RDP 2010-01: Reconciling Microeconomic and Macroeconomic Estimates of Price Stickiness 6. Conclusions

The New-Keynesian Phillips Curve has become a central part of models used for analysis of the macroeconomy and understanding the behaviour of inflation. However, a tension exists between the low frequency of price resetting that the standard estimates of the aggregate NKPC imply and the relatively high estimates from microeconomic data on firms' pricing behaviour. Furthermore, to improve the fit of the NKPC, indexation of prices to past inflation is often introduced, even though there is scant microeconomic evidence of such behaviour.

Using a Monte Carlo approach, we have argued that heterogeneity in the frequency of price resetting amongst firms, which is evident in the microeconomic data, is one source of the differences in the estimates of price stickiness. The presence of heterogeneity leads to econometric complications for the estimation of the NKPC. In particular, while it is necessary to instrument for both expected inflation and real marginal costs, the instruments that have commonly been used in the literature are unlikely to be suitable when heterogeneity exists. The instruments may be likely to be both weak, as previously demonstrated by Mavroeidis (2005), and invalid.

A second source of these differences is the more complex production structure of the economy than is often assumed in macroeconomic models. Roundabout production, whereby the output of each firm can either be consumed or used as an input to production by another firm, is a realistic way of describing the production process. We have argued that when both heterogeneity and roundabout production are present it is no longer appropriate to use the labour share of income as a measure of real marginal costs when estimating the NKPC; doing so is likely to result in estimates which falsely suggest that some firms index their prices to past inflation.

In summary, we have shown that estimates of the aggregate NKPC obtained in the standard way should not be given a structural interpretation. Any inference from such estimates about the frequency with which prices are reset is likely to be misleading. Acknowledging this resolves the tension between the microeconomic and macroeconomic estimates of price stickiness that exists in the literature.