可以用计量经济学的EVIES软件啊，很简单的！

总得建立一个数学模型 才可以用这个软件吧

估计你是应该用个多元线性回归模型就可以了，不过在做模型之前首先要看看有没有多重共线性，并且注意要用个人可支配收入之类的，不能直接用总量数据！

用农业收入、非农业收入来解释可支配收入很容易出现共线性问题.而且,不知道这样的模型想说明或能说明什么问题.建议先把问题表述清楚

hi, I don't think we should firstly considering of changing into another model;
It may be just a problem of estimation, because of serial correlation between the above two regressors.
Why not trying to use other estimation methods beyond least squres methods, such as generalized linear least squres methods, or Instrumentural methods;

If this also turns out to be non-working. I think we could then try to change variables, or go to a new model. For changing variables, we could use
squared orginal variables; for new model, we may update it with a non-linear regression models.
I hope these comments could help some~
Best wishes