Reference metadata describe statistical concepts and methodologies used for the collection and generation of data. They provide information on data quality and, since they are strongly content-oriented, assist users in interpreting the data. Reference metadata, unlike structural metadata, can be decoupled from the data.
The House Price Index (HPI) is a quarterly indicator that measures the changes in transaction prices of dwellings that households acquire on the market. Prices include land value.
The data on the HPI is provided in the form of an overall index number comprising two sub-indices: Purchases of newly built dwellings and purchases of existing dwellings.
The data is based on harmonized definitions and concepts included in the Commission Implementing Regulation (EU) 2023/1470 and described in the OOH Technical Manual.
3.2. Classification system
Index numbers, weights and growth rates are made available at the level of detail established by the Commission Implementing Regulation (EU) 2023/1470.
3.3. Coverage - sector
According to the EU-regulation only dwellings purchased by households should be considered in HPI. Due to the amount of transactions reported, Destatis decided to expand the scope of the HPI by dwellings that were purchased by other sectors than households (namely enterprises and public entities) up until the revision for 2017. Starting from the compilation of Q1 2018 and the revision of 2017, only dwellings purchased by households are taken into account in the calculation of the strata (not published) of new apartments, existing apartments and existing houses. For new houses, purchases by other sectors are still taken into account due to the in any case low number of records for this index.
3.4. Statistical concepts and definitions
HPI data for the purchases of newly built and existing dwellings are compiled on the basis of full transaction prices. The HPI covers all transactions of dwellings (apartments and one/two family houses) made by households regardless of their final use (For new houses, purchases by other sectors are still taken into account due to the low number of records for this index.) The weights for the two sub-indices are estimated to be equal to the total value of dwelling transactions for new and existing dwellings, respectively. Both prices and weights include land value, as recommended by the Manual.
3.5. Statistical unit
Transactions of dwellings which are sold as a package of land and structure to private households (self-built houses and prefabricated houses are not included. These are part of OOH - O.1.1.1.2).
3.6. Statistical population
The target population is the set of all transactions of dwellings purchased by households (in Germany).
3.7. Reference area
The overall HPI index and the two sub-indices are available for Germany as a whole.
The HPI in general is a full census, with the exception of Baden-Wuerttemberg where currently a cut-off sample is applied. This was due to the small-scale organizational structure of the expert committees in property valuation in Baden-Wuerttemberg on the level on municipalities. Until recently, there were about 900 ECPVs in Baden-Wuerttemberg. For reasons of expense, data of ECPVs with a responsibility for municipalities with a population of less than 20,000 have therefore been excluded. About 50% of the population of Baden-Wurttemberg is covered with this cut-off sample. Therefore, the accuracy is expected to be quite good for Germany, even though not all ECPVs in Baden-Wuerttemberg are surveyed. However, ECPVs in Baden-Wuerttemberg have merged in the last two years and are still merging. Therefore, it is planned to include the missing municipalities in Baden-Wuerttemberg in the future.
For comparability over time see 15.2.
Since 2019 indices (base period 2015 Q4) for houses and flats are also available according to so called "districts" in terms of their settlement structure" which distinguish administrative districts along an urban- rural classification. 5 classes are available: Top 7 metropolises, other large cities not attached to a district, urban districts, rural districts with a larger population density, rural districts with a low population density.
3.8. Coverage - Time
Data are available since the first quarter of 2000 for all index series. For comparability over time see 15.2.
3.9. Base period
The HPI is a chain-linked Laspeyres-type price index that uses the year 2015 as reference year.
Index figures with reference year 2015 are available, as well as quarterly and annual growth rates in percentage. Weights are given as parts in 1000.
The compiled quarterly indexes represent the whole calendar quarter and are based on transactions of the quarter.
6.1. Institutional Mandate - legal acts and other agreements
EU legal acts: Framework Regulation (EU) 2016/792 of the European Parliament and of the Council, Commission Implementing Regulation (EU) 2023/1470 that modified the previously applied Commission Implementing Regulation (EU) 2020/1148. Legal acts national level: "Gesetz über die Preisstatistik" (last amended 20 December 2022).
6.2. Institutional Mandate - data sharing
Not applicable for the HPI.
7.1. Confidentiality - policy
Data are published for the HPI and the sub-indices of newly built and existing dwellings.
