Design for QEWS : Exclusion of small businesses and direct movement estimation in the quarterly economy wide survey (Methodology Advisory Committee)
|This document presents the results of an evaluation of an alternative method of estimating quarterly movements in regular surveys of businesses. The method relies on the use of only the common sample between quarters, supplemented by the inclusion of frame births and frame deaths between the two periods. |
The method, termed 'direct movement estimator' (DIME), is a form of composite estimator of movement. It differs from composite estimation, however, in that it does not attempt to possess optimal weights. It shares with the composite estimator the feature that the estimate of movement is not the difference in the two successive estimates of full-sample level.
The direct-movement method is proposed as part of a wider 'package' of estimation for quarterly surveys. This paper does not propose to address all the aspects of this 'package'. This paper confines itself to investigating two major aspects of the direct movement estimator. Firstly, it investigates how well the direct movement estimator performs as an estimator of quarterly movement. Secondly, it also looks at an estimator of level which is derived from the direct-movement estimator. This new estimator of level, termed the 'direct movement estimator of level (DIMEL)', is simply built by adding successive DIME's to an estimate of level calculated at some base period. How this base-period estimate of level is calculated is not addressed in this paper. The DIMEL has the intuitively attractive feature of having the DIME being the difference of successive estimates of level.