Accuracy and reliability

Overall accuracy

The aim of the Producer Price Index for Construction of Dwellings is to illuminate the price development of newly constructed and sold free standing one-family houses incl. producer profits. The target population is therefore all newly constructed free standing one-family houses in Denmark. The index is considered to have a high overall accuracy, although some sources of error exist.

Sampling error: The sample consists of reported price observations by companies selected by a top-down principle, where companies with a high turnover are selected first, in order to achieve as high a market coverage as possible from as few participants as possible. The sample is therefore not a random sample and it is not possible to estimate the overall size of the sampling error.

Non-sampling error: In 2021 and 2022 the sample constituted about 40 pct. of all newly built one-family houses in Denmark, which is considered a satisfactory market share. Data is collected from a small number of typehouse companies, selected by a top-down principle according to company turnover. Therefore, coverage error may occur from the data basis not including unique housing constructions. Furthermore, the applied hedonic model is based on a number of building characteristics, such as information about area size, number of rooms, floors and toilets, and geography etc. However, the model does not consider the quality of applied building materials or the quality of the construction work. About 5-10 pct. of the collected data is removed, mainly due to errors in the reported price observations, which are identified by the manual and computerized troubleshooting. Such non response errors are considered to be random, as no systemic bias has been observed, and it is therefore not believed to significantly impact the accuracy of the statistic.

Overall, the statistic is considered to have a high accuracy compared to the actual price level of newly built one-family houses in Denmark; particularly for standard dwellings, but to a lesser degree for unique houses. It can be argued that unique housing typically has a higher sales price and may use more expensive building materials, and that the statistic therefore may underestimate actual price levels. However, this ascertion is not based on any empirical evidence.

Sampling error

The sample is selected top-down to achieve as high turnover coverage as possible. The sample is therefore not a random sample and it is not possible to estimate the overall size of the sampling error.

Non-sampling error

Coverage error: The statistics is based on 30 to 40 percent of the completed one-family houses i Denmark, which is assessed to be a sufficient market coverage. The data sources are type house construction companies, which specialize in one-family housing, and in the statistic they therefore represent all newly built houses. The statistic therefore does not cover unique housing. It has not been possible to gather information on differences in characteristica between type houses and unique housing and it is therefore not possible to evaluate whether this coverage error leads to over- or underestimations of actual price levels. Data is collected from a small number of typehouse companies, selected by a top-down principle according to company turnover in order maximize coverage while minimizing the reporting burden. This may be a source of error, as prices and building methods may differ between major or minor companies. Coverage errors may in principle occur if type house companies mistakingly report prices of buildings other than free standing one-family houses, such as terraced houses or double houses. However, all collected data is coupled to register information from the BBR either manually or by machine, by which such errors are detected and removed, to ensure that only free standing one-family houses are included in the calculations. There may occur administrative errors in the Danish Buildings and Dwellings Register (BBR) but this is not considered to be a significant source of error.

Measurement errors: Reported prices may be erroneous, but this is not considered to be a significant source of error. All prices are examined manually for extreme values that may lead to exclusion of observations.

Unit non-response rate: About 5-10 % of reported prices are excluded. This is primarily due to reporting errors, such as incorrect building types or building periods, or non-identifiable buildings due to incorrect addresses. Also, a small number of observations may be excluded because building information have not been available in the BBR prior to publication of the index. No systematic bias has been observed amongst lapsed price observations, and it is therefore not considered a significant source of error.

Model assumption error: The price development is calculated using hedonic regression which implies some assumptions with regards to which characteristics influence the price of the construction of a one-family house. This model considers some but not all characteristica relevant to the price. The characteristics used in the calculation are the following: Floor area (residential, conservatory, basement, carport, attic and outhouse), roofing material, number of rooms, number of bathrooms, number of floors, geography, and the construction company involved. However, the model does not consider the quality of applied building materials or the quality of the construction work, as such information cannot be obtained from the administrative registers. This introduces risk of error, as a price development might be driven by changes in quality. However, it is expected that the quality of houses built in two consecutive quarters will have the same quality on average.

Quality management

Statistics Denmark follows the recommendations on organisation and management of quality given in the Code of Practice for European Statistics (CoP) and the implementation guidelines given in the Quality Assurance Framework of the European Statistical System (QAF). A Working Group on Quality and a central quality assurance function have been established to continuously carry through control of products and processes.

Quality assurance

Statistics Denmark follows the principles in the Code of Practice for European Statistics (CoP) and uses the Quality Assurance Framework of the European Statistical System (QAF) for the implementation of the principles. This involves continuous decentralized and central control of products and processes based on documentation following international standards. The central quality assurance function reports to the Working Group on Quality. Reports include suggestions for improvement that are assessed, decided and subsequently implemented.

Quality assessment

An overall quality assessment rests upon a combination of coverage, number reporting businesses and prices in the sample and he quality of the collected prices and the price methods used. The sample is selected in order to cover as large a share of the turnover in the industry as possible. Since the statistics was established in 2015, prices for 30 to 40 per cent of the completed one-family houses each year. The coverage varies according to the combined market share of the reporting companies that has been increasing throughout the time period. Prices are matched with information from the Buildings and Dwellings Register, e.g. floor area, number of bathrooms and roofing materials. This information is used in the calculation of the index using a hedonic regression - a method recommended by Eurostat for calculating price indices for heterogeneous products/services. The quality is monitored continuously and efforts are put in where the quality can be improved. The quality work e.g. consists of supplementing the sample.

Data revision - policy

Statistics Denmark revises published figures in accordance with the Revision Policy for Statistics Denmark. The common procedures and principles of the Revision Policy are for some statistics supplemented by a specific revision practice.

Data revision practice

No revision.