Learning in an Estimated Small Open Economy Model

Expectations of the future play a key role in the transmission of monetary policy. Over recent years, a lot of theoretical and applied macroeconomic research has been based on the assumption of rational expectations. However, estimated models based on this assumption typically fail to capture the dynamics of the economy unless mechanical sources of persistence, such as habit formation in consumption and/or indexation to past prices, are imposed. This paper develops and estimates a small open economy model for Australia assuming two different types of expectations: rational expectations and learning. Learning – where expectations are formed by extrapolating from the historical data – can be an alternative means to generate the persistence observed in the data. The paper has four key findings. First, learning does not reduce the importance of conventional mechanical forms of persistence. Second, despite this, the model with learning is able to generate real exchange rate dynamics that are consistent with empirical models but which are absent in standard theoretical models. Third, there is some tentative evidence that learning is preferred over rational expectations in terms of fitting the data. Fourth, since the adoption of inflation targeting, agents appear to be using a longer history of data to form their expectations, consistent with greater stability of inflation.