Thesis
Name: Manav Chhabra
Student number: 10621148
Specialization: Economics
Field: Labor economics (Macroeconomics)
Number of credits thesis: 12
Title of the thesis: Economic inequality and its effects on
economic growth and development.
Table of Content
Abstract ... 3
Introduction ... 4
Literature review ... 5
1. Credit Market Imperfections ... 6
2. Political economy ... 6
3. Sociopolitical Unrest ... 7
4. Saving rates ... 7
Econometric Model and data ... 10
Empirical Results ... 12
1. The Panel of Developed Countries ... 12
2. The Panel of Developing Countries ... 13
Conclusion ... 15
Bibliography ... 17
Abstract
The purpose of this paper is to investigate the effect of income inequality on GDP per capita growth rates in economies of different development statuses. A dynamic model is used on panel data, taking fixed effects and time effects into account. The Panel data regressions reveal a positive relation of the Gini index with growth rate in developed economies. An increase in inequality causes a marginal increase in growth rates. Developing economies, on the other hand, display a negative relation of the Gini index with economic inequality. The effect of inequality can be correlated to the different regions and development statuses. Human capital index has a positive effect on GDP growth in both panels. An increase in average years of
education increase the growth rate. The initial GDP per capita shows a change in the growth rates. Convergence can be concluded from the paper. Less developed
economies grow faster than more established economies.
JEL Classification: O150
Key words: human capital index, GDP growth rates, Gini index, income inequality, developing, developed
Statement of Originality
This document is written by Student Manav Chhabra, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Introduction
Inequality is a problem that all the countries in the world have to face. Economic inequality, or wealth inequality stands for the increasing gap between the rich and the poor among a population. Not only has the income gap between the rich and the poor been increasing in emerging and developing countries, but inequality has also risen in most advanced economies (E. Dabla-Norris, K. Kochhar, F Ricka, N.
Suphaphiphat, E. Tsounta, 2015, p.8). Some of the countries that experienced a decrease in income inequality had problems of declining accesses to education and health care. People of lower social statuses in many developing countries are
troubled with the problems of assuring sufficient nutrition and paying utility bills. Not only are these resources scarcely allocated, they also prevent them from
participating in social and political events. This “suboptimal” use of human resources and physical labor, not only leads to underrepresentation, but also has serious effects on economic growth (E. Dabla-Norris et al., 2015, p.5).
Getting rid of inequality and unfairness is an important value in modern societies. People like to be treated equally, no matter the race, gender, origin, color or ethnicity. Legal, religious or even economical; equality matters to everyone. According to President Obama, widening income inequality is “the defining challenge of our time” (2013). It is important to understand that inequality is not necessarily something that happens over a course of time and that we are all born equal. A lot of inequalities are beyond our control and are decided even before we are born. Circumstances such as “gender, ethnicity, location of birth and family background” are decisive factors that give or take from us the opportunity to live more equal lives (E. Dabla-Norris et al., 2015, p.6). People from wealthier and socially connected backgrounds have better access to education and tend to have better networks that allow them to get superior jobs. Many people start off the race to equality with one hand tied behind their backs. However, an individual’s acquired skills and effort also influences the ability to get a good job and a stable income.
Many economists argue about the effect of income inequality on economic growth. According to Dabla-Norris et al., some level of inequality provides people with incentive to excel, compete, save and invest to move ahead in life. It offers innovation and triggers entrepreneurship, which is needed in a lot of developing economies. Some people need to be thrusted to invest in education and new
argue that inequality has a positive correlation with volatility and a deterring effect on physical and human capital investment (1999, p.5). Sometimes, inequality has a stimulating and invigorating effect on the economy, while the other times, inequality deteriorates economic growth by preventing efficient human resource allocation. As many of these effects are country-specific, it is important to ask: how does economic inequality affect economic growth in differently developed economies?
