predictive model's scoring issue

#1
Suppose I have built a model to predict default on loans.
In my modeling data set I have three independent variables: income, home(own or rent) and job(manager, sales, self, other). They are all very strong predictors.

For some reason, I won't be able to have the variable "job" in my scoring data set.
Do I still need to build and evaluate the model with the variable "job"? Assume that I need to predict the probability of default and put the candidates in a gains table like below:

decile lift %default cum%default %obs
---------------------------------------------
1 3.4 5% 17% 10%
2 2.1 4.5% 41% 20%
3 1.9 4.1% 52% 30%
4 1.5 3.5% 61% 40%
...
10

Thanks for any input!