Estimating the Primary Energy Demand of the Hungarian Housing Stock*

Released: 30 January 2026

Data Requirements and Background of the Calculations

Assessing the energy demand of the residential building stock and monitoring changes in household energy consumption is a new area in Hungarian statistics. Among the related obligations of the Hungarian Central Statistical Office (HCSO), several development directions designated by Eurostat can be mentioned, aimed at determining the energy quality of the housing stock (Statistics for the European Green Deal, Environmental Accounts, and Waste Statistics), as well as those requirements expected to be set by the EU’s forthcoming new demographic statistical framework (ESOP). The European Union’s Energy Performance of Buildings Directive (EPBD), adopted in 2024, also identifies numerous objectives that necessitate an examination of the condition of the housing stock.

In the following, we present a possible solution for model-based calculation of the energy demand of the housing stock. As a precursor to the current work, in 2023 we prepared an estimate of the energy quality of the housing stock for 2020. For this, the 2016 microcensus dataset was extended with data on dwellings built between 2017 and 2020, and an estimate was produced for the resulting representative sample of dwellings using the energy performance certificates available at the time (Bene et al., 2023).

During the 2022 census, the entire housing stock was enumerated, allowing us to perform calculations not on a representative sample but on the full dataset. Since the census housing enumeration was limited to the physical characteristics of dwellings, our task was to calculate energy demand based on these attributes. Naturally, such information alone is not sufficient to determine the specific energy characteristics of individual dwellings. Therefore, our calculations aimed to estimate the likely energy quality of a given dwelling based on the combination of census variables. For this purpose, we again relied on energy demand values established in available energy performance certificates. These certificates were provided to the HCSO for statistical use by the Lechner Knowledge Centre (LTK), and the data were linked to the census database using property addresses.

From November 2023, the energy certification system was restructured: the previously valid ministerial decree (TNM) was replaced by a decree of the Ministry of Construction and Transport (ÉKM). Accordingly, in our calculations we used only certificates issued between 2020 and 2023 under the TNM decree. The analysis thus refers to the situation in 2022, the census reference year. The inclusion of certificates issued in 2023 does not cause significant distortion, since energy requirements did not change until the end of October 2023 (certificates from November and December 2023 were excluded), and the turnover of the housing stock or renovations within a single year have only negligible impact on the results. The relationship between building characteristics and energy demand was examined using regression methods, while the final estimates were calculated with the random forest method (details are available in the methodological annex).

The results of the calculations originally corresponded to values defined under the TNM decree. However, since it is expected that our results should also be interpretable under the new system, the individual estimates obtained were converted to the ÉKM decree framework. For this conversion, we applied the model proposed by Horváth et al. (2024, 2025).

Comparison of Results for 2020 and 2022

According to the calculations for 2022, the annual energy consumption per square metre of an average Hungarian dwelling was 274 kWh. Two years earlier, the corresponding figure had been 281 kWh. Although the change indicates a decline, this alone is not necessarily sufficient to draw conclusions about the renewal rate of the housing stock. To monitor renewal accurately, further regular investigations are required.

The calculations performed two years apart yielded closely aligned results, with estimated values decreasing across major building types and regions. The energy demand per square metre of detached houses was – as previously – 65% higher than that of apartments in multi-dwelling buildings. Based on data converted to the new certification system, the average energy demand in 2022 was 285 kWh/m² per year.

Based on the calculated energy demand of individual dwellings, under the classification valid until 2023, 9.4% of dwellings fell into the best categories (AA–CC), while 15.3% were placed in the most wasteful categories (II–JJ). Under the new ÉKM classification, somewhat more dwellings were assigned to the extreme categories: 8.1% to class A–C, and 28% to class I (class J does not exist in the new system).

For clarity, all subsequent results are presented on the basis of values estimated using the random forest method and converted according to the ÉKM decree.

Detailed Results

The database covering the entire housing stock makes it possible to present in detail the relationships that are relevant for shaping policy measures. In terms of energy demand, the type and age of buildings are the most decisive factors. The most energy‑intensive dwellings are found in detached houses built between the two world wars, with an average annual energy demand of around 410 kWh/m².

The largest share of the current housing stock was built between 1961 and 1980. Detached houses from this period have extremely high energy demand (375 kWh/m²/a), while multi‑dwelling buildings of the same era already show significant improvement (187 kWh/m²/a). The improvement continues in dwellings built between 1981 and 2000, yet the energy efficiency of detached houses lags considerably behind that of apartments due to the geometric characteristics of the buildings.