There are no national regulations that deal specifically with HPI confidentiality. The national legislation for statistical confidentiality in general is paragraph 16 of the Act on Statistics for Federal Purposes (Gesetz über die Statistik für Bundeszwecke - BStatG) of 22nd January 1987, last updated 20 December 2022 (BGBl. I page 2727).
7.2. Confidentiality - data treatment
Not applicable.
8.1. Release calendar
The indices have to be provided to Eurostat 85 days after the end of the reference quarter. The indices are published nationally approximately one working day before. The exact dates can be accessed via the release calendar.
Search for "Residential property price indices" the Eurostat release calender can be assecced using the following website: Release calendar - Eurostat (europa.eu).
8.3. Release policy - user access
All users may access the data at the same time. A press release is published for the HPI at the date announced in the annual release calendar. At the same time the HPI is available in our database GENESIS-Online. Both can be accessed via the website of the Federal Statistical Office of Germany (DESTATIS).
Quarterly dissemination for the indices and annual update of the weights.
10.1. Dissemination format - News release
A regular press release is published for the HPI at the date announced in the annual release calendar.
10.2. Dissemination format - Publications
Besides the quarterly press release and the articles on methodology in our WISTA journal, the latest publication was in the chapter "housing" of the "Datenreport" which has been published in March 2021. The "Datenreport" is a report on various issues affecting society. It is published by the Federal Agency for Civic Education together with DESTATIS, the Social Science Research Center Berlin, the DIW (German Institute for Economic Research) and the German Federal Institute for Population Research.
The table for annual data has the code 61262-0001, the table for quarterly data has the code 61262-0002.
10.4. Dissemination format - microdata access
Microdata is not disseminated.
10.5. Dissemination format - other
Not relevant.
10.6. Documentation on methodology
The methodologies on which the HPI is based and updated results have been published in various articles of “Wirtschaft und Statistik” (Economics and Statistics), a bimonthly publication of DESTATIS (mainly) in German language, which can be accessed on the DESTATIS website.
The requirements for the work of the statistical offices are defined in the European Statistics Code of Practice and the Quality Assurance Framework. The European Statistics Code of Practice as well as the Quality Assurance Framework serves as the basis for the Quality Manual of the Statistical Offices of the Federation and the Laender in Germany. Furthermore there is a guideline for the handling of mistakes and revisions.
The HPI responds the needs of all users interested in knowing the behavior of prices in the house price market. Important users are the Deutsche Bundesbank and the European Central Bank (ECB). Users would welcome an earlier publication of the HPI. The publication of early estimates has been tested during the Covid-19 pandemic in 2020 (they were followed by larger revisions due to late(r) reporting). Due to a lack of resources this early estimate could not continue to be published after 2020.
12.2. Relevance - User Satisfaction
No regular surveys on user satisfaction are conducted, however regular (in general once a year) exchanges of views take place with the Deutsche Bundesbank and the Federal Institute on Building, Urban Affairs and Spatial Development (BBSR).
12.3. Completeness
All relevant expenditure categories defined in Commission Implementing Regulation (EU) 2023/1470.
13.1. Accuracy - overall
The HPI is perceived to be accurate and to reflect well the overall behavior of the housing market.
13.2. Sampling error
Not applicable.
13.3. Non-sampling error
Non-sampling errors are not quantified for the HPI. Unit and item non-response are monitored. Item non-response by ECPVs and overall non-response are calculated, and the causes of non-response are analyzed. Approximately 45% of all transmitted data can not be processed due to missing information in important variables. About 5% of records are excluded due to plausibility checks.
14.1. Timeliness
HPI data are sent to Eurostat, using eDamis, no later than 85 days after the end of the reference quarter, in accordance with regulation (EU) 2016/792 (usually they are sent approximately two working days before t+85). The indices are published nationally approximately one working day before.
14.2. Punctuality
Typically there is no time lag between the actual delivery of the HPI and the target date when it should have been delivered.
15.1. Comparability - geographical
Not applicable.
15.2. Comparability - over time
The comparability of the HPI is limited by the fact that not all federal states reported data when the HPI was first set up: The geographical coverage for HPI has been improved continuously. From 2010 to 2012 nine federal states reported data for the HPI. In 2013 data collection was expanded to cover the whole economic territory of Germany. Since 2015 ECPVs of all 16 federal states transmit data. Therefore, the HPI before 2015 is not exactly comparable to the HPI after 2015. This information is shared in the metadata and the german quality report to inform users about this break.
