Economic Freedom Benefits Well-Being: an Empirical Study

 

Table of Contents

 

How is Economic Freedom Measured?. 3

Model 1. 4

How is Human Development Measured?. 6

Model 2. 6

Model 3. 8

Model 4. 11

Causality. 13

Limitations. 14

Conclusion. 14

Appendix. 15

Countries Included in the Study. 15

Descriptive Statistics. 16

Model 3 - Park Test for Heteroscedasticity - Variable: Life Expectancy. 17

Model 3 - Park Test for Heteroscedasticity - Variable: Education. 17

Model 4 - Park Test for Heteroscedasticity. 18

Bivariate Correlation Matrix. 19

Test for Multicollinearity (Variation Inflation Factor) Results. 20

References. 21

 

 

Over the years there has been much literature to suggest that a reduction in market restrictions is related to prosperity among people.  This paper extends on the existing literature which has examined the relationships between economic freedom and growth as well as growth and well-being.  This study employs four empirical models which examine relationships between: economic freedom and economic growth in various countries; growth and indicators of human development (longevity and education); and economic freedom and human development, first excluding standard of living within the measure of human development and later including it.  These models have been developed and tested using multiple linear regression.  The findings suggest a relationship in which increasing economic freedom implies greater well-being. 

 

How is Economic Freedom Measured?

 

This study utilizes the research of James Gwartney and Robert Lawson of the Fraser Institute.  The Economic Freedom of the World Index (EFW) which they developed, under the guidance of Milton Friedman, is believed to be the most objective and accurate measure of economic freedom that exists today (Gwartney and Lawson and Emerick 2003).  The EFW is composed of a simple average of the following factors to rate the degree of economic freedom in a country:

 

  .  Size of Government: Expenditures, Taxes, and Enterprises

  .  Legal Structure and Security of Property Rights

  .  Access to Sound Money

  .  Freedom to Trade with Foreigners

  .  Regulation of Credit, Labor, and Business

 

The size of government expenditures, taxes and enterprises is important to economic freedom because it measures the amount of government involvement in consumption.  Government spending means government decision making.  By taxing some people in order to transfer wealth to others, the government takes some individual freedom away from its citizens.  Government enterprises also cause economic freedom to decline because government institutions often operate in protected markets and do not play by the same rules as private institutions.   An independent and impartial judiciary system which provides security and property rights to its citizens is essential to freedom.  Access to sound money is also critical to economic freedom to ensure that inflation does not erode the value of money.  Freedom to trade with foreigners is essential to economic freedom and contributes to higher living standards.  Regulation in any form limits economic freedom.[1] 

 

In my first model[2], the factors which compose the EFW are compared to economic growth using GDP.  GDP is measured on a per capita basis in real terms.  Using purchasing power parity to convert currencies rather than exchange rates themselves provides a more accurate picture of the true measurement that we aim to estimate; quality of life/standard of living because it equalizes the prices of traded goods. 

 

The EFW is based on a zero to ten scale, with ten being most free and zero the least.  All of the research in this paper is based on 2001 data for the 123 countries[3] which have the EFW data. 

 

Model 1

 

Null Hypothesis:                       This is not a good model.

Alternative Hypothesis: Reject the null hypothesis (the model has statistical significance).  

 

N = 123

 

Dependent Variable:                 GDP

Independent Variables:

Size of Government

Legal System and Property Rights

Access to Sound Money

Freedom to Trade with Foreigners

Regulation

 

GDP = function(size of government, legal system and property rights, access to sound money, freedom to trade with foreigners, regulation)

 

 

A p-value of 0.000 indicates that the model is statistically significant and fits deviations in the dependent variable with 99% certainty. 

 

 

The independent variables (the factors of economic freedom) can explain 61% of the variance in GDP.  Additionally, as the sample size (number of countries) increases, the difference between R Square and Adjusted R Square will decline.

 

 

The coefficients table indicates an unstandardized beta of -8150 if the factors of economic freedom were at zero.  However, logic dictates that GDP per capita can never be negative.  A more appropriate measure is the standardized coefficients which forces the constant to zero.   Legal system and property rights have the highest correlation with GDP, followed by the size of government, followed by access to sound money, followed by freedom to trade with foreigners, and finally regulation has the lowest correlation to GDP.  Freedom to trade with foreigners and regulation have rather high dispersion (greater than 0.1) and in order to improve reliability a better model would remove them. 

 

 

The graphs above depict the residuals in standardized form.  In this model they appear to be approximately normally distributed but with a left skew.  In this model, residuals seem to be somewhat leptokurtic, based on visual inspection, and there does appear to be an outlier at the value of 4.5.

