Income Inequality as a Determinant of Economic

Growth: A Cross-Country Analysis

Georgia Institute of Technology

ECON 3161 Econometrics, Fall 2016
Authors: Tiffany Suwoto, Yebo Zhai
Supervisor: Prof. Shatakshee Dhongde

Abstract
In this study, we seek to find a relationship between income inequality and economic growth. Despite
years of extensive research dedicated to finding a relationship between the two, past studies and existing
literature still remain divided on the subject, with some finding a positive relationship and others
declaring a negative relationship. This paper examines the effects of inequality on GDP by using data of
225 countries from 2011. Using this data, simple and multiple linear regression models were formed to
determine the relationship between the two variables. Through empirical analysis, we found that statistical
inference tests supported all variables. This study found a positive relationship between income inequality
and economic growth. If further study is pursued, it should consider evaluating countries separately based
on whether they are developed or developing, and testing different explanatory variables.
1. Introduction
The International Monetary Funds January 2016 World Economic Outlook (WEO) Update reads
that global economic growth is projected to grow 3.4 percent in 2016 and again in 2017 (IMF, 2016). In
both advanced and developing economies, gradual growth is predicted initially but is expected to pick up
in the next two years. However, there are factors that play into this goal that, if not successfully
addressed, could derail economic growth (IMF, 2016). One of these key elements is wealth inequality,
and as global economic interdependence and interconnectedness continues to increase, it becomes vital to
analyze the relation between economic growth and inequality.
At this point in time, income inequality is a topic that is very prominent, especially with the 2016
presidential campaign currently going full force. Each candidate has their own notions and proposals for
how to tackle the disparity in income. Income inequality impacts the poor and underprivileged the most,
because the cost of living and the quality of life people live depends largely on their level of income.
Since income level and quality of life are so interrelated, income inequality impacts the poor in several
negative ways. The primary effect of income inequality is that it prevents capital accumulation (both
human and physical) (Mo, 2000; Kaldor, 1956; Aghion, Caroli and Garcia-Penelosa, 1999). Secondly,
inequality can generate socio-political instability that undermines incentives to save and invest, and would
generate pressure on government (Mo, 2000). Finally, inequality has a detrimental effect on social
mobility: countries with higher levels of inequality show a dependence of childs future earning capacity
on the current earning capacity of their parents (Corak, 2013). Inequality, which currently shows no signs
of undergoing income redistribution, is indisputably a problem that disrupts the balance within a society.
Economic growth has been a popular topic since the financial crisis of 2008. In fact, since the
Great Recession, economic growth has been regarded as a sign of advancement, development, and
recovery. During a period of economic growth, poverty and unemployment is reduced, the standard of
living of the population rises, it incentivizes the young, and the countrys currency appreciates against that
of other countries, giving it more international trade power. Ultimately, all countries desire to experience
economic growth in order to progress. According to a German proverb on the subject of growth,
Stagnant water starts to stink at some point (New York Department of Health, 1909). Economies
cannot stand still; either they can go up for they go down--and everyone wants to go up.
Without a doubt, economic growth is instrumental in poverty reduction in a country, but is
economic growth related positively or negatively in regards to income inequality?
This paper declares a positive relationship between income inequality and economic growth, and
we shall test this prediction empirically with regression analysis. Using cross-country data obtained from
World Bank for the year 2011, we conducted regression analysis of economic growth on income
inequality. Existing studies determined there to be a positive relation between income inequality and
economic growth. This research contributes to the statement and proposes that with higher inequality,
economic growth will continue to accelerate.
This paper is organized as follows. Section II draws literary support and analyzes existing sources
to reinforce and elaborate on the research and hypothesis tests we conducted. Section III introduces the
data and explains the techniques used to conduct our study. Section IV interprets the results from the data
and analysis methods employed, and Section V concludes the findings of this research.

2. Literature Review
Despite there being a magnitude of literature on the link between income inequality and economic
growth, there is a stark divide in findings as to whether there is a positive or a negative relationship.
Numerous differing theories about whether these two variables are related positively or negatively
originate largely from differing explanatory variables, differing years examined (leading to differing
datasets), and differing empirical approaches since the 1950s.

