Subgraph Network Random Effects Error Components Models: Specification and Testing

This paper develops a subgraph random effects error components model for network data linear regression where the unit of observation is the node. In particular, it allows for link and triangle specific components, which serve as a basal model for modeling network effects. It then evaluates the potential effects of ignoring network effects in the estimation of the coefficients’ variance-covariance matrix. It also proposes consistent estimators of the variance components using quadratic forms and Lagrange Multiplier tests for evaluating the appropriate model of random components in networks. Monte Carlo simulations show that the tests have good performance in finite samples. It applies the proposed tests to the Call interbank market in Argentina.

Sur l’estimation des équations de CANDIDE-R

CANDIDE-R is a huge simultaneous macro-economic model which raises estimations difficulties. We avoid the problem of identification assuming that the great number of variables in our model makes it impossible that the necessary condition be not satisfied. We assume that our system converges to a solution solving this way the problem of identification. The core of the paper gives justifications of the procedure we adopted to estimate CANDIDE-R. Because of the presence of regional equations and the limited amount of regional data, we are bound to pool cross sections and time series data. We then justified the use of Zellner's approach instead of the error components models within the class of regional models built on national premises.