The Problem of Estimating Causal Relations Using Models Involving Accounting Semi-Identities

A Cautionary Note on the Use of Accounting Semi-Identity-Based Models

“…Fazzari et al (1988) argued that the higher the explanatory variable coefficient, the more company investments depend on company cash flow to finance them. So, higher coefficients should correspond results, as I demonstrate in Sanchez-Vidal (2007). An ASI is an accounting identity but not a complete one since one or more components of the accounting identity are missing.…”

“…Results of the previous section, taking into account the accounting identity, give an explanation of the strange results of the Fazzari et al (1988) model to date. According to one previous work of mine, Sanchez-Vidal (2007), the ASI problem is probably to blame for the counterintuitive results by Kaplan and Zingales (1997) when they used the FHP (1988) model. Less constrained enterprises are more likely to experience larger investment-cash flow sensitivity since it is easier for them to add long-term debt, increasing the value of "the rest" and resulting in higher estimated coefficients for the cash flow variable, contrary to FHP (1988) model predictions.…”

“…In Sanchez-Vidal (2007), I explained what happens when an ASI is estimated. The predicted coefficients will most likely compensate for the absence of "the rest" or the sum of the accounting identity's missing components.…”

“…In Sanchez-Vidal (2007), I explained that the investment-cash flow sensitivity model as proposed by Fazzari et al (1988) consists simply of regressing the accounting identity (expression (1), but without the "rest":…”