The advantage of multi‑dwelling buildings constructed before the regime change lies in their small specific cooling surfaces and relatively efficient district heating, and partly in the fact that the energy retrofitting of panel buildings was carried out on a mass scale after the transition, albeit not comprehensively (Bene–Szabó, 2019). According to our calculations, the average energy demand of panel apartments was 146 kWh/m²/a in 2022, the lowest among multi‑dwelling buildings. In this respect, there is hardly any difference between the first and second generations of panel apartments: those built between 1961 and 1980 averaged 147, while those built after 1980 averaged 143 kWh/m²/a.

The striking peak around 130–150 kWh/m²/a in the distribution of apartment energy demand by construction period is explained by the approximately 560 thousand panel apartments built during the era of mass housing construction. The energy quality of multi‑dwelling buildings constructed at the same time but without panel walls was less favourable: brick apartment houses of the 1960s and 1970s averaged 238 kWh/m²/a, while medium‑ or large‑block cast‑concrete buildings averaged 194 kWh/m²/a. The difference again stems from building geometry: non‑panel apartment houses are generally smaller, with larger specific cooling surfaces. Between 1980 and 2000, significant improvement is evident in both cases (191 kWh/m²/a for brick and 173 kWh/m²/a for cast‑concrete), yet these values still exceed the energy demand of panel dwellings.

The figure shows that as we move towards newer construction age groups, the energy demand of buildings generally decreases. However, for multi‑dwelling buildings constructed before 1960, the year of construction is virtually irrelevant in terms of energy demand, and only after this period does the improving quality of newer buildings become evident. There is one exception: the age group built between 2011 and 2015 may have been affected both in quality and in number by the housing market crisis. The average energy demand of the mere 22 thousand apartments built in this period slightly exceeded that of the preceding age group. Thereafter, a further decline can be observed, with apartments completed after 2015 showing the most favourable indicator among all categories examined (an average of 115 kWh/m²/a).

The largest age group of detached houses also dates from the 1960s and 1970s. Their distribution curve has a mode of 410 kWh/m²/a and an average of 375 kWh/m²/a, both considerably higher than those of apartments built in the same period. Detached houses built in the 1980s and 1990s represent a transitional stage: the curve is not characterised by a single peak, with most dwellings spread across a broad range between 200 and 400 kWh/m²/a. This means that even among these relatively recent buildings there are about 270 thousand dwellings with energy demand above 310 kWh/m²/a, placing them in class H or worse. The vast majority of detached houses were built of brick, with average energy demand consistently lower than that of contemporaneous adobe houses, a difference stemming mainly from poorer equipment rather than wall material. Ninety‑six percent of adobe dwellings were built before 1981, with energy demand of 410–430 kWh/m²/a. Although the few adobe houses built since then show some improvement, their indicators remain higher than those of detached houses of similar age.

As settlement size increases, the typical energy demand of dwellings gradually decreases for both building types. Considering the average of all buildings, the most favourable indicators were observed in provincial large cities. In Budapest, the quality of detached houses is better than in any other settlement size category, while that of apartments lags behind those in medium‑sized and large provincial cities, as Budapest has the highest proportion of old apartment buildings.

The census database also contains information on whether dwellings are occupied. Although this characteristic was not included among the explanatory variables of our estimation model, results based on other dwelling attributes consistently showed less favourable outcomes for unoccupied dwellings. Nationwide, the average indicator for occupied dwellings was 283 kWh/m²/a, compared with 299 kWh/m²/a for unoccupied ones. As expected, the difference is particularly pronounced in detached houses: while occupied detached houses had an energy demand of 330 kWh/m²/a, unoccupied ones averaged 382 kWh/m²/a, raising questions about cost‑effective renovation (Ertl et al., 2021).

The district‑level estimated average energy demand ranges between 174 and 411 kWh/m²/a. The lowest value characterises Budapest’s 3rd district, followed by the 4th, 11th, and 13th districts, each with averages below 200 kWh/m²/a. These districts are distinguished by a high proportion of housing estates and/or newly built dwellings.

Districts with the highest energy demand are those whose housing stock is dominated by old detached houses (Cigánd, Jászapáti, Pétervására, Devecser, Mezőcsát districts). In addition to disadvantaged areas of the Great Plain, such districts are more frequently identified along border regions, particularly in Northern Hungary and Southern Transdanubia.

Methodology

References

*We thank Tamás Csoknyai (BME) and Áron Horváth (MEHI) for their valuable comments, which contributed to the preparation of this analysis.