Representation and Stability in Testing and Measuring Rational Expectations.

There are at least two elements of theory completion necessary for measurement: (1) a measurement formula and (2) standardization of that representation. Standardization is based on the search for stability. The more stable the correlation which the measurement formula represents is, the less influence other circumstances have. Then, the interconnection between testing, mathematical representation and standardization is of a hierarchical order. By testing a model one tries to find out to what extent the model covers the data of the phenomenon, while to be a candidate for a measurement formula the model must represent the whole data range. And among the possible representations the standard model represents the most stable correlation under different circumstances. Lucas' model of the Phillips curve has been used to investigate this interconnection between testing, representation and stability.