Skip to main content Skip to navigation

Normalisation

The method we use is panel normalisation. To illustrate, suppose the variable is trade. First, the minimum and maximum values of this variable over the years 1970-2001, and over all countries, are found. In the case of the trade variable, it turns out that the maximum is 1448 %, for Netherlands Antilles in 1980, and the minimum is 6%, for Sudan in 1987. Then, if the trade variable for some country (say the UK) in some year is x%, then the panel normalised value of x is

 

            y = (x – 6)/(1448-6)

 

Note that with this normalisation, all values of y lie between zero and one. Note also that,  with this method,  the means and the variances of different variables can still be different to each other, although they will be much closer to each other than before panel normalisation.

Panel normalisation has both advantages and disadvantages. The advantage is that with panel-normalised data, we can make meaningful comparisons over time for a given country or indeed between countries. A disadvantage, discussed in detail in Lockwood (2004), is that when additional years of data are added to the database, the maximum or minimum value of a variable may change, and those variables affected then have to be re-normalised. For example, suppose Sudan’s trade openness variable rose from 6% to 10% in 1987; from the above formula in Section 2, this would change the value of trade openness  for the UK in (say) 1998. This changes countries’ scores in previous years,  and could even change their ranking  in the overall globalization index in previous years.  

Nevertheless, a major objective in constructing our index is to make comparisons between different points in time. So, in spite of the problem just discussed, we use panel normalisation.