In this paper, we review major literature of the past with respect to global income inequality trends and relationship of income inequality with other variables of an economy. Then, we study income dispro- portion by looking at the cross-section of 24 countries from the time period of 2001 to 2014, with the help of the Kuznets Ratio. We seek to find a relationship between income inequality and income (not nec- essarily a direct causation). Then we observe the trend of income distribution for our selected time period. Our conclusion reveals a weak association between income disparity and growth of per capita GDP. Further, we find that the share of total income of the top 10% rich individuals in the world has fallen, hinting at a fall in income inequality.
A substantial number of papers have studied income inequality in the past. These studies have come in various forms. Some examine inter-country and intra-country income disparity trends and some study the linkage of this disparity with variables like growth, development, ‘happiness’1 and even on inequality itself. There are numerous reasons to study income disparity. One reason is that we may be interested in it intrinsically, since large magnitudes of such a disparity may be viewed as unfair. Second, we may look at it as a forecasted outcome of a theory such as the convergence or divergence theory.
Based on empirical evidence, it might help us in judging accuracy of certain theories. Third, it could be used to explain a phenomenon of interest. For example, high inequality may be used to explain high inequality of opportuni- ties available to individuals in an economy or as a key factor affecting crime and political instability. The interesting observation from the pre-existing literature is that there are contrasting arguments for the trend of income distribution and there is no consensus for any statistically significant asso- ciation between such inequality and any of the above-mentioned variables. Since high-income growth is one of the most crucial aims of every economy in the world, in this paper, we try to look for a correlation between global income disproportion and income growth rather than a causal relationship. In particular, this paper asks whether there is any particular relationship between level of income and inequality when we look at the time period from 2001-2014.
Moreover, we move one step further check about the possibil- ity of relationship between growth in inequality and growth in income. We also touch upon the global income distribution and comment regarding the changes taken place therein. We have considered income inequality as a type of inequality for this paper. Hence, the reader should note that whenever the word ‘inequality’ is mentioned in the paper from section 3 onwards, it stands for income inequality. Section 2 contains literature review followed by research objectives in sec- tion 3, data and methodology in section 4, conclusion in section 5, limitations in section 6 and scope for further research in section 7.
Contrasting conclusions for the direction of income disparity lead to difficulty in further analyzing what effect it has on other variables. Based on the data from Zanden (2014), majority of the world lived in absolute poverty in 1800 as of today’s standards. 175 years later, in 1975, a rapid increase in inequality was seen. The world was divided into 2 parts; a destitute, developing part and a developed part which was more than tenfold richer. As of 2015, inequality had declined to a large extent, and majority of the population was above the international poverty line.
Roser (2018) states that from a period of 1988 and 2011, global inequality had increased in the first 2 terciles and has been decreased since then. Hellebrandt and Mauro (2015) endorse that global income disparity has declined, from a Gini coefficient of 68.7 in 2003 to 64.9 in 2013. Cornia (2003) examines that global income inequality increased in the early 1980s and 1990s. Milanovic (2009) on the other hand, calculates Gini indices over time from 1820 to 2002, and observes a consistent increase in it, with a large increase 1980 onwards.
UNICEF Social and Economic Policy Working Paper (2011) shows some reversal of this trend, although it suggests that the inequality situation could have been worsening while the study was conducted due to the then ongoing global economic crisis of that time. The World Inequality Report (2018) asserts that since 1980, income inequality has increased in nearly all countries in recent decades. It has increased enormously in Asia and North America, has amplified to a relatively lesser degree in Europe and has stabilized at exceptionally high levels in the Middle East, Brazil and sub-Saharan Africa. Oxfam (2018) – annual report on global inequality finds that 82% of the wealth that was generated in 2017 went to the richest 1% population of the world.
The report points out that in many countries wage disparity has risen and the share of labour compensation in the GDP has fallen because profits have grown more quickly than wages. Thomas Piketty’s collection of research in his magnum opus Capital in the Twenty-First Century (2013) provided a major breakthrough in the field of research on inequality and spurred a global debate on inequality. His carefully maintained dataset led him establish that the simple reason for rising inequality in the past is due the fact that return on capital has always remained greater than the growth rate of the economy. His findings regarding the effects of taxes on income distribution reflected that inequality is not just the result of economic forces, but also the result of policies and politics.
