Forecasting E-waste generation in data-constrained regions: a parsimonious regression model for Latin America

Abstract

Abstract The rapid increase in Waste Electrical and Electronic Equipment (WEEE) generation, driven by accelerated consumption and shortened product lifespans, has become a critical global environmental challenge. Reliable forecasting tools are essential to support waste management and policy planning, particularly in developing regions where data availability is limited. Using cross-sectional data from 2022, this study proposes a linear-log regression model to estimate WEEE generation across 18 Latin American countries using gross domestic product (GDP) per capita as the primary explanatory variable. Unlike data-intensive approaches such as material flow analysis and time-series models, the proposed model offers a simplified and scalable alternative that requires minimal data inputs, addressing a key limitation in developing contexts. The model demonstrated strong predictive performance, achieving a coefficient of determination (R²) of 0.904 and a mean absolute error (MAE) of 0.870, while satisfying key residual diagnostic checks. Applied to the study sample, predicted values ranged from 1.17 kg per capita (Haiti) to 13.46 kg per capita (Uruguay), closely tracking observed WEEE generation across the full income spectrum of the region. This study demonstrates that regionally calibrated parsimonious regression models can provide reliable WEEE forecasts in data-constrained environments. The model enables policymakers and waste management stakeholders to estimate future WEEE generation and support infrastructure planning and reverse logistics strategies. Future research should incorporate additional socio-economic variables, test the model in other regions, and compare its performance with more complex forecasting techniques.