Abstract

Ramey and Ramey (1995) introduced a non-linear model relating volatility to growth. The solution of this model by generalised computer algorithms for non-linear maximum likelihood estimation encounters the usual difficulties and is, at best, tedious. We propose an algebraic solution for the model that provides fully efficient estimators and is elementary to implement as a standard ordinary least squares procedure. This eliminates issues such as the ‘guesstimation’ of initial values and multiple runs. Our approach also facilitates testing the validity of the Ramey and Ramey (1995) model. We illustrate our approach by reanalysing the R&R data, demonstrating virtually identical results.