Simple tests for social interaction models with network structures

We consider an extended spatial autoregressive model that can incorporate possible endogenous interactions, exogenous interactions, unobserved group fixed effects and the correlation of unobservables. In the generalized method of moments (GMM) and the maximum likelihood (ML) frameworks, we introduce simple gradient-based robust test statistics that can be used to test for the presence of the endogenous effects, the correlation of unobservables and the contextual effects. These test statistics are robust to local parametric misspecifications and only require consistent estimates from a transformed linear regression model to compute. We carry out an extensive Monte Carlo study to investigate the size and power properties of the proposed tests. The results show that the proposed tests have good finite sample properties, and are useful for testing the presence of the various effects in a social interaction model.