Mixed-frequency VAR: a new approach to forecasting migration in Europe using macroeconomic data

Journal article

Forecasting international migration is a challenge that, despite its political and policy salience, has seen a limited success so far. In this proof-of-concept paper, we employ a range of macroeconomic data to represent different drivers of migration. We also take into account the relatively consistent set of migration policies within the European Common Market, with its constituent freedom of movement of labour. Using panel vector autoregressive (VAR) models for mixed-frequency data, we forecast migration in the short- and long-term horizons for 26 of the 32 countries within the Common Market. We demonstrate how the methodology can be used to assess the possible responses of other macroeconomic variables to unforeseen migration events—and vice versa. Our results indicate reasonable in-sample performance of migration forecasts, especially in the short term, although with varying levels of accuracy. They also underline the need for taking country-specific factors into account when constructing forecasting models, with different variables being important across the regions of Europe. For the longer term, the proposed methods, despite high prediction errors, can still be useful as tools for setting coherent migration scenarios and analysing responses to exogenous shocks

Authors: Barker, ER; Bijak, J

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Cambridge University Press
• Journal: Data & Policy
• Volume: 7
• Article number: e3
• Publication date: 2025-01-10
• DOI: 10.1017/dap.2024.82
• EISSN: 2632-3249
• ISSN: 2632-3249
• Language: English
• Keywords: Economics; Econometrics; Vector autoregression