| dc.description.abstract | This paper proposes a heteroskedasticity-robust Breusch-Pagan test of the null
hypothesis of zero cross-section (or contemporaneous) correlation in linear panel data
models. The procedure allows for either xed, strictly exogenous and/or lagged de-
pendent regressor variables, as well as quite general forms of both non-normality and
heteroskedasticity in the error distribution. Whilst the asymptotic validity of the test
procedure, under the null, is predicated on the number of time series observations,
T, being large relative to the number of cross-section units, N, independence of the
cross-sections is not assumed. Across a variety of experimental designs, a Monte Carlo
study suggests that, in general (but not always), the predictions from asymptotic the-
ory provide a good guide to the finite sample behaviour of the test. In particular,
with skewed errors and/or when N=T is not small, discrepancies can occur. However,
for all the experimental designs, any one of three asymptotically valid wild bootstrap
approximations (that are considered in this paper) gives very close agreement between
the nominal and empirical signi cance levels of the test. Moreover, in comparison with
wild bootstrap version of the original Breusch-Pagan test (Godfrey and Yamagata,
2011) the corresponding version of the heteroskedasticity-robust Breusch-Pagan test is
more reliable. As an illustration, the proposed tests are applied to a dynamic growth
model for a panel of 20 OECD countries. | en_GB |
