Does a concern for lowering national Gini coefficients justify redistributionist policies?

Stringham, Edward ; Gonzalez, Rodolfo ; Krishnan, Aanand 等

We address whether a concern for lowering national Gini coefficients justifies implementing redistributionist policies. We explore the simple point that national Gini coefficients are largely determined by political demarcation and calculate the Gini coefficient for two hypothetical countries. First we show that even if U.S. states had perfect equality in each state, current differences in average income between states lead to a countrywide Gini coefficient greater than zero. Next we estimate the Gini coefficient for Europe as a whole by looking at income per quintile in 29 nations. Even though individual European countries typically have low Gini coefficients, the Gini coefficient for Europe as a whole is actually higher than that of the United States. If lowering national Gini coefficients is the ultimate goal, the simplest way to achieve this would be to split countries into numerous small nations with less intrastate inequality.

Jel Codes: D31, 132, H80

Keywords: Gini, redistribution, Tiebout

INTRODUCTION

Numerous authors hold income equality as a policy ideal (Clayton and Williams, (2000); Kawachi and Kennedy, (2002)). One of the most common measures of equality is the Gini coefficient, which enables comparisons of income distribution between countries. A Gini of zero signifies perfect equality; a Gini of one signifies that all income is in the pockets of one person. Although the Gini coefficient is a positive concept, most people discuss it for normative reasons and advocate judging policies based on whether they lower a country's Gini coefficient. One of the more common calls is for redistributionist policies in countries such as the U.S., which has a higher Gini coefficient than almost every country in Europe. For example, Buss, Peterson, and Nantz (1989:13-4) write, "Income is distributed less fairly in the United States," and they conclude, "Income must be reallocated to individuals at the lowest rungs of the distribution."

Other authors judge policies such as immigration laws based on whether they lead to lower Gini coefficients. Borjas (1999) favors restrictions on immigration because open borders have the potential to increase inequality in a nation. Note that these arguments are concerned with national Gini coefficients rather than the Gini coefficient for the entire globe. As Tullock (1997) argues, most first world residents do not favor policies that decrease global inequality because it would entail massive redistribution out of their country. Although some authors, including Firebaugh (1999), Goesling (2001), and Milanovic (2002), are concerned with the global Gini, many others are concerned only with the Gini within nations (Glaeser, Scheinkman and Shleifer, 2003).

This paper addresses the scholars concerned with lowering national Gini coefficients. Their policy prescriptions vary, but most tend to argue against classical liberal policies, which are hands off on the issue of income distribution. In this article we address whether a concern for lowering national Gini coefficients justifies implementing redistributionist policies. Whereas authors such as Tullock (1997) question the desirability of equality altogether, we take that goal as given. We argue, however, that even if one accepts low national Gini coefficients as the ideal, redistributionist policies remain unjustified; Gini coefficients can be lowered by other means. We explore the simple point that Gini coefficients are largely determined by political demarcation and they can be lowered by changes in the same.

In countries with many regions or states, such as the US and Brazil, national inequality is determined by interstate inequality in addition to intrastate inequality. Ceteris paribus, when political boundaries are defined across large areas with significant regional differences, the Gini will increase. If lowering national Gini coefficients is the ultimate goal, the simplest way to achieve this would be to split countries into numerous small nations with less intrastate inequality. Nozick proposed a world where people are free to leave and join political units with people similar to them. Such a system would be approximated by Tiebout competition. If people chose to be in jurisdictions whose members have similar incomes, Gini coefficients would instantly decrease. Such a policy of political decentralization does not affect global inequality, only national inequality. One might conclude that policies should not be judged on whether they change individual national Gini coefficients (because the global Gini coefficient may remain unchanged). If one takes this position, however, one must abandon the use of national Gini coefficients to judge policy. If, on the other hand, one wishes to hold onto national Gini coefficients as a normative goal, then one might as well advocate radical decentralization.

WHY GINI COEFFICIENTS ARE AFFECTED BY POLITICAL DEMARCATION

Let us consider how political demarcation of boundaries affects Gini coefficients. Notably, most European countries have lower Gini coefficients than the U.S. One cause is that countries in Europe tend to have more redistribution than the U.S. (Gwartney, Lawson, and Block 1996), but another possible cause is the relatively small size of nations. Europe is comprised of numerous countries that have fewer regional differences than the U.S. does. Depending on one's definition of Europe, it has 29 nations for 588 million residents; the U.S. has one country for 294 million residents. The average European country contains only 20 million residents, which is about three times the population of the average U.S. state. If a country's residents are more alike and few regional differences exist, ceteris paribus their Gini will be lower.

In contrast to smaller European countries, the U.S. has large regional differences within its borders, and income varies significantly between states. Even if each state had perfect equality within its borders, as long as regional differences exist, the national Gini coefficient cannot be zero. Let us consider a sample calculation. Imagine a nation with 50 equally sized states, each with a Gini coefficient of zero but with average income by state as varied as in the current U.S. Table 1 shows the per capita income by state; the unweighted average is $20,767 with per capita income ranging from $15,853 in Mississippi to $28,766 in Connecticut.

We can calculate the Gini coefficient for a country with the following formula:

G [approximately equal to] 1/[square root of 3] [[sigma].sub.y]/[bar.y] [rho] (y, [r.sub.y])

where s is the standard deviation of the income; y is the income point for each individual; [bar.y] is the average income for the country; [r.sub.y] is the rank corresponding to the income (1); and r is the correlation coefficient between the rank and income distributions of the population.

Knowing income and population for each hypothetical state, we can map the cumulative population and income curves. The Gini coefficient can then be calculated by comparing the cumulative earnings to the equality diagonal. Figure 1 shows the Lorenz curve for this hypothetical country; its Gini coefficient is 0.08. This is much lower than the Gini coefficient of the current U.S., which is 0.408 (Central Intelligence Agency, 1997), but this hypothetical coefficient is still not zero.

Supposing that lowering national Gini coefficients is the ultimate goal, one policy would be redistribution between the states. A much simpler way to decrease the Gini coefficients, however, would be to split this hypothetical country into fifty separate nations. Although this situation does not exactly mirror the U.S., we want to emphasize that Gini coefficients are the result of numerous factors, one of which is demarcation of political boundaries. If regional differences are important, a region's Gini coefficient may be lower than the Gini coefficient of the larger political unit to which it belongs.

[FIGURE 1 OMITTED]

We believe that this situation may exist in Europe. Although individual countries in Europe typically have high equality, (2) the Gini coefficient for the continent as a whole may be significantly less equalitarian.

Calculating the Gini coefficient for Europe is imprecise, but we believe that we can obtain a fairly good estimate. Even though we do not have income data for every person in Europe, one can estimate the Gini by looking at average income per quintile for each country. If one knew the average income per quintile in two countries, one would have ten points to map the cumulative population and cumulative income curves. As one knows more points, the curves become more accurate. Because we have average income per quintile data for 29 countries, we have 145 points for our calculations. We have no commitment to any particular geographical definition of Europe, but we excluded smaller counties with missing data (3) as well as Russia (and three former Soviet Republics) because Russia is large, mostly in Asia, and often not considered part of Europe. (4) Table 2 shows population, average income, and percentage share of national income per quintile for each of the countries on our list.

We obtained the number of people in each country's quintiles by dividing the country's population by five. We then arranged the 145 data points by income and calculated the cumulative population and income at each point. For example, if the lowest income category in Europe was in a country with 8 million people, the lowest point has 1.6 million people; if the next lowest income category in Europe was in a country with 22 million, the next lowest point would have 4.4 million, and so on. By looking at total income and total population of Europe, we can calculate the amount of cumulative income at each point if income distribution were equal. By comparing this to actual cumulative income, we can calculate the Gini. Figure 2 shows the Lorenz curve for Europe as a whole; its Gini coefficient is 0.472.

The result is that Europe has a higher Gini coefficient than the U.S. Individual countries in Europe have lower Gini coefficients and more equality than the U.S, but that does not mean that Europe as a whole has more equality than in the U.S. One might rightfully respond that the final Gini coefficient for Europe depends on how one defines Europe. If we exclude all of Eastern Europe from the sample, the Gini coefficient of Europe is lower than the U.S. As one limits the analysis to more homogenous groups of countries, such as those in Western Europe, the Gini coefficient decreases. We agree, but that proves our point: Gini coefficients are partially determined by how one defines political units. If Europe were one large nation, its Gini coefficient would be higher than that of the U.S.

