- Email: [email protected]
Recent trends in the global distribution of income
Journal of Policy Modeling 23 (2001) 497 – 501
Recent trends in the global distribution of income Donghyun Park Nanyang Business School, Nanyang Technological University, Room No. S3-B1A-10, Singapore 639798, Singapore
Abstract We examine trends in the global distribution of income over the period 1960 – 1992 as a simple means of checking the validity of the convergence hypothesis. Our data set consists of 133 countries and is derived from the Penn World Tables (PWT). Our two main findings are that while the global distribution of income has not become more equal during 1960 – 1992 as a whole, it has been on a secular decline during 1976 – 1992. Thus, although the overall data fail to lend support to the convergence hypothesis, more recent data do so. D 2001 Society for Policy Modeling. Published by Elsevier Science Inc.
1. Introduction We analyze the trends in the global income distribution over the period 1960– 1992 in this article. For this purpose, we use data from the Penn World Tables (PWT). We divide the global population into percentiles in terms of per capita income and estimate the share of global income accruing to each percentile. We then use those income shares to estimate global Gini coefficients. Convergence hypothesis, which is based on neoclassical growth theory, implies that the living standards of the countries of the world will converge over time. In order for convergence to occur, the poor countries must experience higher rates of growth than richer ones. Whether or not convergence has been taking place is a relevant issue for policymakers, particularly in developed countries, since we can expect economic inequality among the countries of the world to be a significant source of international friction and conflict. Absence of convergence would strengthen the case for more active trade, investment, and aid 0161-8938/01/$ – see front matter D 2001 Society for Policy Modeling. PII: S 0 1 6 1 - 8 9 3 8 ( 0 1 ) 0 0 0 5 9 - X
498
D. Park / Journal of Policy Modeling 23 (2001) 497–501
policies by rich countries to help poor countries. As De la Fuente (1997) notes in a comprehensive review of the literature, the existing empirical evidence does not lend support to the convergence hypothesis.
2. Data and methodology Our data set is the PWT (version 5.6). For a comprehensive explanation of this data set, please refer to Summers and Heston (1991). As is well known, PWT’s great advantage is that all the economic variables are expressed in a common set of prices and in a common currency. PWT thus allows for more meaningful international comparisons of income. Our sample consists of 133 countries. The sole criterion for our sample selection was the availability of PWT data. Our sample covers well over 97% of the global population. Our variables of interest are the population (POP in PWT) and per capita income (RGDPC in PWT) of each country. We examine the data annually from 1960 to 1992. In this article, global income inequality refers to the inequality among the nations of the world rather than the individuals of the world. The key assumptions we make in this connection are that all the individuals of a country earn the same level of income and that all the countries in our sample constitute a single world economy.
3. Empirical evidence We now report the principal trends in international income inequality for the period 1960– 1992 implied by our data set. 3.1. Percentile shares of global income We divide the global population into fifths and tenths. Our first step is to rank all countries by their per capita income. Thus, in 1960, Ethiopia is at one end and the US at the other. For the case of tenths or 10 percentiles, we first divide the global population by 10. For example, if there were 4 billion people in the world, each tenth would consist of 400 million. In constructing the poorest tenth, we would include all Ethiopians as well as the populations of the next poorest countries until 400 million people living in the poorest countries are included. We repeat the exercise for the other 10 percentiles. Countries at cut-off points will have a part of their population included in one-tenth and another part included in another tenth. In Table 1, q1 refers to the percentage share of global income accruing to the poorest tenth of the global population, as defined above, while q10 indicates the percentage share of global income accruing to the richest tenth