This paper will focus on the relationship between GDP growth and economic inequality in countries of different development status. The paper will be structured as follows: Firstly, the paper will provide a brief literature review on papers from similar fields and studies. Here, I will discuss the theoretical effects of inequality on growth and investment. The purpose will be to understand modern approaches and theories on income inequality and how these affect societies of different income status. The second part of the paper will entail the research method and hypothesis. An econometric model will be displayed, furthermore, panel data setting will be set up and elaborated on in this section. Thirdly, the Analysis will be presented by
comparing the outcome of the different panels. In the fourth section, I will conclude and summarize the results of the analysis. Finally, a fifth section will contain tables and graphs in an appendix.
Literature review
In the late 20th century, according to the conventional approach, inequality was considered good for incentives and growth. Many development economists
countered this approach. “Greater equality in developing countries may in fact be a condition for self-sustaining economic growth” (M. Todaro, 1997). Economists like Michael Todaro researched that inequality may have a greater and more negative impact on the economy than previously assumed. According to him, inequality would lead to unproductive investment by the rich and more biased demand for local goods by the poor. Inequality can have different effects on different macroeconomic
settings. It is important to understand and assess these macroeconomic impacts and relations between inequality and growth. Robert J. Barro classifies this relation in four categories (2000). These are as follows:
1. Credit Market Imperfections
This approach models an imperfect credit market. There is a limit on the amount that can be borrowed. The rates of return on investment opportunities differ from the margin. This is typically caused by asymmetric information. Barro uses the example of difficulty in obtaining defaulted loans because of imperfections in law enforcement or bankruptcy laws (2000, p.2). While the rich and the poor are not equally likely to obtain a credit based on their incomes and assets, the rich receive loans for further investments while the poor relinquish their “human-capital investments that offer relatively high rates of return” (R. Barro, 2000, p.2). Inequality causes average productivity of investment to decrease. Reducing inequality would therefore, increase economic growth. The development of markets and legal institutions improve with an economies’ development. This makes it more important to improve these imperfections in poor economies in comparison to rich economies.
2. Political economy
The political economy approach is based on redistribution with a focus on a political vs. economical approach. When the mean income in an economy exceeds the median income, the majority of people will favor redistribution of resources from the rich to the poor. In an economy, the amount of poor people that fall beneath the mean income is much larger than the amount of rich people above it. This is mostly because of unfavorable income distribution. This theory starts with the assumption that the richest people in a country, increase the mean income by a large margin. Given that most people fall under this mean; a
redistribution of income becomes more appealing. This can take place in the form of public-expenditure programs and regulatory policies (Barro, 2000, p.2). While this reallocation of income is motivated through the political process, it is not necessarily backed by economic theory. The relocating of income can discourage work effort by richer people, instigating a decrease in economic growth. As higher income inequality will cause more redistribution, it can be assumed that inequality would decrease growth. This distinction can be simplified by thinking of it as uniform political power (one-person/one-vote democracy) being unequal to economic power. If more economic power would imply more political power, the rich would prevent redistributive policies by consuming resources and corrupting
the political system. Thus, it can be said that inequality can have a negative impact on growth through politics, no matter if redistribution takes place or not. 3. Sociopolitical Unrest
The third point is based on the theory that income inequality will motivate people of poorer backgrounds to get involved in crime, riots or other disruptive activities (Barro, 2000, p.3). This can threaten political stability and cause more uncertainty and less longevity of the legal system. Not only is this a waste of potential human resources, but the time and effort that criminals put into crime and other activities cause a loss of potential productivity. Threats to property rights will also cause a loss of prospective investment. More inequality will cause more sociopolitical unrest and decrease economic growth. A better allocation of income will theoretically decrease crime rates, and better redistribution towards education and health care would further diminish this unrest. Redistribution would be the decisive factor in disrupting sociopolitical conflict.
4. Saving rates
Keynes’s General Theory models a direct relation of investment and savings (I=S). While the national income function is described by investment, an increase in income should increase investment, and thus, increase savings. If this is accurate, then redistribution of income from the rich to the poor will lower the aggregate rate of saving by decreasing investment (Barro, 2000, p.4). An
increase in inequality should increase investment, which in turn, should increase economic growth.