[FIGURE 2 OMITTED]

CONCLUSION

Calculating the Gini coefficients for two hypothetical countries shows that the coefficients depend on demarcation of national boundaries. A desire to lower national Gini coefficients does not justify income redistribution. The simplest way to lower national Gini coefficients is to redefine political jurisdictions to eliminate regional and population differences. One might conclude that large nations should not be split into many small states, but if one does, the use of national Gini coefficients as a normative benchmark must be abandoned. If, on the other hand, one wants to continue using national Gini coefficients for normative reasons, one should embrace a policy of radical decentralization. Rather than justifying various policies at odds with classical liberalism, a concern for lowering Gini coefficients justifies the policy prescription in Nozick's third section of Anarchy, State, and Utopia (1974). A policy of political decentralization would lower Gini coefficients instantly. (5)

REFERENCES

Buss, J., Peterson, G.P., and Nantz, K. (1989), "A Comparison of Distributive Justice in OECD Countries" Review of Social Economy, Vol. 47, 1-13.

Central Intelligence Agency (1997), CIA World Factbook. Washington, DC: Central Intelligence Agency.

Central Intelligence Agency (2003), CIA World Factbook. Washington, DC: Central Intelhgence Agency.

Clayton, Matthew and Williams, Andrew (eds.) (2002), The Ideal of Equality. New York: Palgrave Macmillan.

Firebaugh, G. (1999), "Empirics of world income inequality" American Journal of Sociology, Vol. 104, 1597-1630.

Glaeser, E., Scheinkman, J., and Shleifer, A. (2003), "The Injustice of Inequality" Journal of Monetary Economics, Vol. 50, 199-222.

Goesling, Brian (2001), "Changing Income Inequalities within and between Nations: New Evidence" American Sociological Review, Vol. 66, 745-761.

Gwartney, James, Lawson, Robert and Block, Walter (1996), Economic Freedom of the World: 1975-1995. Vancouver: Fraser Institute.

Kawachi, Ichiro and Kennedy, Bruce (2002), The Health of Nations: Why Inequality is Harmful to Your Health. New York: New Press.

Milanovic, Branko (1997), "A simple way to calculate Gini coefficient and some implications ." Economics Letters, Vol. 56, 45-49.

Milanovic, Branko (2002), "True World Income Distribution, 1988 and 1993: First Calculations Based on Household Surveys Alone" Economic Journal, Vol. 112, 51-92.

Nozick, Robert (1974), Anarchy, State, and Utopia. New York: Basic Books.

Tullock, Gordon (1997), Economics of Income Redistribution, 2nd ed. Boston: Kluwer Academic Publishers.

NOTES

(1.) For example, the lowest income would have a rank of 1 and the highest income would have a rank of 'N,' where N is the population of the region.

(2.) The Gini coefficient in almost every European country is lower than that of the U.S. Countries such as Italy, Norway, Finland, Sweden, and Denmark have Gini coefficients under 0.3 (Central Intelligence Agency, 2003).

(3.) We did not have data for Albania, Andorra, Bosnia and Herzegovina, Gibraltar, Iceland, Liechtenstein, Malta, Monaco, San Marino, Serbia, and Vatican City.

(4.) University of Edinburgh Professors Jamieson and Grundy (2002:11) report, "The majority of Edinburgh residents rated all the countries [in Eastern and Western Europe] as part of Europe except Russia."

(5.) In fact, a Gini coefficient of 0 for every country could be achieved if the world were split up into 6 billion countries.

