A “True” Time Series and its Indicators

Abstract
A problem that economic statisticians frequently face is to estimate the true movement in a time series on the basis of two or more imperfect indicators. A well-known example is nonagricultural employment as indicated (a) by a monthly survey of households and (b) by a monthly survey of employers. This article describes a procedure for estimating true changes in a time series as a linear combination of two indicators, with weights for indicators chosen so as to minimize errors. It applies the procedure to two examples, nonagricultural employment and capital goods prices.

This publication has 1 reference indexed in Scilit: