Abstract:
Forecasting with macroeconomic variables brings about several challenges, one of them being the presence of data revisions. When new sample information is available to National Agencies, economic estimates are revised because macroeconomic variables are characterised by hard-to-process samples that need to be often updated as new data is gathered by governments. The relationship between revised (i.e., latest available) and unrevised (i.e., real-time) data represents the fulcrum of the thesis, which is investigated and applied to the unemployment variable. The estimating sample consists of 28 countries, and unemployment data has been gathered from the OECD’s Economic Outlook issues from 1996 to 2019. Regarding the work’s structure, a description of the basic introductory concepts is followed by a preliminary analysis, aiming at shedding light on the impact of data revisions and whether such revisions translate into lower time-series volatility. The analysis carries on with an empirical exercise, developed using STATA, in which unemployment data is forecasted for the years 2019, 2020, and 2021 and then compared to the OECD’s forecasts. Ultimately, the thesis aims to examine if, and if so, how unemployment forecasting is affected by data revisions.