In 2019 the HPI was revised back until 2016 due to methodological changes in the calculation. Therefore there is a break in the time series of the HPI before 2016 and from 2016 on. The revision and an explanation why it was done was published in German together with the press release in 2019 and can still be accessed by the following website: Revision des Häuserpreisindex - Statistisches Bundesamt (destatis.de).
The HPI was revised due to methodological problems with the variable representing the location of the dwelling in the regression. The location had been represented by a classified version of the “standard ground value” (SLV) , that is calculated by the ECPVs for areas within a municipality. This SLV is adapted by the ECPVs usually on a yearly basis, based on several variables including the development of the prices. To avoid that one and the same house would change SLV-classes between two periods and therewith cause an underestimation of the price development, the calculation was adapted. Instead of using the classes of the standard land value, another variable depicting the location is now used for metropolises, where the prices had been rising to the largest extent.
15.3. Coherence - cross domain
The HPI partly overlaps the scope of the other owner-occupied house price indices described in Commission Implementing Regulation (EU) 2023/1470. Contrary to these other indices, the HPI does not follow the net acquisitions concept and the price of land is always included in prices and weights of the HPI. Any comparison of the HPI with the other OOH price indices should be done with this in mind.
The VdP ("Verband deutscher Pfandbriefbanken") publishes an index called "vdp Price Indices for Owner Occupied Housing", however it is conceptually more comparable to the HPI than the OOHPI. Up until now the index shows a similar trend to the overall HPI on a national level. The HPI is based on transaction data from the notaries (via the ECPVs) while the VdP index is based on bank data. The HPI therefore covers also transactions that where not financed by a bank loan.
15.4. Coherence - internal
All figures are checked for internal consistency.
A software module was developed that extracts data from the administrative database of the ECPVs and delivers the extract to Destatis automatically. Therefore the burden is considered minimal. However those ECPVs that do not use one of the major software products, have to extract data from their respective database. The data itself should in principle be recorded by the ECPVs due to their own tasks, even without Destatis demanding it.
17.1. Data revision - policy
Planned revisions: The HPI is revised regularly. The previous quarter is revised with the first calculation of the current quarter. The year is usually revised with the calculation of the first quarter of the next year.
Method change-related revisions: The changeover to a new index base takes place every ten years. The last change was in 2017, where the new base year 2015 was introduced.
17.2. Data revision - practice
Planned revisions: The main reason for planned revisions is additional data that are transmitted by the ECPVs.
Unplanned revisions: So far, the HPI has been revised unplanned once for the full years 2014 and 2015, as there was a significant improvement in the data situation as a result of subsequent deliveries following the renewed consultation of the ECPVs.
18.1. Source data
See below.
18.1.1. Prices
Prices: Type of data set: administrative data (transaction prices - both cash and mortgage) for new (turnkey-ready) dwellings and existing dwellings (census data in areas where possible). New (turnkey-ready) dwellings data and existing dwellings data are collected from the local Expert Committees for Property Valuation ('Gutachterausschüsse für Grundstückswerte'). The main tasks of these expert committees is on the one hand to keep their data bases on real estate transactions up-to-date by recording information from the notary deeds and enrich these data with information from surveys of the respective buyer of the property. Some ECPVs (especially those of larger cities) also provide market reports for their area of responsibility. On the other hand it is their task to estimate current market values of dwellings and land.
18.1.2. Weights
House Price Index: The weight, i.e. the total expenditure (total revenue) in the residential property market is mainly derived from data taken from GEWOS (Institute for City, Regional and Housing Research). Price index for (the purchase of) new dwellings: weight is derived from NA data (gross fixed capital formation according to housing). The additional mark-up for land is derived from data on revenues on residential building land, published by GEWOS. Existing dwellings index: difference between total and new dwellings.
Weights for the strata (houses/flats) below the sub-indices are calculated from the ECPV transaction data.
18.1.3. Source data - House Sales indicators
Not applicable.
18.2. Frequency of data collection
All data are collected quarterly.
18.3. Data collection
Data are reported electronically via "eSTATISIK.core".
18.4. Data validation
Various procedures ensure that errors are minimized in the data. The following validation checks are made on each record in the source data:
• Extreme values and obviously erroneous data are identified and eliminated by cut-off values based on the analysis of yearly data, • Multidimensional outliers are identified by Least Trimmed Squares and eliminated, • ECPVs which sent extraordinarily more/less data in a quarter compared to the previous quarters are contacted to ensure that no technical errors have occured, • If a lot of records of an ECPV show conspicuous patterns in scatter plots, then the ECPV is contacted to find out the reasons.
