Dear all,
Many thanks for all your feedback which is indeed very
useful and interesting.
The reason why I was asking about how to check for
heteroskedasticity is the following. In a recent
paper, Santos Silva and Tenreyro (2007) show that
"under heteroskedasticity, the parameters of
log-linearized models estimated by OLS lead to biased
estimates of the true elasticities". They show that
the Poisson estimator is more appropriate.
I am estimating a model similar to a gravity equation
of trade, where the dependent and independent
variables are in logs and the data are time series
cross-sectional. I wanted to check that
heteroskedasticity is indeed present in order to
justify the use of the Poisson estimator instead.
