Statistical processing

Source data

The key sources are:

• The balance of payments
• The quarterly/annual (product) national accounts
• Quarterly/annual government finance statistics
• Quarterly non-financial sector accounts for financial corporations
• Sector industry matrix (see section 2.4 for further details)
• Accounts statistics, VAT statistics, agricultural statistics and income statistics
• Annual earnings, employment and hours
• Non-profit institutions serving households (NPISH)
• Securities statistics, MFI statistics and finance company statistics
• Digital accounts in the digital format XBRL (eXtensible Business Reporting Language - XBRL is a programming language that transforms electronic communication from the business community to financial data)

Frequency of data collection

Data is collected quarterly, except for certain sources that are collected annually. Annual data comes from e.g. accounts statistics, income statistics, agricultural statistics, sector industry matrix and digital annual reports (xrbl data).

Data collection

Data for these statistics is provided through internal deliveries.

Data validation

Data validation is a tiered process:

1. The first phase takes place even before we receive data from our suppliers, since our suppliers have already subjected input data to validation and quality assurance.
2. The next phase takes place when we have received data from our suppliers, where we subject the input data to additional troubleshooting and transformation. Transformation means that the input data is classified and coded according to ESA 2010.
3. In the third phase, macro corrections and imputations are made, provided errors or gaps have been identified during the second phase.
4. In the fourth phase, the data is validated by matching the sources. Because the institutional sectors are a reconciled system, the internal consistency is a quality control on its own in the current production, since it is possible to compare different sources containing the same information.
5. In the final phase, micro corrections can be added, if the development in a time series deviates a lot (outlier) without any explanation.

Data compilation

First and foremost, the compilation of data for this set of statistics consists in integrating data from the many data sources to obtain a harmonized data set where the collected data is transformed by a coherent set of sector accounts.

In this phase, data is created for the household sector, since data from the other sectors provide contra sector information that is used to determine the household sector’s transactions.

Subsequently, the non-financial corporate sector is created. With just a few exceptions, the non-financial corporate sector is calculated residually and consequently, is subject to the highest number of adjustments.

The integration of data takes place in a customised statistical calculation system, SBS, which is used to integrate the many data sources systematically. The running of SBS takes place in two rounds. In the first SBS run, input data is integrated and calculations, imputations, transaction codes, corrections and benchmarking methods are created.

In the second SBS run, the chart of accounts is created. The chart of accounts makes it possible to see the complete accounting system for each sector, so that it is possible to look at trends in the time series and in revisions (compared to the latest release) and not least ensure that the institutional sectors are consistent and balanced.

Imputation is the part of the troubleshooting process that replaces missing or invalid values with estimated values.

• Deterministic imputation is used for calculation of most transactions in the non-financial corporate sector. With just a few exceptions, the non-financial corporate sector is calculated residually, and consequently, it is subject to the highest number of adjustments.
• Simple imputation is used e.g. when input data only exists as annual data. In such cases, the simple method would be just to divide the annual figure by four (used especially if data is used as part of a major calculation).
• Another imputation method applied is to let annual data follow the trend in another time series, when these are closely correlated. This method is also used for the forecasting of data.
• The current transfers in the household sector are based on an estimate, which has been firmly fixed for the last five years.

Adjustment

For the institutional sectors, seasonally adjusted figures are published in Statbank Denmark for the specific series (consumption, disposable income, correction for changes in pension rights, gross savings and consumption of fixed capital). Several seasonally adjusted series are published in Eurostat, cf. section 7.1.