Innovation Under Financial Constraint: A Causal Machine Learning Analysis of Smes in Emerging Economies

Abstract

This study investigates how financial constraints influence product innovation across small and medium-sized enterprises (SMEs) in five emerging economies. The authors employ Causal Forests to estimate both average and heterogeneous treatment effects of perceived financing obstacles based on the Rubin potential outcomes framework. The analysis reveals that financial constraints affect firms in highly uneven ways, with innovation outcomes varying significantly by firm age, size, financing structure, export orientation, and exposure to corruption or infrastructure challenges. To enhance interpretability, the authors integrate SHAP (Shapley Additive Explanations), decomposing prediction scores to identify the relative contribution of firm-level characteristics to innovation under constraint. The findings show that in some contexts, financial constraints act as barriers that lead to innovation abandonment or failure; in others, they serve as pressure points that catalyze adaptive innovation. This combination of quasi-experimental methods with explainable machine learning offers a novel approach to identifying which firms innovate under pressure and why, supporting more targeted innovation policies and financing strategies in developing economies.

Keywords: SMEs, Financial constraints, Product Innovation, Causal Machine Learning, SHAP