Illustration of Denton’s Temporal Disaggregation Methods with the Application of Annual GDP of Bangladesh for Quarterly Benchmarking and therefore Forecasting

Abstract

Different methods of temporal Disaggregation are discussed in detail, mainly Denton’s methods developed by Denton in 1971, and others are purely mathematical. In the beginning, Denton's original article describes the original method that Denton developed and its solution. Annual export (low-frequency series) of Bangladesh is disaggregated into quarterly series (high-frequency series) by the methods of Denton’s additive and proportional (first and second difference), and Denton-Cholette additive and proportional (first and second difference) methods considering quarterly imports of capital goods & others as the indicator series for the fiscal year FY2009 to fiscal year FY2019. The R package “tempdisagg” [47] for temporal disaggregation of time series has been used for the temporal disaggregation of quarterly data. By comparing the estimated series with the real quarterly export with the aid of root-mean-squared errors (RMSEs), it comes to an end that the Denton-Cholette additive method (the first difference) performs better than the methods that are considered