 

Results conclude that the factors which make up economic freedom are highly correlated to standard of living.  Perhaps instead of taking a simple average of the factors, a better model for determining the Economic Freedom of the World Index would be to weigh the various factors according to their impact on GDP, such that:

 

GDP per capita = -.142(Size of Government) + .645(Legal System and Property Rights) + .152(Access to Sound Money) + .102(Freedom to Trade with Foreigners) + -.084(Regulation)

 

You’ll note that the size of government as well as regulation coefficients are negative, implying that as government gets larger or enforces more regulation, GDP will decline. 

 

 How is Human Development Measured?

 

This study utilizes the research of the United Nations Development Programme (UNDP) which currently compiles data for 175 countries.  The following research only examines data on the 123 nations that are served by the EFW.  The UNDP takes a simple average of the following three factors to generate the Human Development Index (HDI) for each country:

 

  .  Longevity

  .  Education

  .  Standard of Living

 

Longevity is measured by life expectancy at birth.  This measure can be viewed as a generalized assessment of the healthcare system.  It does however implicitly assume that all people would have the same genetic and cultural life expectancy.  The education index is determined by taking a 2/3 weight on adult literacy rate, and a 1/3 weight on gross enrollment in primary, secondary, and tertiary school combined.[4]  Standard of living is measured through GDP per capita using purchasing power parity as before. 

 

In order to compare the Human Development Index to GDP, I had to separate the factors that makeup the HDI.  This involved removing the standard of living factor that is used in determining the HDI in order to prevent simultaneity bias (a type of mis-specification which in effect creates a circular reference by preventing GDP from truly existing as an independent variable).  In my second model, longevity and education are compared to GDP. 

 

Model 2

 

Null Hypothesis:                       This is not a good model.

Alternative Hypothesis: Reject the null hypothesis (the model has statistical significance).  

 

N = 123

 

Dependent Variable:                 GDP

Independent Variables:

Education Index

Life Expectancy Index

 

GDP = function(education, life expectancy)

 

 

A p-value of 0.000 indicates that the model is statistically significant and fits deviations in the dependent variable with 99% certainty. 

 

 

The independent variables (education and life expectancy) can explain 45% of the variance in GDP. 

 

The coefficients table indicates an unstandardized beta of -18207 if life expectancy and education were at zero.  Again, although the rate of change in GDP can be negative in years of declining productivity, GDP per capita can never be negative.  A more appropriate measure is the standardized coefficients which forces the constant term to zero.   Life expectancy has the highest correlation with GDP, followed by education.  Both variables are very significant with almost equal importance.    

 

 

The graphs above depict the residuals which appear to have a left skew and exhibit kurtosis.  In this model, based on visual inspection, residuals seem to be somewhat leptokurtic, and there does appear to be an outlier at the value of 5.

 

Results conclude that these indicators of human well-being are highly correlated to standard of living.  A model for determining GDP as a function of the UNDP’s life expectancy and education indices would weigh the factors according to their impact on GDP, such that:

 

GDP per capita = .378 (Life Expectancy Index) + .347(Education Index)

 

You’ll note that life expectancy and education are almost equally important when measuring standard of living.  Intuitively we would expect there to be some collinearity among factors in all of the models in this paper and I will address this issue in the limitations section. 

 

My third model begins to address the variables that are at the heart of my argument.  This model compares factors of the Human Development Index, again removing the standard of living factor for reasons aforementioned.  In this model, longevity and education are compared to the Economic Freedom of the World Index. 

 

Model 3

 

Null Hypothesis:                       This is not a good model.

Alternative Hypothesis: Reject the null hypothesis (the model has statistical significance).  

 

N = 123

 

Dependent Variable:                 Economic Freedom Index

Independent Variables:

Education Index

Life Expectancy Index

 

Economic Freedom Index = function(education, life expectancy)

 

 

A p-value of 0.000 indicates that the model is statistically significant and fits deviations in the dependent variable with 99% certainty. 

 

 

The independent variables (life expectancy and education) can explain 27% of the variance in economic freedom.  Additionally, as the sample size (number of countries) increases, the difference between R Square and Adjusted R Square will decline.

 

 

Education has the highest correlation with GDP, followed by life expectancy.  Both variables are very significant even at the 95% confidence interval.      

 

 

The graphs above depict the standardized residuals which appear to be near-perfectly normally distributed.  There do not appear to be any outliers.  The graph on the following page depicts the unstandardized squared residuals for this model.

 

 

Among cross sectional data the classical assumption of error terms having a constant variance is often violated; a caveat known as heteroscedasticity.  If heteroscedasticity is found, there is an implication that the interpretation of standard errors might be misleading.  In the case of this model, the graph above seems to depict a pattern among the errors.  Upon employing the Park Test, we cannot reject the idea that there may be heteroscedasticity for the life expectancy and/or the education coefficient.[5]  Thus, the parameter estimates in this model are still consistent but may be inefficient.  An effort to increase efficiency would weigh the least squares according to their differing variances, and/or to increase the number of independent variables in the model. 

