Based on the National Housing Policy Plan of the Republic of Croatia, the application of subsidy measures for the acquisition of first residential real estate for citizens under the age of 45 started in June 2025. The measure continues, and so far over 4000 applications have been received.
To calculate subsidies for the acquisition of first residential real estate under contracts concluded in 2026, APN will use new average prices in accordance with the Study developed by the Institute of Economics, Zagreb.
Taking into account that in 2025 only 142 local self-government units recorded a sufficient number of apartment purchase and sale transactions to calculate an average apartment price, while in the remaining 414 units there were either no transactions or an insufficient number of them to calculate an average price, the Institute of Economics conducted an analysis through several stages of the research process in order to estimate average prices in units where it was not possible to calculate the price directly from the data.
In all stages of the analysis by the Institute of Economics, data from the eReal Estate information system were used. The analysis included the achieved purchase and sale prices of apartments, whereby VAT was excluded for real estate subject to the value added tax (VAT) system, since precisely the price excluding VAT constitutes the basis for determining eligibility for a tax refund under the Programme of subsidies for the acquisition of first residential real estate.
The first analysis stage included local self-government units which had over 100 apartment purchase and sale transactions recorded in the eReal Estate base for the first 11 months of 2025. There were 33 such units. These are also the largest and the most developed local real estate markets in Croatia, including cities such as Zagreb, Split and Rijeka, as well as larger coastal and continental centres.
This dataset was used to determine the minimum number of transactions necessary for a reliable calculation of an average price. The bootstrap analysis suggests that, in average, a minimum of twenty transactions is necessary to consider the calculation reliable. Hence, an average apartment price for those local units which recorded twenty and more transactions in the first eleven months was calculated directly from the transactional data available through the eReal Estate information system. This means that for 98 largest local self-government units in Croatia, an average apartment price excluding VAT was calculated directly from available transactional data.
For local self-government units with less than 20 transactions or with an insufficient number of calculations for direct calculation, average prices were estimated based on contemporary statistical methods and machine learning methods. The initial seven models of machine learning were trained by using 20 demographic, economic, spatial and market variables available for all units in Croatia, and they included spatial relations between neighbouring markets as well. Based on the results of training these models, five key variables were selected that best explain and predict the movement of average apartment prices in certain local self-government units where it is possible to calculate the apartment price directly.
As the final model to estimate an average apartment price, used to estimate average prices in local units with no sufficient number of transactions, a multiple linear mixed-effects regression model with five key variables was selected. It demonstrated the best balance of predictive accuracy, stability and interpretability. The selection criterion was the root mean square error (RMSE), which particularly penalises larger deviations and is suitable for official estimations in public policy.
By applying said methodology, complete spatial coverage of all local self-government units in Croatia was ensured, as well as the largest possible degree of internal consistency and congruency of estimates, which can be achieved considering the substantially large number of local units for which there are no directly recorded data on average apartment prices.
In the table of average prices, prices are thus available for all local self-government units, while the City of Zagreb is further divided into cadastral municipalities due to significant differences among them.
It is important to note that for all contracts concluded in 2025, the Table of average prices referring to 2024 will be used, regardless of when citizens submit their application.
News |
Information
