Wojtek, This is a bit of a muddle. Ormerod was an accomplished econometric modeler in Britain. Built up a business around it, then sold it for big bucks, so he knows what he's talking about, but you tripped up in a few places. >> His argument can be summarized as follows: Prediction in economics means translating cause-effect relationships into equations (cf. y=a+bx where y -represent the outcome x represents the cause and b is the coefficient specifiying the relationship between cause and the outcome). Since outcomes have multiple causes, realty is modeled by a system of equations each one corresponding to a relatively simple underlying cause. >> No. A single equation can contain multiple factors which are thought of as 'causes.' None of these need be 'simple.' >> That multiplicity creates a problem that in linear algebra is called "underdtermined systems." In plain english, it means that there is not enough information in the system to find unique solution for all the equations. >> Multiplicity per se has no necessary implication for 'underdetermined' systems. What matters is how the system of equations is constructed. For X unknowns you need X equations, but even then the equations cannot, to keep terminology simple, be redundant in the mathematical sense. Too many non-redundant equations, on the other hand, can also be a problem, since mathematically the system can be over-determined, which means conceptually the model is not internally consistent. Your illustration is correct, though its equations only had one 'cause' each. >> One solution of the problem involves restricting certain parameters . . . >> The more likely solution is to complete the model! In other words, to fill in the missing equation(s). The rest of your post is valid, in general. There are lots of problems with econometrics, including its practical use. Ormerod is worth reading. (another liberal, BTW). One of my favorite professors, Mancur Olson (a prolific neo-classical economist), used to tell us that econometrics never settled anything. He had little use for it. Marxists have some models like this too. I ignore all of that stuff (Neo-C and marxist), and so do many neo-classical types. On the other hand, the business uses of econometrics in forecasting are typically based on Keynesian models, hence a better guide to the economy than much of what is being done in the name of macro-theory. (I'm told the new, right-wing theories can't be implemented in practical forms.) mbs
