Modelling The Cobb-Douglas Production Function

I’m currently doing some regression analysis on the US aggregate production function. I’m using time-series data for the period 1954-2005. I’ve modeled the US aggregate production function using a log-transformed Cobb-Douglas production function. Using the Ramsey RESET test, I’ve found my model to be misspecified at the 1% level. I found this surprising. Whilst I understand that a simple Cobb-Douglass production function can’t model reality in its entirety, I didn’t expect such strong evidence in favour of misspecification. Obviously, I’ll be carrying out further tests to find the cause of the problem.