2.1 Inequality and Economic Growth: The Perspective of the New Growth Theories
There is a consensus among many authors of literature that there is a negative relationship
between the average rate of economic growth and the measure of inequality. (Aghion, Caroli,
Garcia-Penalosa, 1999). Aghion, Caroli, and Garcia-Penalosa (1999) examined case studies of South
Korea and the Philippines. According to their research, the ratio of the income share of the top 20% of the
bottom 40% of the population in Philippines was almost twice as large as in South Korea. Despite their
differences in degree of income inequality. these two countries demonstrated similar levels of
macroeconomic health (through GDP per capita, investment per capita, average saving rates, etc) at the
beginning of the study. Over the course of 30 years, however, Aghion, Caroli, and Garcia-Penalosa
(1999) found a marked difference in the rate of growth between the two countries (Aghion, Caroli,
Garcia-Penalosa 1999). They ascertained that South Koreas output level underwent a five-fold increase,
while that of the Philippines barely doubled, demonstrating that the country with a higher level of income
inequality grew at a slower rate. After they determined these results in a case study, they conducted
research on redistribution to find whether redistribution fosters or hinders growth. Aghion, Caroli, and
Garcia-Penalosa (1999) found that income inequality was found to be positively correlated with volatility,
and through a series of cross-country regressions found that greater volatility reduces the average rate of
growth during a set period. Their findings were bolstered with results declaring that redistribution has
stimulating effect on economic growth, therefore determining that inequality has a negative impact on
economic growth. These results coincide with other literature declaring a negative relationship between
income inequality and economic growth.

2.2 A Non-Parametric Measure of Poverty Elasticity

In a study that yielded similar results, Chambers and Dhongde (2011) pursued a non-parametric
approach to examine an extensive and up-to-date dataset from the World Bank, inclusive of 1977 through
2007, representing more than 96% of the population of the developing world. Rather than GDP,
Chambers and Dhongde(2011) measured the growth elasticity of poverty (GEP) and found that countries
with higher levels of inequality had lower GEP, and countries with lower inequality had higher GEP.
Through more extensive research (and their non-parametric approach), they studied the typical linear
model to measure the relationship between poverty, mean income, and the Gini index and found evidence
that the relationship between income inequality and growth is best described as non-linear. Chambers and
Dhongde (2011), by analyzing a model which considers the nonlinearity of the growth-poverty-inequality
neux, found that poverty declines rapidly with higher mean income, but slowly with lower values of the
Gini index. In short, their results were obtained using data that was much more comprehensive and
methods that were more robust than those of most studies. Their findings reflect those of Aghion, Caroli,
and Garcia-Penalosa (1999) as well as many others that have also found a negative relationship between
economic growth and income inequality.

2.3 Income Inequality is Not Harmful for Growth: Theory and Evidence
While there seems to be insurmountable evidence in favor of a negative relationship between
income inequality and economic growth, there are numerous studies that yielded a positive connection
between the two variables. In an analysis conducted by Li and Zou (1998), the results stated that empirical
evidence revealed through a regression of GDP growth rate on the Gini coefficient that income inequality
is positively associated with economic growth. Following in previous literatures footsteps, Li and Zou
(1998) followed Alesnia and Rodrik (1994) and Barro (1990) to find income inequalitys relationship
with economic growth by dividing government spending into production services and consumption
services. However, in contrast with Alesnia and Rodrik (1994) and Barro (1990) according to their
results, income inequality can lead to fast economic growth when government spending is wholly driven
by public consumption. In fact, by using this extension of government spending, Li and Zou (1998) found
that since government spending is all for consumption, individuals will try to allocate resources between
public and private consumption. Therefore Li and Zou (1998) state that income inequality can generate
high savings rates and growth rates if the rich have a larger share of income, or if income is more
unequally distributed in the economy.

2.4 Income inequality and Economic Growth (Shin)

While some literature declare a positive relationship and others support a negative one, there are
some studies in which no position is taken and both sides of the debate are examined and analyzed (Shin,
2012). Shin (2012) chose not to pursue a particular stance on the topic but rather chose to examine
reasons why this disparity exists. According to Shin (2012), there is a correlation between the
positive/negative relationship between inequality and economic growth and whether or not the country is
developed or not. Shin (2012) performed a case study of East Asian and South American countries, which
are developing countries. The findings revealed a negative relationship between income inequality and
economic growth in those countries. Conversely, in a case study of the United States and France, which
are developed countries, a positive relationship between income inequality and econ growth was found. In
an agreement with Barro (2000), Shin (2012) declared that the effect of income inequality on economic
growth was contingent on the state of economic development. Specifically, Shin (2012) found that
income inequality in poor countries retards economic growth; that is, in countries with GDP per capita
below 2070, the effect of income inequality is negative. According to Shin (2012), this is caused by a lack
of opportunity to invest by the population of a developed country. This in turn would lead to political and
social instability, which contributes towards economic growth decline. Therefore income inequality
reduces economic growth. In contrast, income inequality in rich countries encourages growth; that is, in
countries with GDP per capita over 2070, the effect is positive. Income redistribution from the rich to the
poor reduces the saving rate of the economy which would lower the incentive for the rich to work hard.
So, income equality would reduce economic growth. It can be inferred from this paper that the result of
income inequality on economic growth varies depending on whether the country is developed or not.
As we stated before, there is a large divide in literature as to if income inequality and economic
growth are related through a positive or negative relationship. The purpose of this paper is to evaluate the
effect of income inequality on economic growth and to contribute relevant findings to the discussion by
examining extensive datasets from the World Bank ranging from 1981 to 2014, which enables us to do a
long-term comparison case study. The world has been undergoing constant economic change,, and global
interconnectedness and interdependence grows and changes each year. To better analyze our data, we
incorporate some other important variables that may have an impact (helpful or detrimental) on the
relationship between economic growth and inequality.
3. Data
We chose the Gini coefficient (pre-tax) for the explanatory variable (x) in our simple regression
line. The Gini coefficient was chosen for this model because it is a common measure of income inequality
across many countries that represents the income distribution of a countrys residents, where 0 represents
perfect equality and 100 represents max inequality, and is recognized and used in much of the literature.
Annual growth percentage of gross domestic product (GDP) was the dependent variable (y). The Gini
coefficient and GDP growth datasets in this paper were obtained from the World Banks Development
Research Group (World Bank, 2011). We chose to regress GDP growth on the Gini coefficient because
most of the literature we referenced found income inequality to have a more marked effect on GDP
growth than GDP growth on income inequality. Our ultimate objective was to find the relationship
between income inequality and economic growth. However, there are numerous variables that may affect
economic growth, including urbanization ratio, population growth rate, financial development (M2/GDP),
openness (export/GDP), etc (Li and Zou, 1998). In order to better understand and analyze the effect of
income inequality on GDP growth, we controlled for other factors that had the most significant impacts
on economic growth in an economy. These variables were gross savings, unemployment rate, education
(mean school years), and fertility rate. Gross savings (World Bank, 2011) is one of the most common
indicators of the growth of a country because it reflects the countrys ability to consume and save.
Fertility rate was included because research has shown that lower fertility rates lead to economic growth.
Unemployment rate (World Bank, 2011) represents the long term unemployment rate, or natural rate of
unemployment, in a country. Unemployment rate is an obvious indicator of a countrys economic
well-being. The mean school years are also expected to have an impact on economic growth. The more
educated a country, the more growth is to be expected because of the capacity for high-skilled laborers.
Finally, a dummy variable was used to measure if the level of development of a country would affect their
economic growth. These two categories (developed and developing) were classified according to the
World Bank classification system.
A summary of the variables is provided in Table 1 below.

Table 1: Variable Descriptions

grgdp Growth of Gross Domestic Product

Gini Gini Coefficient (measure of inequality)

gsav Gross savings

fertil Fertility rate

unemp Unemployment rate

educ Mean school years

3.1 Summary Statistics

Table 2 shows the summary statistics for the data. This study was conducted using 225 countries.
Because a countrys economy can regress, the fact that the minimum of grgdp is a negative number is not
a huge concern.
Table 2: Summary Statistics

Variable Observations Mean Standard Minimum Maximum

Deviation

grgdp 184 3.909663 3.158448 -7.304 12.615

Gini 105 38.0155 8.6441 24.70333 63.38

gsav 160 21.36649 12.6191 -11.5887 60.00212

fertil 184 2.876358 1.4462 1.205333 7.655

unemp 170 8.658712 5.921921 0.3 31.46667

educ 103 8.9109 2.980666 .055963 13.72269

dev 186 0.1827 0.3875 0 1

3.2 Gauss Markov Assumptions

This section tests whether the data meets the Gauss Markov Assumptions. For the sake of
accuracy and effectiveness, the data and models were required to fit the Gauss-Markov assumptions so
that it is ensured that the Ordinary Least Squares (OLS) estimates are accurate, linear, and unbiased. This
way, we can see if our data is justifies our multiple linear regression models.
MLR 1: The model is linear in parameters. Y = 0 + 1X1 + + kXk + u, thus our model
meets assumption one.
MLR 2: There is a random sampling of regressors. Countries selected at random without a
particular reason yield a random sampling. We collected data from random countries in the world
according to the World Bank and obtained our sample from whatever data points were available during
the year 2011, our year of study. Thus, our model meets assumption two.
 MLR 3: There is no perfect collinearity between any of the regressors Table 3 illustrates that
there is no perfect collinearity between any of the regressors, therefore our model meets assumption three.

Table 3: Correlation Among Variables

grgdp Gini gsav fertil unemp educ

grgdp 1.0000

Gini 0.3943 1.0000

gsav 0.3769 -0.0118 1.0000

fertil 0.4549 0.3233 -0.0422 1.0000

unemp -0.3815 0.0274 -0.3224 -0.2558 1.0000

educ -0.4028 -0.4369 0.1541 -0.6355 0.1802 1.0000

MLR 4: According to the zero conditional mean, the expected value of error given all
explanatory values equals 0. Through calculation of the residuals, this was tested and proven. Figure 1
shows the mean of the residuals for the multiple linear regression model tested was about zero.
MLR 5: The error u has the same variance given any value of the explanatory variables.
The residual distribution must approximate a normal curve. Our model should reflect the best linear
unbiased estimators (B. L. U. E. s). Sowe conducted several multiple regression models as well as plot the
residuals. The residual distribution in Figure 1 approximates a normal curve, so our model fulfills the fifth
assumption.
Figure 1: Residuals PDF

PDF of residuals from regressing grgdp on gini, gsav, unemp, educ, and fertil.
4. Results
4.1 Simple Linear Regression Model
The purpose of the simple linear regression model is to test the relationship between GDP growth
and the Gini coefficient. To test this relationship, GDP Growth was only regressed on the Gini coefficient.

Model 1: grgdp = 0 + 1Gini + u

The results of this regression are shown in the following table, Table 4 (see also Table A1, Appendix 2).

Table 4: Results of Regression Estimation for Model 1

OLS: n=105 Dependent Variable: grgdp

Variable Coefficient SE t-value p-value Significance

Gini .1106 .03303 3.35 0.001 ***

Constant -.3067 1.2874 -.24 0.812

*, **, *** denotes significance of coefficients at 10%, 5%, and 1% respectively.

The results showed a positive relationship between the Gini coefficient and GDP growth, which
can be seen in Figure 2 with a scatterplot of GDP grown (5) on the Gini Coefficient. This indicates that
for one unit increase in Gini coefficient, the GDP growth rate increases by 11.06 percent. Since the
intercept is negative, this means that with zero inequality (Gini equals zero), there would be negative
growth. This is a reasonable inference because perfect inequality, which is what is assumed be no
inequality, would allow the assumption of negative growth. The p-value of Gini was 0.001, indicating a
very high statistical significance. Also, the R2 found is 0.0981, which means the Gini coefficient only
explains 9.8 percent of the GDP growth in the model--a low value. We found this rather unsatisfactory.
The reason could be our sample is too diverse or applies for too many different countries since different
countries situation may vary. For instance, one cannot explain the economic growth of some countries
with a universal model. Or, this could indicate a non-linear relationship. In our subsequent research, we
will build more models using different sets of datas, hoping to find a theory to explain it.
 Figure 2: Scatterplot of GDP Growth (%) vs Gini Coefficient