18.5. Data compilation
See below.
18.5.1. Calculation and Aggregation
See below.
18.5.1.1. Index formulae
The HPI should be a Laspeyres-type price index. Currently a price updating of the weights (based on data from t-1) between t-1 and q4 of t-1 is not yet implemented. It is therefore a Young index. Price updating will be implemented with the next updating of the weights in June 2024.
18.5.1.2. Aggregation method
Below the level of newly built and existing dwellings, the data is stratified for houses and apartments. The hedonic imputation approach is used to adjust for quality differences in the calculation of each of those sub-indices. Price relatives between the current quarter and the fourth quarter of the previous year are calculated. Subsequently, a geometric mean is calculated of these price relatives (Jevons index) for new houses, existing houses, new apartments, existing apartments.The weights described in 18.1.2 are then applied. The weights for the houses/ flats strata are currently updated less than yearly (last update 2017). The aim is to update these stratum weights on a yearly basis as well once house sales indicators are received by the ECPVs (EU funded project ongoing). The weights for the sub-indices of new and existing dwellings are updated on a yearly basis as requested by the OOH/HPI regulation.
18.5.1.3. Derivation of sub-index and elementary aggregate weights
The weights for the houses/flats strata are currently updated less than yearly (last update 2017). The aim is to update these stratum weights on a yearly basis as well once house sales indicators are received by the ECPVs (EU funded project ongoing). The weights for the sub-indices of new and existing dwellings are updated on a yearly basis as requested by the OOH/HPI regulation.
18.5.1.4. Price updating
Weights are not yet adjusted with a price change between the year t-1 and the fourth quarter of the year t-1. This will be implemented with the next update of the weights in June 2024.
18.5.1.5. Chaining and linking method
The indices are chain-linked every year with the fourth quarter of the last year.
18.5.1.6. Compilation of sub-indices
Hedonic double imputation.
18.5.1.6.1. Hedonic method
The hedonic imputation approach is used to adjust for quality differences in the calculation of each of those sub-indices. The explanatory variables in the regressions are (a polynomial of) the age of the building, (the logarithm of) the floor area, (the logarithm of) the living space and several dummy variables for the location. The dependent variable is the log price of the dwelling. The estimated prices from these two models are then aggregated using the Jevons formula to new houses, existing houses, new apartments, existing apartments.
18.5.1.6.2. SPAR method
Not applicable.
18.5.1.6.3. Stratification
Below the level of newly built and existing dwellings, the data is stratified for houses and apartments.
18.5.2. Other processing issues
See below.
18.5.2.1. Timing for pricing
The price of the dwelling is recorded when the binding contract is signed. The transaction is used for the respective quarter, in which the contract was signed.
18.5.2.2. Treatment of VAT
VAT is not included in the price of new dwellings.
18.5.2.3. Treatment of other taxes
Taxes other than VAT are not included in the price of dwellings.
18.5.2.4. Treatment of government subsidies
Government subsidies are relevant for financing of the transaction, but not for the transaction prices.
18.5.2.5. Treatment of land
The price of land is included in both prices and weights.
18.5.2.6. Housing cooperatives
Housing cooperatives are not covered in the HPI.
18.5.2.7. Treatment of non-market transactions
The following cases of property transactions are not to be transmitted by the ECPVs: transfer by the way of gift, expropriations, hereditary building right, inheritances, exchanges of properties.
18.5.2.8. Treatment of multi-object transactions
The ECPVs do not transmit multi-object transactions to us if they can not be split up into single units.
18.5.2.9. Treatment of fractional transactions
The german HPI covers freehold flats, where single apartments in a multi apartment building are sold. Other fractional transactions are not included.
18.5.2.10. Outliers detection
Outliers, such as dwellings with a particularly small or large floor area or very high or low prices per m2 etc. are excluded from the index. The limits for these assessments are based on the HPI data of the previous year and updated every year. The indices are then calculated always based on the same plausibility limits in the current and the reference period in order to avoid an impact of the plausibility limits to the indices. Starting from June 2018 for the revision of 2017 and the provisional results for Q1 2018 Least Trimmed Squares is implemented as a further measure to detect and exclude multidimensional outliers. Furthermore, after the automatic processes described above Scatterplots are checked for further noticeable records.