 

These results lead to the conclusion that the factors which make up human development are highly correlated to economic freedom even if they only explain 27% of the variance in economic freedom.  A model for determining the EFW as a function of the UNDP’s life expectancy and education indices would weigh the factors according to their impact on economic freedom, such that:

 

Economic Freedom Index = .324 (Life Expectancy Index) + .244(Education Index)

 

You’ll note that life expectancy has more of an impact on economic freedom than education when measuring standard of living.  In essence this model has measured the Economic Freedom of the World Index compared to the Human Development Index, removing GDP as a factor.  In this model, EFW is the independent variable. 

 

The final model compares the Human Development Index directly to the Economic Freedom of the World Index.  In this model, the HDI is the independent variable. 

 

Model 4

 

Null Hypothesis:                       This is not a good model.

Alternative Hypothesis: Reject the null hypothesis (the model has statistical significance).  

 

N = 123

 

Dependent Variable:                 Human Development Index

Independent Variable:               Economic Freedom Index

 

Human Development Index = function(Economic Freedom Index)

 

 

 

A p-value of 0.000 indicates that the model is statistically significant and fits deviations in the dependent variable with 99% certainty. 

 

 

Economic freedom can explain 33% of the variance in human development. 

 

 

The coefficients data reinforces conclusions that economic freedom is very significant in determining human development.

 

 

The graphs above depict the residuals which appear to be normally distributed with a right skew.  There do not appear to be any outliers.  The graph below depicts the unstandardized squared residuals for this model.

 

 

In the case of this model, the Park Test was again employed and there seems to be no heteroscedasticity from the economic freedom index.

 

These results lead to the conclusion that human development is highly correlated to economic freedom.  There is a 58% correlation between the Economic Freedom of the World Index and the United Nations Development Programme’s Human Development Index. 

 

Causality

 

Studies have shown (Farr and Lord and Wolfenbarger 1998) that while a statistically significant relationship does not exist directly between economic and political freedom, a Granger-causal relationship exists implying that economic freedom impacts GDP per capita, which impacts political freedom.  It was determined that a multivariate Granger-causal relationship between GDP per capita and economic freedom exists while a univariate line of Granger-causation exists between GDP per capita and political freedom.  This multivariate relationship between economic freedom and growth is important because it implies that not only does economic freedom drive growth (GDP), but growth additionally enhances economic freedom. 

 

Granger-causality tests suggest that (Heckelman 2000) economic freedom precedes growth however when examining the components of economic freedom individually, the reverse relationship may actually exist between growth and government expenditures and enterprises, and no causal relationship exists between growth and freedom to trade with foreigners and growth and taxation.[6]  This might explain earlier studies’ results which concluded a multivariate relationship between economic freedom and growth.  Also, this suggests that taxation should not be lumped together with government expenditures and enterprises under the label ‘size of government’. 

 

Heckelman’s inability to find a causal relationship between growth and openness (freedom to trade with foreigners) supports my low correlation finding in Model 1 between freedom to trade and GDP.  The data in this subfactor of the Economic Freedom of the World Index is largely subjective and mostly relies on survey data from the “Global Competitiveness Report”[7].  In particular, many of the countries in the study have hidden administrative restraints and capital controls which are difficult to quantify.  The subjective nature of the data makes it nearly impossible to enumerate the extent of trade restrictions necessary for an accurate measure of the freedom to trade.  This suggests that forms of protectionism prohibit trade from currently having a significant effect on economic growth.  It is impossible to determine the true significance of opening borders to trade and the effect of doing so on growth using only the data in this study. 

 

In light of the evidence that has been suggested over the years on the benefits of reduction of market restrictions, many countries have moved toward increasing their degree of economic freedom particularly in the areas of international trade and monetary stability.  However, there has also been a push for larger government expenditures, which directly contradicts economic freedom by increasing the size of government.^[8]  Studies have shown that (Gwartney and Holcombe and Lawson 1998) when government expenditures exceed the level of 15% of GDP, there is a negative impact on economic growth (GDP).  These conflicting ideas are suggestive of Robert Nozick’s “minimal state” concept[9] which requires some minimal level of government expenditures to serve the basic needs of its citizens, but limits interference beyond that threshold, thus avoiding infringement on liberty. 

