Modelling loans to non-financial corporations within the Eurozone : a long-memory approach

This paper uses fractional integration and cointegration methods to analyse the determinants of the amount of loans provided to non-financial corporations (NFCs) during the last three decades in four Eurozone countries, namely Germany, France, Italy and Spain. More specifically, ARFIMA (AutoRegressive Fractionally Integrated Moving Average) and FCVAR (Fractionally Cointegrated Vector Autoregression) models are estimated and then forecasts are also produced. All series are found to be highly persistent and long-run equilibrium relationships between them are also identified, confirming the role of real GDP and real gross investment as determinants of loans to NFCs. The forecasting accuracy of the FCVAR was also assessed by comparing it to that of the ARFIMA specifications, and the former were found to outperform the latter in all cases.