Conventional macroeconomic time series models generally only include aggregate variables (e.g., output, inflation, unemployment), and can only be used to address “aggregate questions”. In this paper we bridge aggregate (macro) and micro data in a functional-Vector Autoregression (fVAR). Whereas these tools have been employed already for structural analysis, we extend the model to appropriately consider the mixed-frequency nature of the data, and we evaluate the nowcasting and forecasting performance of the fVAR. In an application to UK data, we derive appropriate inter-temporal restrictions to link unobserved quarterly income distribution to observed annual income distribution, and we perform a real-time out-of-sample forecasting exercise. We find that correctly exploiting the mixted-frequency of the data can deliver improvements in the predictive ability of the model.
© 2024 | Università degli studi di Verona
******** CSS e script comuni siti DOL - frase 9957 ********p>