Yitzhaki (1997) argued that different methods of calculating income and income variation would lead to different results and perhaps this might be the factor that leads to compelling yet contrasting conclusions. Research papers of the past have especially tried to find a direct relation- ship between income growth and income inequality, that is, whether such disparity is deleterious or beneficial for income growth and if growth has any impact on income distribution. Although it turns out that there is no con- sensus in literature for any statistically significant association among these 2 variables, many papers also make some strong related observations.
Kaldor (1957) argued that a high profit to wage ratio, implying greater income dis- parity, would lead to greater growth. Earliest measure of income inequality in growth regressions, in 1990s, found a significant negative coefficient, imply- ing negative effects of income inequality on growth outweigh positive effects. The issue is these papers relied on ordinary least squares or instrumental variables regression. Later studies found that omitted variables in these pa- pers may have biased the OLS coefficients. The literature in and after this time was mainly divided into 2 models.
First model stated the effects of in- come disparity and imperfect capital markets on investment activities. There is a positive wealth threshold, below which individuals choose not to invest in human capital. A higher number of people below the threshold leads to lower aggregate wealth in equilibrium and lower growth. Poor people have a low opportunity cost of having children (low wage rate) and therefore have greater number of children. Credit market restraints do not allow them to have ‘high quality’ children, increasing the supply of unskilled workers, re- ducing their wage rate and further increasing inequality. This may suggest that poverty may itself lead to increased poverty and decreased growth.
The second model talked about the political economy, where inequality leads to taxation and spending decisions that may not match with the socially optimal decisions. Barro (2000) considered the possibility that the effect of inequality on growth might differ between rich and poor countries. While no significant relationship is found for the whole sample, he reports a significantly negative relationship for the poorer countries and a positive relationship among richer countries when the sample is split.
The paper also states that the ‘inverted U hypothesis’, as proposed by Simon Kuznets, seems to be an empirical reg- ularity for nations. Benabou (2000) asserted that since greater voting power is exercised by wealthier people, inequality might lead to lesser rather than higher taxation. Milanovic (2005) finds that at low average income levels, the income share of the poor is smaller in economies that are more open to trade. Ravallion (2012) claims that a higher level of poverty rather than inequality is strongly associated with lower economic growth.
Given this literature backdrop, we try to answer 3 particular questions in this paper. First, we seek to understand the dynamics between inequality and income by looking at the cross-section of 24 countries from the time period 2001-2014. In particular, we try to answer whether there is any particular relationship between inequality and income. After looking at the issue of inequality and income, we move forward and see if the rate of change in inequality depends upon the changes in the rate of growth of per capita income. Lastly, we touch upon the global income distribution and see the changes that took place in the shares of income at every decile of the distribution in the given time period. The third observation becomes crucial in the care- fully interpreting the recent debates in the popular press regarding global inequality.
Data, Methodology and Findings
The entire data for the current analysis has been extracted from the World Inequality Database which is maintained by Thomas Piketty, Lucas Chancel, Gabriel Zucman and their colleagues. To answer the first two questions, we use Kuznets Ratios as a measure of inequality. As mentioned in the literature review, Kuznets Ratio looks at inequality through the ratio of share of income of top x% rich and bottom y% poor. For this paper, we have defined Kuznets ratio as the ratio of share of income of top 10% rich and bottom 50% poor.
An increase in Kuznets ratio means increase in income inequality, while decrease in Kuznets ratio means decrease in income inequality. Before moving forward, it is important to understand why this defini- tion of Kuznets ratio is used and why the stated number of countries and the respective time period is selected. Throughout the process of preparing this paper, our emphasis has been to keep our data as much comparable across countries as possible. Therefore, during the process of cleaning up the dataset, we rejected the countries whose data was not directly comparable to the most available data of countries.
The issue that arises is that the World Inequality Database contains dif- ferent types of income variables like national income, gross domestic product, net domestic product and fiscal income and not all income variables are avail- able for all the countries. For some countries, it provides the data for national income, while for the others, it provides data for net domestic product. The problem becomes more complicated when the database uses different crite- ria of calculation of income variables for different countries.
