unreasonable result

currently, i wrote a program utilizing the cmlmt library to estimate an interest rate model (the CKLS model), but the result that GAUSS gave is really unreasonable. Result  reads as follows:

********************************************************************************

return code = 0
normal convergence

Log-likelihood 9966.54
Number of cases 2058

Covariance of the parameters computed by the following method:
ML covariance matrix
Parameters Estimates Std. err. Est./s.e. Prob. Gradient
---------------------------------------------------------------------
params[1,1] 0.0003    0.0071      0.045     0.9645    0.1369
params[2,1] 43.3941    974.7281    0.045    0.9645     0.0000
params[3,1] 50.0000     .                      .             0.0000     0.1226
params[4,1] 1.9391       0.0041     477.249   0.0000    -0.0008

Correlation matrix of the parameters
1             -0.99987408         .           -0.00080509574
-0.99987408      1               .                0.00065732412
.                                 .                .                 .
-0.0008050958         0.00065732417             .            1

Wald Confidence Limits

0.95 confidence limits
Parameters Estimates Lower Limit Upper Limit Gradient
----------------------------------------------------------------------
params[1,1] 0.0003           -0.0135       0.0142      0.1369
params[2,1] 43.3941         -1868.1644    1954.9525      0.0000
params[3,1] 50.0000        .                        .                          0.1226
params[4,1] 1.9391           1.9311            1.9471             -0.0008

Number of iterations 1000
Minutes to convergence 0.21132

79636975.79535         576.89366            0.00548          20464.99581
576.89366                      0.00418                  0.00000         0.14808
0.00548                             0.00000                1.65375           -314.76628
20464.99581                  0.14808                  -314.76628      60579.31980

********************************************************************************

then i reviewed my program, no mistake regarding the loglikelihood function was found after checking many times. and, the gradCheck from GAUSS said the gradient and hessian part of the code is right. so, i dont know where the problem might located at.

1 Answer

0

Post your command file and I'll look at it.  Or better yet, put your command and data files into a compressed .zip file and attached it here and I'll be able to look at it in detail.

link

RonSchoenberg

210