Statistical processing
Contact info
Section for Earnings and Absence, Social StatisticsHege Susanne Hauglund
Get as PDF
Data are collected for more or less all persons employed in the public sector in Denmark and refer to the second month of the quarter in interest. Before production of the index is started, the data are roughly searched for errors. But there are also performed search for errors later in the process, e.g. by looking at the rate of increase in the average wages for each company or organisation. Each employment is given a weight after the share of hours worked in relation to a full-timer’s normal hours, which is used when adding observations to calculate the rate of increase for an enterprise or branch of economic activity.
Source data
Data is primarily collected from the public pay transfer systems. Also, a small number of enterprises for the public sector transmit data through their own IT-systems for payroll administration.
Frequency of data collection
Data is collected quarterly, although only information for the second month is used in calculating the index.
Data collection
Data is transmitted from the public pay transfer systems by the use of a so-called system-to-system method. Information on earnings for government employees is delivered from the Agency for Modernisation under the Ministry of Finance, as well as the Defense Command Denmark (Forsvaret) and Silkeborg Data. Data on regional and municipal employees are mainly sent from the Office of Earnings for Regional and Municipal employees (KRL). Besides that, data is also obtained from some smaller enterprises through their own IT-systems for payroll administration.
Data validation
A search for errors is performed with the use of an upper and lower limit of the average earnings per hour worked. The limits are calculated at the start of each year by finding the average increase in earnings from last year and adding these to the old limits. If the observations do not fall within these limits they are removed from the dataset going into production. This validation is necessary as it makes sure that extreme observations do not distort the results. Another validation performed on the dataset later in the process, is of the rate of increase in average earnings for enterprises/institutions or class of industries are either too high or too low. If this occurs, it is checked whether this is caused by large changes in the composition of labour in an enterprise or class of industry as a whole. The respective enterprises are either removed or analyzed further. This is done as the purpose of the index is to depict real changes in what employees receive and not the changes in earnings resulting from changes in labour composition.
Data compilation
When data is received it is validated as described under Data validation. The average earnings for each observation is weighted according to his/her share of working time in relation to a full-timers working time. A full-timer is weighted as 1, while a part-timer working half the time compared to a full-timer is weighted as 0.5. Since the index more or less covers all employees in the public sector, there is made no correction for fall-out of observations.
Adjustment
Only enterprises that have transmitted data that can be used in both the quarter being measured and the previous quarter are included in the calculations of the index. This means that whenever data from an enterprise is omitted due to quality issues in a certain quarter, the enterprise will as a minimum first be included in the index after the preceding quarter.