Derived Measurement in Macroeconomics: Two Approaches for Measuring the NAIRU Considered

This paper investigates two different procedures for the measurement of the NAIRU; one
based on structural modeling while the other is a statistical approach using Vector Auto
Regression (VAR)-models. Both measurement procedures are assessed by confronting
them with the dominant theory of measurement, the Representation Theory of Measurement, which states that for sound measurement a strict isomorphism (strict one-to-one mapping) is needed between variations in the phenomenon (the NAIRU) and numbers. The paper argues that shifts of the Phillips-curve are not a problem for the structural approach to measurement of the NAIRU, as the NAIRU itself is a time-varying concept. It is however, the impossibility to identify the exact shape of the Phillips-curve that causes problems of multiple empirical, relational forms and hence non-unique isomorphic mappings for measurement. While VAR-models are being accused of being ‘atheoretical macroeconometrics’ in the literature, the Wold decomposition theorem applied to the VAR brings out a stable correspondence between variance of the phenomenon (the NAIRU) and numbers and turns the set of equations into an isomorphic mapping that can serve as a useful foundation for the construction of a measuring instrument.