The first prize was awarded to Igors Tatarinovs and Reinis Fals (SSE Riga) for their paper titled Phillips Curve in the Eurozone: Forecasting Inflation and the Output Gap Using Explainable Neural Networks.
"This paper introduces the use of Hemisphere Neural Networks for forecasting the inflation and output gap in the eurozone within the macroeconomic model of Phillips Curve. The use of this method is challenging due to the complexity of data structure of this region. The study implements advanced data preprocessing, machine learning techniques and optimisation algorithms and scales the Hemisphere Neural Networks for forecasting output gap and inflation components, augmenting the research with comparison with conventional models," explained LB.
Hemisphere Neural Networks model is an innovative method, firstly introduced by P. G. Coulombe in 2022 in a study for the US. This paper adapts the method for the eurozone.
"Authors show that all tested HNN models surpass the benchmark linear regression and univariate ARIMA models in predictive accuracy for inflation forecasting and help with the interpretation of results. The results for predicting output gap showed models' ability to track general economic trends," said LB.
Second prizes were awarded to Luīze Mestere and Anastasija Sergeļa (SSE Riga) for Navigating Climate Risks: Assessing Cross-Border Spillovers of African Climate Shocks on European Union through International Trade and to Abigeila Keirāne and Jekaterina Šagejeva (SSE Riga) for Latvian Labour Market under the Microscope: Minimum Wage Impacts on Income Disparities Across Different Occupations.
Several third prizes and 'honourable mentions' were also awarded.
Mārtiņš Kazāks, Governor of Latvijas Banka, congratulated the authors of the winning papers who "received money prizes" according to a release from LB.
33 papers were submitted in all from 6 higher educational institutions. The winning papers will be published on the central bank's website.