4.2 Multiple Linear Regression Models

We constructed several more multiple regression models to account for other factors or variables
with economic significance that may affect economic growth, and to remove any omitted variable bias.
These new variables were chosen to control for the Gini coefficient. GDP growth was regressed on the
Gini coefficient and 4 new explanatory variables. Table 5 shows the regression estimates for each model
and whether they are significant at 10%, 5%, and 1% (*, **, and *** respectively). The additional
variables were gross savings, unemployment rate, years of education, and fertility rate. The Gini
coefficient was consistently maintaining a positive relationship with GDP growth, as shown in Figure 2.
All the variables had positive relationships with GDP growth except for unemployment, which had a
negative relationship. While the intercept was consistently negative, this could be due to a strong effect
from the Gini coefficient or due to fluctuations in the magnitude of the intercept. The R-squared values
did not fluctuate too widely (with the exception of the estimates related to the dummy variable).
Model 2, our first multiple regression model, included the Gini coefficient and the gross savings
rate.
Model 2: grgdp = 0 + 1Gini + 2gsav + u
 Model 2 in Table 5 shows the regression estimation equation results. In Model 2, both
independent variables were positive and significant at the 1% level. The R2 value was 0.199, which
increased from the R2 value of 0.098 for the simple regression model Model 1.

In the following model, Model 3, the variable fertil, for fertility rate, was added to the preexisting
variables Gini coefficient and gross savings.

Model 3: grgdp = 0 + 1Gini + 2gsav + 3fertil + u

The table yields results that show that fertility rate was also a positive and significant relation to
GDP growth. The Gini coefficient and the gross savings rate retained significance in Model 3.The Gini
coefficient is now significant at the 5% level, while gross savings and fertility rate were significant at the
1% level. The R2 value increased to 0.385, which means that the variables explain 38.5% of the variation
in grgdp. This makes sense because as we control for more variables, the larger R2 will be.
In Model 4 we added the unemployment rate, which although proved to be significant alongside
the other variables, had a negative relationship with GDP growth. The Gini coefficient maintained
significance at the 5% level, like the unemployment rate, while gross savings and fertility rate remained
significant at the 1% level. The R2 value for this model increased once more to 0.423.

Model 4: grgdp = 0 + 1Gini + 2gsav + 3fertil + 4unemp + u

In Model 5 we incorporated the variable educ, which represents mean years of education. The
mean years of education had a negative relationship with GDP growth. This new variable differed from
all the other variables because it was not statistically significant at the 10%, 5%, or 1% levels. The Gini
coefficient maintained significance at the 5% and 10% levels, while gsav, fertil, and unemp all retained
their statistical significance at the 1% level. In addition, the intercept was not statistically significant at
any level in this model, unlike the other models. Therefore, we can conclude from these results that the
Gini coefficient, gross savings, fertility rate, and unemployment all have an impact on GDP growth, while
no conclusions can be made about education. The R2 value rose once more to 0.478, which means that
47.8% of the variation can be explained by the model. Model 5 is the model we chose as our restricted
model after testing for correlation of variables. A value of positive or negative one would be a perfect
correlation while a value of zero is no correlation. These results are shown in Appendix 2 Table A7.
 Model 5: grgdp = 0 + 1Gini + 2gsav + 3fertil + 4unemp + 5educ + u

After we had constructed and analyzed these models, we decided to add a dummy variable to
show the difference between developed and developing countries. This dummy variable, dev, is shown
in Model 6s regression. According to Model 6, compared to the intercept of the developing countries of
-0.645, the developed countries had an intercept of -3.995. This model also had the highest R2 value of
0.632 and the smallest number of observations. This is much larger than the previous values, but
expected, as increasing the number of variables always increases the R2 value. These differences in
information gathering may be the cause of some of the differences in models. The correlation among
variables with the inclusion of the dummy variable can be found in the Appendix 2, Table A8.

Model 6: grgdp = 0 + 1Gini + 2gsav + 3fertil + 4unemp + 5educ + 6dev + u

Table 5: OLS Regression Estimates for Models 1-6

Dependent Variable grgdp

Independent SLR Model MLR Model MLR Model MLR Model MLR Model MLR Model
Variables 1 2 3 4 5 6

Gini 0.111*** 0.114*** 0.067** 0.073** 0.0885** 0.046

(0.033) (0.317) (0.029) (0.029) (0.035) (0.031)
gsav 0.103*** 0.134*** 0.115*** 0.118*** 0.115***
(0.029) (0.026) (0.027) (0.033) (0.028)
fertil 1.033*** 0.893*** 0.656* 0.655**
(0.193) (0.205) (0.330) (0.28)
unemp -0.098** -0.093* -0.111**
(0.047) (0.051) (0.436)
educ -0.165 0.075
(0.129) (0.119)
dev -3.349***
(0.654)
Intercept -0.307 -2.644* -4.107*** -2.751* -1.376 -0.645
(1.287) (1.402) (1.265) (1.413) (2.422) (2.056)
No. of obs 105 99 99 97 70 70
R-square 0.098 0.199 0.385 0.423 0.478 0.632
The quantities in parentheses are standard errors. *, **, *** denotes significance of coefficients at 10%,
5%, and 1% respectively.

The results yielded from the regressions support our hypothesis that GDP growth and the Gini
coefficient are positively related. Depending on the model used, a one point increase in the Gini
coefficient can result in about an 11% increase in GDP growth. This may be caused by an unequal
distribution of wealth in an economy with income inequality. Essentially, as inequality increases, the
majority of the wealth of the economy is concentrated in the hands of the top percentage of the people.
This can then increase GDP growth through investment. Unsurprisingly, gross savings and the GDP
growth in an economy are positively associated. With a 1% increase in gross savings, there is (depending
on the model) a 10-13% increase in GDP growth. These findings support Shins (2012) and Malinens
(2013) research that an increase in the level of saving in an economy will enhance growth. These results
also support Aghion, Comin, Howitt and Tecu (2009) which states that increased savings may increase
innovation and therefore foreign investment in technology, which in turn would have a positive effect on
the economy. Fertility also has a positive impact on the economy. As the population of a country grows,
more people are added to the labor force and the country is more productive. In fact, a fertile population
of a country signifies health and potential for growth as well. According to our findings, fertility rate is
actually one of the more influential variables of an economys GDP growth. Unemployment,
unsurprisingly, has a negative correlation with GDP growth. An increase in unemployment results in a
decrease in a countrys GDP growth, and vice versa. As unemployment rate increases in a countrys
economy, there are social and economic implications and repercussions. Generally, unemployment is
negatively related to disposable income as well. This results in reduced consumption which will lead to
reduced economic growth. Finally, the statistics show that mean years of education does not have
statistical significance in these models. Interestingly, the correlation between mean years of education
changes from negative to positive when the dummy variable is added. This model including the dummy
variable is something that should be further investigated.
4.3 Statistical Inferences
Looking at the regression models created, we can see which factors have a positive impact on
economic growth and which factors have a negative impact on economic growth. Our models
unanimously demonstrated that the Gini coefficient, the gross savings rate, and fertility rate had a positive
effect on GDP growth, while unemployment and education had a negative correlation with economic
growth (not encompassing the model including the dummy variable). Also, for each regression, two-tailed
t-tests were performed on each variable. The null stated that the coefficient of the variable equaled zero,
and the alternative hypothesis stated that it did not equal zero. The tests were then examined at the 1%,
5%, and 10% significance level. The t-values and p-values that resulted are in the appendix. In
conjunction with the simple regression model, we found that the Gini coefficient was statistically
significant at all three levels, and decreased in significance slightly (5% and 1%) when independent
variables were added for the construction of Models 2-6. The variables with a positive effect on GDP
growth (gross savings and fertility rate) consistently were statistically significant at all three levels, while
unemployment, the negatively correlated variable, was consistently significant at the 5% and 10% levels.
The significance of the intercept varied widely throughout the tests, and so we cannot conclude much
about its statistical significance with our current research results. However, we can conclude that from
this model, all variables had an impact on GDP growth.
Looking at the growth rate of developing countries compared to developed countries, on average,
developing countries had higher GDP growth rate than developed countries. This could be explained by
inequality in those countries. Unequal distribution can result in more economic mobility, especially in
developing countries or countries in the early stages of development (Aghion, Caroli, Garcia-Penalosa
1999). While this may be a possible explanation, it is not something that we have adequate research or
results to back a claim at this point, as it was not our focus for this research. However, our hypothesis was
supported by the simple and multiple regression models performed on the data collected. Unsurprisingly,
gross savings, and fertility had a positive relationship with economic equality, and unemployment and
education had a negative relationship. On average the developing countries had a higher economic growth
regardless of the initial GDP per capita as shown by our data.
4.4 Robustness
All of our explanatory variables proved to be statistically significant when using the t-test.
However, just in case that we had missed any other possible relations among our variables, we conducted
the f-test in order to check whether our control variables had an impact on GDP growth. The null
hypothesis and alternative hypothesis did not change: the null stated that the coefficients on the control
variables equaled zero and the alternative hypothesis stated that at least one was not equal to zero. For our
f-test, the restricted model was Model 5, which put out a sum of squared residuals (SSRR) of 356.810. Our
unrestricted model was Model 6, the model including the dummy variable, which yielded a sum of
squared residuals (SSR) of 251.894. With these findings, we calculated the f-statistic for both the
restricted and unrestricted models using the equation F = [(SSRR - SSRUR)/q]/[SSRUR/(n-k-1)], where q
represents the number of restrictions imposed on the restricted model (4 for our model). The degrees of
freedom are represented by (n-k-1) in the unrestricted model, which is 63 for this model. This equation
yields an f-statistic of 18.01. This is a fairly large f-statistic, so we can conclude that our variables may be
jointly significant at a very low level. In other words, the model is useful in predicting GDP growth.
Thus, although education is not individually significant, it has a joint effect on GDP growth in
conjunction with the other control variables.
5. Conclusion
In general, considering all the variables, the OLS regression models show a positive relationship
between income inequality and economic growth. For our restricted model, we found that a one point
increase in the Gini coefficient leads to an increase in GDP growth by 8.85%, and for the unrestricted
model, a one point increase in the Gini coefficient leads to an increase in GDP growth by 4.56%. Another
interesting result was that the Gini coefficient and gross savings rate were statistically significant at the
!% level, while fertility rate and unemployment were statistically significant at at least the 10% level.
Education, however, was not statistically significant in this study, and therefore we needed to conduct an
f-test to determine if mean years of education had a joint impact on economic growth with the other
control variables. The results of the f-test revealed to us that education still had an impact on GDP growth
in conjunction with the other variables despite being individually insignificant.
We decided for our purposes, Model 5 was the best representation of our result for this study.
Further research needs to be conducted to investigate the negative relationship between mean years of
education and economic growth. In addition, more variables should be added and studied to see if there
are other forces that influence economic growth in a country. Finally, further study into how country
development can or may impact the GDP growth should be investigated.
In conclusion, we recognize that income inequality is a topic that is frequently discussed right
now, especially with the Presidential election of 2016. Arguments for increase of minimum wage to
reduce income inequality is a topic brought up frequently. Even according to existing literature, there are
examples of how income inequality can improve the conditions of the inhabitants of a country as well as
data that shows income inequality is related to many economic and social dilemmas that a country may
face. Therefore, a country should seek to attain a good balance between income inequality and the
repercussions so as to achieve the most optimal economic growth.
References (APA format)

