Definition of simultaneous equation modelling I have been asked to comment on a statistical analysis proposal that suggests using simultaneous equation modelling  as a method. I have a basic general knowledge of statistics (Masters level) but very little experience with econometrics-specific modelling.
Does this technique go by any other name?
If it is solely used in econometrics, I'm looking for a somewhat non-theoretical applied definition.
I'm also looking for an indication of which software it can be implemented in out of a choice between Stata/SAS/SPSS.
I have looked through various online sources and am either not convinced of their accuracy or they present the information too theoretically.
 A: Take a look at this chapter in "Baby" Wooldridge on Simultaneous Equation Models. In particular, example 16.1 on estimating the relationship between police force and the murder rate in very illuminating. Briefly and very roughly, you need for each equation to have a ceteris paribus (all else equal) behavioral interpretation. For example, the equations may describe the causal, behavioral relationships of different economic agents (the potential murderers and city officials). Secondly, these equations are simultaneous, in the sense that the decisions are interlinked. Presumably, criminals murder less when there are lots of cops around, and police forces grow when there are long-run crime sprees.
Economic theory imposes certain restrictions/relationships on the parameters that allow you to recover the parameters of these behavioral equations in several ways from data (these are mentioned in the Wiki entry). Sometimes this only happen when the assumptions are so strong that no one believes them. Other times the assumptions are sufficiently confusing so that no one understands them and the analyst can get away with murder (pun intended). Sometimes only functions of parameters can be recovered. In short, lots of heavy lifting needs to happen before you fire up your statistical software.
Finally, I think of simultaneous equations as a type of structural equation model. These can easily be handled with Stata, R, SPSS and SAS). There's also specialized software like LISREL. SEM models pop up in other social sciences, such as psychology, especially in psychometrics.
