Mandated CSR spending and tax aggressiveness: A machine learning-driven analysis

We explore if mandated Corporate Social Responsibility (CSR) spending encourages tax aggressiveness in firms. We use actual CSR spending amounts instead of CSR proxy scores and consider India, which has mandated CSR spending regulation, as opposed to voluntary CSR regimes in existing works. As an outlying feature, we use the Conditional Inference Tree (CIT), a machine learning approach, uniquely to gain insights into the relationship and check the robustness of standard regression analysis outcomes. We find that tax aggressiveness is negatively associated with CSR spending (characterized by CSR spending growth) and positively influenced by other factors, like a greater focus on investors and higher returns. Our findings indicate that socially responsible companies with increased CSR spending are less likely to demonstrate tax aggressiveness even in a mandatory CSR regime. The finding that firms’ earnings manipulation influences tax aggressiveness is also a prominent contribution of our study.