Sup material for DATA-Mdpi - A Dataset Examining Problem of ASIs

The use of accounting semi-identity-based (ASI) models in econometrics can lead to severe problems, with OLS regressions potentially misrepresenting causal relationships. This study uses Monte Carlo simulations to control causal links and illustrate issues that real-world samples cannot reveal, due to unknown factors behind firms' investment decisions. Simulations allow precise assessment of estimator accuracy. The investment-cash flow sensitivity model (Fazzari, Hubbard, Petersen, 1988) is used as an example of an ASI model. Randomly generated variables (cash flows and Tobin’s Q) evaluate ASIs’ effects on regression reliability. The paper aims to both explain ASI issues in more depth and promote the use of Monte Carlo simulations to model complex economic variable interactions and assess problems like multicollinearity, heteroskedasticity, and endogeneity.