Essays on Cross-Sectionally Dependent Panel Data
with an Application to Fiscal Policy in the European
Monetary Union

Abstract

In the past two decades, macroeconomists have used panel data to study the merits of fiscal policy for economic stabilisation. The datasets considered in these studies typically consist of a small number of time series corresponding to countries. This configuration does not match with the archetypical survey-style panel ataset for which a large literature concerning estimation and hypothesis testing exists. This PhD develops an estimation methodology that is catered towards macroeconomists: in four self-contained chapters, we develop a methodology for the estimation of dynamic models in the small N, large T framework in the presence of cross-sectional dependence in the error term.
In the first chapter we examine the effect of factors on the point estimates of several commonly-used estimators in the empirical literature and we find that these estimators are inconsistent. We also propose an estimator that is consistent for the parameters for of the model studied in that chapter.
In the second chapter we develop consistent quasi-difference GMM estimators and inferential procedures
for the small N, large T dynamic panel data model with factor error structures. We also prove consistency
and mixed-normality of the estimator when the number of factors is over-estimated.
In the third chapter we consider the large N, large T framework and show the first eigenvalues of the
covariance matrix of an approximate factor model are dominated by the factors whereas the remainder is
controlled by the residual noise. We show that this result is the basis for any consistent inferential procedure about R and continues to hold when R grows large, when the factors are weak and, importantly, in the large N, large T interactive fixed effects model.
In the fourth chapter we study fiscal policy using the methods developed in the thesis. We estimate
vector autoregressions from European countries and restrict the impulse-response functions to adhere to the
Stability and Growth Pact. We find that this one-size-fits-all approach is not appropriate for stabilization of
the European economy.