D. Park / Journal of Policy Modeling 23 (2001) 497–501
499
Table 1 Shares of global income by 10 percentiles (unit: %)
1960 1964 1968 1972 1976 1980 1984 1988 1992
q1
q2
q3
q4
q5
q6
q7
q8
q9
q10
2.21 1.87 1.73 1.91 1.57 1.88 1.88 1.68 1.63
2.53 2.04 1.92 2.20 2.08 2.33 2.57 2.85 2.96
2.57 2.19 2.01 2.25 2.13 2.52 2.76 2.95 3.08
3.18 3.06 2.49 2.43 2.32 2.56 2.90 3.14 3.49
3.43 3.33 2.67 2.53 2.57 2.72 2.90 3.14 3.49
5.36 4.75 4.42 3.93 3.95 4.06 4.05 3.86 4.07
8.91 9.02 8.69 8.79 9.34 8.90 8.36 7.56 7.31
11.8 12.4 13.5 13.9 15.1 14.9 14.2 13.5 12.5
21.2 22.5 23.5 24.1 23.8 23.5 23.5 23.8 23.0
38.8 38.8 39.1 38.0 37.1 36.6 36.9 37.5 38.4
of the global population. Table 1 also allows us to infer trends in the share of global income accruing to each fifth or 20 percentile of the global population — the sum of q1 and q2 gives the income share of the world’s poorest fifth, and so forth. 3.2. Gini coefficient Gini coefficient is the most well-known and widely used measure of inequality. It is based on the Lorenz curve, which plots the cumulative share of total income against the cumulative share of total population and is shown for the 20% case in Fig. 1. The smaller is the area between the 45° line and the actual income distribution, the smaller is the degree of inequality. The Gini coefficient is a convenient one-number summary of inequality in the sense of the Lorenz curve
Fig. 1. Lorenz curve for q = 20%, selected years.
500
D. Park / Journal of Policy Modeling 23 (2001) 497–501
and it satisfies scale independence, population size, and the Pigou – Dalton condition. The formula for the Gini coefficient is: G¼1þ
1 2 ðqn þ 2qn1 þ . . . þ nq1 Þ n n2 q¯
where q1 is the income share of i-th percentile and q1 q2 . . . qn. Making use of this, we obtain the values of the Gini coefficient for 20 percentiles and 10 percentiles of the global population. These are shown in Table 2. Note that a higher value of the Gini coefficient denotes greater inequality. Regardless of whether we use 20 or 10 percentiles, the Gini coefficient exhibits the same pattern — a secular rise during 1960 –1968, a period of volatility during 1968– 1976 and a secular decline for 1976 –1992. For the entire Table 2 Global Gini coefficients for 1960 – 1992 Year
Gini, 10%
Gini, 20%
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992
.5255 .5427 .5415 .5470 .5462 .5498 .5519 .5573 .5652 .5599 .5508 .5506 .5566 .5596 .5562 .5474 .5594 .5517 .5507 .5516 .5431 .5421 .5374 .5351 .5362 .5328 .5339 .5363 .5360 .5387 .5370 .5341 .5310
.5022 .5201 .5190 .5246 .5240 .5269 .5287 .5347 .5425 .5381 .5297 .5298 .5358 .5387 .5359 .5270 .5381 .5306 .5286 .5303 .5220 .5205 .5170 .5144 .5150 .5116 .5126 .5147 .5143 .5146 .5145 .5113 .5080
D. Park / Journal of Policy Modeling 23 (2001) 497–501
501
Fig. 2. Trends in the Gini coefficient, 1960 – 1992.
period of our study, however, the Gini coefficient does not fall and in fact, rises slightly. We can see these trends more clearly in Fig. 2.
4. Concluding remarks Our examination of the convergence hypothesis yields two main findings. First, our evidence indicates that the global distribution of income did not become more equal during 1960 –1992. Second, the same distribution does appear to have fallen continuously during the subperiod 1976 – 1992. Therefore, our evidence does supports the convergence hypothesis during the second half of our sample period although it fails to do so during the whole sample period. In addition, there appears to be a period of secular rise in global inequality during the subperiod 1960– 1968 as well as a period of volatile movements in the Gini coefficient during the subperiod 1968 –1976. It would thus be fairly accurate to say that the period under study consists of three distinct phases in terms of changes in the world distribution of income.
References De la Fuente, A. (1997). The empirics of growth and convergence: a selective review. Journal of Economic Dynamics and Control, 21, 23 – 73. Summers, R., & Heston, A. (1991). The Penn World Table (Mark 5): an expanded set of international comparisons, 1950 – 1988. Quarterly Journal of Economics, 105, 327 – 368.
