Hockey stick type curve fit

[yiwu ye <yiwu21111958@yahoo.com>]

 Dear list,

I used MASS liberary's optim function to fit a Hockey stick-like curve, i.e. for x below a certain level, y is a linear function of x,but for x above the level, y levels off. I can successfully estimate the three parameters ( one for the critical level, two for the linear regression below the critical level), but cannot use hessian matrix to estimate standard errors for the three parameters. Of course, I can repeat the regression for the y and x below the estimated critical level to estimate parameter standard errors. But, is it possible to estimate the SE for the parameter of the critical level? How to explain it?

Thanks for your help

Yiwu

---

Love cheap thrills? Enjoy PC-to-Phone  calls to 30+ countries for just 2¢/min with Yahoo! Messenger with Voice.