)

Abstract

A recent revision to the preliminary measurement of GDP(E) growth for 2003Q2 caused considerable press attention, provoked a public enquiry and prompted a number of reforms to UK statistical reporting procedures. In this article, we compute the probability of 'substantial revisions' that are greater (in absolute value) than the controversial 2003 revision. The predictive densities are derived from Bayesian model averaging over a wide set of forecasting models including linear, structural break and regime-switching models with and without heteroscedasticity. Ignoring the nonlinearities and model uncertainty yields misleading predictives and obscures recent improvements in the quality of preliminary UK macroeconomic measurements.
Original languageEnglish
Pages (from-to)1128-1144
Number of pages17
JournalEconomic Journal
Volume118
Issue number530
DOIs
Publication statusPublished - Jul 2008

Fingerprint

Data revisions
Model uncertainty
Bayesian model averaging
Macroeconomics
Regime-switching model
Heteroscedasticity
Predictive density
Nonlinearity
Structural breaks

Keywords

  • structural breaks
  • regime switching
  • model uncertainty
  • Bayesian model averaging
  • predictive densities
  • econometrics

Cite this

Koop, G.M. ; Garratt, Anthony ; Vahey, Shaun. / Forecasting substantial data revisions in the presence of model uncertainty. In: Economic Journal. 2008 ; Vol. 118, No. 530. pp. 1128-1144.
@article{bdbc922eca054d12b9e6bfa12f18d00e,
title = "Forecasting substantial data revisions in the presence of model uncertainty",
abstract = "A recent revision to the preliminary measurement of GDP(E) growth for 2003Q2 caused considerable press attention, provoked a public enquiry and prompted a number of reforms to UK statistical reporting procedures. In this article, we compute the probability of 'substantial revisions' that are greater (in absolute value) than the controversial 2003 revision. The predictive densities are derived from Bayesian model averaging over a wide set of forecasting models including linear, structural break and regime-switching models with and without heteroscedasticity. Ignoring the nonlinearities and model uncertainty yields misleading predictives and obscures recent improvements in the quality of preliminary UK macroeconomic measurements.",
keywords = "structural breaks, regime switching, model uncertainty, Bayesian model averaging, predictive densities, econometrics",
author = "G.M. Koop and Anthony Garratt and Shaun Vahey",
note = "Working paper version",
year = "2008",
month = "7",
doi = "10.1111/j.1468-0297.2008.02163.x",
language = "English",
volume = "118",
pages = "1128--1144",
journal = "Economic Journal",
issn = "0013-0133",
number = "530",

}

Forecasting substantial data revisions in the presence of model uncertainty. / Koop, G.M.; Garratt, Anthony; Vahey, Shaun.

In: Economic Journal, Vol. 118, No. 530, 07.2008, p. 1128-1144.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Forecasting substantial data revisions in the presence of model uncertainty

AU - Koop, G.M.

AU - Garratt, Anthony

AU - Vahey, Shaun

N1 - Working paper version

PY - 2008/7

Y1 - 2008/7

N2 - A recent revision to the preliminary measurement of GDP(E) growth for 2003Q2 caused considerable press attention, provoked a public enquiry and prompted a number of reforms to UK statistical reporting procedures. In this article, we compute the probability of 'substantial revisions' that are greater (in absolute value) than the controversial 2003 revision. The predictive densities are derived from Bayesian model averaging over a wide set of forecasting models including linear, structural break and regime-switching models with and without heteroscedasticity. Ignoring the nonlinearities and model uncertainty yields misleading predictives and obscures recent improvements in the quality of preliminary UK macroeconomic measurements.

AB - A recent revision to the preliminary measurement of GDP(E) growth for 2003Q2 caused considerable press attention, provoked a public enquiry and prompted a number of reforms to UK statistical reporting procedures. In this article, we compute the probability of 'substantial revisions' that are greater (in absolute value) than the controversial 2003 revision. The predictive densities are derived from Bayesian model averaging over a wide set of forecasting models including linear, structural break and regime-switching models with and without heteroscedasticity. Ignoring the nonlinearities and model uncertainty yields misleading predictives and obscures recent improvements in the quality of preliminary UK macroeconomic measurements.

KW - structural breaks

KW - regime switching

KW - model uncertainty

KW - Bayesian model averaging

KW - predictive densities

KW - econometrics

UR - http://personal.strath.ac.uk/gary.koop/GKV-article.pdf

U2 - 10.1111/j.1468-0297.2008.02163.x

DO - 10.1111/j.1468-0297.2008.02163.x

M3 - Article

VL - 118

SP - 1128

EP - 1144

JO - Economic Journal

JF - Economic Journal

SN - 0013-0133

IS - 530

ER -