Statistical processing

Source data

Information on sales of individual cooperative housing units comes from quarterly reports. For the period 2015Q1 to 2021Q2 sales data is collected using a voluntary sample through the sector for cooperative housing. Sales data from 2021Q3 and onwards is collected through andelsboliginfo.dk where reporting is mandatory by law.

For the cooperative housing associations themselves, data originates from yearly reports through andelsboliginfo.dk, which has been mandatory since 2014.

Frequency of data collection

Quarterly.

Data collection

Data for this statistics is gathered electronically fra administrative registers at andelsboliginfo.dk and the Buildings and dwellings Register.

Data validation

The reports from andelsboliginfo.dk can be deficient and the data is only validated scarcely at andelsboliginfo. The purpose of andelsboliginfo is to support the cooperative housing associations with their reporting duties of the unit transactions. It is the associations' responsibility that the key figures are reported correctly. The reported data belongs to the associations and they are responsible for the quality of the data. Andelsboliginfo does have some checks to ensure that the reported figures are sensible. When Statistics Denmark receives data from andelsboliginfo extensive controls are made on 11 selected variables that are essential for the calculation of the statistics. On avereage 12 % of the observations are deemed unfit for inclusion in the calculations. To avoid exclusion of all 12 pct. of the sales imputations and corrections are made. This effort is focused on three essential variables. The methods to impute and correct depends on the variable, but generally five methods are used: 1. Replacement with a value from a different register 2. Calculation of the variable using other variables, that forms a theoretic identity (eg. area=unit value/association value pr. m2) 3. Correction of 10-, 100- and 1000 reporting errors by comparison with other variables that are usually strongly correlated with the erroneous variable 4. Replacement with a value from an almost perfectly correlated variable 5. Adoption of the most frequent value from the associations other accounts When then corrections are completer all checks are run again for all the variables that have been either corrected/imputed and the distribution is reassessed. Then the price concept is calculated as follows: Price concept=F2*distribution+Z18-Z3, Where F2 is the property value of the association by the used valuation principle, Z18 is the sales price and Z3 is the unit’s share value. The price concept must be within set thresholds. The data correction for 2015Q1 – 2021Q2 is more limited in order to keep as many observations as possible as the data for this period is based on a smaller sample. The data correction from 2021Q3 and onwards is more thorough as the data comes from the mandatory reports on andelsbolig.info.

Data compilation

Combining sales data and association data: Firstly, the information on sales of individual cooperative housing units with information on the cooperative housing association as a whole. There is no single correct way of making this combination so pragmatic solutions are chosen that includes as many sales a possible. Furthermore, it is prioritized to use the newest information on the association as possible. The following rules are applied: - The combination is conducted on the basis of the accounting period which is combined with the sale - The accounting period lies between 0 and 24 months before the date of the sale - If there are several accounting periods within 0-24 months before the sales data, the newest information is chosen Loan information (to include interest swap as an explanatory variable): According to law, the credit institutions must report the information in January along with the publication of the loan or financial agreement in question. I.e., in January 2022 the loans of December 31st 2022 are reported. When a new loan is reported in the system it is automatically attached to the information about the cooperative housing association as a whole. The following rules are applied: - The sales that have taken place in e.g. 2022 combined with the loans reported by December 31st 2021. I.e. sales from 2022 are combined with loans reported by December 31st 2021. Identification of duplicated sales from 2021Q3 and onwards (through andelsboliginfo.dk): If a sale matches the address of one or more sales, where the dates of sale are within 30 days of each other, these checks are run in this prioritised order to decide which sales to keep. If the first check is undecided it moves on to the next, etc.: 1. The sale where the most essential variables overall passes the consistency check 2. The sale with most individually passed the consistency checks 3. The sale where the most variables are within the threshold values 4. The sale with the most filled in variables in total 5. The sale with the newest date of sale Identification of duplicate sales in the periode 2015Q1-2021Q2 (sample data): Sales data for the period 2015Q1 – 2021Q2 has been gathered from several sources and thus, the same sales could have been reported to Statistics Denmark multiple times. If the address and the date of sale is the same but the sales prices does not match the observations are excluded as it is not possible to be sure which information is correct. Hedonic price index: Statistics Denmark has chosen use a hedonic regression for the calculation of the price index for cooperative housing. This is different from the method used for the calculation of the house price index where the property values can be used to control for differences in the quality of the housing units. This is not an option for cooperative housing. The development of a hedonic price index involves both regression and construction of quality corrected price indices. This approach is suitable when the prices must be corrected for several conditions as is the case for cooperative housing. The chosen hedonic method is the rolling time dummy (RTD) method and it is described further in 6.2.2 in the Technical manual on Owner-Occupied Housing and House Price Indices.

Adjustment

When calculating a price development you want to compare the price of the exactly same item in two different periods. The challenge with cooperative housing units is that they are very different and, thus, calculating a price development is not easy. To accommodate this hedonic regression is the chosen method. With hedonic regression an observable price can be related to a range of variables and estimate their effects on the price. E.g. a 0/1 dummy can be used to see if there "exclusively is cooperative housing unit in the housing association" and a variable specifying the "share of rented squaremeters in the association". By giving these variables the average value in all observations you can construct a price index corrected for the effects of a disproportionate large or small share of the cooperative housing units in the quarterly turnover.

No seasonal correction is made.