 

Limitations

 

Intuitively we would anticipate some collinearity to exist between variables of this nature.  The degree of collinearity, measured by the Pearson Correlation, between variables is not high enough (0.8) to bias the estimates for any of the four models.  I also measured the variation inflation factor for each of the models and determined that no multicollinearity exists.[10] 

 

Limitations within this study lie in the collection of data.  The data used in the models is believed by many experts in the industry to be the most objective measures of standard of living, economic freedom, and human development in existence.  However, whenever researchers have to rely on information gathered through potentially biased parties the possibility of flawed data exists.  A certain amount of subjectivity has been used in measuring several of the indices’ subfactors, and in particular there is evidence suggesting that this subjectivity has affected the accuracy of the level of significance of the freedom to trade with foreigners.   

 

Additionally, the use of simple average of the factors in order to generate the EFW and the HDI indices may not be as accurate as models of a form similar to that which I generated in this paper.  Further studies should also be conducted on the subfactors that makeup the factors of the EFW and the HDI. 

 

Conclusion

 

The focus of this study has been on economic freedom and socioeconomic conditions across nations.  While a relationship does exist, these socioeconomic conditions cannot be explained entirely by economic freedom.  We must not ignore the role that other factors play in socioeconomic conditions. 

 

The evidence presented in this paper concludes that policy which imposes government restrictions often aimed at providing assistance to its citizens actually hinders well-being.  Efforts to redistribute income contribute to a reduction in economic freedom, which in turn reduces the society’s overall well-being as measured by standard of living, education, and life expectancy.   The calculations presented here support the idea that improving well-being among people[11] can be accomplished through a reduction of market restrictions.    

 

 

Appendix

 

Countries Included in the Study

 

Descriptive Statistics

 

Model 1:

 

Model 2:

 

Model 3:

 

Model 4:

Model 3 - Park Test for Heteroscedasticity - Variable: Life Expectancy

 

 

 

 

 

Model 3 - Park Test for Heteroscedasticity - Variable: Education

 

 

 

Model 4 - Park Test for Heteroscedasticity

 

 

 

Bivariate Correlation Matrix

 

Pearson Correlation
	Size of Government	Legal System and Property Rights	Sound Money	Freedom to Trade with Foreigners	Regulation	Economic Freedom Index	GDP per Capita (USD in PPP terms)	Life Expectancy Index	Education Index
Legal System and Property Rights	-0.213*
Sound Money	0.144	.580**
Freedom to Trade with Foreigners	0.038	.556**	.488**
Regulation	0.137	.624**	.491**	.423**
Economic Freedom Index	.284**	.780**	.842**	.712**	.746**
GDP per Capita (USD in PPP terms)	-.265**	.767**	.513**	.493**	.416**	.606**
Life Expectancy Index	-0.033	.513**	.451**	.474**	.233**	.503**	.633**
Education Index	-0.174	.547**	.384**	.479**	.330**	.482**	.624**	.734**
Human Development Index	-0.137	.642**	.489**	.540**	.344**	.580**	.768**	.926**	.905**
	N = 123
	* = Correlation is significant at the 0.05 level (2-tailed).
	** = Correlation is significant at the 0.01 level (2-tailed).

Test for Multicollinearity (Variation Inflation Factor) Results

 

Model 1:          2.667

Model 2:          1.84

Model 3:          1.4

Model 4:          1.51

 

References

 

Barro, R. J., and J.-W. Lee. (2001). “International Data on Educational Attainment: Updates and Implications.” Oxford Economic Papers, 53 (July): 541-63.

 

Farr, W.K., R.A. Lord, and J.L. Wolfenbarger (1998). “Economic Freedom, Political Freedom, and Economic Well-being: A Causality Analysis”, Cato Journal 18: 247-262.

 

Grubel, Herbert G.  (1998). “Economic Freedom and Human Welfare: Some Empirical Findings”. Cato Journal: 18 287-304.

 

Gwartney, J., R. Lawson, with Neil Emerick. (2003). Economic Freedom of the World 2003 Annual Report. Vancouver, B.C.: Fraser Institute.

 

Gwartney, J.D., R.A. Lawson and W. Block. (1996) Economic Freedom of the World: 1975–1995. Vancouver, B.C., Canada: Fraser Institute.

 

Gwartney, James, Randall Holcombe, and Robert Lawson. (1998). “The Scope of Government and the Wealth of Nations”. Cato Journal: 18 163-190.

 

Heckelman, Jac C. (2000).  “Economic Freedom and Economic Growth: a Short-Run Causal Investigation”.  Journal of Applied Economics III: 71-91. 

 

Nozick, Robert.  (1974).  Anarchy, State, and Utopia.  Basic Books Inc.

 

UNDP (2003). “Human Development Report 2003”. New York: Oxford University Press.

 

World Economic Forum (2003).  “Global Competitiveness Report 2002-2003”.  New York: Oxford University Press.