These theories provide a useful insight into assessing the impact of inequality on growth. However, the total net effect of income inequality on growth cannot be completely predicted with these theories. They also do not explain how inequality affects growth over time. To understand income inequality over the process of development, the Kuznets hypothesis (1963) provides an interesting insight.
According to cross-country data and time series research on the matter, a U-shaped relationship was discovered between inequality and GNP. Kuznets found “the
evolution of the distribution of income over the transition from a rural to an industrial economy”. Inequality should increase during the early stage of development due to urbanization and industrialization and foster growth. The industries would attract a
large segment of rural labor force and cause inequality to recede during later stages. The idea was to understand how inequality changes over time on different
development stages. The first part is focused on an early agricultural state. This stage models low income per capita and little inequality. According to Kuznets, less developed and less industrialized economies will have higher levels of intersectoral inequality of income per capita (1963, p.21). While there is development towards early industrial- and urbanization, income per capita in the sector starts increasing, therefore, increasing inequality. This advancement triggers a movement of resources from the agricultural to the industrial sector. Although initially, a small group of people in the industrial sector start off with high incomes, causing more inequality, more and more of the local population move towards the industrial areas. The lower demand for working in the agricultural sector drives up the relative wages. This movement towards industrial areas and increase in relative wages for agricultural workers combine to decrease overall inequality at a more developed stage of the economy. While this approach worked to predict inequality to some extent in the late 20th century, wage inequality has sharply increased over the last two decades (Aghion, Caroli, Garcia-Penalosa, 1999, p. 1666). Aghion et al explain this increase as a product of trade liberalization, technological change and new organizational forms. A good insight can be provided by thinking of it in terms of differences in technology between rich and poor areas. While poorer areas utilize more outdated and older technology in comparison to richer areas, inequality rises faster in richer areas. This can be supported by Kuznets’ theory of industrialization. New technologies in
industrial areas, while increasing income per capita, also increased output by using machines that can easily make the work of certain human labor obsolete. Innovations such as the factory system, electrical power, computers and the internet initially raise the inequality levels (Barro, 2000, p.5). As these technologies become available to a larger population, inequality tends to decrease again. This is because people tend to catch up with these new technologies eventually. Fractions of the people that fall behind on these technologies see an increase in relative wages as demand
increases for those certain sectors. Therefore, the growth in inequality is related to the time technological innovations are introduced to the economy. As the older models such as the Kuznets hypothesis do not necessarily explain the effect of income inequality on developed economies to its previous precision anymore, it is important to change the approach to adapt to modern times. Due to these
technological advancements and changes, data of over 25-30 years ago should be disregarded for an effective analysis.
In a research paper written by David H. Autor, the income share of the richest and poorest people is questioned. In his paper, he finds that the earnings premium for education has been steadily increasing in advanced economies, contributing to an increase in net growth of earnings inequality (2014, p.2). It hints towards the rising skill premium in advancing economies. Autor finds that the difference in earnings between college and high school graduates in the last 30 years has more than doubled in the U.S. Another interesting outcome of his paper is his findings on the increasing income of the top percentile in U.S. households. In his analysis, it was discovered that if the yearly profits of the top percentile from 1979 to 2012 were redistributed to the remaining population, every household would receive $7,107. This shows the impact of inequality on an advanced economy such as the U.S. This makes it essential to discover the impact of unequal distribution and inequality on developed economies.
An interesting approach was used in the study of Benabou (1996, p.2) where he compared developing countries such as South Korea and Philippines. These
countries were comparable with similar macroeconomic indicators such as GDP per capita, investment per capita and average saving rates in the 1960s. At the time, income share of the richest 20% to poorest 40% was twice as high in the Philippines as in South Korea. Thus, income distribution was more unequal in the Philippines than in South Korea. After an analysis with data over 30 years, it was discovered that output had increased over five times in South Korea. The Philippines, on the other hand, had a bare increase of twice its previous output. The more unequal country had a slower growth. This outcome refutes the traditional approach where inequality is considered to trigger economic growth.
