Abstract:
This paper develops a Kaleckian-inspired reduced-form framework for analysing wage-income dynamics in Morocco, focusing on the interaction between macroeconomic conditions, fiscal policy, and labour-market pressure. Because consistent annual wage-income data are unavailable, household final consumption expenditure as a percentage of GDP is used as a reduced-form proxy. GDP growth serves as a profitability proxy, while a constructed institutional-fiscal indicator combines tax revenue and unemployment. The empirical analysis applies XGBoost, Random Forest, and LSTM models to capture nonlinear predictive relationships that conventional linear specifications may overlook. Using annual Moroccan data for 2000-2022, the study evaluates model performance through chronological out-ofsample prediction and considers both the level and year-on-year change of the proxy. The contribution is a reduced-form empirical framework linking Kaleckian distributional theory with nonlinear predictive methods to explore how fiscal and labour-market conditions interact with income dynamics.
