Dear all,

I tray to analyze the determinants of loan repayment performance in a given organization. The outcome variable was classified into three categories namely ‘paid on time’ for the clients who repaid loan before the due date, ‘delinquency’ for clients who repaid late from the due date or repaid less than the appropriate amount of their most recent loan, and ‘default’ for the clients who did not pay after three months of the due date. After I run the model using mlogit command i found the following result. But I have some question regarding the model and the iia test.

1. Is the overall estimation result good as per the title ?

2. What is the rationality behind choosing base category for specific model?

3. iia test doesnt work for this model what is the problem with it?

I need your help.

Thank you very much.

Ermiyas

I tray to analyze the determinants of loan repayment performance in a given organization. The outcome variable was classified into three categories namely ‘paid on time’ for the clients who repaid loan before the due date, ‘delinquency’ for clients who repaid late from the due date or repaid less than the appropriate amount of their most recent loan, and ‘default’ for the clients who did not pay after three months of the due date. After I run the model using mlogit command i found the following result. But I have some question regarding the model and the iia test.

1. Is the overall estimation result good as per the title ?

2. What is the rationality behind choosing base category for specific model?

3. iia test doesnt work for this model what is the problem with it?

I need your help.

Code:

mlogit loanstatus age loansize income area rooms floor tenur hhsize sex educ Iteration 0: log likelihood = -145.5761 Iteration 1: log likelihood = -75.804603 Iteration 2: log likelihood = -72.239623 Iteration 3: log likelihood = -64.135462 Iteration 4: log likelihood = -62.519257 Iteration 5: log likelihood = -62.061845 Iteration 6: log likelihood = -62.058576 Iteration 7: log likelihood = -62.058576 Multinomial logistic regression Number of obs = 155 LR chi2(20) = 167.04 Prob > chi2 = 0.0000 Log likelihood = -62.058576 Pseudo R2 = 0.5737 ------------------------------------------------------------------------------ loanstatus | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- Default | age | .0900269 .0497059 1.81 0.070 -.0073949 .1874487 loansize | .0000514 .0000259 1.99 0.047 7.49e-07 .0001021 income | -.0014927 .0004514 -3.31 0.001 -.0023774 -.0006079 area | .0167898 .0949439 0.18 0.860 -.1692968 .2028765 rooms | -3.228844 1.858627 -1.74 0.082 -6.871686 .4139975 floor | .0920056 .2192906 0.42 0.675 -.3377962 .5218073 tenur | -.0978784 .3479336 -0.28 0.778 -.7798156 .5840589 hhsize | -.1732183 .3867932 -0.45 0.654 -.931319 .5848824 sex | -.199167 .9524633 -0.21 0.834 -2.065961 1.667627 educ | -1.596048 .6462633 -2.47 0.014 -2.862701 -.3293956 _cons | .5467191 8.059603 0.07 0.946 -15.24981 16.34325 -------------+---------------------------------------------------------------- Delinquent | age | .0536494 .0354863 1.51 0.131 -.0159024 .1232012 loansize | .0000102 .0000203 0.50 0.615 -.0000296 .0000501 income | -.0002777 .0001202 -2.31 0.021 -.0005133 -.0000422 area | .06586 .0851608 0.77 0.439 -.1010522 .2327721 rooms | -1.079745 1.246765 -0.87 0.386 -3.52336 1.363869 floor | .3715418 .1467483 2.53 0.011 .0839205 .6591631 tenur | -.1872231 .2649991 -0.71 0.480 -.7066118 .3321656 hhsize | .0925946 .2187679 0.42 0.672 -.3361827 .5213719 sex | .473651 .6225617 0.76 0.447 -.7465475 1.69385 educ | -.8049165 .4456839 -1.81 0.071 -1.678441 .0686079 _cons | -2.80576 5.615922 -0.50 0.617 -13.81276 8.201244 -------------+---------------------------------------------------------------- Paid_on_time | (base outcome) ------------------------------------------------------------------------------ . mlogtest, iia Problem determining number of categories. **** Hausman tests of IIA assumption Ho: Odds(Outcome-J vs Outcome-K) are independent of other alternatives. You used the old syntax of hausman. Click here to learn about the new syntax. (storing estimation results as _HAUSMAN) flat region resulting in a missing likelihood r(430);

Ermiyas

## Comment