Learning about Monetary Union: An analysis of boundedly rational learning in European labour rraarkets

When the conduct of monetary policy changes, for example when monetary union is formed, economic agents, for example wage bargainers, must learn how to forecast inflation in the new environment. The strong version of the rational expectations hypothesis is that agents act as if they knew the true structure of the world and hence make no systematic mistakes in forecasting; the weak version would be that they learn about the structure of the world efficiently by observing how the monetary authorities behave. This paper compares the two versions of the hypothesis by embedding forward‐looking wage equations in the National Institute Global Econometric Model. It develops the work of Stephen Hall on learning by introducing data based Kalman filter expectations equations into the model. It is assumed that individuals in European labour markets are boundedly rational, and have to form expectations based on the information available to them. The reactions of the European economies under the assumption of model consistent expectations are compared to those with learning. Three possible realignments of the ERM are discussed, along with responses to an oil price shock. In the UK and France expectations based on learning are slower to react to shocks than are those based on model consistent expectations. The reverse is the case in Italy, suggesting that our estimated Kalman filter equation captures some of the lack of credibility of the Italian authorities. In all cases the European Central Bank is able to stabilise the price level, although this may not always be easy if individuals learn only slowly. The work reported here was undertaken in collaboration with the London Business School, using the world model which is maintained by the two teams jointly. Contact NIESR if copy required