Automatic locally stationary time series forecasting with application to predicting UK gross value added time series

Rebecca Killick; Marina I. Knight; Guy P. Nason; Matthew Nunes; Idris A. Eckley

doi:10.1093/jrsssc/qlae043

Automatic locally stationary time series forecasting with application to predicting UK gross value added time series

Rebecca Killick
, Marina I. Knight
, Guy P. Nason
, Matthew Nunes
, Idris A. Eckley

Lancaster University
York University
Imperial College London

Research output: Contribution to journal › Article › peer-review

4 Citations (SciVal)

Abstract

Accurate forecasting of the U.K. gross value added (GVA) is fundamental for measuring the growth of the U.K. economy. A common nonstationarity in GVA data, such as the ABML series, is its increase in variance over time due to inflation. Transformed or inflation-adjusted series can still be challenging for classical stationarity-assuming forecasters. We adopt a different approach that works directly with the GVA series by advancing recent forecasting methods for locally stationary time series. Our approach results in more accurate and reliable forecasts, and continues to work well even when the ABML series becomes highly variable during the COVID pandemic.

Original language	English
Article number	qlae043
Pages (from-to)	18-33
Number of pages	16
Journal	Journal of the Royal Statistical Society: Series C - Applied Statistics
Volume	74
Issue number	1
Early online date	23 Aug 2024
DOIs	https://doi.org/10.1093/jrsssc/qlae043
Publication status	Published - 31 Jan 2025

Bibliographical note

Publisher Copyright:
© 2024 The Royal Statistical Society.

Data Availability Statement

The ABML data can be obtained from the Office for National Statistics website https://www.ons.gov.uk. The (differenced) ABML dataset for the period examined in this article, as well as the forecasting method, is available within the forecast LSW R package on CRAN.

Funding

The authors were partially supported by the Research Councils UK (RCUK) Energy Programme. The Energy Programme is an RCUK cross-council initiative led by Engineering and Physical Sciences Research Council (EPSRC) and contributed to by Economic and Social Research Council (ESRC), Natural Environment Research Council (NERC), Biotechnology and Biological Sciences Research Council (BBSRC), and Science and Technology Facilities Council (STFC). G.P.N. gratefully acknowledges support from EPSRC grants EP/I01697X/1 and K020951/1. I.A.E. gratefully acknowledges support from EPSRC grant EP/I016368/1.

Funders	Funder number
Natural Environment Research Council
Economic and Social Research Council
Research Councils UK
Biotechnology and Biological Sciences Research Council
Engineering and Physical Sciences Research Council
Science and Technology Facilities Council	K020951/1, EP/I016368/1, EP/I01697X/1

Keywords

COVID shocks
classical forecasting
local partial autocorrelation
spectral estimation
wavelets

ASJC Scopus subject areas

Statistics and Probability
Statistics, Probability and Uncertainty

Access to Document

10.1093/jrsssc/qlae043Licence: CC BY

Cite this

@article{7a7eef50d5e645d3a9f0a035305a98e7,

title = "Automatic locally stationary time series forecasting with application to predicting UK gross value added time series",

abstract = "Accurate forecasting of the U.K. gross value added (GVA) is fundamental for measuring the growth of the U.K. economy. A common nonstationarity in GVA data, such as the ABML series, is its increase in variance over time due to inflation. Transformed or inflation-adjusted series can still be challenging for classical stationarity-assuming forecasters. We adopt a different approach that works directly with the GVA series by advancing recent forecasting methods for locally stationary time series. Our approach results in more accurate and reliable forecasts, and continues to work well even when the ABML series becomes highly variable during the COVID pandemic.",

keywords = "COVID shocks, classical forecasting, local partial autocorrelation, spectral estimation, wavelets",

author = "Rebecca Killick and Knight, \{Marina I.\} and Nason, \{Guy P.\} and Matthew Nunes and Eckley, \{Idris A.\}",

note = "Publisher Copyright: {\textcopyright} 2024 The Royal Statistical Society.",

year = "2025",

month = jan,

day = "31",

doi = "10.1093/jrsssc/qlae043",

language = "English",

volume = "74",

pages = "18--33",

journal = "Journal of the Royal Statistical Society: Series C - Applied Statistics",

issn = "0035-9254",

publisher = "Oxford University Press",

number = "1",

}

TY - JOUR

T1 - Automatic locally stationary time series forecasting with application to predicting UK gross value added time series

AU - Killick, Rebecca

AU - Knight, Marina I.

AU - Nason, Guy P.

AU - Nunes, Matthew

AU - Eckley, Idris A.

PY - 2025/1/31

Y1 - 2025/1/31

N2 - Accurate forecasting of the U.K. gross value added (GVA) is fundamental for measuring the growth of the U.K. economy. A common nonstationarity in GVA data, such as the ABML series, is its increase in variance over time due to inflation. Transformed or inflation-adjusted series can still be challenging for classical stationarity-assuming forecasters. We adopt a different approach that works directly with the GVA series by advancing recent forecasting methods for locally stationary time series. Our approach results in more accurate and reliable forecasts, and continues to work well even when the ABML series becomes highly variable during the COVID pandemic.

AB - Accurate forecasting of the U.K. gross value added (GVA) is fundamental for measuring the growth of the U.K. economy. A common nonstationarity in GVA data, such as the ABML series, is its increase in variance over time due to inflation. Transformed or inflation-adjusted series can still be challenging for classical stationarity-assuming forecasters. We adopt a different approach that works directly with the GVA series by advancing recent forecasting methods for locally stationary time series. Our approach results in more accurate and reliable forecasts, and continues to work well even when the ABML series becomes highly variable during the COVID pandemic.

KW - COVID shocks

KW - classical forecasting

KW - local partial autocorrelation

KW - spectral estimation

KW - wavelets

UR - https://www.scopus.com/pages/publications/85214882844

U2 - 10.1093/jrsssc/qlae043

DO - 10.1093/jrsssc/qlae043

M3 - Article

SN - 0035-9254

VL - 74

SP - 18

EP - 33

JO - Journal of the Royal Statistical Society: Series C - Applied Statistics

JF - Journal of the Royal Statistical Society: Series C - Applied Statistics

IS - 1

M1 - qlae043

ER -

Automatic locally stationary time series forecasting with application to predicting UK gross value added time series

Abstract

Bibliographical note

Data Availability Statement

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this