Modified Joint Test of Heteroscedasticity and Serial Correlation in a Random Effect Panel Data Model under Restricted Alternatives

Abstract

Abstract In time series econometric theory, the sign of the parameter may be pre-assigned, that is the parameter may be ordered or restricted. Estimation of this restricted parameter is then a vital problem. If we know the parameters are restricted by some constraints, it is reasonable to expect that we should be able to do better by incorporating such additional information than by ignoring them. The paper concerned with estimating and testing restricted hypotheses in Panel data regression model. The aim of the paper: constructing a modified LM test statistic to test the joint effect of heteroscedasticity and serial correlation in a random effect error component model, where the hypotheses are restricted. Monte Carlo results show that the tests have good size and power under various forms of heteroscedasticity and serial correlation. Keywords: Panel Data Model, heteroscedasticity, serial correlation, unconstraint and constraint based LM test