Forecasters have frequently been concerned with designing seasonal adjustment procedures that satisfy particular theoretical criteria (e.g. orthogonality, idempotency, symmetry, Lovell). In evaluating the merits of a particular technique, Monte Carlo studies are often undertaken and the results are then compared to those derived from the Census Bureau's X—11 routine (Wallis, Stephenson, Grether). However, many practical questions have not been addressed, such as to what extent can seasonal routines affect parameter estimates, forecast values, and policy scenarios? The purpose of this article is to focus upon these questions. Data from a short‐term petroleum demand model is seasonally adjusted six different ways. The seasonally adjusted data is then used to estimate the demand relationships of the model using the same structural equation in each case. The results of these estimations provide illuminating information about how seasonality affects parameter values. For policy purposes, this information can be crucial as various policies can be predicated upon an estimated response to a particular variable (e.g. the price of gasoline). The question answered here is how sensitive are the expected policy results to the type of seasonal routine employed in making the estimations.
BOPP, A. and DURST, M. (1978), "A COMPARISON OF THE EFFECT OF SEASONAL ADJUSTMENT ON PARAMETER VALUES, FORECASTS AND POLICY ANALYSIS", Journal of Economic Studies, Vol. 5 No. 1, pp. 20-30. https://doi.org/10.1108/eb008070Download as .RIS
MCB UP Ltd
Copyright © 1978, MCB UP Limited