Understanding the recent behaviour of inflation: an empirical study of wage and price developments in eight countries 1

An important and surprising characteristic of the economies in industrialised countries in the 1990s was the extent to which prices decelerated in an environment of generally rising economic activity and tightening labour markets. The coexistence of a healthy economic environment and low inflation can appropriately be described as good news. However, economists tended to overstate underlying inflation pressures in many of these countries during most of the decade, and these prediction errors demonstrate our lack of understanding of the inflation process and raise questions about the appropriate stance of monetary policy in such an uncertain environment. In this paper, we investigate the magnitudes and potential sources of inflation forecast errors during the 1990s in a sample of eight industrialised countries. Our analysis consists of two separate approaches. First, we examine the errors in official OECD forecasts, which we take to be representative of the mainstream of macroeconomic analysis during that time. Second, we document and analyse prediction errors in our own set of econometric specifications, which are loosely based on the Phillips curve model of the inflation process. Our analysis of OECD forecast errors is indicative of persistent overpredictions of price inflation for most of the countries in our sample. In contrast, little bias is evident in the OECD forecasts of wage inflation. The combination of the forecast errors for wages and prices thus implies that real wage growth has been unexpectedly strong during the 1990s and that the major sources of the forecast errors are likely to be located in that part of the Phillips curve framework which models firms’ prices as a mark-up on costs. More precisely, the unexpectedly slow rise in prices relative to wages could indicate that firms have benefited from favourable supply shocks or that they lost pricing power during the 1990s in that they were not able to fully pass on wage cost increases into their prices. Our own models for consumer prices also consistently overpredict inflation in nearly every country. Nonetheless, there is little statistical evidence of parameter instability. The one exception is a decline in the intercept term in the 1990s, a finding indicative of structural change but, unfortunately, not particularly helpful in identifying the source of the change. In contrast, the parameters in our wage models appear to have changed in about half of the countries. However, no single coefficient stands out as particularly sensitive to the addition of the more recent data.