A0299
Title: Shrinkage estimations for social interactions models
Authors: Hon Ho Kwok - National Taiwan University (Taiwan) [presenting]
Abstract: The aim is to give an in-depth analysis of shrinkage estimations for social interactions models. Identification and consistent estimations for social interactions models have been well developed. The statistical properties of the instrumental generalized method of moments (GMM) and maximum likelihood (ML) estimators were thoroughly analyzed in the past few decades. However, the properties of shrinkage estimators have yet been fully understood. A general class of linear and partially nonlinear social interactions models, and its shrinkage estimators, which are relatively computationally simple, are considered. The connection between consistency and shrinkage is developed. The estimation quality is rigorously investigated, for example, in terms of mean squared errors (MSE) and prediction errors. The mathematical conditions for acceptable quality are derived. Lastly, the theory developed would be extended to a decent model selection theory for social interactions models.
