GMM Dynamic Panel Data Analysis that examine the linkages between macroeconomic factors and income inequality and income distribution of the Asian countries
Demographic transition and the rise of wealth inequality
Impact of Macroeconomic Factors on Income Inequality and Income Distribution in Asian Countries
1. ADBI Workshop on Structural Transformation and Inclusive Growth
20-21 September 2016
Asian Development Bank Institute, Tokyo
Japan
Ravindra Deyshappriya
Research Director
Lakshman Kadirgamar Institute
Sri Lanka
ravindra@lki.lk
ravipdn@yahoo.com
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 1
2. 9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 2
Outline of the Presentation
Introduction
Trends of Inequality in Asia
Matters arising from
reviewing the literature
Methods & Results
Conclusion &
Recommendations
3. Introduction – Background of the Study
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 3
• In terms of growth dynamics and poverty reduction mechanism, the Asian
region has achieved this in a remarkable degree, more than any other
region in the world.
• However, income inequality across the Asian region is increasing and the
inequality of more than half of the Asian countries is above the average
inequality of Asia.
• Inequality reduces the pace at which growth translates in to poverty
reduction as well (Bourguignon, 2004; Kakwani, 1993)
• According to Zhuang et al (2014), factors (technological improvements,
market-oriented reforms and globalization) that drive Asia’s accelerated
economic growth have themselves caused inequality.
• There is a lack of updated scholarly works on macro-inequality linkages,
and even existing knowledge does not provide a consensus on the linkages.
• Methodological issues of existing literature.
4. Introduction – Objectives of the Study
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 4
1. Examine the key determinants of income inequality of the Asian
Countries.
Observe how;
1. Macroeconomic factors
2. Political Economic factors
3. Demographic factors
affect income inequality in Asian countries
2. Analyze the determinants of quantile-wise income distribution of
Asian countries.
5. Trends of Inequality in Asia
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 5
• The average Gini coefficient of the Asian region has increased from 34.5 in
the 1980s to 38.3 by the 2000s
• In the 2000s, the inequality of 16 countries out of the selected 33 Asian
countries is higher than that of the average inequality of the Asian region
(38.3).
6. Trends of Inequality in Asia
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 6
• Lower middle income countries have a remarkably high inequality growth
rate followed by high income countries.
• The majority of the upper middle income countries have negative growth
rate in the Gini index, except, notably, China.
-5 0 5 10 15 20 -20 0 20 40
South Korea
Singapore
Russian
Japan
Israel
Cyprus
Tajikistan
Nepal
Maldives
Cambodia
Bangladesh
Yemen
Vietnam
Uzbekistan
Sri Lanka
Philippines
Pakistan
Mongolia
Latvia
Indonesia
India
Georgia
Turkmenistan
Turkey
Thailand
Malaysia
Lebanon
Kazakhstan
Jordan
Iran
China
Azerbaijan
High Income Low Income
Lower Middle Income Upper Middle Income
Growth Rate of Gini Index 1990s-2000s
7. Matters arising from reviewing the literature
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 7
• Contradictory findings.
• The well-known Kuznet (1955) argument has been questioned by
Bruno et al (1996), Fishlow (1995) and Deininger and Squire (1997).
• Calderon and Serven (2004) and Chatterjee and Turnovsky (2012)
arrived at contradictory findings on government expenditure.
• Kraay (2004) argued that international trade benefited the poor
and reduced inequality. However, Barro (2000), Bourguignon and
Morrisson (1990) and Milanovic (2005) provided counter arguments.
• Have considered only very limited macroeconomic factors.
• Inflation and exchange rate (Bulir and Gulde, 1995)
• government debt (You and Dutt, 1996).
8. Matters Arising from Reviewing the Literature
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 8
• Methodological Issues
• Time Series analysis. Maestri and Roventini (2012)
• Cross Sectional analysis. Sarel (1997)
• Exclusion of factors other than Macroeconomic variables. Maestri
and Roventini (2012)
• Suggestions for future researchers by Sare (1997)
• Include fiscal policy variables and demographic variables.
• Employ panel data analysis.
This study attempts to overcome the issues attached to the literature by
incorporating the suggestions of Sare (1997) by using a large number of
countries (33) over 1990-2013
9. Methodology – Sample Countries
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 9
• Lbhlblbnpn
33 Countries
(1990 – 2013)
Armenia Malaysia
Azerbaijan Maldives
Bangladesh Mongolia
Cambodia Nepal
China Pakistan
Cyprus Philippines
Georgia Russia
India Singapore
Indonesia Sri Lanka
Iran Tajikistan
Israel Thailand
Japan Turkey
Jordan Turkmenistan
Kazakhstan Uzbekistan
South Korea Vietnam
Latvia Yemen
Lebanon
10. 9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 10
Methodology – Variables
Variable Name Variable used for the Study Data Source
Dependent Variables
Inequality Gini Index (WIID)
Income Distribution Income Shares owned by Quantiles (WIID)
Independent Variables – Macroeconomic Variables
National Production Gross Domestic Product World Bank Data
Investment Capital Formation World Bank Data
Changes in Price Level Inflation World Bank Data
Unemployment Unemployment World Bank Data
Trade Terms of Trade World Bank Data
Debt Level Government Debt as Percentage of GDP World Bank Data
Independent Variables - Political Economy Variables
Corruption Corruption Perception Index Transparency International
Political Instability Political Risk Index www.prsgroup.com
Development Assistance Official Development Assistance World Bank Data
Independent Variables - Demographic Variables
Education Gross Enrollment Ratio Secondary Education World Bank Data
Labor Force Participation Labor Force Participation World Bank Data
Population Population Growth Rate World Bank Data
11. Methodology – Empirical Models
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 11
• Model 01 – Measures the impact of macroeconomic factors on
income inequality
• Model 02 – Measures the impact of macroeconomic factors on
income distribution among income quantiles
• 𝑋 - Vector of Macroeconomic Factors
• 𝑌 – Vector of Political Economy Factors
• 𝑍 – Vector of Demographic Factors
𝐺𝑅𝐺𝑖𝑛𝑖𝑖,𝑡 = 𝛽1′𝑋𝑖,𝑡 + 𝛽2′𝑌𝑖,𝑡 + 𝛽2′𝑍𝑖,𝑡 + 𝛿1,𝑖 + 𝜖1𝑖,𝑡
(𝑄1)𝑖,𝑡…….(𝑄5)𝑖,𝑡= 𝛽1′𝑋𝑖,𝑡 + 𝛽2′𝑌𝑖,𝑡 + 𝛽2′𝑍𝑖,𝑡 + 𝛿1,𝑖 + 𝜖1𝑖,𝑡
12. Methodology – Analytical Techniques
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 12
• Employed Generalized Method of Moment (GMM)
estimation technique.
• Overcomes the weakness of cross-country and time
series analysis.
• Applied the first difference of the regression
equation.
• Avoid country specific omitted variable bias.
• Used appropriate instruments.
• Overcome endogeneity issue.
13. Results – Income Inequality
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 13
Factors that widen inequality
• Increase in lnGDP
• Higher Inflation
• Higher Unemployment
• Higher Terms of Trade
• Higher Political Risk
Factors that reduce inequality
• Increase in the Square of
lnGDP
• Higher Official Development
Assistance
• Higher Education
• Higher Labour Force
Participation
14. Results – Income Distribution
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 14
• The initial increase in GDP
redistributes income from the
bottom 20% of people to the
middle class and the richest
group.
• Further increases in GDP
redistribute the income from
the top 20% to the middle
income and the poor groups.
• Higher inflation reduces the
income share of the 1st and
2nd quantiles and increases
only the share of the richest.
• Unemployment reduces the
income share of all quantiles,
except the richest.
• Higher education
redistributes the income from
richest quantile to third and
the poorest.
15. Conclusions and Recommendations
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 15
• The descriptive analysis identified that inequality in countries such as Japan,
Russia, Israel, Singapore, Sri Lanka, China, India, Latvia, Bangladesh and
Indonesia has been continuously increasing since the 1990s. In contrast,
countries such as Iran, South Korea, Malaysia, Turkey and Pakistan have
been experiencing declining inequality.
• There is an inverted U shape (parabolic) relationship between GDP and
income inequality which is similar to Kuznets (1955)
• Higher inflation, terms of trade and unemployment increase inequality in
Asian countries. Similarly, political risk may marginally increase income
inequality.
• ODA, education and labour force participation reduce income equality
significantly in Asian countries.
16. Conclusions and Recommendations
9/28/2016 Lakshman Kadirgamar Institute, Sri Lanka 16
• Capital formation, growth rate of debt, corruption and growth rate of
population do not significantly affect income inequality in Asian countries.
• An initial increase in GDP, higher inflation, and unemployment reduce the
income share of relatively poor quantiles while increasing the income share
of the quantiles with relatively higher income.
• Further increases in GDP and higher educational attainment redistributes
income from higher income quantiles to lower income quantiles.
• Higher and long term economic growth is essential for more equal
distribution together with efficient fiscal instruments that ensure fair
allocation of the benefits of growth.
• Education should be promoted while generating more employment
opportunities.
• Price stability and political stability should also be ensured.
Good afternoon all. I am Ravindra from Sri Lanka. Its great pleasure for me to address the distinguish gathering at the ADBI workshop on Structural Transformation and Inclusive Growth. Today, my research focuses on the impacts of macroeconomic factors on income inequality in Asian countries. Moving to the outline of the session…
First, I will provide a brief background information for the study and followed by the trends of inequality in Asia. After that, the key issues attached to the literature will be highlighted. Then methodology and results will be explained and finally the conclusion and possible recommendation will be elaborated. Let me start from the introdcutoin….
It is well-known fact that this century is known as “Asia’s Century” and of course Asia has accomplished significant achievements in terms of economic growth and poverty reduction. This achievements of Asian is in fact higher than any other region in the world. However, income inequality across the Asia has been growing over the time and more than half of Asian countries have income inequality which is higher than the average inequality of Asia.
The growing inequality in Asian is in fact a sever economic issue and empirical studies have proven that growing inequality reduces the pace at which growth translates in to poverty reduction. Apart from that, some studies have observed that Asia’s growth drivers themselves widen the income inequality. For instance, technological improvements, market-oriented reforms and globalization have been recognized as the drivers of the economic growth and these factors themselves increase the income inequality. Consequently, policy makers and up with a trade off whether to stimulate these factors or to discourage. According to this example, it is apparent that there should be a link between income inequality and macroeconomic factors. However, There is a lack of updated scholarly works on macro-inequality linkages, and even existing knowledge does not provide a consensus on the linkages. Apart from that existing literature has critical methodological issues that needed to be addressed through appropriate and updated research.
Having considered about this situation I formed the objective of this research as follows. The first objective is to examine the key determinants of income inequality of Asia, highlighting the impact of macroeconomic, political economy an demographic factors. Secondly, I analyzed how income has distributed among each quantiles and what factors have caused for quantile-wise income distribution of Asian countries.
Taking the trends of inequality in Asia…The average inequality in Asia has been increasing up to 38.3 by 2000s from 34.5 in 1980s. Similarly, inequality of approximately 50% of Asian countries are higher than the average inequality of Asia in 2000s. Especially, courtiers such as China, Sri Lanka, Singapore, Malaysia, Philippines and Russia have significantly higher income inequality in the new millennium.
This illustration shows further analysis of income inequality by categorizing the countries in to four sections based on income per capita. In fact, this graph has considered GR of income inequality. The red colored vertical reference lines indicate the average growth rate of I
Income equality of each category of countries. Lower middle income countries have a remarkably high inequality growth rate followed by high income countries. Especially, in India, Yemen and Indonesia have significantly higher income inequality. In high income countries, Japan has the highest income inequality in Asian region. Conversely, most of the in Upper Middle Income countries have negative growth in income inequality. However, China has dramatically high inequality growth rate which is the highest in the Asia.
After reviewing the literature, I found 4 key points. The first is contradictory findings. Mainly, there is no consensuses on the relationship between economic growth rate and inequality, government expenditure and inequality an trade and inequality.
The second key point is most of the empirical studies have considered limited explanatory variables only.
The third key point is about the methodological issues such as issues attached to time series and cross sectional analyses. Mainly, the findings from time series analyses cannot be generalized and also cross sectional analyses take all heterogeneous individual countries as homogeneous units. Apart from that, some studies have not considered non-macroeconomic factors. Even though, the main focus in on macroeconomic factors and inequality, it is essential to include other relevant non-marco variables in order to observe more robust estimations.
The 4th key point is about the suggestions for future researchers. IMF study has highlighted that future research on this field should incorporate fiscal and demographic factors and at the same time panel data technique should be used.
The study based on 33 Asian countries and mainly the countries were selected based on the availability of the data over the period of 1990-2013.
This section summarizes the data sources for all the variables include in the models. More importantly, data for both inequality and income distribution variables were collected from WIID.
In order to overcome the methodological issues attached to the literature, panel data analysis was employed and the empirical models can be expressed as follows. The first model explains the impacts of macroeconomic factors on income inequality while the second model observes the determinants of income distribution among the each quantiles.
The dynamic panel data analysis along with GMM estimation technique was applied to estimate both empirical models. Moreover, derived the first difference of the regression equation in order to avoid the country specific omitted variables bias and lag independent variables used as the instruments to overcome endogeneity issue.
This section indicates the results on income inequality. According to the table, GDP is one of the crucial factors of income inequality in Asian region. The lnGDP (log of GDP) particularly positively affects the income inequality while the square of lnGDP has a negative effect in all three models with higher levels of statistical significance. This relationship clearly indicates the existence of a parabolic linkage between GDP and Income inequality. Particularly, income inequality initially increases with the increase of GDP and reduces thereafter with further increase of GDP. Thus, this finding is consistent with Kuznets (1955) and Barro (2000).
Apart from that, the study found that inflation, terms of trade, political risk and unemployment increase the income inequality in Asian countries and the results are statistically significant.
In contrast, long term economic expansion, ODA, edu and labor force participation have been identified as the factors that reduce the income inequality.
The goodness of fit of the model was evaluated using the Sargan Test and the Serial Correlation Test. These tests respectively verify the appropriateness of the instruments and non-existence of serial correlation among error terms. The higher p-values attached to these tests clearly indicates acceptance of the null hypotheses that explain over-identified restrictions are valid and error terms are not serially correlated respectively.
This slide summarizes the analysis related to income distribution of Asia. So, this analysis provides a better understanding on how macroeconomic factors affect the income distribution of Asian countries. As the result indicates, an increase in GDP initially may redistribute income from poor people (1st and 2nd quantiles) to middle class (3rd and 4th quantiles) or richest group (5th quantile). However, further increases in GDP (considering lnGDP2) decreases the income share of the richest group and increases the income shares of all other quantiles. Education also has similar impact on income share of the quantiles and hence both education and further increases in GDP redistribute the income from richest group to middle and poor income groups.
Apart from that, higher inflation redues the income share of 1st and 2nd quantiles and increase the share of income only for the richest group. Similarly, benefits of trade also increase only the income share of the richest group. In contrast, unemployment reduces the income share of all quantiles, except the richest quantile.
The goodness of fit of the this model also at a higher level, as the null hypotheses of both sargan test and serial correlation tests cannot be rejected.
Moving to the conclusion…
The descriptive statistics indicated that the average inequality in Asia is increasing and nearly 50% of Asian countries have inequality that is higher than Average level of Asia.
The econometric analysis related to inequality found that there is an inverted U shape relationship between GDP and income inequality in Asia. Further, Higher inflation, terms of trade, political risk and unemployment increase inequality in Asian countries while ODA, education and labour force participation reduce income equality significantly in Asian countries
The econometric analysis focus on income distribution highlighted that An initial increase in GDP, higher inflation, and unemployment reduce the income share of relatively poor quantiles while increasing the income share of the quantiles with relatively higher income. Similarly, Further increases in GDP and higher educational attainment redistributes income from higher income quantiles to lower income quantiles.
The current study recommends that in order to have more equal income distribution, it is essential to have higher and steady economic growth together with appropriate fiscal instruments that can assure fair distribution of growth benefits across the society. Similarly, higher educational attainments, price and political stability are also highly recommended to reduce growing income inequality in the Asian region.