To be precise, it uses 3 different population age group and 3 different population category group to derive a value of a particular income. The 3 population age groups are adults-including elderly (20+), adults-excluding elderly (20-65) and all ages. The remaining 3 population categories are individuals, tax units and equal-split adults (i.e. income or wealth divided equally among spouses). Each income variable has to belong to a particular age group and a par- ticular category group. Hence, a total of 9 different variants are available for one income variable, and again, not every variant is available for every country.
At most two or three variants for share of national income were available. Hence, among all the different permutations of income variable, their population ages and categories; we made sure that we take maximum number countries with maximum time period of data of common income variable with same population age and same category. Doing so, we arrived at a panel data of 24 countries spanning from 2001 to 2014 with income share data of top 10% rich and bottom 50% poor.
The income variable selected is the share of pre-tax national income (of top 10% and bottom 50%) of age group adults-including elderly (20+) and category equal-split adults. Moving forward, we try to analyze the first question. We use scatterplot and correlation coefficient as our tool. Our hypothesis is that if there exists a definite relationship between inequality and income in our panel of selected countries, then that relationship will be reflected in the scatterplot with values of Kuznets Ratio and log of per capita GDP on each of its respective axes.
Moreover, if the relationship turns out to be linear, it will also be reflected in the values of correlation coefficient, whose value will be close to -1 or 1. Even if there exists some non-linear relationship, scatterplot is a useful tool to capture that. The correlation coefficient of Kuznets Ratio and per capita GDP for the cross-section of our sample from 2001 to 2014: As the table depicts, we don’t find the correlation coefficient to be signif- icantly strong for any year within the considered time period.
Usually, the correlation co-efficient of greater than or equal to 0.75, or less than or equal to -0.75 is considered to be a strong correlation between two variables. One can notice from the table that most values of correlation coefficient are closer to zero, with the value of just 2006 barely crossing the 0.5 mark. Moreover, it is intriguing to note that the value of correlation coefficient is positive for all the years. This leads us to conclude that income and inequality usually move in the same direction, but the relationship of this movement seems to be weak.
Also, the chances for the relationship to be linear gets rejected due to the lower values of correlation coefficient. Further, the possibility of presence of any non-linear relationship is checked by preparing the scatter plots for the cross-section of all 24 countries from 2001-2014. Mentioned below is the scatterplot for the year 2001. We notice some movements in the data points of the plot during each respective year, however, the shape of the scatter mostly remains the same throughout our time period. Hence, looking at the scatter plot for 2001 can give a good idea regarding the shape of the scatter throughout the time period.
The first impression allows us to comment that the possibility of noticing an ‘inverted U’ is completely rejected for our sample of countries since countries like Kuwait (KW), Saudi Arabia (SA), Oman (OM), Qatar (QA) and United Arab Emirates (AE) show higher values of Kuznets Ratio even at higher levels of per capita GDP. Moreover, if there are chances of existence of a non-linear relationship between income and Kuznets Ratio, the direction of that relationship seems positive with the increase in income and not the other way around. Allowing us to digress a little, when we track the movement of countries in the scatter plots year after year, we notice that not all countries move in the same direction.
For example, India is one of the very few countries among our sample whose values of Kuznets ratio rose every year. While on the other hand, country like Thailand registered a continuous fall in its Kuznets ratio from 2004 onwards. Moreover, the movement of the Kuznets Ratio also seems to be affected by the political situations within a country. Kuwait registered a steep fall in its Kuznets Ratio during the 2011 due to its political crisis2. This decline, however, was a result of the decrease in wealth of its top 10% rich population and not due to the increase in the share of bottom 50%.
The Moving on to the second question, we try to find out whether any trade-off exists between growth and inequality. It is important to note the distinction between this question and the earlier one. Earlier, we asked whether there exists a relationship between inequality and level of income. Here, we ask whether high growth of income comes at the cost of higher inequality. To be precise, we examine the relationship between growth of per capita GDP and growth of Kuznets Ratio in this sub-section. To test this relationship, we plot average growth of Kuznets Ratio of countries over the years against the average growth of per capita GDP.
The reason behind taking average growth of both the variables is that inequality may respond slowly to the changes in level of per capita GDP. Therefore, to ensure that the lag response is captured, we take average growth rate so that all the observations of the dataset are utilized while preparing the scatter plot. The Figure 3 hints us about the existence of the inverted-U relationship over here.
