This paper describes a short-term projection model for French economic activity, OPTIM, the aim of which is twofold. First it gives an early estimate of real GDP growth for the previous quarter, when no figure has yet been released by Insee, the French National Statistical Institute, along with flash estimates for main GDP components (consumption, investment, inventories and external trade) together with a breakdown by sectors (services, manufacturing, construction, equipment, agri-food). This appears particularly useful for the short-run analysis. In this respect OPTIM may be considered as a traditional bridge equation model since it links a particular indicator available generally ahead of the release of the quarterly national accounts with a quarterly aggregate like GDP, consumption… Second, this tool supplies also estimates for GDP growth and its main components for the current quarter and for the next quarter (i.e two and three quarters respectively following the latest reference period of Insee's GDP data release). A pool of (mainly) monthly variables is used, which are, sometimes, directly introduced in the specification but, more often, summarised by the implementation of a principal component analysis (PCA). The largest part of the set of indicators comprises survey data together with monthly traditional indicators (industrial production, consumption in manufactured goods…). But other data (in particular financial data) are also introduced. The outcomes of OPTIM rely on a relatively complex procedure involving about twenty equations and mixing two alternative approaches: a supply approach consisting in a direct modelling of GDP and a demand approach where GDP is the sum of consumption, investment, changes in stocks and net trade (exports minus imports). The discrepancy between these two estimates is distributed according to an original method, yielding a unique GDP estimation. The paper is organised as follows. Section 1 presents the main features of OPTIM. Section 2 deals with data description while section 3 addresses the data assessment's issue. In section 4, the main equations are described. Section 5 presents a general assessment of OPTIM in terms of forecasting record. Finally section 6 concludes and proposes some avenues for further developments
