# st: Why does TRUNCREG cannot find the MLE estimate?

 From "Li, Shaling (lis1)" To Subject st: Why does TRUNCREG cannot find the MLE estimate? Date Sat, 3 Sep 2005 22:00:38 +0100

```Dear Statalist members,

I use TRUNCREG command to run regression with truncated sample. My regression is to regress return on deps, a bivariate regression. Return is right-skewed with a fat tail and deps is left-skewed also with a fat tail.

Model: Return = a+ b*deps + error ;
Truncation point: return = 0;
My command is "truncreg return deps, ll(0)" and "truncreg return deps, ul(0)"

For the right-truncated sample, it works well and produce the same results as I could get from SAS. But for the left-truncated sample, STATA shows that it "could not calculate numerical derivatives, missing values encountered" (showed below). SAS also fails to generate any stable results for the left-truncated sample (either fail to generate a result as does STATA, or generate random results). For my research design, it would be very difficult to interpret the coefficient if I transform both variables into variables more close to normal distribution. I tried to use option (robust, or bootstrap) to control the deviation from normality of the variables but STATA still could not work it out. How to solve this problem?

Variable	Obs	        Mean	    Std. Dev.	   Min	        Max

return	75602	       .0730425	  .5803403	   -.8676471	    3.823529
deps	     75602	       .0048605	  .1726828	   -1.312448	   .3449057

Report from STATA
truncreg return deps, ll(0)
(note: 19778 obs. truncated)

Fitting full model:

Iteration 0:   log likelihood = -12310.067
Iteration 1:   log likelihood = -5697.8827
Iteration 2:   log likelihood = -5401.2288
Iteration 3:   log likelihood = -5369.5397
Iteration 4:   log likelihood = -5363.2702  (not concave)
Iteration 5:   log likelihood = -5349.9794  (not concave)
Iteration 6:   log likelihood = -5348.4234
.......
Iteration 50:  log likelihood = -4851.4588	(not	concave)
numerical derivatives are approximate
nearby values are missing
numerical derivatives are approximate
nearby values are missing
Iteration 51:  log likelihood = -4851.4588	(not	concave)
numerical derivatives are approximate
nearby values are missing
numerical derivatives are approximate
nearby values are missing
Iteration 52:  log likelihood = -4851.4588	(not	concave)
could not calculate numerical derivatives missing values encountered
r(430);

.

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```