Dear experts,

For my master thesis I'm trying to replicate the results of Boswijk, Hommes and Manzan (2007). They estimate the following model:
[ATTACH=CONFIG]temp_15829_1569666325994_470[/ATTACH]

R* and x are known. Phi 1, Phi 2 and Beta* have to be estimated using non linear regression. My weekly data corresponds well with the monthly data of the paper.

If n(t) is substituted in the first equation. The model could be estimated using NLLS. In Stata 15, I ran the following command:

nl (Rx = {b1}*(x_min_1*(1/(1+exp(-1*{b3}*(({b1}-{b2})*x_min_3*(x_min_1-R_star*x_min_2)))))) + {b2} * x_min_1 *(1-(1/(1+exp(-1*{b3}*(({b1}-{b2})*x_min_3*(x_min_1-R_star*x_min_2)))))))

[ATTACH=CONFIG]temp_15830_1569666728917_24[/ATTACH]

b1 corresponds to Phi 1, b2 to Phi 2, and b3 to Beta*. I tried several initial values. And also the estimated values of Boswijk et al. However, I do not get any standard deviations, t values, p values and CI for b2. This happens with different initial values and only for b2.

So, does anyone know what could cause the blanks for b2?

Thank you in advance for any suggestions,

Thomas van der Meiden