Inflation expectations are less important than you may think
A while ago I wrote about the sensational paper by Jeremy Rudd who thoroughly criticized the common belief in the importance of inflation expectations for inflation and monetary policy. In short, central banks focus on keeping inflation expectations “well-anchored” around the long-term inflation target, fearing that if inflation expectations move higher, consumers will demand higher wages and ignite a wage-price spiral. Yet, the empirical evidence for the relevance of inflation expectations is rather weak and today, I want to discuss two recent research projects into that topics and what they mean for our belief in the importance of inflation expectations.
Let’s start with a paper by Elmar Mertens from the Deutsche Bundesbank (arguably the institution most worried about inflation anywhere in the world) and James Nason from North Carolina State University. They look at US inflation and the expected inflation from the Survey of Professional Forecasters. They look at inflation and the relationship with expected inflation over a very long time from 1968 to 2018 and show that long-term inflation forecasts slightly improve the estimates of long-term future inflation. However, the paper also finds that shorter-horizon inflation expectations have no bearing on the forecast accuracy of models or any relationship with realised future inflation. Most importantly, the relationship between inflation expectations and future inflation changes over time, not just in degrees, but in sign, meaning that sometimes higher inflation expectations lead to higher inflation and other times they lead to lower inflation.
Another paper by economists from the ECB, the Bank of Spain and the Deutsche Bundesbank looked at Eurozone inflation between 2005 and 2019. They checked if inflation expectations can improve inflation forecasts for the coming one and two years. The interesting thing about their research is that they used different ways of measuring inflation expectations. They found that inflation expectations derived from bond markets (the preferred method of the Fed, for example) has no information about future inflation and hence does not improve inflation forecasts. Similarly, and most importantly, inflation forecasts from businesses or households do not inform future inflation or the performance of inflation forecast models. To quote from the paper: “[Inflation expectations of firms and households] … appear to contain contemporaneous rather than forward looking information”. Translated into plain English: Businesses and households think that future inflation will be the same as today’s inflation.
However, the study also finds that inflation expectations of professional forecasters slightly improve inflation forecast accuracy similar to the US case. My interpretation of this finding is that the consensus of professional forecasters improves inflation forecasts because it acts like an average of many different inflation forecast models. Professional forecasters are experts and most of them will have a quantitative inflation model to make their predictions. By using the consensus of professional forecasters, one is essentially averaging the forecasts of a number of different models, which is known to improve forecasting accuracy in all kinds of finance applications.
But does that mean that the inflation expectations of professional forecasters have any bearing on future inflation? I doubt it. While these two papers are not a smoking gun, they demonstrate at least that inflation expectations generally have no or at best a very small impact on future inflation.