Danish GFS compilation system Use of primary data in national accounts By Kristina Stæhr Vest Gitte Frej Knudsen Martin Rasmussen

1.Primary sources and the GFS system  Budgetary sources  Extra budgetary sources  Data structure in the GFS system  Subsectors  Financial/non-financial accounting data 2.Other entries 3.The classification process 4.Time adjustment 5.Accounting principles 6.Micro-macro link 7.Output (non-financial) 8.Software (non-financial) Overview 2 9.Continual revisions 10.Who compiles the GFS? 11.Data relationships in financial accounts  Ensuring consistency  The Bottleneck issue 12.Software (financial)

1.1 Budgetary sources  Central government accounts  Municipality and regional accounts  Accounts for Social security funds  Unemployment funds 1. Primary sources and the GFS system 3

1.2 Extra budgetary sources  Central government extra budgetary units  Universities  Secondary schools  State Church  Local government extra budgetary units Nationwide organization for the municipalities (KL) Nationwide organization for the regions (Danish Regions) Tourist information agencies 1. Primary sources and the GFS system 4

1.3 Data structure in the GFS system 1. Primary sources and the GFS system 5 Central governmentSocial securityLocal government § 81-89Various national account calculations and imputations § 80Municipalities accounts § 71-79Various national account calculations and imputations § 64-69Various national account calculations and imputations § 70Regions accounts § 51Consolidation § 52-56Central government national account calculations § 56-59Extra budgetary units § 61State Church national account calculations § 62State Church accounts § 63Social security funds § 01-45Central government accounts entries 150 entries entries entries entries

1.4 Delimitation of subsectors  Extra budgetary institutions are added to the dataset.  Entries outside the GFS definition of general government in the budgetary information concerning public quasi corporations are not removed but classified as “Sector delimitation corrections”.  The primary data set can be reconstructed by leaving out the Extra budgetary institutions in GFS system. 1. Primary sources and the GFS system 6

1.5 Delimitation of financial/non-financial accounting data  All financial entries (according to ESA95) originating from the primary non-financial sources are not removed from the data but classified as “Financial transactions”.  This practice enables us to keep track on the transition from primary data to national accounts statistics/GFS. 1. Primary sources and the GFS system 7

Other sources  Taxes  Income transfers to households (quarterly government finance statistic)  Extra-budgetary units National accounts compilation Primary accounting data + Extra-budgetary units -Public quasi corporations -All financial entries 2. Other entries and calculations 8

Automated coding process of data  Codes are applied using a matching process:  type of transaction  COFOG  Consolidation  Trivial entries are coded automatically using conversion rules from primary accounts to GFS/national accounts Manual process  Manual coding is applied on residual entries Validation of data  Logic tests  Inspection of time series at macro level and micro level 3. The classification process 9

 All time adjustments are made as supplementary corrections to the original entries and not by substituting the entries  Taxes  Income transfers (quarterly government finance statistic)  This again helps keeping track on the transition from primary data to national accounts variables and aggregate/GFS statistics 4. Time adjustments 10

 Micro accounting on a accrual or commitment basis  Commitment basis used as a proxy for accrual basis  For some entries commitment basis = cash (e.g. wages and salaries)  Accrual principle  Income and expenditure attributed to the period accrued on an annual basis  On a quarterly basis more time adjustments needed, for example:  Interests  Consumption of fixed capital 5. Accounting principles 11

Input data Primary accounts data Entry no 1 Entry no 2 DIOR Coding process Output statistics GFS/national accounts (non financial) Category i Category i+1 Category i+n 6. Micro-macro link 12 DIOR = Detailed Integrated Government Accounts Categories defined by: Sector, type of transaction, COFOG, industry

Types of questions to be answered:  How has a given transaction in a certain government institution been classified in GFS/national accounts?  Which effect has a given transaction on government consumption, deficit etc.?  Where does an expenditure/revenue category in GFS have its origin in primary accounts? 6. Micro-macro link 13

 Type of transaction (SNA/ESA) table  Revenue, expenses and net lending/borrowing  On subsectors  COFOG table  Cross compilation table – COFOG and type of transaction  Tax, specification table  Income transfers, specification table  Subsidies, specification table 7. Output (non-financial accounts) 14

 The GFS compilation system is developed in Oracle  The system generating tables is programmed using SAS  The GFS compilation system is basically a database where calculations is manually added to the entries in the system (Excel) 8. Software (non-financial accounts) 15

Timeline for revisions in the Danish GFS Prelimenary/final accounting data  The March-version is based on preliminary accounting data.  When the March-version is published the data is flagged as preliminary.  If there are problems with the quality of data in a version it is described in the publication article and in the StatBank Denmark.  Final GFS version in November (t+3). 9. Continual revisions 16

The compilation of the government account is split into two different processes in Statistics Denmark:  Financial accounts  Non-financial accounts -Special issue: Balancing the two compilation systems  Non-financial accounts are compiled by Gitte Frej Knudsen and Martin Rasmussen.  Financial accounts are compiled by Louise Kristensen and Kristina S. Vest. 10. Who compiles the GFS? 17

11. Data relationships in financial accounts

Ensuring consistency between financial and non- financial accounts:  Initially the systems are based on the same datasources  Changes with an effect on the working balance made on one side of the system must be mirrored completely on the other side of the system  When replacing datasources on one side of the system consistency must be assured.  Issues about balancing the systems 11. Data relationships in financial accounts 19

The Bottleneck issue:  Support of upstream dataproducers  Refering questions  Metadata and documentation  Explaining differences  Consistency checks 11. Data relationships in financial accounts 20

 At present most of the system is excel based  Some dataprocessing takes place in SAS  In future: Oracle based system  Pros and cons of the two systems 12. Software (financial accounts) 21