Identifying Identical Distributed Lag Structures by the Use of Prior SumConstraints
NBER Working Paper No. 179
This paper derives an estimation procedure which, when the same distributed lag appears twice in an equation to be estimated by least-squares regression, identifies all of the relevant coefficients and lag weights and also constrains the two sets of individual lag weights to be identical. The procedure for solving this identification-constraint problem involves prior imposition of a restriction on the lag weight sum -- i.e., it is necessary to impose the sum restriction before estimating the equation. A further useful feature of the derived procedure is that it facilitates conveniently imposing the sum restriction on all of the weights in a distributed lag even if the leading weight is independent of a polynomial restriction imposed on the others.
Document Object Identifier (DOI): 10.3386/w0179