Household Consumption, Poverty Depth and Inequality in Emerging Economies: A Nonlinear Predictive Analysis

Understanding how household consumption varies with poverty depth, inequality, and price instability is relevant to assessing economic vulnerability and informing discussions of Sustainable Development Goal 1 (SDG 1: No Poverty). This study examines these relationships across 13 emerging economies over 2000–2024, using a completed analytical panel of 325 country-year observations. Seven linear and nonlinear models are compared using country-demeaned variables and five-fold GroupKFold cross-validation grouped by country. The benchmark Extra Trees model achieves a mean cross-validated R² of 0.6886, compared with 0.3727 for the linear baseline, representing an 84.76% relative increase in R². Corresponding RMSE values are 0.1421 and 0.1697. Shapley Additive Explanations (SHAP) identify poverty depth (POV_GAP) as the most prominent predictor in the benchmark attribution profile, accounting for 54.89% of normalized mean absolute SHAP attribution. Supplementary analyses using a separate Extra Trees specification describe nonlinear attribution patterns and an aggregate consumer price inflation attribution-magnitude ratio of 2.21 between observations above and at or below their country-specific means. This ratio summarizes differences in absolute model contributions rather than behavioral loss aversion or responses to equivalent inflation shocks. Cross-country SHAP attribution profiles further illustrate heterogeneity in the predictors associated with consumption deviations, without establishing distinct structural regimes. Because preprocessing uses full-period information, the validation results concern retrospective country-centered variation rather than prospective forecasting. The findings contribute evidence on nonlinear consumption–poverty associations and motivate attention to poverty depth and country context in research on household vulnerability, without identifying causal mechanisms or the effectiveness of specific policy interventions.