Geometric combination lags as flexible infinite distributed lag estimators (Q1114286)

This paper considers a new class of distributed lag estimators, the geometric combination lag (GCL), with several desirable properties. The estimated weight structure is flexible in practice and satisfies a theoretical approximation theorem. The truncated remainder terms can be estimated rather than simply being dropped. Estimation requires very little computational effort and, under the maintained assumptions, has the properties of OLS. These features will make the GCL attractive to applied researchers.