18.5.2.11. Treatment of incomplete data source coverage
Not applicable (Eurostat project on this issue is ongoing).
18.6. Adjustment
Not applicable.
No comment.
The House Price Index (HPI) is a quarterly indicator that measures the changes in transaction prices of dwellings that households acquire on the market. Prices include land value.
The data on the HPI is provided in the form of an overall index number comprising two sub-indices: Purchases of newly built dwellings and purchases of existing dwellings.
The data is based on harmonized definitions and concepts included in the Commission Implementing Regulation (EU) 2023/1470 and described in the OOH Technical Manual.
13 November 2024
HPI data for the purchases of newly built and existing dwellings are compiled on the basis of full transaction prices. The HPI covers all transactions of dwellings (apartments and one/two family houses) made by households regardless of their final use (For new houses, purchases by other sectors are still taken into account due to the low number of records for this index.) The weights for the two sub-indices are estimated to be equal to the total value of dwelling transactions for new and existing dwellings, respectively. Both prices and weights include land value, as recommended by the Manual.
Transactions of dwellings which are sold as a package of land and structure to private households (self-built houses and prefabricated houses are not included. These are part of OOH - O.1.1.1.2).
The target population is the set of all transactions of dwellings purchased by households (in Germany).
The overall HPI index and the two sub-indices are available for Germany as a whole.
The HPI in general is a full census, with the exception of Baden-Wuerttemberg where currently a cut-off sample is applied. This was due to the small-scale organizational structure of the expert committees in property valuation in Baden-Wuerttemberg on the level on municipalities. Until recently, there were about 900 ECPVs in Baden-Wuerttemberg. For reasons of expense, data of ECPVs with a responsibility for municipalities with a population of less than 20,000 have therefore been excluded. About 50% of the population of Baden-Wurttemberg is covered with this cut-off sample. Therefore, the accuracy is expected to be quite good for Germany, even though not all ECPVs in Baden-Wuerttemberg are surveyed. However, ECPVs in Baden-Wuerttemberg have merged in the last two years and are still merging. Therefore, it is planned to include the missing municipalities in Baden-Wuerttemberg in the future.
For comparability over time see 15.2.
Since 2019 indices (base period 2015 Q4) for houses and flats are also available according to so called "districts" in terms of their settlement structure" which distinguish administrative districts along an urban- rural classification. 5 classes are available: Top 7 metropolises, other large cities not attached to a district, urban districts, rural districts with a larger population density, rural districts with a low population density.
The compiled quarterly indexes represent the whole calendar quarter and are based on transactions of the quarter.
The HPI is perceived to be accurate and to reflect well the overall behavior of the housing market.
Index figures with reference year 2015 are available, as well as quarterly and annual growth rates in percentage. Weights are given as parts in 1000.
See below.
See below.
Quarterly dissemination for the indices and annual update of the weights.
HPI data are sent to Eurostat, using eDamis, no later than 85 days after the end of the reference quarter, in accordance with regulation (EU) 2016/792 (usually they are sent approximately two working days before t+85). The indices are published nationally approximately one working day before.
Not applicable.
The comparability of the HPI is limited by the fact that not all federal states reported data when the HPI was first set up: The geographical coverage for HPI has been improved continuously. From 2010 to 2012 nine federal states reported data for the HPI. In 2013 data collection was expanded to cover the whole economic territory of Germany. Since 2015 ECPVs of all 16 federal states transmit data. Therefore, the HPI before 2015 is not exactly comparable to the HPI after 2015. This information is shared in the metadata and the german quality report to inform users about this break.
In 2019 the HPI was revised back until 2016 due to methodological changes in the calculation. Therefore there is a break in the time series of the HPI before 2016 and from 2016 on. The revision and an explanation why it was done was published in German together with the press release in 2019 and can still be accessed by the following website: Revision des Häuserpreisindex - Statistisches Bundesamt (destatis.de).
The HPI was revised due to methodological problems with the variable representing the location of the dwelling in the regression. The location had been represented by a classified version of the “standard ground value” (SLV) , that is calculated by the ECPVs for areas within a municipality. This SLV is adapted by the ECPVs usually on a yearly basis, based on several variables including the development of the prices. To avoid that one and the same house would change SLV-classes between two periods and therewith cause an underestimation of the price development, the calculation was adapted. Instead of using the classes of the standard land value, another variable depicting the location is now used for metropolises, where the prices had been rising to the largest extent.
