Econometric analysis of economic growth and income inequality through the lens of Kuznets theory: insights across diverse economic groups (2004-2019)
Description
Research Hypothesis The research investigates the relationship between economic growth and income inequality, drawing on Kuznets' theory of an inverted U-shaped relationship. The central hypotheses are: H0: Income inequality is not affected by GDP growth, indicating no relationship between economic growth and income inequality. H1: GDP growth influences income inequality, which may increase or decrease depending on societal and economic contexts. H2: GDP growth positively affects income inequality, widening income disparities. H3: GDP growth negatively affects income inequality, reducing disparities and promoting equitable distribution. H4: In lower-middle-income countries, GDP growth reduces income inequality. Description of Data The study utilizes data from the World Bank for 39 countries spanning the years 2004 to 2019. The dataset includes: Gross Domestic Product (GDP): Measured in constant local currency units (LOGGDP), used as a proxy for economic growth. Gini Index: A standardized measure of income inequality, ranging from 0 (perfect equality) to 100 (maximum inequality). Income Categories: Countries are grouped into high, upper-middle, and lower-middle income categories based on the World Bank’s GNI per capita classification. Methodology and Data Gathering Selection Criteria: Countries were selected to represent diverse income groups, ensuring a balanced and comprehensive analysis of varying economic contexts. Data Source: All data were sourced from the World Bank’s publicly available databases. Data Analysis: Correlation analysis to explore the general relationship between GDP and inequality. Linear regression models to identify causal relationships across income categories. Group-specific analysis to investigate how GDP impacts inequality within high-, upper-middle-, and lower-middle-income countries. Notable Findings Overall Trends: Across all countries, a positive correlation was observed between GDP and the Gini index, indicating that GDP growth is generally associated with increasing income inequality. The regression model (GINI = 23.931 + 0.937 × LOGGDP) confirmed a statistically significant relationship, with an F-value (p < 0.05) supporting the model’s validity. Income Group Analysis: High-Income Countries: No statistically significant relationship between GDP growth and inequality. Upper-Middle-Income Countries: A weak relationship was observed, but it lacked statistical significance. Lower-Middle-Income Countries: A significant negative relationship was identified (β = -22.291, p < 0.001), suggesting that in these countries, GDP growth reduces income inequality. Interpretation and Use of Data: The findings can be interpreted in light of Kuznets' hypothesis, which posits that inequality first rises and then falls as economies develop.
Files
Steps to reproduce
Data Sources Primary Data Sources: Data was obtained from the World Bank’s publicly available databases. Variables: GDP: Measured in constant local currency units (LOGGDP) for consistency over time and across regions. Gini Index: A widely accepted measure of income inequality. Income Categories: Based on the World Bank’s GNI per capita classifications for high-, upper-middle-, and lower-middle-income countries. Timeframe: 2004–2019. Justification: The dataset covers diverse income groups and a sufficient timeframe for analyzing trends and relationships. Methods and Protocols 5.1 Statistical Analyses Software: Statistical analyses were conducted using SPSS (version 28). Techniques: Descriptive Statistics: Used to summarize GDP and Gini data (means, medians, standard deviations). Correlation Analysis: Explored relationships between GDP (LOGGDP) and Gini indices across the entire sample and subgroups. Pearson’s correlation coefficient was calculated for continuous variables. Linear Regression: Equation: GINI=β0+β1×LOGGDP+ϵGINI=β0+β1×LOGGDP+ϵ Separate regressions for high-, upper-middle-, and lower-middle-income countries. Regression diagnostics included variance inflation factors (VIF) to test for multicollinearity. ANOVA (Analysis of Variance): Used to validate the significance of the regression models across subgroups. Residual Analysis: Examined to ensure homoscedasticity and normality. 5.2 Income Group Stratification The dataset was divided into three income categories as defined by the World Bank’s GNI per capita thresholds. Group-specific analyses ensured tailored insights into how GDP growth impacts inequality within economic contexts. Workflow Step 1: Data Extraction: Access and download GDP and Gini data from the World Bank portal. Extract metadata for each country (e.g., income group classification). Step 2: Data Cleaning: Identify and handle missing data, outliers, and inconsistencies. Standardize GDP values through logarithmic transformation. Step 3: Exploratory Data Analysis: Use scatterplots, histograms, and correlation matrices to explore initial relationships. Step 4: Regression Modeling: Test the linearity of relationships between LOGGDP and Gini indices. Conduct regression analysis and ANOVA. Step 5: Subgroup Analysis: Perform separate analyses for high-, upper-middle-, and lower-middle-income countries. Step 6: Validation and Diagnostics: Check regression assumptions using residual plots and statistical tests. Ensure model robustness through sensitivity analyses. Step 7: Interpretation: Compare findings with existing literature.