Roberto Perotti (1996) supplies a quite practical approach to GDP growth analysis. The average growth rate is regressed with GDP, purchasing power parity value of investment deflator, income share and average marginal tax over a certain time period (Perotti, 1996, p.160). He also considers average years of secondary
schooling in male and female population to represent human capital. Human capital measurement is an important variable in this regression. It can show whether the growth in GDP can be somewhat explained by increasing literacy rates and
employment.
Cultivating on the above-mentioned researches, a cross-country panel data research will be executed and analyzed.
Econometric Model and data
Recent studies on discovering the effect of income inequality on economic growth have focused on cross-country panel data of certain time series. Given the
technological advancements of the last two decades, the time series has been limited to about 25 years per country. In this study, two panels were created, one
representing developing and newly industrialized countries, and the other
representing developed countries. The panel of developing countries consists of the
BRICS countries. Namely, Brazil, Russia, India, China and South Africa. These
economies have contributed to “changing the global balance of power, raising hopes of a more egalitarian global governance architecture through international trade and development co-operation” (C. Ivins, 2013, p.2). While these countries with growing influences provide stability and encourage investment in developing countries, four out of these five countries have seen a steep increase in inequality. Ivins argues that reducing these figures can contribute to decreasing crime rates and building stronger trust and cohesion (2013, p.2). Furthermore, it can improve the general health of the population by providing access to the basic necessities. The developed panel
consists of France, Germany, the Netherlands, UK and the US. The purpose of using panel data regression is to run a two-dimensional analysis with time series and
cross-country data.
While cross-section analysis is extremely useful in performing an analysis on multiple dimensions, there are some inconsistencies. For one, it cannot control for certain country fixed factors such as technology, tastes, climates or institutions (A. Castello, p.5, 2007). These factors can cause some omitted variable bias that will affect the coefficients of the estimates. The model used in this paper will include a partial solution to this problem. While some of these fixed effects will be checked for in the model, there are a lot more variables that affect GDP growth. A lot of these variables such as political instability, consumer confidence and level of infrastructure affect GDP growth, but cannot be measured easily. Another discrepancy in this paper is in
the panel of developing economies. Data on Gini indices in Russia only dates back to 1990. These three missing observations make the data somewhat weakly balanced. The dataset is still balanced and the missing Gini indexes were corrected for by excluding parallel values for the other variables. The panel of developing countries is strongly balanced and includes data from a time series of 1988-2012 for each
country. The following standard growth model used in this paper is as follows: 𝑙𝑛(𝐺𝐷𝑃𝑖,𝑡) − 𝑙𝑛(𝐺𝐷𝑃𝑖,𝑡−1) = 𝑐 + 𝛽1∗ ln(𝐺𝐷𝑃𝑖,𝑡−1) + 𝛽2∗ 𝐺𝐼𝑁𝐼𝑖,𝑡−1+ 𝛽3∗ 𝐻𝐶𝐼𝑖,𝑡+ 𝛾𝑡+ 𝛿𝑖+ 𝜀𝑖𝑡 (1) This can be rewritten as a dynamic model:
𝑙𝑛(𝐺𝐷𝑃𝑖,𝑡) = 𝑐 + 𝛽̃1∗ ln(𝐺𝐷𝑃𝑖,𝑡−1) + 𝛽̃2∗ 𝐺𝐼𝑁𝐼𝑖,𝑡−1+ 𝛽̃3∗ 𝐻𝐶𝐼𝑖,𝑡+ 𝛾̃𝑡+ 𝛿̃𝑖+ 𝜀̃𝑖𝑡 (2) The dependent variable is 𝑙𝑛(𝐺𝐷𝑃𝑖,𝑡), which stands for the percentage growth rate in
GDP per capita of country i at time t. The first independent variable is the GDP per capita of last year. This variable shows the average percentage change of the GDP per capita per year. The GDP per capita indexes were obtained from un.data.org which correlates with the figures from the World Bank and is measured in US dollars. Secondly, Gini coefficients are used as a measure of inequality. Gini coefficients are statistical dispersions used to estimate the distribution of wealth among residents in a country. It is a frequency distribution that is measured from 0 to 1. The figure 0
represents perfect equality, where everyone obtains the same income. The number 1, on the other hand, represents perfect inequality, where a single resident earns the wealth of the entire nation. These are extreme figures and not really applicable to any country. The higher the Gini index, the more inequality there is in a nation. The
purpose of the Gini index is to find out what impact inequality has on differently developed economies. In the model, Gini is used at time t-1 to represent the Gini at the start of the year. The Gini coefficients were obtained from the SWIID database. This stands for Standardized World Income Inequality database. The indexes used, represent net income (post-tax) Gini’s. The third independent variable HCI stands for human capital index. The human capital index is obtained from the PENN World tables and is measured by average years of schooling relative to returns to education. It measures how well human capital resources are allocated in an
organization. It is included in the regression to find out how higher education affects GDP growth. The term 𝛾𝑡 represents a time specific effect. As mentioned before, one
of the inconsistencies of economic growth theory is omitted variable bias. The time specific effects are dummy variables that control for time shocks that occur in a certain year. It helps prevent omitted variable bias to some extent. Furthermore, fixed effects are used in this model. This is represented by 𝛿𝑖. Fixed effects control for country specific effects that are constant over time. This can be things that are explicit to a country, like languages or the legal system. Like with time effects, controlling for fixed effects means to create dummy variables for every observation and including them in the standard regression to control for cases of fixed effects. It uses the change in variables over time to estimate the effects of the independent variable on the dependent variable. Finally, 𝜀𝑖𝑡 represents the error term that differs across countries and over time.
The following hypothesis will be tested in the panel data regression:
𝐻0: 𝛽1,𝛽2, 𝛽3 = 0 𝐻1: any of the variables is unequal to 0
Empirical Results
1. The Panel of Developed Countries:
To begin with, it is important to analyze the data. In appendix 3A, a test for outliers in the data is conducted. The test revealed that while there are some outliers in the data, there are no severe outliers. There are no measurement errors caused by large outliers in the data. Subsequently, a Woolridge test on the autocorrelation is
conducted. This can be seen in appendix 4A. While serial correlation is not always necessarily a big problem in micro panel datasets, they are necessary in macro panels where the time series exceeds 20 years. The p-value of 0.0037 is small enough to reject the null hypothesis. From this we can conclude that there is first-order serial correlation in the data which could cause the standard errors of the coefficients to be smaller than they are (Stock & Watson, 2015, p.649).
Consequently, a modified Wald test is conducted to check for group wise
heteroscedasticity in the fixed effect model. The results in appendix 5A reveal a presence of heteroscedasticity as the null is rejected at 5% significance. Thus, the data is autocorrelated and heteroscedastic. The following regression is conducted considering these conditions.
The first estimated coefficient to be considered here is the initial per capita income. The estimates can be seen in appendix 1A:
Appendix 1A: Coefficients - Developed
GDP per capita Coef. Robust Std. Error t 𝑃 > |𝑡| 95% Confidence Interval Past GDP -5.69e-06 2.12e-06 -2.68 0.055 -0.0000116 2.05e-07 GINI 0.0055559 0.0046922 1.18 0.302 -0.0074717 0.0185834 HCI 0.047923 0.0706518 0.68 0.535 -0.1482379 0.2440838
Note: The regression is estimated while controlling for heteroscedasticity and autocorrelation
The initial GDP per capita changes the growth rate by -0.000569%. It is significant for a 10% significance. As expected, the coefficient is negative. This estimator will
mainly be used to check for convergence between the two panels. The results
displayed in the table also show a positive effect of the Gini index on the growth rate. A change in the Gini index by a percentage point changes the growth rate by 0.56 percent. This implies that inequality is growth-enhancing. However, from the table we can determine that the effect of the Gini coefficient is individually not statistically significant. Lastly, the estimator for human capital index is also positively related to the growth rate, but like the Gini index, it is not statistically significant. An additional year of secondary schooling increases the growth rate by 4.79 percent. Appendix 2A provides an overview of several regressions. Regressions (1) and (2) are conducted without time and fixed effects. The figures for the adjusted 𝑅2 are relatively low. When time effects and fixed effects are included in the model, much more of the variations in the growth rate are explained. Additionally, table 6A shows a positive correlation between the Gini coefficient and human capital index. This is because higher education is typically associated with better paying jobs. This differs across countries, but in developed economies there is mostly a positive correlation between the two variables. To test for joint significance of the estimators on GDP growth, an F-test is conducted (see appendix 7A). The F-test outcome of 2.62 has a p-value of 0.1874. At a 5% significance level, there is not enough statistical evidence to reject the null hypothesis. The effects of the independent variables on growth rate are not significant.
Identically to the panel data of developed countries, tests on the outliers,
autocorrelation and heteroscedasticity need to be conducted. Appendix 3B shows that similar to the developed panel, there are no severe outliers in the developing panel. The Woolridge test on autocorrelation from appendix 4B displays a p-value of 0.0059. This value is small and implies a rejection of the null hypothesis. There is autocorrelation in the data. Subsequently, the modified Wald test on
heteroscedasticity shows an insignificant p-value that is rejected at the 5%
significance. Again, the data is autocorrelated and heteroscedastic. The regression controls for robustness and autocorrelation.
Appendix 1B: Coefficients - Developing
GDP per capita Coef. Robust Std. Error t 𝑃 > |𝑡| 95% Confidence Interval Past GDP -0.0000264 0.0000102 -2.57 0.062 -0.0000548 2.07e-06 GINI -0.0029305 0.0036527 -0.80 0.467 -0.0130721 0.0072111 HCI 0.0876265 0.1570425 0.56 0.607 -0.3483935 0.5236465
Note: The regression is estimated while controlling for heteroscedasticity and autocorrelation
Comparably to the first regression, the initial GDP per capita has a negative correlation with the GDP growth rate. It causes the growth rate to change by an average of -0.00264 percent per year. This already shows that an increase in the GINI index would tend to decrease economic growth. It is also within the 10%
significance level. Since the estimator of change in income per capita is larger than in the developed economy panel, it supports the convergence or so-called catch-up effect (Stock & Watson, 2015, p.94). In theory, it means that the income per capita in poorer economies will increase faster when compared to richer economies. This can be caused by the less developed economies reproducing technologies and
institutions from the more developed economies. Eventually, the per capita income will converge towards a certain income per capita, closer to the income per capita of the developed countries. In contrast to the panel of developed economies, the Gini index is negatively related to the GDP growth. While it is not individually significant, a percentile change in the Gini index causes a change of -0.2931 percent on the
growth rate. This means that more inequality will decrease the GDP growth rate in emerging economies. The human capital index is not significant but demonstrates a positive relation with the GDP growth. An additional year of schooling in the BRICS countries increase growth by nearly 8.8%. While checking for the adjusted 𝑅2 in appendix 2B, similar outcomes as in the panel for developed economies can be
observed. In (1) and (2), the detected values are relatively low. Whereas, when the model controls for time effects and fixed effects, the 𝑅2 is increased to 0.5027. While checking the correlation between the Gini and the human capital index, it can be witnessed that there is a negative correlation between the two variables (see
appendix 6B). More education in developed countries does not necessarily imply less inequality. The other way around, it implies that more inequality would lead to more restricted access to higher education. The F-test on joint significance of the
independent variables displays an F-value of 6.27 and a p-value of 0.0541. At the significance level of 0.05, the null hypothesis cannot be rejected as there is not enough statistical evidence. With 90% confidence, the null hypothesis can be rejected.
Even though the results from both panels are not significant, they do provide an insight into the relation between inequality and GDP growth. The analysis reveals that there are indeed differences in the effect of inequality on growth. Developing countries are more prone to see a decrease in growth rates when the Gini increases than the developed countries. Some of these outcomes can be explained by the 4 major approaches by Barro (2000).
Conclusion
In this paper, the effects of income inequality on economic growth in differently developed countries was analyzed. After controlling for fixed effects and time effects, estimation of the model showed different effects of inequality and human capital on growth. It revealed that these variables have different effects on different regions with diverse development levels. A negative impact of inequality in less developed
economies, and positive impact of inequality in more developed economies was discovered on economic growth. Human capital was positively related to economic growth in both panels. More years of average education tends to have a positive effect on per capita GDP growth. Additionally, the impact of initial GDP per capita shows a higher in GDP growth in developing economies than in developed
economies. Richer and more established economies will see more initial
technological advancements than some emerging economies. This will cause the inequality levels to fluctuate. Moreover, richer economies will have better
are typically associated with GDP growth. There are many factors causing increases or decreases in economic growth. This makes research on this matter essential to further improve econometric models, so that we can understand unjust income distribution and poverty. Furthermore, it is important so more economic stability can be achieved.
Bibliography
Aghion, P., Caroli, E., & Garcia-Penalosa, C. (1999). Inequality and economic growth: the perspective of the new growth theories. Journal of Economic
literature, 37(4), 1615-1660.
Barro, R. J. (2000). Inequality and Growth in a Panel of Countries. Journal of
economic growth, 5(1), 5-32.
Benabou, R. (1996). Inequality and growth. NBER macroeconomics annual, 11, 11-74.
Castelló-Climent, A. (2010). Inequality and growth in advanced economies: an empirical investigation. Journal of Economic Inequality, 8(3), 293-321.
Dabla-Norris, M. E., Kochhar, M. K., Suphaphiphat, M. N., Ricka, M. F., & Tsounta, E. (2015). Causes and consequences of income inequality: a global perspective. International Monetary Fund.
David, H. (2014). Skills, education, and the rise of earnings inequality among the" other 99 percent".
Ivins, C. (2013). Inequality matters: BRICS inequalities fact sheet. Oxfam Policy and
Practice: Climate Change and Resilience, 9(1), 39-50.
Kuznets, S. (1963). Quantitative aspects of the economic growth of nations: VIII. Distribution of income by size. Economic development and cultural change, 11(2, Part 2), 1-80.
Perotti, R. (1996). Growth, income distribution, and democracy: what the data say. Journal of Economic growth, 1(2), 149-187.
Stock, J. H., & Watson, M. W. (2015). Introduction to econometrics. Boston: Addison Wesley.
Todaro, M. P. (1997). Urbanization unemployment and migration in Africa: theory and policy.
Appendix:
Developed panel:
Appendix 2A - Developed Regressions
