For collective bargaining we used national union coverage,
since countries like France have low unionization rates but a high percentage
of coverage via the collective bargaining agreements for workers who do not belong to a union.
Since unions and works councils both influence
the ability of managers to make decisions, we used a measure of the interaction
of intensity of mandated employee representation in the country and
the level of unionization For our labor relations law index, we used an international index that
includes laws on collective bargaining, workers’ participation in the company’s
management, and collective disputes, which are then aggregated into
one index 。As part of our measures of the industrial
relations system, we presented the level of bargaining centralization and
coordination, which reflects whether there is establishment, local or
national wage bargaining in a manner developed by Calmfors and Driffill
Since we assume that countries with similar industrial relations
systems may want to invest in nations that have similar or complementary
industrial relations systems, we controlled for this interaction by adding a
dummy variable for whether the sending and host nations have in the same
quintile in the industrial relations system using the summated rating scale.
Using maximum likelihood estimation techniques produces a value
normed at zero for the Rasch value of the industrial relations system. Our
estimates for the partial credit model for the Rasch approach found results
that are consistent with this scaling approach.
Since there is no overriding
statistical or theoretical reason to use only one approach, we presented both
the summated rating scale and the Rasch values in most of our analysis.
Statistically, these measures of the industrial relations system are highly
correlated. For the composite index measure we found that Cronbach’s
alpha measure of the intercorrelation of the industrial relations variables
was 0.74, which was well beyond the acceptable statistical threshold of 0.60
for the intercorrelation of these variables 。
Table 2 shows the simple correlations of the industrial relations system variables. They are
generally highly correlated, except for the labor law index. We nevertheless
included the variable since it has theoretical importance for the industrial
relations system and is likely to impact FDI decisions.
Industrial Relations Institutions, Income Inequality, and FDI
Figure 1 shows the basic country relationships between our measures of
industrial relations institutions and income inequality as measured by the Gini coefficients. The measure of the industrial relations system is the summated
ratings’ value for each nation. Consistent with findings in other studies, the
slope of the line is negative, suggesting that numerous and more intense
levels of industrial relations institutions are associated with less inequality
in the country .