Statistical processing

Source data

Producer Price Index for Services is calculated on the basis of price reports directly from selected service producing companies. The sample consists of ca. 2.300 prices divided into a number of industry groups set out by the European Regulation that the statistic follows. Within each industry group, the largest companies are selected, based on turnover value (top-down) to achieve as high an industry turnover coverage as possible. By selecting the largest companies within a given product group, it is assumed that as per basic market theory these larger companies, as price influencers, have both a large direct share of transactions but also an even larger indirect representativeness which helps to ensure a representative sample for the entire product group. Each of these companies is requested to select their most representative service(s), i.e. those products that most represent their sales within a given product group. Since there is a big difference between the industries, they are treated independently with their own sample selection.

Each selected company is determined by their legal entity registration. In Denmark's Statistics Business Statistical Register, the legal entities are determined by their CVR number. In cases where a company has multiple CVR numbers sharing the same activity these can be combined into one unit from which prices are collected, for example, franchises. The companies must, as far as possible, report the average price for the quarter, if that is unavailable they must report the price in force on the 15th of the mid-month of the quarter. For industries or products with high variation within the quarter – an average price for the whole month is always preferred.

Weights for aggregation purposes come from the National Accounts product balances, based on several internal primary sources (industrial commodity statistics, account statistics and foreign trade statistics). This covers the main part of all trading in the first turnover stage in Denmark, within the delimitation of the purpose of the statistics. The weights used are implemented with a lag of 4 years.

Frequency of data collection

Quarterly.

Data collection

Prices are collected through an electronic reporting form, which is sent to relevant contacts within the selected companies.

Data validation

The first validation of price data happens when prices arrive to Statistics Denmark. Here they are auto tested for unusual developments. The prices which development is greater than a predetermined threshold value, are checked manually by the staff and accepted only if the companies can verify the change. When all prices are received, the system generates a list that includes all price changes and a measure of how these affects the elementary aggregates. The final validation is a qualitative inspection of all calculated index tables. The last validation is a manual inspection of all calculated indices, where index movements over a longer period, and in different markets, are reviewed to ensure that any atypical developments are also investigated.

Data compilation

The Producer Price Index for Services are calculated in a hierarchical system, where the prices collected are divided into CPA groups, which are the most detailed groups of services. The CPA groups are then aggregated into into so-called elementary aggregates (EAs) for which basic prices can be calculated as geometric Jevons Indices. The EAs are the most detailed groups of commodities. Each period an average current period price is calculated and compared with the previous period using a geometric mean of matched observations. The aggregate price development is used to update the index level of the EA and corresponds to the total price development for a given 6-digit DB07 industry.

Subsequently, EAs are aggregated together using a Laspeyres-type index (weighted arithmetic mean), where the composition of the commodity basket remains fixed between reference period and current period. In a Laspeyres-type index the price reference period does not coincide with the weight reference period. The price reference period is the period you compare the current period's prices. The weight reference period is the period that the weights are from. Typically the weights reference period comes before the price reference period due to data availability constraints.

This calculation is explained in more detail in the publication Index Calculations in Statistics Denmark (Indeksberegninger i Danmarks Statistik). Which only exist in a Danish version currently.

Weights are assigned to every EA and used for weighting the base indices together to produce prescribed aggregate level indicators. Weights are based on National Accounts data.

Where non-response is experienced for essential commodities, imputation is used in which the price development of known observations is used to represent missing observations (class mean imputation). In other cases, prices are considered unchanged (carry-forward method). The monthly non-response rate is less than 1 percent and is not considered a significant source of error.

Adjustment

There are no corrections of data beyond what has already been described during data validation and data processing.