Testing the specification
The main tools of evaluation of models like the NPCM have been the GMM test of validity of overidentifying restrictions (i. e. the xj-test earlier) and measures and graphs of goodness-of-fit.9 Neither of these tests is easy to interpret. First, the xj may have low power. Second, the estimation results reported by GG and GGL yield values of bp1 + bp1 close to 1 while the coefficient of the wage share is numerically small. This means that the apparently good fit is in fact no better (or worse) than a model in the double differences (e. g. a random walk); see Bardsen et al. (2002b). There is thus a need for other evaluation methods, and in the rest of this chapter we test the NPCM specification against alternative models of the inflation process.