explained by the regression specification. Interestingly, looking at latitude, countries with greater
distance to the equator receive more FDI. While air-departures remain significantly positive in
the regression, a negative and individually significant impact of roads and international telephone
circuits can be observed. A test for the joint significance of the three infrastructure variables has
a p-value of 0.000; therefore the variables are also jointly highly significant. The arising question
is why roads and international telephone circuits become negative in sign. One explanation
is that the correlations between the infrastructure variables, especially between air-departures
and international telephone circuits, contribute to the changes in sign. The findings in Table 6
suggest that the level of air-departures is an important determinant of stock of FDI in a cross-
section of countries. As before, openness is important for explaining the stock of international
capital. This time openness is negatively correlated to FDI. The results obtained suggest that
also natural resources have a negative impact on the attraction of FDI stocks. An interesting
finding, illustrated in Table 6, is that countries with access to coastal areas, i.e. ports, are able
to attract more FDI liabilities. The analysis of the portfolio equity stock does show no effects
overall. Therefore, results are omitted from the discussion.
The relationship between the average stock of debt and infrastructure is documented in Table
7.12 Out of the three bivariate specifications in columns (1) to (3) the variable on international
telephone circuits is significant at the one percent level and explains 30 percent of the variation
in the stock of debt across countries. A one percentage point improvement in the level of inter-
national telephone circuits is associated with an increase of 2.35 percentage points in the stock
of debt. Controlling for cross-sectional variations in wealth and size in columns (4) to (6), even
international telephone circuits lose their statistical significance. In all three columns the size
variable, GDP, has a negative and individually highly significant impact on the average stock
of debt. Introducing the remaining regressors in columns (7) to (9) does not alter the charac-
teristics of the infrastructure variables. None of the infrastructure variables enters significantly
in our specification, even though they keep their positive sign. However, now also the wealth of
the country plays a negative and individually significant role in determining the average stock
of debt. This suggests that poorer countries hold higher stocks of debt. As it is the case for the
stock of total liabilities, openness is positively related to the countries’ stock of debt liabilities.
This is confirmed in columns (7) to (10). Natural resources enter marginally significantly in the
specification for roads and international telephone circuits and with a negative sign for the three
infrastructure variables. The joint specification in column (10) reveals the importance of the
geographical position of countries. Countries closer to the equator receive more debt. A joint
test provides a p-value of only 0.924, implying no joint significance of the infrastructure variables
12South Africa is excluded due to missing observations.
10