inputs from the “North” and 70 percent from “Asia” and then it sells 40 percent of its
output domestically and 60 percent to the “North”, we have that the “Asia-North” flow has
the highest weight. In fact this firm is mostly characterized by having upstream contacts
with Asia and downstream contacts with the North. Even though the other flows40 are not
excluded from the analysis but they enter with a lower weight.
From the combinations of the four origin/destinations, sixteen kinds of flows are
generated. However for each flow there are too little non-zero observations to perform a
parametric analysis. For this reason we choose to group these flows according to different
criteria.
Firstly we concentrate on those flows that have the same origin and destination. This is to
test the idea that specialization towards a specific market generates the necessary
knowledge to overcome information and search costs. Therefore it permits to find the
most appropriate inputs and to better know the standards required to satisfy local demand
in order increase efficiency.
Thus, we construct an index that groups all the flows for which j=i excluding the domestic
cases41. Table 8a reports the results from the estimations obtained including these indexes
in equation (8). As expected, the index that measures the magnitude of import and export
flows to the same area shows a positive sign and it is statistically different from zero. The
sign and statistical significance is maintained also when controls such as import share,
export share or IE index are introduced. In addition, column (4) displays the coefficient of
the ratio between the index referring to the same origin and destination and the IE index.
This term gives a measure of the weight of those international activities concentrated on
the same areas with respect to all the international activities. Such coefficient is positive
and statistically significant indicating that the correlation between performance and
concentration persists regardless the amount of international trade the firm is involved in
as long as it is spatially concentrated. However when we introduce lagged values of the
variables as instruments, to solve the possible endogeneity bias, the number of
observations becomes fairly small and the results become weaker (columns (6) to (10)).
40 “Asia-Home”, “North-Home” and “North-North”.
41 When j=Home and I=Home
24