From: Ben on
Dear Matlab users,

Currently I am estimating Smooth Transition Regression models in Matlab using the procedure of Teräsvirta (1998): a grid search delivers the initial values for the transition function parameters and subsequently the model is estimated by nonlinear least squares (nlinfit). Does anyone know if there is a more efficient procedure to estimate these models in Matlab?

Regards,
Ben

Reference: Teräsvirta, T. (1994), "Specification, Estimation, and Evaluation of
Smooth Transition Autoregressive Models," Journal of the American Statistical Association, 89 (425), 208-218.