RDP 9502: Price Stickiness and Inflation 5. Conclusions

At the outset, we contrasted two explanations for the observation that prices are “stickier” when going down than when going up. The traditional explanation is that this asymmetry arises from social customs or some special aversion by firms to allowing their output prices to fall. This traditional view implies that asymmetric price stickiness is not amenable to microeconomic analysis. The alternative view, associated with the New Keynesian literature, starts with the observation that changing price is not costless. Once this is recognised, asymmetric price stickiness follows from the optimising response of price-setting firms to their microeconomic environment and the shocks they face, rather than from an aversion to reducing their prices.

We have two strands of evidence supporting the latter, New Keynesian, analysis of price stickiness. The first strand of evidence is very simple: our industry price data do not reveal any aggregate evidence that firms have a special aversion to price falls.

The second strand of evidence is more involved. It requires detailed analysis of the Ball-Mankiw model of firm's price-setting behaviour. Ball and Mankiw established novel implications for the relationship between the economy-wide distribution of price changes and inflation in the short-run. Our main contribution is to demonstrate the importance of expected inflation for this story. We derive the detailed implications of the Ball-Mankiw model for the relationship between expected inflation, the economy-wide distribution of industry price changes and actual inflation and show that both US and Australian industry-price data strongly support these implications.

While the correlation between relative price dispersion and inflation has been well-known and widely-analysed (see e.g., Fischer 1981) the correlation between inflation and the skewness of price changes was an empirical curiosity in search of an explanation. Ball and Mankiw (1992b) provided an explanation for the correlation, and established that the relationship between skewness and inflation is stronger, empirically, than the dispersion-inflation relationship in the post-WWII United States.

We establish a further implication of the Ball-Mankiw model: that the relative strength of the skewness-inflation and dispersion-inflation relationships depends critically on expected inflation. When expected inflation is low, the skewness-inflation relationship is the stronger of the two. As expected inflation rises, however, the dispersion-inflation relationship becomes increasingly important. The skewness-inflation relationship is estimated to be the stronger for expected inflation below about 4 to 5 per cent per annum, while the dispersion-inflation relationship is the stronger for higher rates of expected inflation.

The Ball and Mankiw result that the skewness-inflation relationship is the stronger in the post-WWII United States is then a consequence of the low average rate of expected inflation in the United States over this time. By contrast, over our twenty year sample of Australian data, inflation (and, by inference, expected inflation) averaged 8 per cent per annum. At this rate of expected inflation, the relationship between dispersion and inflation is stronger empirically than the skewness-inflation relationship.

As we have seen, the empirical evidence from industry price data supports the detailed predictions of the model, both at different rates of expected inflation and in two different industrial economies. This confirms and strengthens the argument that the aggregate implications of price stickiness can be understood in terms of the optimising behaviour of firms with market power responding to the shocks they face. It increases our confidence in our understanding of the microeconomics of firm pricing behaviour and its implications for aggregate inflation.

To conclude, we discuss the policy relevance of the analysis. As we have seen, the inflationary impact of relative price shocks depends strongly on expected inflation. When expected inflation is high, a rise in the economy-wide dispersion of shocks is inflationary in the short-run. By contrast, when expected inflation is low, a rise in the dispersion of shocks has minimal impact on inflation. Economy-wide relative price shocks, like terms of trade shocks, are an unavoidable feature of the economic landscape. Their disruptive effect on inflation is minimal, however, when average inflation, and therefore average expected inflation, is kept low.