- After the upside surprise to hourly wages earlier this month, the consumer and producer price inflation numbers have also comein higher than expected
- Anticipating inflation dynamics has become very difficult, as the Phillips curve has become less apparent
- This forces investors and policy makers to pay more attention to recent data than to long-term relationships
When you’re desperate to get something, the marginal utility of receiving what you want is huge. In economics classes, teachers tend to illustrate this with the relief of a pint of beer when you’re thirsty. The opposite alsoapplies: getting more and more of something you don’t like generates mounting despair. Think of homework after school or of inflation.
Except for those who would have found the perfect hedge, inflation at some point in the cycle turns into a headache: the outlook for monetary policy changes and with some delay the growth outlook as well (downward revisions). Are we in the very early stages of such a path in the US considering that hourly wages, consumer and producer price inflation have surprised to the upside? How will the Fed react to such a “Three in a row”? The data in the run-up to its March 20-21 meeting will be particularly important as well as the post-meeting press conference of the new Fed chair, Jerome Powell. Central bank guidance has become particularly important considering that the traditional paradigm, the Phillips curve – a trade-off between unemployment and wages, has broken down. The upper part of the chart shows the evolution of the beta coefficient of rolling regressions (60-month window) of the monthly change in hourly wages as a function of the unemployment rate (for the narrow and broad definition: U3 and U6). It’s a very basic version of the Phillips curve but judging by the R² and the t-statistic, it quite often did a satisfactory job in the past.
However, since the Great Recession the beta has collapsed, implying little reaction of wage growth to changes in unemployment. In addition, the R² has basically dropped to zero in recent years. Perhaps more sophisticated estimations could do a better job in terms of fit although research by the BIS has shown that the beta coefficient has indeed become very small. This is a challenge for markets as much as for central banks: if long-term relationships are no longer statistically significant, one is left with scrutinising the recent data even more. And greater data-dependency also implies more room for surprises as the past two weeks have shown.