CREATES

On long memory origins and forecast horizons

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

Standard

On long memory origins and forecast horizons. / Vera-Valdés, J. Eduardo.

In: Journal of Forecasting, Vol. 39, No. 5, 08.2020, p. 811-826.

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

Harvard

Vera-Valdés, JE 2020, 'On long memory origins and forecast horizons', Journal of Forecasting, vol. 39, no. 5, pp. 811-826. https://doi.org/10.1002/for.2651

APA

Vera-Valdés, J. E. (2020). On long memory origins and forecast horizons. Journal of Forecasting, 39(5), 811-826. https://doi.org/10.1002/for.2651

CBE

Vera-Valdés JE. 2020. On long memory origins and forecast horizons. Journal of Forecasting. 39(5):811-826. https://doi.org/10.1002/for.2651

MLA

Vera-Valdés, J. Eduardo. "On long memory origins and forecast horizons". Journal of Forecasting. 2020, 39(5). 811-826. https://doi.org/10.1002/for.2651

Vancouver

Vera-Valdés JE. On long memory origins and forecast horizons. Journal of Forecasting. 2020 Aug;39(5):811-826. https://doi.org/10.1002/for.2651

Author

Vera-Valdés, J. Eduardo. / On long memory origins and forecast horizons. In: Journal of Forecasting. 2020 ; Vol. 39, No. 5. pp. 811-826.

Bibtex

@article{eea43143ca254ddf864c4198a8aff532,
title = "On long memory origins and forecast horizons",
abstract = "Most long memory forecasting studies assume that long memory is generated by the fractional difference operator. We argue that the most cited theoretical arguments for the presence of long memory do not imply the fractional difference operator and assess the performance of the autoregressive fractionally integrated moving average (ARFIMA) model when forecasting series with long memory generated by nonfractional models. We find that ARFIMA models dominate in forecast performance regardless of the long memory generating mechanism and forecast horizon. Nonetheless, forecasting uncertainty at the shortest forecast horizon could make short memory models provide suitable forecast performance, particularly for smaller degrees of memory. Additionally, we analyze the forecasting performance of the heterogeneous autoregressive (HAR) model, which imposes restrictions on high-order AR models. We find that the structure imposed by the HAR model produces better short and medium horizon forecasts than unconstrained AR models of the same order. Our results have implications for, among others, climate econometrics and financial econometrics models dealing with long memory series at different forecast horizons.",
keywords = "ARFIMA, cross-sectional aggregation, forecasting, HAR model, long memory, nonfractional memory",
author = "Vera-Vald{\'e}s, {J. Eduardo}",
year = "2020",
month = aug,
doi = "10.1002/for.2651",
language = "English",
volume = "39",
pages = "811--826",
journal = "Journal of Forecasting",
issn = "0277-6693",
publisher = "JohnWiley & Sons Ltd.",
number = "5",

}

RIS

TY - JOUR

T1 - On long memory origins and forecast horizons

AU - Vera-Valdés, J. Eduardo

PY - 2020/8

Y1 - 2020/8

N2 - Most long memory forecasting studies assume that long memory is generated by the fractional difference operator. We argue that the most cited theoretical arguments for the presence of long memory do not imply the fractional difference operator and assess the performance of the autoregressive fractionally integrated moving average (ARFIMA) model when forecasting series with long memory generated by nonfractional models. We find that ARFIMA models dominate in forecast performance regardless of the long memory generating mechanism and forecast horizon. Nonetheless, forecasting uncertainty at the shortest forecast horizon could make short memory models provide suitable forecast performance, particularly for smaller degrees of memory. Additionally, we analyze the forecasting performance of the heterogeneous autoregressive (HAR) model, which imposes restrictions on high-order AR models. We find that the structure imposed by the HAR model produces better short and medium horizon forecasts than unconstrained AR models of the same order. Our results have implications for, among others, climate econometrics and financial econometrics models dealing with long memory series at different forecast horizons.

AB - Most long memory forecasting studies assume that long memory is generated by the fractional difference operator. We argue that the most cited theoretical arguments for the presence of long memory do not imply the fractional difference operator and assess the performance of the autoregressive fractionally integrated moving average (ARFIMA) model when forecasting series with long memory generated by nonfractional models. We find that ARFIMA models dominate in forecast performance regardless of the long memory generating mechanism and forecast horizon. Nonetheless, forecasting uncertainty at the shortest forecast horizon could make short memory models provide suitable forecast performance, particularly for smaller degrees of memory. Additionally, we analyze the forecasting performance of the heterogeneous autoregressive (HAR) model, which imposes restrictions on high-order AR models. We find that the structure imposed by the HAR model produces better short and medium horizon forecasts than unconstrained AR models of the same order. Our results have implications for, among others, climate econometrics and financial econometrics models dealing with long memory series at different forecast horizons.

KW - ARFIMA

KW - cross-sectional aggregation

KW - forecasting

KW - HAR model

KW - long memory

KW - nonfractional memory

UR - http://www.scopus.com/inward/record.url?scp=85079450775&partnerID=8YFLogxK

U2 - 10.1002/for.2651

DO - 10.1002/for.2651

M3 - Journal article

AN - SCOPUS:85079450775

VL - 39

SP - 811

EP - 826

JO - Journal of Forecasting

JF - Journal of Forecasting

SN - 0277-6693

IS - 5

ER -