Regression with Kuznets ratios (top/bottom decile) (base + school)

WAS KUZNETS RIGHT? A STUDENTS’ VIEW
This text is the collective effort of a class (International Economics advanced ac. year 2011-2012). Even if individual names are indicated for specific sections,
most of the work done (related to coordination problems: e.g, data collection)
remain “invisible”.
ABSTRACT
(authors: Lucia Bozzi, Laura Alzapiedi, Valeria Brunetti, Francesca Giuliani)
The purpose of this paper is to examine the existence of a Kuznets Curve relationship between
per capita income and inequality, using data of several countries from 1990 to 2009, coming from
the World Databank and the UNU-Wider dataset. The main variables determining inequality, which
this project is focused on, are the following:

development index (per capita income);

openness (FDI inflow and outflow, trade, communication and computer…);

demography (age dependency, pop. growth);

population structure (female labour force, urban population growth, urban
population share);

finance (domestic credit, M2);

macro stability (inflation rate, exchange rate $).
Firstly, the work involves a brief explanation of the Kuznets hypothesis and a comprehensive
summarize on the main conclusions reached by other authors that have already tackled this matter.
Afterwards, the technique through which the already mentioned variables have been chosen, their
meaning and their functionality will be clearly discussed. In order to estimate the Kuznets curve, the
project uses some descriptive statistical tools, such as the correlation and regression models.
INTRODUCTION
(authors: ? Kamila, Kia, Petra, Alberto ?)
Nowadays inequality and world income distribution are in the centre of public debate but also
concerning economists and researchers. Level of inequality changes throughout the years and
depends on the per capita income of every country as well as the pace of their development. In
order to diminish the problems connected with divergence in the world, it is essential to know
where they come from and what they are truly about.
The aim of this paper is to estimate the econometric relationship between index of inequality and
per capita income of selected countries during the period of 1990-2009. The project is based on
building a Kuznets curve by using multiple variables described later on.
The explanatory part of the project consists of the theory of the Kuznets hypothesis and the review
of the literature concerning the subject, including several papers by writers such as R. Barro and J.
Lee. Different aspects of inequality and development are discussed and summarized.
The following part is devoted to explaining the choice of variables used in the project. In this
section data sources are enlisted and every variable is described in details, including the methods of
gathering the information. The variables were chosen according to several predetermined indexes
which in this case are development index, openness, demography, structure, finance and macro
stability. Data retrieved for the purpose of the project was collected from the World Data Bank.
The last sections describe methodology of processing data and statistical operations used to analyze
it. The main focus is on the regression model which is explained in detail later. The observations
and results of the data collected are presented later as well.
The Kuznets hypothesis
(Authors: Jasmine, Orsi, Marta, Diego)
The economist Simon Kuznets developed the so called Kuznets Curve which is based on the level
of economic development and inequality. For his work he won the Nobel Price1. As you can see in
the following picture, in the y-axis we have income inquality and on the x-axis per capita income.
Pic. Example of the Kuznets Curve
Explanation
The relationship between inequality and economic development may take the form of an inverted“U” .This shape comes from progression in the development of individual countries. According to
the hypothesis at first the inequality increases in the early stages of development, reaches a
maximum at an intermediete level of income and then declines as the country achieves a high level
of per capita income. We can devide the process of development in 3 stages which are developing
economies, turning point income and developed economies. In the next paragraphs we will explain
each of the stages more in detail.
1) Developing economies:
1
Andrew Sharpe, Linkages between economic growth and inequality: introduction and overview, 2003 p. 9-10
The majority of people is working in agricultural sector with a low level of productivity which
explains the low level of inequality because of homogenity of wealth. The inequality will rise due to
the fact that people move out of agriculture and passing to industrial sector.
2) Turning point income:
In this point the inequality reaches the maximum level because some people move to industry and
some still remain in agriculture.Moreover, government does not need to invest longer in
infrastructure. People have the opportunity to get education and better jobs. This is because of
increasing per capita income.