EDWARD STRINGHAM, RODOLFO GONZALEZ, and AANAND KRISHNAN

San Jose State University, San Jose

Table 1
Per Capita Income in the United States by State

Alabama 18,189
Alaska 22,660
Arizona 20,275
Arkansas 16,904
California 22,711
Colorado 24,049
Connecticut 28,766
Delaware 23,305
Florida 21,557
Georgia 21,154
Hawaii 21,525
Idaho 17,841
Illinois 23,104
Indiana 20,397
Iowa 19,674
Kansas 20,506
Kentucky 18,093
Louisiana 16,912
Maine 19,533
Maryland 25,614
Massachusetts 25,952
Michigan 22,168
Minnesota 23,198
Mississippi 15,853
Missouri 19,936
Montana 17,151
Nebraska 19,613
Nevada 21,989
New Hampshire 23,844
New Jersey 27,006
New Mexico 17,261
New York 23,389
North Carolina 20,307
North Dakota 17,769
Ohio 21,003
Oklahoma 17,646
Oregon 20,940
Pennsylvania 20,880
Rhode Island 21,688
South Carolina 18,795
South Dakota 17,562
Tennessee 19,393
Texas 19,617
Utah 18,185
Vermont 20,625
Virginia 23,975
Washington 22,973
West Virginia 16,477
Wisconsin 21,271
Wyoming 19,134

Source: US Bureau of the Census, 2000 Summary File 3.

Table 2
European Population, Income, and Percentage share of Income
per Quintile

 Population Average
 (Mid-Year) income 1st 2nd

Austria 8,069,876 28,170 6.90 13.20
Belgium 10,199,787 26,870 8.30 13.90
Bulgaria 8,066,057 1,180 10.10 13.90
Croatia 4,319,632 4,750 8.80 13.30
Czech Republic 10,300,707 5,280 10.30 14.50
Denmark 5,283,663 34,280 9.60 14.90
Estonia 1,458,065 3,400 7.00 11.00
Finland 5,134,406 25,620 10.00 14.20
France 58,623,428 25,850 7.20 12.60
Germany 82,011,073 28,580 8.20 13.20
Greece 10,502,372 12,280 7.50 12.40
Hungary 10,244,684 4,510 10.00 14.70
Ireland 3,667,233 19,780 6.70 11.60
Italy 57,479,469 22,000 8.70 14.00
Latvia 2,470,454 2,300 7.60 12.90
Lithuania 3,651,923 2,220 7.80 12.60
Luxembourg 421,014 46,460 9.40 13.80
Netherlands 15,604,464 27,250 7.30 12.70
Norway 4,405,672 36,380 9.70 14.30
Poland 38,655,842 3,810 7.80 12.80
Portugal 9,994,921 11,260 7.30 11.60
Romania 22,562,458 1,520 8.00 13.10
Slovakia 5,383,010 3,860 11.90 15.80
Slovenia 1,917,851 9,870 9.10 13.40
Spain 39,855,442 15,380 7.50 12.60
Sweden 8,897,619 27,950 9.60 14.50
Switzerland 7,193,761 43,840 6.90 12.70
Turkey 63,047,647 3,180 5.80 10.20
United Kingdom 58,808,266 21,510 6.10 11.60

 3rd 4th 5th

Austria 18.10 23.90 38.00
Belgium 18.00 22.60 37.30
Bulgaria 17.40 21.90 36.80
Croatia 17.40 22.60 38.00
Czech Republic 17.70 21.70 35.90
Denmark 18.30 22.70 34.50
Estonia 15.30 21.60 45.10
Finland 17.60 22.30 35.80
France 17.20 22.80 40.20
Germany 17.50 22.70 38.50
Greece 16.90 22.80 40.30
Hungary 18.30 22.70 34.40
Ireland 16.40 22.40 42.90
Italy 18.10 22.90 36.30
Latvia 17.10 22.10 40.30
Lithuania 16.80 22.40 40.30
Luxembourg 17.70 22.60 36.50
Netherlands 17.20 22.80 40.10
Norway 17.90 22.20 35.80
Poland 17.10 22.60 39.70
Portugal 15.90 21.80 43.40
Romania 17.20 22.30 39.50
Slovakia 18.80 22.20 31.40
Slovenia 17.30 22.50 37.70
Spain 17.00 22.60 40.30
Sweden 18.10 23.20 34.50
Switzerland 17.30 22.90 40.30
Turkey 14.80 21.60 47.70
United Kingdom 16.40 22.70 43.20

Sources: World Bank 2002World Development Indicators; U.S. Census
Bureau, 1997 Population Division, International Programs Center.