However, this presence of inverted-U relationship is again a weak relationship since there are only three countries growing above the rate of 7% in our sample and they happen to lie on a downward slope. If we remove these 3 countries from the graph, then the remaining part of the scatter doesn’t provide us with any indication of the presence of a definite relationship. If there actually is an inverted-U relationship present for all countries, then its interpretation would be that initially, growth can be brought by triggering increase in the incomes of the top deciles of the population, but further increase in growth rate requires a country to trigger increase in income across all the deciles of population.
It might be possible to comment that growing at higher rate can only occur if that growth is inclusive in nature. Thinking regarding the tradeoff between high growth and inequality, the scatter plot does not provide us with a definite answer, but just hints towards a weak inverted U relationship. Lastly, moving to the question of global inequality, we deal with the findings of “Elephant curve” which was first noted by Branko Milanovic and then, subsequently constructed by various authors in the World Inequality Report. The Elephant Curve of World Inequality Report 2018 showed that much of the total increase in income during 1980-2016 period was captured by the top 1% rich population of the global income distribution.
While the population from 20th percentile to 70th percentile did notice some increase in their real incomes, it was the population from 70th to 90th percentile which noticed by very little increase in their real incomes. If one visualizes the shape of the curve which shows the increase in real income at each percentile, then that shape resembles to that of the trunk of an elephant and hence, called the Elephant curve. Milanovic used this curve to show the winners and losers of globalisation, while the World Inequality Report uses it to show the dynamics of global inequality.
The time period used by his study was 1988- 2008. Over here, we show that when we choose the time-period of 2001-2014, the one which we used for the earlier analysis, we don’t find the existence of the elephant curve. The percentage change in average income across every decile of the global income distribution. Instead of noticing the sharp increase in income at the top decile, we find that the increase in income is, in fact, the lowest in the top 10% across the global income distribution. This lowest increase also brings down the share of top 10% in the income distribution and thereby, increasing the share at the lower deciles. This decrease in share at higher levels of income distribution and increase at lower levels can be seen as a progressive transfer.
The famous Dalton principle in the field of inequality measurement states that inequality in a given distribution should decrease in presence of a progressive transfer3. Such decrease in global inequality is evident from the Kuznets Ratio of the global income distribution which fell from 6.33 in 2001 to 5.42 in 2014. However, it is important to note that the share of total income of the poorest 10% has hardly changed. Our analysis shows that inequality in the global context has improved from 2001-2014.
While the elephant curves of Milanovic and World Inequality 3A progressive transfer, as originally formulated by Dalton (1920), is a transfer of income from a richer individual to a poorer individual. In our case, we relate it to the fact that the income share of the richer individuals has decreased and the share of poorer individuals has risen. Report tells us that the picture is not so good if we look at the 1980-2016- time period. This led us to conclude that much damage to the global income distribution must have occurred during the time period 1980-2000.
Our analysis brings us to the following conclusion regarding the inequality and growth for the cross-section of our selected sample of countries from 2001-2014:
- There remains a weak positive correlation between inequality and levels of per capita GDP for all the years.
- Despite the correlation, no definite relationship between inequality and income is noticed. Hence, Kuznets’ ‘Inverted-U’ hypothesis is not seen as an empirical regularity.
- There seems to be a weak ‘Inverted-U’ relationship between growth of inequality and growth of per capita GDP.
- No shape resembling to ‘Elephant curve’ is noted when the changes in income across deciles of global income distribution is plotted for the time period 2001-2014. Since the ‘Elephant curve’ is mostly observed from the time period of 1980-2016, we are convinced to argue that much damage (in terms of inequality) to the global income distribution took place during the time period around 1980-2000. While the condition of global inequality seems to be improved from 2001-2014 as noticed by the decline in the Kuznets Ratio of the global income distribution.
- The share of total income of the top 10% rich people in the global income distribution has declined from 2001-2014.
