ESSnet Admin Data Alison Pritchard ONS, UK
Quick overview of ESSnet Results from WP3 and WP7 Future access to results
Introduction ESSnet Admin Data ran for 4 years (2009 2013); Partnership of 8 NSIs BE, DE, EE, IT, LT, NL, PT and UK; Very ambitious work programme, divided between 8 workpackages; True partnership, with at least 3 co-partner NSIs involved in each work package. Collaboration - when it works, it s both satisfying and interesting!
Aims of ESSnet Admin Data To recommend best practices for dealing with the following identified problems: Poor timeliness of admin data for STS (WP4); SBS definitions differ from those used in admin and accounts data (WP7); Some variables are not available at all in admin sources (WP3); No quality measures (WP6); Errors in admin data (WP2).
SBS v. IFRS definitions Aim of WP7 - to determine how useful accounts data are, as replacements for SBS survey data. Definitions in International Accounting Standards were compared with SBS, building on the results in ICON Institute s Taxonomy Report. Actual accounts were compared with SBS survey - do the definitional differences matter in practice? If they do, WP3 team members then endeavoured to estimate the variables in other ways using admin data.
SBS Turnover v. IAS/IFRS Revenue Excise Duties: IFRS Revenue includes only the gross inflows of economic benefits; Amounts collected on behalf of third parties are not economic benefits; Therefore, Revenue generally excludes excise duties (except, theoretically, in very rare cases); SBS Turnover includes all duties and taxes on goods and services invoiced by the unit, except for VAT, unlike the requirements of IAS/IFRS.
WP7 s Conclusions re Turnover/Revenue Recommendation to exclude Excise Duties from SBS Turnover definition, so that company accounts can be used directly. Impact of the other definitional differences re grants, interest, royalties and dividends - was shown to be insignificant for SBS results.
Purchases SBS variable Total purchases of goods and services does not exist in Financial Statements, but calculation is possible where nature of expenses method is used, because definitional differences are not significant for SBS results. Where function of expenses method is used, must also extract information from the Notes.
Production value - SBS v. IAS/IFRS SBS IAS/IFRS Turnover (12110) +/- Change in stocks of finished products and work in progress (13213) +/- Change in stocks of goods and services purchased for resale (13211) - Purchases of goods and services purchased for resale (13120) Revenue Is recorded in the Income Statement based on the nature of expense method or calculated from the classified inventories which are disclosed either in the Balance sheet or in the Notes. The calculation of the variable is complicated and basically impossible.
Personnel costs v. Employee benefits SBS defines Personnel costs as: Total remuneration, in cash or in kind, payable by employer to employee in return for work done during the reference period. Personnel costs are made up of Wages and salaries and Social security costs, including taxes and employer's compulsory and voluntary social contributions. IFRS splits Employee benefits into 4 categories: 1. short-term employee benefits, such as wages, salaries and social security contributions; 2. post-employment benefits, such as pensions; 3. other long-term employee benefits, including longservice leave, sabbatical leave etc.; and 4. termination benefits.
Accounting data v. Social Security data Total labour cost in the annual profit & loss account tends to be higher than in admin source. Possible reasons: Inclusion of certain overhead costs; cost of compliance with labour legislation outsourcing salary administration/ executive search... Foreign staff on the payroll; Tax avoidance (especially for highest paid employees). Conclusions from further analysis of the metadata: content of social security office s records is subject to modification, resulting from legislative changes; Failure to comply biased personnel cost estimates.
Investment variables Values for total fixed assets at end (and start) of year are available in Balance sheets; Acquisitions, disposals, transfers and value adjustments of fixed assets may be in the Notes. Acquisitions are measured at fair value rather than cost, but in practice this difference is not significant. Detailed data about the nature of the fixed asset invested in (land/buildings/improvements/plant) are usually not available.
Sales of fixed assets Analysing company accounts is very timeconsuming, in the absence of xbrl data; The disposal value of a fixed asset is not the same as the amount deducted in the balance sheet, because accounts use book value, not sale proceeds; But the sales proceeds of investment goods may be found in the Cash Flow Statement.
WP7 s recommendations Revenue can be used for SBS variable Turnover, as discrepancies arising from differences between IAS/IFRS and SBS requirements are small (this excludes the impact of excise duty which needs to be measured individually). The definition of Turnover should be amended to exclude excise duties; The data from the Notes to the Financial Statements should be made available to EU NSIs in a useable format (e.g. the IFRS xbrl Taxonomy is a likely source).
Estimating difficult SBS variables (WP3) Change in stocks Purchases Payments for agency workers FTE Gross investment (and components) and Sales of tangible investment goods Production value (derived variable)
Investment: UK Cut-off sampling Unit level regression modelling produces largest sample size savings: GITG model fitted using sample data from cut-off band in last year survey was run model parameters fixed over time STIG model fitted using sample data from above the cut-off band model parameters can be updated annually Simple ratio adjustment method works well too; Sample size saving depends on size of divisions where method works well.
WP3 Scenario Admin variables are available with strong correlation with SBS variable Multiple admin variables are available with reasonable correlation with SBS variable Single available admin variable has reasonable correlation with SBS variable Estimation methods to try Regression modelling for whole population Cut-off sampling with: Simple ratio adjustment; or Regression modelling Cut-off sampling with simple ratio adjustment To make a saving on an existing ratio estimation survey approach. Inflation of survey weights
Estimating for Components Admin variables are available for the total, but not for components. Work done on components of total gross investment in tangible goods (Lithuania). Work done on components of change in stocks (Italy). Recommend using admin data for totals, and survey questionnaires (sent only to largest enterprises) for components; then can estimate components for smaller enterprises.
Derived variable production value (UK) Production value = turnover - purchases for resale + change in stocks + capitalised production + other income. Three admin sources available in UK: Social security employment; VAT turnover and expenditure; and Company accounts Investigated use of cut-off sampling for: Estimating Production Value total; and Estimating the components and then combining them to give the Production Value total
20
Results estimating components Simple ratio adjustment method applied to each component Number of divisions with acceptable results: Ratio Auxiliary variable Production value Turnover Purchases Change in stocks Capitalised production Harder to estimate for some of the components, very small estimates & lots of zeroes. Other income Overall Register turnover 32 33 1 3 7 8 Register Overall employment 31 25 3 3 8 10 Division Register turnover 31 33 13 3 9 14 Register Division employment 30 30 10 3 9 15 21
WP3 Scenario Admin data are available for the total, but have to estimate components for some businesses To estimate a total and related components for some businesses To estimate a derived variable Estimation methods to try Estimate breakdown from similar businesses and apply to total, using - Nearest neighbour donor imputation; or Cut-off sampling with simple ratio adjustment Nearest neighbour donor imputation; or Cut-off sampling with simple ratio adjustment. Estimate directly; or Retain some components in survey and estimate the rest separately.
Top 25 IC user countries, in the last year 20 May 2012 to 19 May 2013 4,427 visits from 95 countries Sequence Country Visits 1. United Kingdom 846 2. Italy 720 3. Belgium 259 4. Estonia 250 5. Netherlands 233 6. Poland 142 7. Germany 131 8. Lithuania 122 9. Bulgaria 112 10. Austria 109 11. Finland 108 12. New Zealand 104 13. Cyprus 88 14. Hungary 84 15. Switzerland 80 16. Sweden 67 17. Slovakia 61 18. Spain 55 19. