Published Paper Details:

Income and wealth inequality continue to pose deep-rooted structural challenges in India, carrying far-reaching consequences for economic stability, inclusive development, and effective policymaking. Consequently, accurate forecasting of inequality trends is crucial for evidence-based policy decisions. This study evaluates the suitability and comparative predictive performance of nonlinear machine learning and time-series techniques for forecasting income and wealth inequality in India. Departing from the constraints of traditional linear econometric models, the analysis employs Random Forest (RF), Autoregressive Integrated Moving Average (ARIMA), Support Vector Regression (SVR), Extreme Gradient Boosting (XGBoost), and Long Short-Term Memory (LSTM) models to better capture the nonlinear relationships and temporal dynamics that characterize inequality processes. Drawing on historical inequality measures obtained from nationally representative secondary datasets, the forecasting models are developed and tested within a common analytical framework. Model performance is evaluated using Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). The empirical evidence demonstrates that machine learning-driven approaches most notably XGBoost and LSTM outperform conventional ARIMA and other benchmark models in predictive accuracy. This advantage stems from their enhanced capacity to capture nonlinear dynamics and long-range temporal dependencies. The results highlight the growing importance of advanced machine learning methods as reliable and complementary instruments for forecasting inequality in emerging economies. Overall, the study makes a methodological contribution to the inequality forecasting literature and provides actionable insights for policymakers aiming to adopt data-driven strategies to track and mitigate income and wealth inequality in India.