Recent trends in the global distribution of income Donghyun Park Nanyang Business School, Nanyang Technological University, Room No. S3-B1A-10, Singapore 639798, Singapore
Abstract We examine trends in the global distribution of income over the period 1960 – 1992 as a simple means of checking the validity of the convergence hypothesis. Our data set consists of 133 countries and is derived from the Penn World Tables (PWT). Our two main findings are that while the global distribution of income has not become more equal during 1960 – 1992 as a whole, it has been on a secular decline during 1976 – 1992. Thus, although the overall data fail to lend support to the convergence hypothesis, more recent data do so. D 2001 Society for Policy Modeling. Published by Elsevier Science Inc.
1. Introduction We analyze the trends in the global income distribution over the period 1960– 1992 in this article. For this purpose, we use data from the Penn World Tables (PWT). We divide the global population into percentiles in terms of per capita income and estimate the share of global income accruing to each percentile. We then use those income shares to estimate global Gini coefficients. Convergence hypothesis, which is based on neoclassical growth theory, implies that the living standards of the countries of the world will converge over time. In order for convergence to occur, the poor countries must experience higher rates of growth than richer ones. Whether or not convergence has been taking place is a relevant issue for policymakers, particularly in developed countries, since we can expect economic inequality among the countries of the world to be a significant source of international friction and conflict. Absence of convergence would strengthen the case for more active trade, investment, and aid 0161-8938/01/$ – see front matter D 2001 Society for Policy Modeling. PII: S 0 1 6 1 - 8 9 3 8 ( 0 1 ) 0 0 0 5 9 - X
498
D. Park / Journal of Policy Modeling 23 (2001) 497–501
policies by rich countries to help poor countries. As De la Fuente (1997) notes in a comprehensive review of the literature, the existing empirical evidence does not lend support to the convergence hypothesis.
2. Data and methodology Our data set is the PWT (version 5.6). For a comprehensive explanation of this data set, please refer to Summers and Heston (1991). As is well known, PWT’s great advantage is that all the economic variables are expressed in a common set of prices and in a common currency. PWT thus allows for more meaningful international comparisons of income. Our sample consists of 133 countries. The sole criterion for our sample selection was the availability of PWT data. Our sample covers well over 97% of the global population. Our variables of interest are the population (POP in PWT) and per capita income (RGDPC in PWT) of each country. We examine the data annually from 1960 to 1992. In this article, global income inequality refers to the inequality among the nations of the world rather than the individuals of the world. The key assumptions we make in this connection are that all the individuals of a country earn the same level of income and that all the countries in our sample constitute a single world economy.
3. Empirical evidence We now report the principal trends in international income inequality for the period 1960– 1992 implied by our data set. 3.1. Percentile shares of global income We divide the global population into fifths and tenths. Our first step is to rank all countries by their per capita income. Thus, in 1960, Ethiopia is at one end and the US at the other. For the case of tenths or 10 percentiles, we first divide the global population by 10. For example, if there were 4 billion people in the world, each tenth would consist of 400 million. In constructing the poorest tenth, we would include all Ethiopians as well as the populations of the next poorest countries until 400 million people living in the poorest countries are included. We repeat the exercise for the other 10 percentiles. Countries at cut-off points will have a part of their population included in one-tenth and another part included in another tenth. In Table 1, q1 refers to the percentage share of global income accruing to the poorest tenth of the global population, as defined above, while q10 indicates the percentage share of global income accruing to the richest tenth
