Construction of the 2005 Index of Globalization

To construct the indices of globalization, each original variable has been transformed to an

index on a zero to ten scale, where ten is the maximum value for a specific variable over the

period 1970-2003, and zero is the minimum value. 1 Higher values denote more globalization.

When higher values of the original variable indicate higher globalization, the formula ((Vi-

Vmin)/(Vmax-Vmin)*10) has been used for transformation. Conversely, when higher values

indicate less globalization, the formula is ((Vmax-Vi)/(Vmax-Vmin)*10). The weights for

calculating the sub-indices are determined using principal components analysis. The analysis

partitions the variance of the variables used in each sub-group. The weights are then

determined in a way that maximizes the variation of the resulting principal component, so that

the indices capture the variation as fully as possible. 2 The same procedure is applied to the

sub-indices in order to derive the overall index of globalization.

Data are calculated on a yearly basis. However, not all data are available for all

countries and years. In calculating the indices the weights are readjusted to correct for this. 3

Some variables without much time series variation have been linearly interpolated before

applying the weighting procedure. 4 Data for sub-indices and the overall index of globalization

are not calculated, if they rely on a small range of variables in a specific year and country. 5

1
Note that there is a methodological change as compared to the 2002 index of globalization, where the variables
have been normalized between zero and ten for each particular year. Normalizing over the whole period implies
the advantage of better comparability over time. As one drawback, the resulting globalization index is affected
by the inclusion of additional countries.
2
In the 2002 release of the globalization index the weights have been calculated for the year 2000 and have then
been used to calculate the indices back to 1971. The present version of the index calculates the weights on the
basis of all data available.
3
Observations with value zero do not represent missing data (and enter the index with weight zero).
4
Specifically, missing values have been interpolated for hidden import barriers, the mean tariff rate, capital
account restrictions, the number of embassies in a country, membership in international organizations, foreign
population, costs of telephone call to the US, and the number of McDonalds restaurants.
5
Observations for the actual economic flows area are reported as missing when data for more than two variables
are unavailable. The same is true for restrictions. Data on personal contacts and, respectively, information flows
are reported as missing when the cumulative weights of the variables available is less than 0.4. The overall index
is based on at least two sub-indices.