3) Developed economies:
In this stage employment rate is high in the industrial sector and the productivity is increasing due
to technological changes. Because of this the wealth is distributed more uniformly. The inequality
in this stage reaches the minimum bacause also people employed in agriculture move now to
industrial sector and earning roughly similar wages . This means that all people have a high level of
income and productivity which explains the decreasing of inequality. Most of people are now
employed in industry and they reach the highest level of per capita income.
Criticism
However, also some critics appeared. Some argue that “U”-shape comes rather from historical
differences between countries. There can also be a different outcome if considering additional
variables.2
To conclude, not all countries may follow Kuznets inverted “U” curve, but the majority of countries
developed along this path.
EMPIRICAL LITERATURE REVIEW
(authors: Bucari and … which others??)
The objective of this section is to recall some previous studies concerning the interconnections
between inequality & others feature of economic development explained by Kuznets using the
2
http://www.ijesar.org/docs/volume3_issue2/kuznets.pdf [11.12.2011]
inverted U curve and leaded by several others scholars. Paukert in 1973 used the Gini index as a
measure of inequality and collected data from 56 countries categorized by GDP per-capita in 1965
usd while Ahluwala sampled data of 60 countries; the populations were divided in quintiles of GNP
per-capita in 1970 usd. In 1993 Anand & Kanbul introduced a crucial problem for the cross-section
models which required some assumptions: pooling together different countries required the
assumption of the existence of the same inequality-income relationship for all the countries. The
first ones who used dummy variable to isolate particular characteristics of some countries were
Lindert & Williamson in 1985. In 1996 Deininger & Squire run their model first pooling all
countries together and then they repeated the experiment introducing some dummy variables: doing
like that they found that the inverted-U largely vanishes and also the coefficients turn to be no more
significant.
Barro in his work entitled “ Inequality and Growth revisited ” used the regression system to
estimate the Kuznets curve considering the period between 1960s and 2000s. He chose the Gini
coefficient, the lowest quintile share and the highest quintile share of the income distribution as
dependent variables of the regression, while the explanatory variables were the log of per capita
GDP, the square of the log of per capita GDP, the openness variable and some dummy variables
like the ones for Latin America and sub – Saharan countries, for former colonies, for the net
income/export. Considering the Gini coefficient as the dependent variable Barro showed that the
effect of the log of per capita GDP on the Gini coefficient was positive while the square of the log
of per capita GDP had a negative impact on this dependent variable. The consequences of this was
the Kuznets curve.
In the paper of Jong-Eun Lee, the author attempted to investigate the impact of globalization on the
income inequality in the European Union (EU) from 1951–1992 applying the Kuznets formula.
According to Lee, income inequality is a multifaceted concept and in this paper he expanded the
typical Kuznets-type quadratics as:
GINI it   it  y it  yit2   it tradeit   ( FDI ) it   ( X ) it   it
where X is other control variables and εit is the disturbance term.
Globalization effect is characterized by two variables, trade and FDI; trade is measured by (export +
import)/GDP, and FDI is defined by the value of (net FDI inflow/gross fixed capital formation
×100%). For Lee those were major channels of introducing skill-biased technology and thus had an
implication on income distribution by altering the relative demand for skilled and unskilled labours.
From the study it came out that each old member country was found to have already passed the
turning point of their Kuznets curve. Lee assumed that, given the current position of each member
country on the Kuznets curve, growth must accompany globalization in Europe not merely for
growth in itself, but also for controlling inequality.
The authors of the paper “A semi-parametric partially linear investigation of the Kuznets’
hypothesis” applied a semi-parametric partially linear regression to investigate the existence of the
inverted-U relationship between inequality and development, finding evidence in support of an
inverted-U curve and confirming the validity of the Kuznets’ hypothesis for a large sample of
countries. Using different model specifications and alternative inequality measures, in particular,
the key variables in the parametric regression, the Authors found considerable support for the
Kuznets’ hypothesis and also the graphical outcomes showed clear evidence of an inverted-U
relationship between income inequality and per capita As a robustness check, they used the ratio of
incomes between the most developed and the least developed regions within a country and repeated
the test using this alternative inequality measure, moreover they estimated the unknown relationship
between this new measure and GDP per capita using the semi-parametric approach, with or without
the inclusion of other control variables; in all the parametric cases, the signs and significances of the
crucial coefficients remained virtually unchanged across alternative conditioning sets. The authors
Matthew Higgins and Jeffrey G. Williamson explored three hypotheses regarding sources of
inequality: (1) the effect of demographic conditions (cohort size), (2) the effect of development
(Kuznets Curve), and (3) the effect of globalization (degree of openness in trade and migration).
The analysis reported strong evidence that inequality follows the Kuznets’ inverted-U pattern,
tending to rise as a country passes through the early stages of development, and tending to fall as a
country passes through the later stages. This work differed from most previous studies of the
Kuznets hypothesis, as it examined the inequality-development relationship conditional on other
variables. The authors extended their analysis to clarify its implications for the recent debate about
rising wage inequality in the United States and other OECD economies in the 1980s. They found
little support for the hypothesis that a policy commitment to globalization has an impact on
inequality.
METHODOLOGY AND DATA CHOICE
(Authors: Belardinelli, Mineo, Belardinelli, Mancinelli)
METHODOLOGY
Regression model is used to predict one variable from one or more other variables. Its aim is to
explain how the typical value of the dependent variable changes when any one of the independent
variables is varied, while the other independent variables are held fixed.
This model is expressed by the following formula:
Y = β0 + β1X1 + β2X2 + β3X3 +……+ βnXn + ε
where Y represents the Gini index, β is the coefficient of the independent variable X and ε is the
random variable representing the error.
To develop our project we created a sample of countries whose population is higher than 300000
inhabitants and for each country we have calculated some specific variables to describe the general
level of development of the country. Among them, the countries without sufficient information
about the index of inequality have been deleted; from the resultant subsample we obtained a data
matrix where we estimated the regression model through the pooled O.L.S. method.
In order to apply this method we have chosen the most promising variables which are the ones with
higher correlation with Gini index 3. Nevertheless in some cases we constructed smaller subsamples
in order to retain information on relevant variables4 that were lacking for several years and/or
countries.
DATA CHOICE
We chose the variables to estimate an econometric relationship between indices of inequality and
per capita income and to estimate the Kuznets curve. We took them from the website World
Databank.
3
Variables: Icome and Inequality; Openess; State; Demography; Employment structure; Education; Unemployment;
Urbanization; Resource endowment – oil production; Credit; Macroeconomic Environment; Insitutional
4
State: Tax revenue as % GDP, Expense %, Social contribute. %, Central government debt % GDP
Resource endowment: Oil rents %GDP; Fuel exports %; Ores and metals exports
The first variable of the matrix is the Gini index that measures the extent to which the distribution
of income among individuals or households within an economy deviates from a perfectly equal
distribution. A Lorenz curve plots the cumulative percentages of total income received against the
cumulative number of recipients, starting with the poorest individual or household. The Gini index
measures the area between the Lorenz curve and a hypothetical line of absolute equality, expressed
as a percentage of the maximum area under the line. Thus a Gini index of 0 represents perfect
equality, while an index of 100 implies perfect inequality.
The other variables have been grouped in 9 broad categories that explain the main characteristics of
the economic structure of countries.
The broad categories are:
1) Country’s openness
2) Development
3) Demographic structure
4) Employment and unemployment structure
5) Education
6) Oil production
7) Financial sector
8) Macro stability
9) Role of the state
Don’t belong to these categories the 2 dummy variables that we put in the last two columns of the
matrix: they should catch Latin America and Sub-saharan Africa specific behaviour.
A dummy variable is one that takes the values 0 or 1 to indicate the absence or presence of some
categorical effect that may be expected to shift the outcome.
DESCRIPTIVE STATISTICS
(authors: Cecarini, Giovannangeli, Giacomini, Hysa)
In the following, we present the five highest and lowest values for our base variables: Gini index,
top and bottom income deciles, per capita income
GINI
Country name
Azerbaijan
Romania
Year
2005
1991
5 Lowest
GINI index
Country name
16,83 Namibia
20,5 South Africa
1993
5 Highest
GINI index
74,33
2004
67,4
Year
Czech Republic
1991
Czech Republic
Belarus
1992
1993
Comoros
2004
Maldives
21,6 Lesotho
1998
1994
21,2
21,4
64,3
63,27
63,16
According to the data, we realized that countries with the highest GINI index are mainly African.
On the other hand, we figured out that ex URSS countries have a lower value of GINI index .
TOP DECILE
Country name
Azerbaijan
Belarus
Austria
Austria
Austria
Year
2005
1993
1997
1998
2001
5 Lowest
Income share held
by highest 10%
17,69
19,4
20
20
20
Country name
Year
Namibia
Namibia
1993
1993
2006
2004
2003
South Africa
Comoros
Ecuador
5 Highest
income share held
by highest 10%
65,00
65,00
57,54
55,19
52,6
The table above confirm the data of the first table. The difference is that on one side we have the
addiction of one European country like Austria, while on the other side we notice one new country
from South America like Ecuador.
BOTTOM DECILE
Country name
Lesotho
Paraguay
Year
Maldives
5 Lowest
income share held
by lowest 10%
Country name
1994
0,46 Azerbaijan
1998
0,49 Belarus
1998
2005
2007
Bolivia
Honduras
Year
0,49 Japan
5 Highest
income share held
by lowest 10%
2005
6,21
1993
4,87
1993
1993
1997
0,5 Czech Republic
0,53 Pakistan
4,78
4,64
4,52
The general trend of the data is confirmed also on this matrix because of the presence of Latin
countries on one side and ex URSS countries with the addiction of Japan and Pakistan on the other
one.
PER CAPITA INCOME
Country name
Congo, Dem.
Rep.
Burundi
5 Lowest
GDP per capita,
PPP (constant
2005 international
$)
Year
Country name
5 Highest
GDP per capita,
PPP (constant 2005
international $)
Year
2006
279,4144276 Luxembourg
2003
2006
353,012825 Luxembourg
2004
74113,99203
73848,95971
Liberia
2007
365,08 Qatar
2007
72813,88333
Liberia
2007
365,08 Luxembourg
2001
1998
368,9714083 Luxembourg
2000
68319,63721
65800,21708
Burundi
Countries with highest GDP per capita are characterized by a low population and an high level of
development as well as natural resources. On the other hand, there are African countries which are
still involved in developing issues and afflicted by instability of the governments (civil wars,
dictatorship etc.)
RESULTS AND CONCLUSION
(authors: Lucia Bozzi, Laura Alzapiedi, Valeria Brunetti, Francesca Giuliani)
The unconditional and conditional linear regression of Gini index illustrates that only few
variables are significant, because it is possible to comment those variables which have in relative
terms a t-value greater than 2 and -2.
The unbiased variables which are listed in the already mentioned table, are:
 GDP per capita, that shows an inverse correlation to the Gini index;
 Population growth (annual %), that shows a direct correlation to the Gini index;
 Latin America (dummy variable);
 Sub-Saharan Africa (dummy variable).
A particular attention has to be deserved to the dummy variables because Latin America and
Sub-Saharan Africa are characterized by a specific political background which do not fuel the
welfare system.
R squared has a value of 0.6, which means that the model is totally correct.
The analysis of the tables which compare the unconditional Gini index with other variables
such as tax revenues % of GDP, expenses % of GDP, social contributions % of revenues, central
government debts, total % of GDP and natural resources, showed out some uncommon results.
The first technical inconsistency comes from the elaboration of dataset, having considered
different range of values according to their correlation.
The first table that takes into consideration State variables do not contain the “Money and
quasi money (M2) as % of GDP” variable, and therefore the regression of this values is not
comparable with the one of the unconditional linear regression.
On one hand, the most meaningful outcome of this study suggests that variables concerning
the role of the State are of great importance. As far as tax revenues, expenses and central
government debts are concerned, the more a country is liable to taxes and the higher the amount of
its business trades is, the lower the inequality index is. On the other hand, social contribution has a
different impact, because it is calculated in reference to the percentage of government revenues.
The last table is related on the impact natural resources have on the Gini index; while if oil
rents increase, the inequality diminishes, a boost of ores and metal exports makes the Gini index
increase.
Results of regression of Kuznets ratio are not evaluable because all the variables, except the
dummy ones, aren’t significant; that might be because the valued sample is too small.
In conclusion, we were not able to compose a model that could have strengthened the
hypothesis of Kuznets.
SHORT BIBLIOGRAPHY
(Ray Debraj citations - to be completed)
Barro R. (2008), Inequality and growth revisited, Asian Development Bank, working papers series
on regional economic integration n. 11
Lin S., Huang H., Weng H. (2006), A semi-parametric partially linear investigation of the Kuznets'
hypothesis, Journal of Comparative Economics Volume 34, Issue 3, September 2006, Pages 634647
Higgins M., Williamson J. (2009), Explaining Inequality the World Round: Cohort Size, Kuznets Curves,
and Openness, NBER Working Paper No. 7224, July
Lee J. (2006), Inequality and globalization in Europe, Journal of Policy Modeling n. 28, pp. 791-796
APPENDIXES
VARIABLES LIST
The first variable is the Gini index as a measure of inequality.
We grouped the other variables in 9 categories:
1) Country’s openness:
- foreign direct investment inflows/ outflows;
- trade (% of GDP);
- ores and metal exports;
- international migrant stock (% of population).
2) Development:
- GDP per capita PPP (constant 2005 international $);
- Income share held by highest/lowest 10%.
3) Demographic structure:
- Age dependency;
- Population growth (annual %);
- Urban population growth (annual %);
- Urban population (% of total).
4) Employment and unemployment structure:
- Labor force, female (% of total labor force);
- Employment in services (% of total employment);
- Employers, total (% of employment);
- Unemployment, total (% of total labor force);
5) Education:
- Public spending on education, total (% of GDP);
- Progression to secondary school (%);
- School enrollment, mean primary secondary (% net);
Literacy rate, adult total (% of people ages 15 and above).
6) Oil production:
- Oil rents (% of GDP);
- Fuel exports (% of merchandise exports);
7) Financial sector:
- Domestic credit to private sector (% of GDP);
- Money and quasi money (M2) as % of GDP;
8) Macro stability:
- Inflation, GDP deflator (annual %);
- Official exchange rate (LCU per US$, period average);
- Tax revenue (% of GDP);
- Expense (% of GDP);
- Social contributions (% of revenue);
- Central government debt, total (% of GDP)
9) Role of the state:
- CPIA transparency, accountability, and corruption in the public sector rating (1=low to
6=high);
- CPIA quality of public administration rating (1=low to 6=high);
- CPIA financial sector rating (1=low to 6=high);
- CPIA property rights and rule-based governance rating (1=low to 6=high);
- CPIA policies for social inclusion/equity cluster average (1=low to 6=high).
The last two columns of the matrix are 2 dummy variables: Latin America and Sub-saharan Africa.
Descriptive Statistics
Minimum Maximum
Mean
GINI index
16,83
74,33
39,115
628
Income share held by highest 10%
17,69
65
32,466
8371
Income share held by lowest 10%
0,46
6,21
2,4393
9418
279,4144
74113,99
11460,
5877
0
5,49E+09
275944
122
Foreign direct investment, net inflows (% of GDP)
-6,94567
564,9163
7,3987
2357
Foreign direct investment, net outflows (% of GDP)
-1,41282
344,5264
14,015
1133
Trade (% of GDP)
0,321968
316,6345
76,579
8083
International migrant stock (% of population)
0,045274
43,10309
9,5597
8167
0,23042
48,40184
16,653
4704
0,234025
72,31654
27,315
6405
Social contributions (% of revenue)
0
60,22698
21,670
5841
Central government debt, total (% of GDP)
0
5,18E+11
456122
6156
Age dependency ratio (% of working-age population)
21,62633
121,3429
61,133
8452
Population growth (annual %)
-2,85097
18,58832
1,0531
5988
Labor force, female (% of total labor force)
13,19451
52,80833
41,750
2193
0,5
79
49,763
5516
0
99,76501
16,983
1898
0,3
99,7672
46,664
0586
GDP per capita, PPP (constant 2005 international $)
Squared GDP per capita, PPP (constant 2005 international $)
Tax revenue (% of GDP)
Expense (% of GDP)
Employment in services (% of total employment)
Employers, total (% of employment)
Literacy rate, adult total (% of people ages 15 and above)
Public spending on education, total (% of GDP)
1,28822
99,43074
17,789
5598
Progression to secondary school (%)
2,60753
188,9822
85,516
0766
0
188,98
60,803
9653
0,5
59,5
9,9241
914
-2,9683
18,67207
1,6387
5961
6,66
100
59,477
2396
Oil rents (% of GDP)
0
84,05463
2,4759
7865
Fuel exports (% of merchandise exports)
0
99,73948
10,294
9669
9,25E-05
88,81229
7,2314
2688
Domestic credit to private sector (% of GDP)
0
319,4676
45,604
3292
Money and quasi money (M2) as % of GDP
0
636,5102
48,947
7962
Inflation, GDP deflator (annual %)
-23,4789
6836,881
43,366
86
Official exchange rate (LCU per US$, period average)
2,96E-05
16302,25
491,70
0972
CPIA transparency, accountability, and corruption in the public sector rating
(1=low to 6=high)
2
4
2,8454
5455
CPIA quality of public administration rating (1=low to 6=high)
2
4
3,1
CPIA financial sector rating (1=low to 6=high)
2
4,5
3,2727
2727
CPIA property rights and rule-based governance rating (1=low to 6=high)
1,5
3,5
2,9090
9091
CPIA policies for social inclusion/equity cluster average (1=low to 6=high)
2,2
4,2
3,4076
9231
School enrollment, mean primary secondary (% net)
Unemployment, total (% of total labor force)
Urban population growth (annual %)
Urban population (% of total)
Ores and metals exports (% of merchandise exports)
Detailed Results – tables
Regression with Gini index (base)
(authors: D’Alesio, Carestia, Tombolini, Andreucci)
Unconditional regression
Intercept
GDP per capita
Squared GDP per capita
R squared
F
Coefficients
t-value
44,38994295
51,23129799
-0,00048842
-4,399897443
4,32908E-09
1,912215145
0,091089069
20,8453059
Conditional regression
Intercept
GDP per capita
Squared GDP per capita
Trade (% of GDP)
Age dependency ratio (% of working-age population)
Population growth (annual %)
Labor force, female (% of total labor force)
Urban population growth (annual %)
Money and quasi money (M2) as % of GDP
LA
SSA
R squared
F
Coefficients
38,45313881
-0,000205011
2,7785E-09
-0,010710175
-0,045425096
1,413881474
0,006313543
0,290770678
-0,007130703
15,61234313
9,581151173
0,641229257
72,92164773
t-value
9,995997112
-2,190328742
1,478805129
-1,249506661
-1,168065557
2,018507212
0,11495379
0,714718831
-0,936022496
19,14550351
6,308765523
Regression with Kuznets ratios (top/bottom decile) (base +
school)
(Authors: Carbonari, Cecchi, Magni, Bigi)
Unconditional Regression
Intercept
GDP per capita
Squared GDP per capita
R Squared
F
Coeff.
22,73
0,00
0,00
t
9,54
-0,77
-0,38
0,04
4,18
Conditional regression
Coeff.
Intercept
GDP per capita
Squared GDP per capita
Urb. Pop.
School Enr.
Lab. Female %
Pop Growth
Age dep.
LA
SSA
R Squared
F
17,12
-0,000071
0,0
1,24
-0,02
-0,11
-0,90
-0,03
24,19
19,44
0,52
26,22
t
1,65
-0,18
0,15
1,20
-0,77
-0,75
-0,48
-0,26
11,08
4,87
Regressions base + State variables - GINI
(authors: Blomquist, Aquilanti, Piancatelli, De Lellis)
UNCONDITIONAL REGRESSION
Intercept
GDP per capita
Squared GDP per capita
R^2
F
Coefficients
t-value
43,90214668
24,6405
-0,000540459
-3,14821
5,00573E-09
1,875391
0,175115
9,128482
CONDITIONAL REGRESSION (Tax revenue % of GDP)
Intercept
GDP per capita
Squared GDP per capita
Trade (% of GDP)
Tax Revenue (% of GDP)
Age dependency ratio (% of working-age
population)
Population growth (annual %)
Labor force, female (% of total labor force)
Urban population growth (annual %)
Latin America
Sub-Saharan Africa
R^2
F
Coefficients t-value
44,85388
5,83695
-0,00039
-3,514429
4,62E-09
2,7666
-0,0386
-3,507408
-0,48753
-4,22679
-0,02632
-0,322038
2,181218
0,14562
0,617961
10,59667
17,11924
1,417855
1,147735
0,61875
6,133907
3,407775
0,782142
28,00311
CONDITIONAL REGRESSION (Expense % of GDP)
Intercept
GDP per capita
Squared GDP per capita
Trade (% of GDP)
Expense (% of GDP)
Age dependency ratio (% of working-age
population)
Population growth (annual %)
Labor force, female (% of total labor force)
Urban population growth (annual %)
Latin America
Sub-Saharan Africa
R^2
t- value
Coefficients t-value
53,35565
6,1694
-0,00025
-2,024893
2,69E-09
1,485494
-0,03037
-2,493915
-0,33593
-4,073163
-0,07443
-0,867282
0,758625
-0,00042
0,536005
11,21235
17,93739
0,779205
27,5269
0,489317
-0,003107
0,532388
6,295739
3,503296
CONDITIONAL REGRESSION (Central government debt, total, % of GDP)
Intercept
GDP per capita
Squared GDP per capita
Trade (% of GDP)
Central government debt, total (% of GDP)
Age dependency ratio (% of working-age
population)
Population growth (annual %)
Labor force, female (% of total labor force)
Urban population growth (annual %)
Latin America
Sub-Saharan Africa
R^2
F
Coefficients
26,24724
-0,00035
3,73E-09
-0,06701
-3,3E-11
0,072177
t-value
3,626956
-3,207675
2,282291
-6,917981
-5,060542
0,938792
4,661294
0,348829
-0,26431
7,775967
7,344405
2,91457
2,767185
-0,268674
4,734054
1,537035
0,798424
30,89502
CONDITIONAL REGRESSION (Social contributions % of revenue)
Intercept
GDP per capita
Squared GDP per capita
Trade (% of GDP)
Social contributions (% of GDP)
Age dependency ratio (% of working-age
population)
Population growth (annual %)
Labor force, female (% of total labor force)
Urban population growth (annual %)
Latin America
Sub-Saharan Africa
R^2
F
Coefficients
33,3707631
-0,00053
6,2E-09
-0,06074
0,041091
0,057265
t-value
3,93928
-3,6087
2,989277
-5,282209
0,59312
0,640407
1,572879
0,196182
1,079273
8,943667
11,77218
0,926418
1,384775
0,917335
4,773007
2,178017
0,733444
21,46215
Regressions base + natural reosurces variables - GINI
Authors (Pelatelli, Sottanelli, Mendes, Ippodimonte)
Coefficients
Intercept
GDP per capita
Squared GDP per capita
T value
7.919669
30.25263
-9.77E-05
-0.9601498
4.097E-10
0.24790182
Age dependency ratio (% of working-age
population)
0.0274103
0.5508649
Population growth (annual %)
Urban population growth (annual %)
1.0969966
1.2125582
0.729391
1.135434
Urban population (% of total)
0.0312
0.9608
Oil rents (% of GDP)
Fuel exports (% of merchandise exports)
Ores and metals exports (% of
merchandise exports)
Official exchange rate (LCU per US$, period
average)
-0.207
-2.069
0.02856
0.77749
0.0849
1.9092
3.1E-05
0.1061
LA
SSA
14.457681
13.7354528
9.4824677
5.1800235
R2
F
0.6976759
57.115776