Aghion, P., Caroli, E., & Garca-Pealosa, C. (1999). Inequality and economic growth: The perspective of
the new growth theories. Journal of Economic Literature, 37(4), 16151660.

Aghion, Philippe, Diego Comin, Peter Howitt, and Isabel Tecu, 2009. When Does Domestic Saving
Matter for Economic Growth?(Harvard Business School, Harvard Business School Working Papers:
09-080).

Alesina, Alberto, and Dani Rodrik, 1994. Distributive Politics and Economic Growth, Quarterly Journal
of Economics 109, 465-490.

Barro, Robert J. Determinants of Economic Growth: A Cross-Country Empirical Study. NBER Working
Paper Series: Economic Fluctuations and Growth 5698 (1996): n.pag. Print.

CHAMBERS, D., & DHONGDE, S. (2011). A NON-PARAMETRIC MEASURE OF POVERTY

ELASTICITY. Review of Income and Wealth, 57(4), 683703.

Corak, Miles, 2013. Income Inequality, Equality of Opportunity, and Intergenerational Mobility, Journal
of Economic Perspectives 27, 79-102.

IMF world economic outlook (WEO) update, January 2016: (2016, January 19). Retrieved October 4,
2016, from International Monetary Fund, http://www.imf.org/external/pubs/ft/weo/2016/update/01/

Kaldor, N. (1956). Alternative Theories of Distribution, Review of Economic Studies. 23: 83100.

Li, H., & Zou, H. (1998). Income inequality is not harmful for growth: Theory and evidence. Review of
Development Economics, 2(3), 318334.

Malinen, T. (2012). INEQUALITY AND GROWTH: ANOTHER LOOK WITH A NEW MEASURE
AND METHOD. Journal of International Development, 25(1), 122138.

Mo, P. H. (2000). Income inequality and economic growth. Kyklos, 53(3), 293315.
New York (State). Dept. of Health (1909). Annual Statistical Report of the Department of Health (Vol 29,
Part 1 ed.). New York: New York State Department of Health.

Shin, I. (2012). Income inequality and economic growth. Economic Modelling, 29(5), 20492057.

World Bank, 2010-2011. GDP growth (annual %) (The World Bank).

World Bank, 2010-2011. GINI index (World Bank estimate) (The World Bank).

World Bank, 2010-2011. Gross Savings (% of GDP) (The World Bank).

World Bank, 2010-2011. Total Fertility Rate (births per woman) (The World Bank).

World Bank, 2010-2011. Total Unemployment (% of total labor force) (The World Bank).
Appendices

Appendix 1: List of Countries

Bosnia
and Montene
Andorra Comoros Fiji Indonesia Lao PDR Peru Serbia Tuvalu
Herzegov gro
ina
Afghanist Cabo Philippin South
Belrus France India Lebanon Mongolia Tanzania
an Verde es Sudan
Sao
Papua
Costa Mozambi Tome
Angola Belize Gabon Ireland Liberia New Uganda
Rica que and
Guinea
Principe
Iran,
United Mauritani
Albania Bermuda Cuba Islamic St. Lucia Poland Suriname Ukraine
Kingdom a
Rep
United
Puerto Slovak
Arab Bolivia Cyprus Georgia Iraq Sri Lanka Mauritius Uruguay
Rico Republic
Emirates
Argentin Czech United
Brazil Ghana Iceland Lesotho Malawi Portugal Slovania
a Republic States
Uzbekist
Armenia Barbados Germany Guinea Israel Lithuania Malaysia Paraguay Sweden
an
Antigua Qatar Venuzuel
Luxembo Swazilan
and Bhutan Djibouti Gambia Italy Namibia a, RB
urg d
Barbuda
Guinea-B Romania Seychelle Vietnam
Australia Botswana Dominica Jamaica Latvia Niger
issau s
Central Russian Vanuatu
Austria African Denmark Greece Jordan Macao Nigeria Federatio Chad
Republic n
Dominica Rwanda Samoa
Azerbaija Nicaragu
Canada n Grenada Japan Morocco Togo
n a
Republic
Switzerla Guatemal Kazakhst Netherlan Saudi Yemen,
Burundi Algeria Moldova Thailand
nd a an ds Arabia Rep
Madagas Sudan Tajikista South
Belgium Chile Ecuador Guyana Kenya Norway
car n Africa
Egypt, Hong Kyrgyz Senegal Timor-Le Congo,
Benin China Mexico Nepal
Arab Rep Kong Republic ste Dem Rep
Burkina Cote Cambodi Marshall New Singapor Zambia
Spain Honduras Tonga
Faso d'Ivoire a Islands Zealand e
Solomon Trinidad Zimbabw
Banglade Cameroo Korea, Macedon
Estonia Croatia Oman Islands and e
sh n Rep ia, FYR
Tobago
Congo, Sierra
Bulgaria Ethiopia Haiti Kosovo Mali Pakistan Tunisia
Rep Leone
El
Bahrain Colombia Finland Hungary Kuwait Malta Panama Turkey
Salvador
Appendix 2: STATA Regression Outputs

Table A1: Model 1 regression

Table A2: Model 2 regression

Table A3: Model 3 regression

Table A4: Model 4 regression

Table A5: Model 5 regression

Table A6: Model 6 (with dummy variable) regression

Table A7: Correlation Table Stata Output

Table A8: Correlation Table Stata Output (with dummy variable)

Summary Tables of Variables: