We examine problems associated with the identification and estimation of simultaneous equation models, and introduce a new approach to identify such models together with their parameter estimation. In this approach we separate the model into two partial models, i.e., one is called the endogenous model which contains only endogenous variables and the other is called mixed model which contains endogenous and exogenous variables. The key idea of our method is to represent the endogenous model as a recursive system based on a set of orthogonal transformations. We consider identifiability of the simultaneous equation models based on the recursive system. The scope of model identinability can be expanded sometimes. The model identification and parameter estimation can be practice with ease. For linear and Gaussian models, the estimator for our method is equal to the maximum likelihood estimator. Results from numerical simulations show performance advantages when using our method.
|Number of pages||20|
|Publication status||Published - 1 Jan 2016|
- Indirect least squares method
- Recursive system
- Simultaneous equation models
- Two-stage least squares method