Edward B. Douthett
George Mason University
Linda M. Parsons
George Mason University
Abstract: This paper examines the issue of simultaneous equations bias in government and nonprofit accounting research. Simultaneous equations bias arises when single equation estimation (OLS) includes an explanatory endogenous variable, potentially rendering results biased and inconsistent. We illustrate specification testing for simultaneity bias when the system includes a nonlinear equation or a limited dependent variable. Using data from a previous study, our specification testing confirms the presence of simultaneity bias in a system of equations used to estimate auditor reputation and fees in the government sector. Unexpectedly, we also confirm the existence of simultaneity bias in a preliminary system of equations used to estimate charitable donations and an efficiency measure of nonprofit organizations. We conclude by offering suggestions for research where simultaneous equations may be appropriate in other government and nonprofit settings.
Back to Program
