Several parameter estimation methods of Econometrics

A more general parameter estimation method:

1, ordinary least squares: suitable for the classical hypothesis, but the equation model;

2, weighted least squares: Suitable for variance data, the essence of weighting is to use a variable divided by the error term, so that the variance of the error term into a constant;

3. Tool variable method: It is suitable for the explanatory variables to be random, and the explanatory variables are related to the error term. The idea is to find a variable associated with the explanatory variable that is not related to the error term, to be a tool variable, and to replace the explanatory variable with a tool variable for regression. Y=AX+B+U,X=A1Z+B1+U1, bring the rear type into the front, y=a (A1X+B1+U1) +b+u=aa1x+ (ab1+b) + (au1+u). Many estimation methods can be considered as tool variables. When I first learned this method, I wondered where to look for the right tool variable, until I learned that the pre-determined variable was a ready-made tool variable when I was learning a simultaneous equation model.

4. Indirect least square method: it is suitable for precisely recognizing simultaneous equations, that is, estimating the reduction form of simultaneous equations, and then solving the linear equations to obtain the original parameters.

5. Two-stage least squares method: It is suitable for the exact recognition and over-recognition of simultaneous equations, using the linear combination of the pre-determined variables as the tool variables of each endogenous variable. In the specific operation of a certain equation, the inside of the internal variables to all the pre-determined variable regression, with the results obtained the estimated value of the endogenous variable, and then used to replace the original endogenous variable and then a regression.

6, three-stage least squares: Simultaneous equation Model system estimation method, not understand.

7, the maximum likelihood estimate: the likelihood function to calculate the maximum value.

8, the generalized moment method: not carefully studied.

Several parameter estimation methods of Econometrics