This article investigates the non-linear causality of tax burdens on personal income and corporate income to GDP growth per capita. It examines the period 1948 to 2008 and employs the tax revenues GDP data observed in these years to act as the basis for the non-linear causality tests.
This paper helped me to gain supporting data for my expected results and provided a source of credibility to my regression results. This article did not suggest any issues that I have regarding the assumptions of the classic linear model.
Parker Delmolino's Econ Blog 
Have you come across any other research where there have been other functional forms tried? I have similar data and am considering running a log linear regression. Maybe you would want to try that and see what happens?