Next we follow a similar exercise to that used by Girma et. al. (2005) and Wagner
(2006) to test the rank ordering of the TFP distribution of firms that differ in their
involvement in international markets.17 Calculating a two-sided Kolmogorov-Smirnov
statistic, it is possible to test whether the productivity distribution of one sub-group of
firms lies to the right of another sub-group. If so, there is shown to be first-order
stochastic dominance between such variables.
Table 2 (the first two blocks of results) shows that firstly, in every industry examined
firms that export have a distribution that lies significantly to the right of non-
exporters.18 We also look at the TFP levels for foreign-owned firms, comparing them
to both UK exporters and non-exporters. As shown in the third and forth blocks of
results, the distribution of TFP for foreign-owned subsidiaries dominates that of UK-
owned non-exporters, but that foreign-owned firms are not always better than UK-
owned exporters (UK-owned exporters unambiguously dominate foreign-owned firms
in 10 out of the 30 industries examined). Lastly, the final set of results suggest that the
TFP distributions of foreign-owned exporters are generally to the right of those of
foreign-owned non-exporters in a majority of industry groups but not all. 19 Overall,
this suggests that foreign-owned firms operating in the UK are less useful as a
comparator sub-group when considering whether exporters have relatively higher
productivity, since non-exporting foreign-owned firms have productivity advantages
that do not necessarily stem from exporting to overseas markets (indeed FDI itself is
an alternative means of internationalisation - see Head and Ries, 2003; Helpman et.
17 See Section IV for details on the estimation of TFP.
18 However, for three industries (financial intermediation; real estate; and other business services) it is
also possible to reject the null that the distribution for exporters is more favourable. A closer
examination shows that in these industries, exporters dominate non-exporters for a large part of the
distribution of TFP values, but at some levels (usually at high levels of TFP) there is a cross-over and
non-exporters dominate exporters.
19 Our results therefore confirm those presented by Girma et. al. (op. cit.) that the productivity
distribution of exporters dominate that of non-exporters in the UK (although we also cover non-
manufacturing in this study).
14