On the sequential benchmarking of subannual series to annual totals. (5th September 2018)

• Record Type: Journal Article
• Authors: Daalmans, Jacco
• Other Names: Chen Baoline guestEditor.; Di Fonzo Tommaso guestEditor.; Mushkudiani Nino guestEditor.
• Is Part Of: Statistica Neerlandica. Volume 72:Number 4(2018)
• Page Start: 406
• Page End: 420
• Publication Date: 2018-09-05
• Subjects: consistency -- data integration -- Denton -- reconciliation -- time series; Statistics -- Periodicals; 519.5; 314.92
• DOI: 10.1111/stan.12163
• Languages: English
• ISSNs: 0039-0402

Abstract

Abstract : Temporal benchmarking according to the least squares methods of Denton is widely used for official statistics. The purpose of Denton methods is to achieve consistency between high‐ and low‐frequency data, for example, quarterly and annual data. The high‐frequency data are adjusted to align with low‐frequency data, while preserving as much as possible the short‐term movements of the preliminary high‐frequency data. Theoretically, it is best to benchmark all available data for all periods at once. Practically, such a simultaneous approach is often not feasible, due to the impossibility of changing results that have already been published. Therefore, benchmarking is often applied according to a sequential approach. This paper demonstrates that a popular Denton method is not always appropriate for sequential benchmarking. Undesirable abrupt changes of benchmarking corrections can occur. This paper proposes solutions that better preserve the short‐term movements between sequentially benchmarked series.