Title: Network identification methods based on change of basis
Authors: Hon Ho Kwok - National Taiwan University (Taiwan) [presenting]
Abstract: Methods for identifying the parameters and networks in linear social interactions models are developed. Three situations are considered: the samples span Rn (long panels); the samples span a proper subspace of Rn (short panels); the models have multiple equilibria. For the short panel situation, the sample vectors are proposed to be represented with respect to a basis of a lower-dimensional space, so that we have fewer regression coefficients, and hence some reduced form submatrices. These submatrices provide equations for identifying the parameters and networks. For the multiple equilibria situation, a method based on matrix triangularization is developed.