(1) (2) (3) (4) last year's GDP per capita -0.000*** -0.000*** -0.000*** -0.000* (-2.6) (-2.6) (-3.2) (-2.7) Gini index 0.000 -0.000 0.006 0.006 (0.2) (-0.1) (1.0) (1.2) Human capital index 0.020 0.048 0.048 (0.6) (0.4) (0.7) year=1989 0.000 0.000 (.) (.) year=1990 0.159*** 0.159** (5.4) (3.1) year=1991 0.030 0.030 (1.0) (1.9) year=1992 0.084*** 0.084* (2.7) (2.7) year=1993 -0.025 -0.025 (-0.8) (-1.1) year=1994 0.065** 0.065** (2.0) (4.2) year=1995 0.129*** 0.129** (3.9) (3.1) year=1996 0.040 0.040** (1.1) (3.5) year=1997 -0.002 -0.002 (-0.1) (-0.1) year=1998 0.061 0.061** (1.7) (4.1)
year=1999 0.040 0.040** (1.0) (2.8) year=2000 -0.018 -0.018 (-0.4) (-0.5) year=2001 0.033 0.033* (0.8) (2.1) year=2002 0.092** 0.092** (2.3) (4.5) year=2003 0.188*** 0.188*** (4.4) (5.3) year=2004 0.177*** 0.177*** (3.8) (6.6) year=2005 0.116** 0.116** (2.3) (4.4) year=2006 0.139** 0.139*** (2.6) (5.8) year=2007 0.210*** 0.210*** (3.7) (8.3) year=2008 0.166** 0.166*** (2.6) (6.3) year=2009 0.034 0.034 (0.5) (0.8) year=2010 0.109* 0.109** (1.7) (3.9) year=2011 0.173*** 0.173*** (2.7) (12.0) year=2012 0.090 0.090 (1.3) (2.0)
Fixed effects No No Yes Yes
Time effects No No Yes Yes
Observations 120 120 120 120
Adjusted R2 0.0604 0.0645 0.7298 0.7298
Notes: t-statistics in parentheses
Appendix 3A – Test for outliers
Appendix 4A – Test for autocorrelation
Appendix 5A – Test for heteroscedasticity
Appendix 6A – Correlation between variables
Variables GDP per capita Past GDP GINI Human Capital GDP per capita 1
Past GDP -0.2337 1
GINI -0.0344 0.2286 1
Appendix 7A – F-test
Graph 1A – Plot of average Gini indexes
Developing panel:
Appendix 2B - Developing Regressions
(1) (2) (3) (4) last year's GDP per capita 0.000 -0.000 -0.000** -0.000* (0.2) (-1.2) (-2.2) (-2.6) Gini index 0.002 0.006* -0.003 -0.003 (0.7) (1.7) (-0.6) (-0.8) Human capital index 0.129 ** 0.088 0.088 (2.4) (0.4) (0.6) year=1986 0.000 0.000 (.) (.) year=1987 0.009 0.009* (0.1) (2.4) year=1988 -0.010 -0.010 (-0.1) (-0.5)
year=1989 -0.000 -0.000 (-0.0) (-0.0) year=1990 -0.083 -0.083 (-0.6) (-0.8) year=1991 -0.140 -0.140** (-1.0) (-3.1) year=1992 -0.100 -0.100* (-0.7) (-2.3) year=1993 -0.066 -0.066 (-0.5) (-1.6) year=1994 -0.052 -0.052 (-0.4) (-0.5) year=1995 0.041 0.041 (0.3) (0.6) year=1996 -0.058 -0.058 (-0.4) (-1.1) year=1997 -0.048 -0.048 (-0.3) (-1.2) year=1998 -0.200 -0.200* (-1.3) (-2.3) year=1999 -0.233 -0.233** (-1.5) (-3.9) year=2000 -0.032 -0.032 (-0.2) (-0.3) year=2001 -0.124 -0.124 (-0.7) (-1.7) year=2002 -0.112 -0.112 (-0.7) (-1.8) year=2003 0.064 0.064 (0.4) (0.7) year=2004 0.097 0.097 (0.5) (1.1)
year=2005 0.085 0.085 (0.5) (0.8) year=2006 0.072 0.072 (0.4) (0.6) year=2007 0.140 0.140 (0.7) (1.2) year=2008 0.089 0.089 (0.4) (0.5) year=2009 -0.064 -0.064 (-0.3) (-0.5) year=2010 0.190 0.190 (0.9) (1.3) year=2011 0.181 0.181 (0.8) (1.1) year=2012 0.034 0.034 (0.1) (0.2) year=2013 0.094 0.094 (0.4) (0.6)
Fixed effects No No Yes Yes
Time effects No No Yes Yes
Observations 117 117 117 117
Adjusted R2 0.0205 0.0867 0.5027 0.5027
Notes: t statistics in parentheses
* p < 0.10, ** p < 0.05, *** p < 0.01
Appendix 4B – Test for autocorrelation
Appendix 5B – Test for heteroscedasticity
Appendix 6B – Correlation between variables
Variables GDP per capita Past GDP GINI Human capital GDP per capita 1
Past GDP 0.0128 1
GINI 0.0474 0.0497 1
Human capital 0.1063 0.5566 -0.4470 1
Graph 1b – Plot of average Gini indexes
