The Institute of Public Finance and the Croatian National Bank presented, on 14 May, 2026, the award-winning papers of the Hanžeković Foundation competition for 2025.

Frane Banić, from the Research Area and Economic Analysis Department at the Croatian National Bank, presented the winning paper in the regular category (co-authored by Guzmán González-Torres from the Economics, Statistics and Research Department of the Italy’s Central Bank) which provides new insights into the behaviour of inflation in the smaller economies of the euro area. 

The authors examine recent inflationary pressures in Croatia, using firmlevel survey data on production constraints and capacity utilization to untangle supply and demand factors in services and manufacturing, considering sectoral inflation divergence. Based on European Commission survey data on factors limiting production, the paper disentangles supply and demand effects in the services and manufacturing sectors, taking into account sectoral divergences during the inflationary period.

The results indicate that traditional output gap measures do not sufficiently capture inflation developments in the post-pandemic period, whereas the sectoral approach based on survey data provides a more reliable indicator of inflationary pressures. In the services sector, inflationary pressures were primarily driven by strong demand and labour shortages, while in manufacturing, alongside demand, supply chain disruptions and shortages of materials and equipment played a key role during 2021 and 2022. After supply constraints began to ease, inflationary pressures in industry gradually declined, whereas in services they remained elevated due to persistently strong demand and ongoing labour shortages.

Mihael Brusan, a student at the Faculty of Economics and Business, University of Zagreb, and Jakov Čorak, an intern at the Directorate for Economic Analyses of the Croatian National Bank, presented the award-winning paper in the student category, which for the first time examines the use of machine learning methods for forecasting inflation in Croatia, thereby introducing modern methods into the domestic research framework and heralding a new era of macroeconomic forecasting. 

Out-of-sample forecasts are generated for multiple horizons using ten models and four alternative sets of input features comprising lags of the target variable, conventional macroeconomic indicators, unconventional variables including Google Trends data, and a combined feature set. 

The results indicate that no single model consistently outperforms others across all settings. Nevertheless, SARIMA, a traditional statistical model for time-series analysis and forecasting, proved to be the most reliable and accurate model. As more data and new macroeconomic conditions become available over time, models such as LightGBM and other tree-based models are expected to play an increasingly important role in inflation forecasting.

The awarded papers were published and available at the Public Sector Economics, journal of the Institute of Public Finance. The competition for the 2026 year is open, the deadline for submitting papers is 1 June, more information is available on the Foundation's website.

Photo: CNB