Construction of the Index

Counting Czechoslovakia in 1990, 125 nations were included in this study. However, as the result of incomplete data, we were only able to derive summary ratings for 119 countries in 1997 and 111 in 1990. Data were assembled for each of the 25 components of the index. Since we wanted the ratings to be easily comparable across countries and time periods, they were placed on a scale from zero to 10. Higher ratings are indicative of institutions and policies more consistent with economic freedom.

How were the ratings derived? The ratings for 13 of the 25 components in the index reflect various categorical characteristics; those for the remaining 12 are based on continuous data. Countries with categorical characteristics more consistent with economic freedom are given higher ratings. For example, countries with few government enterprises are given higher ratings than those with widespread use of such enterprises. Similarly, countries where price controls are absent (or apply in only a few markets) are given higher ratings than countries where these controls are extensively applied.

Depending on whether higher values are indicative of more or less economic freedom, alternative formulas are used to transform the 12 continuous variables to a zero-to-10 scale. When higher values are indicative of more economic freedom, the formula used to derive the zero-to-10 ratings is: (V i - Vmin) / (Vmax - Vmin) multiplied by 10. Vi is the country's actual value for the component, Vmax the maximum value for a country during the 1990 base year, and Vmin the minimum base-year value for the component. This formula is used to derive the ratings for all years. A country's rating will be close to 10 when its value for the component is near the base-year maximum. In contrast, the rating will be near zero when the observation for a country is near the base-year minimum. As the actual values exceed the base-year minimum by larger and larger amounts, ratings will rise from zero toward 10. Whenever the actual value for the component is equal to or greater than the base-year maximum, a rating of 10 is assigned. When the actual value is equal to or less than the base-year minimum, the rating is zero.

Higher actual values are often indicative of less economic freedom. Inflation and size of the transfer sector provide examples. Increases in these variables reflect reductions in economic freedom. When higher values for a component are indicative of less economic freedom, the formula used to derive the zero-to-10 ratings is: (Vmax - Vi) / (Vmax - Vmin) multiplied by 10. This formula will assign higher ratings to countries with actual values closer to the base-year minimum. In some cases, component values of zero represent an ideal, a benchmark that should be required for a rating of 10. For example, a zero mean tariff rate and a zero rate of inflation (perfect price stability) are benchmark outcomes representing maximum economic freedom. When zero represents an ideal benchmark value, this value was included as Vmin in the formula even if no country actually achieved this ideal during the base year. In some cases where extreme component values are present (for example, a 10,000 percent rate of inflation), Vmax is constrained at a level clearly warranting a rating of zero even if this was not the maximum observed value during the base year. If this had not been done, extreme observations would have created such a large range that the ratings would have been concentrated near 10. The precise formula used to derive the zero-to-10 ratings for each component is presented in the section Explanatory Notes and Data Sou rces below (page 40).

The procedures used to convert the continuous component values to the zero-to-10 ratings have two important characteristics. First, if all (or most) countries improve (or regress) with the passage of time, the ratings will reflect the change. Second, the distribution of the country ratings along the zero-to-10 scale reflects the distribution of the actual values among the countries.

Principal component analysis was used to determine the weight given to each component in the construction of the area index. This procedure partitions the variance of a set of variables and uses it to determine the linear combination--the weights--of these variables that maximizes the variation of the newly constructed principal component. In effect, the newly constructed principal component--an area rating, for example--is the variable that most fully captures the variation of the underlying components. It is an objective method of combining a set of variables into a single variable that best reflects the original data. The procedure is particularly appropriate when several sub-components measure different elements of a principal component. This is precisely the case with our index. Economic theory is a road map indicating components that are likely to capture various elements of a broader area (a principal component). In turn, principal component analysis indicates the permissibility of grouping components together and the weights most appropriate to combine a set of sub-components into a principal component. The component weights derived by this procedure are shown in parentheses , e.g. (.500), in Exhibit 1. The same procedure was also used to derive the weights for the area components in the construction of what we will refer to as the "weighted" summary index. The weights for each of the seven areas in Exhibit 1 are shown in brackets , e.g. [11.0%] .

Alternatively, equal weight could have been assigned to each of the seven areas. We also derived summary ratings based on this procedure. We will refer to the summary index that assigns equal weight to each of the seven areas as the "unweighted" summary index. The ratings and rankings derived by these two alternative procedures were quite similar.

info@fraserinstitute.ca You can contact us at the above email address for any comments or information requests. Please report any dead links or technical problems.

If you know someone who would be interested in this web page, please enter their email address below, and we will forward this URL to them: Email Address:

Last Modified: Wednesday, October 20, 1999.