determinants in the gravity theory (see Frankel and Romer ,1999).
3 Evidence on the Linkage of Infrastructure and Capital
Flows
In this section a cross-country data set, exhibited in Tables 1 and 2, is utilised to test the
prediction that the inflows of external capital to countries are positively related to the level of
infrastructure in the economy.5 The relationship becomes vital if the level of infrastructure is
able to promote the inflow of capital to a certain extent.
The regression analysis makes use of a cross-sectional approach. This type of approach allows
to abstract oneself from short-run fluctuations in the external capital liabilities which occur due to
temporary shocks in the world capital markets. Another motivation for a cross-sectional analysis
is that measures of infrastructure are more convincing in capturing cross-country variations
rather than tracking changes in the level of infrastructure of an individual country over time.
The cross sectional analysis consists of 30 countries from Asia, Africa as well as Latin America
and uses data from 1990 to 1995 (refer to Table 3).6 With reference to the stock of total liabilities
and debt the sample size is restricted to 29 observations due to the fact that South Africa did
not report on these items. The results on longer time series, including periods from the 1970s
onwards, were similar to the ones reported here. A sample which also containing industrialised
countries was also experimented with. However, the most interesting results occurred for the
developing country sample.
The developing country sample appears to be appropriate since variations in infrastructure
between developing countries may have a stronger impact on the attraction of capital inflows.
The definition of developing countries in this paper is broadly consistent with those countries
likely to be elected for developing country treatment by the World Trade Organization (WTO).
The group therefore includes the Republic of Korea and Singapore, which were clearly devel-
oping economies in the last decades but are now classified by the World Bank as high-income.
However, Singapore is not included in the list of developing countries.7 Since Singapore has an
5In an earlier version of this paper public investment and its relationship with external capital inflows was also
considered. The measurement was obtained by the construction of a perpetual public capital stock. However,
since one cannot say whether public investment is used efficiently, those results are not reported.
6Multivariate outliers were tested for using the procedure suggested by Hadi (1992, 1994) in the regression
specification.
7In general an outlier like Singapore is not necessarily bad. The variation in the data is exactly what allows
the identification of relationships. However, if an outlier is non-representative due to factors that make it different
from the rest of the sample, it makes sense to exclude it from the sample.