One possible limitation of our analysis is that the sample of selected countries may not properly represent all the countries of the world. Almost half the size of our sample is occupied by West Asian countries like Saudi Arabia, UAE and others. Moreover, as per income group classification provided by the World Bank, much part of our sample is occupied by high income and upper middle-income countries. It is important to note that these limitations are noted with respect to the findings of question 1 and 2 of the analysis and does not apply to the findings about the global income distribution, since that data is separately provided in the World Inequality Database. Moreover, our opinion is that these lim- itations do not invalidate our findings about the cross-country comparisons completely.
If we treat the sample as representative of upper middle income and high-income countries, then the as per the postulated ‘Inverted-U’ hy- pothesis by Kuznets, we should have noticed downward trend in inequality with increase income, which was not noticed. Hence, based on this line of reasoning, the possibility of Kuznets hypothesis still gets rejected. It is important to note that our choice of the sample arrives from the fact that we laid utmost importance to use data which can be readily comparable across countries. Hence, there is no unclarity regarding our findings when it comes to the comparability of countries within the dataset, since all the observation arrive from the same income variable.
Scope for further research
The findings of our research could be further refined by using econometric techniques like running a panel regression across dataset. Moreover, the size of the data set could be increased by using regression to estimate the values of income shares at various deciles for the countries whose income variables are different from the one considered in this paper.
- Anand, Sudhir, and Paul Segal. ”What Do We Know about Global In- come Inequality?” Journal of Economic Literature 46, no. 1 (2008): 57-94. https://www.jstor.org/stable/27646947
- van Zanden, J., et al. (eds.) (2014), “How Was Life?: Global Well-being since 1820”, OECD Publishing, Paris, https://doi.org/10.1787/9789264214262-en.
- Roser, Max. (2018). ”Global Economic Inequality”. OurWorldInData.org.’https://ourworldindata.org/global-economic-inequality’
- Hellebrandt, Tomas and Mauro, Paolo. (2015). “The Future of World-wide Income Distribution”. Peterson Institute for International Economics Working Paper No. 15-7. http://dx.doi.org/10.2139/ssrn.2593894.
- Cornia, Giovanni, (2003), “The Impact of Liberalisation and Globalisation on Income Inequality in Developing and Transitional Economies”, No 843, CESifo Working Paper Series, CESifo Group Munich
- Milanovic, B. (2009). “Global Inequality and the Global Inequality Extraction Ratio: The Story of the Past Two Centuries”. Policy Research Working Paper 5044. Washington D.C.: World Bank.
- Ortiz, Isabel & Cummins, Matthew. (2011). “Global Inequality: Beyond the Bottom Billion – A Rapid Review of Income Distribution in 141 Countries”. SSRN Electronic Journal. 10.2139/ssrn.1805046.
- Alvaredo, Facundo, Lucas Chancel, Thomas Piketty, Emmanuel Saez, and Gabriel Zucman. (2018). World Inequality Report. World Inequality Lab.
- Oxfam. (2018). Reward Work, Not Wealth. Oxfam GB.
- Yitzhaki, Shlomo. (1997). “More than a Dozen Alternative Ways of Spelling Gini”, Hebrew University
- Kaldor, N., (1957). “A Model of Economic Growth,” Economic Journal, 67, 591–624.
- Ferreira, F. H., Lakner, C., Lugo, M. A. and Özler, B. (2018), “Inequality of Opportunity and Economic Growth: How Much Can Cross-Country Regressions Really Tell Us?”. Review of Income and Wealth, 64: 800-827. doi:10.1111/roiw.12311
- Barro, Robert J. ”Inequality and Growth in a Panel of Countries.” Journal of Economic Growth 5, no. 1 (2000): 5-32.
- Kuznets, Simon. ”Economic Growth and Income Inequality.” The American Economic Review 45, no. 1 (1955): 1-28.
- Benabou, R., (2000). “Unequal Societies: Income Distribution and the Social Contract,” American Economic Review, 90, 96–129.
- Milanovic, Branko. ”Can We Discern the Effect of Globalization on Income Distribution? Evidence from Household Surveys.” The World Bank Economic Review 19, no. 1 (2005): 21-44.
- Ravallion, Martin. ”Why Don’t We See Poverty Convergence?” The American Economic Review 102, no. 1 (2012): 504-23.
- Dalton, Hugh. ”The Measurement of the Inequality of Incomes.” The Economic Journal 30, no. 119 (1920): 348-61. doi:10.2307/2223525.