This paper describes a forecasting model of the French consumer price index. Its purpose is to allow for rapid and detailed analysis of recent inflation developments, as well as frequent forecasts. Its characteristics are therefore the following: a small number of equations, a monthly frequency, and a fairly detailed sectoral breakdown of the price index. A feature of our forecasting procedure is also to systematically supplement the forecast with confidence intervals. The " core " part of the CPI (i.e. food excluding sensitive products, manufactured goods and private sector services ) is modeled using reduced-form Phillips curves Sensitive product components, especially fresh products, are then modelled. Oil products are linked in en ECM-form to crude oil price, which is considered as exogenous. The remainder of the price index is regarded as exogenous. Last, we consider methods for constructing the confidence intervals associated with the forecast, looking at three of them in particular: computation of RMSE, bootstrap simulation, and the approach developed by the Bank of England.
Éric Jondeau, Hervé Le Bihan and Franck Sédillot
Updated on: 06/12/2018 11:09