D. Park / Journal of Policy Modeling 23 (2001) 497–501
499
Table 1 Shares of global income by 10 percentiles (unit: %)
1960 1964 1968 1972 1976 1980 1984 1988 1992
q1
q2
q3
q4
q5
q6
q7
q8
q9
q10
2.21 1.87 1.73 1.91 1.57 1.88 1.88 1.68 1.63
2.53 2.04 1.92 2.20 2.08 2.33 2.57 2.85 2.96
2.57 2.19 2.01 2.25 2.13 2.52 2.76 2.95 3.08
3.18 3.06 2.49 2.43 2.32 2.56 2.90 3.14 3.49
3.43 3.33 2.67 2.53 2.57 2.72 2.90 3.14 3.49
5.36 4.75 4.42 3.93 3.95 4.06 4.05 3.86 4.07
8.91 9.02 8.69 8.79 9.34 8.90 8.36 7.56 7.31
11.8 12.4 13.5 13.9 15.1 14.9 14.2 13.5 12.5
21.2 22.5 23.5 24.1 23.8 23.5 23.5 23.8 23.0
38.8 38.8 39.1 38.0 37.1 36.6 36.9 37.5 38.4
of the global population. Table 1 also allows us to infer trends in the share of global income accruing to each fifth or 20 percentile of the global population — the sum of q1 and q2 gives the income share of the world’s poorest fifth, and so forth. 3.2. Gini coefficient Gini coefficient is the most well-known and widely used measure of inequality. It is based on the Lorenz curve, which plots the cumulative share of total income against the cumulative share of total population and is shown for the 20% case in Fig. 1. The smaller is the area between the 45° line and the actual income distribution, the smaller is the degree of inequality. The Gini coefficient is a convenient one-number summary of inequality in the sense of the Lorenz curve
Fig. 1. Lorenz curve for q = 20%, selected years.
500
D. Park / Journal of Policy Modeling 23 (2001) 497–501
and it satisfies scale independence, population size, and the Pigou – Dalton condition. The formula for the Gini coefficient is: G¼1þ
1 2 ðqn þ 2qn1 þ . . . þ nq1 Þ n n2 q¯
where q1 is the income share of i-th percentile and q1 q2 . . . qn. Making use of this, we obtain the values of the Gini coefficient for 20 percentiles and 10 percentiles of the global population. These are shown in Table 2. Note that a higher value of the Gini coefficient denotes greater inequality. Regardless of whether we use 20 or 10 percentiles, the Gini coefficient exhibits the same pattern — a secular rise during 1960 –1968, a period of volatility during 1968– 1976 and a secular decline for 1976 –1992. For the entire Table 2 Global Gini coefficients for 1960 – 1992 Year
Gini, 10%
Gini, 20%
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992
.5255 .5427 .5415 .5470 .5462 .5498 .5519 .5573 .5652 .5599 .5508 .5506 .5566 .5596 .5562 .5474 .5594 .5517 .5507 .5516 .5431 .5421 .5374 .5351 .5362 .5328 .5339 .5363 .5360 .5387 .5370 .5341 .5310
.5022 .5201 .5190 .5246 .5240 .5269 .5287 .5347 .5425 .5381 .5297 .5298 .5358 .5387 .5359 .5270 .5381 .5306 .5286 .5303 .5220 .5205 .5170 .5144 .5150 .5116 .5126 .5147 .5143 .5146 .5145 .5113 .5080
D. Park / Journal of Policy Modeling 23 (2001) 497–501
501
Fig. 2. Trends in the Gini coefficient, 1960 – 1992.
period of our study, however, the Gini coefficient does not fall and in fact, rises slightly. We can see these trends more clearly in Fig. 2.
4. Concluding remarks Our examination of the convergence hypothesis yields two main findings. First, our evidence indicates that the global distribution of income did not become more equal during 1960 –1992. Second, the same distribution does appear to have fallen continuously during the subperiod 1976 – 1992. Therefore, our evidence does supports the convergence hypothesis during the second half of our sample period although it fails to do so during the whole sample period. In addition, there appears to be a period of secular rise in global inequality during the subperiod 1960– 1968 as well as a period of volatile movements in the Gini coefficient during the subperiod 1968 –1976. It would thus be fairly accurate to say that the period under study consists of three distinct phases in terms of changes in the world distribution of income.
References De la Fuente, A. (1997). The empirics of growth and convergence: a selective review. Journal of Economic Dynamics and Control, 21, 23 – 73. Summers, R., & Heston, A. (1991). The Penn World Table (Mark 5): an expanded set of international comparisons, 1950 – 1988. Quarterly Journal of Economics, 105, 327 – 368.