A Benchmarking Approach to Temporal Disaggregation of Economic Time Series by Related Series

Abstract

Several methods have been devised to deal with the problem of temporal disaggregation of economic time series (a) either when related series are available or (b) when only aggregate figures exist. In this article, we propose a statistical model-based approach to temporal disaggregation of economic time series by related series. The proposed approach is performed in two stages. In the first stage, we evaluate a preliminary estimate of the disaggregated series using a regression model for the disaggregated series and related series observed in the same frequency. The preliminary estimate of disaggregated series obtained in the first step is not consistent with aggregate figures. To ensure consistency we propose in the second stage, the use of a modified benchmarking approach based on signal extraction (Hillmer and Trabelsi, 1987; Trabelsi and Hillmer, 1990) to adjust the preliminary estimate of disaggregate series. The approach developed here is used for Seasonally Adjusted (SA) and Not Seasonally Adjusted (NSA) data. A comparison with previous temporal disaggregation methods has been done.

Keywords:

• ARIMA model-based approach (AMB)
• Benchmarking
• Consistency
• Signal extraction
• Temporal disaggregation
• Unknown components ARIMA (UCARIMA) model

Mathematics Subject Classification:

• 37M10
• 62M10
• 65D05
• 62D05

Notes

We use the NSA GDP in current prices and not in contant prices because the BEA does not produce the NSA Implicit Price Deflators for Gross Domestic Product.