Until now expectations on inflation have been formed with a mechanical rule whereby the past realized inflation rate is used as the best prediction of the next period one so that in symbols: πte= πt−1. The reasons for adopting such a simple rule are primarily empirical. As Carlin and Soskice (2006, 2014) argue, it is a specification that fits the data well over the period of the great moderation and it captures the empirical observation of a certain inertia in inflation, whereby the current rate of inflation in any one period has a bearing on what the next period's rate of inflation can be. Assuming that households and firms operate under such a rule has a certain plausibility, provided that the general context in which expectations are formed be a relatively stable one. These conditions are obviously matched in the sample period of the great moderation and it is therefore plausible to imagine that many economic agents would resort to such a simple mechanical rule, or collectively act as if they had adopted such a rule, when the actual oscillations of the inflation rate around the target rate are too minimal to warrant more informed and more costly assessments of what is to be expected. Nonetheless there is no denying that the adoption of such a mechanical rule has serious theoretical drawbacks that largely match what prompted the abandonment of the adaptive expectations mechanism in the 1970s. In practice, there is in fact little way of discriminating between the inertial specification of the Phillips curve and the one adopted by Friedman (1968) with an adaptive expectations mechanism, whereby expectations are formed on the basis of some update or correction on past forecasting errors. A degree of observational equivalence is inevitable and therefore the theoretical criticisms apply equally. These in particular are the following.

In the first instance, there is a degree of undesirable asymmetry whereby the central bank is forward-looking or is able to predict what is going to happen, in particular the future period inertia-augmented Phillips curve, whereas firms and households are entirely backward-looking and are reacting to past events rather than looking forward to forthcoming ones.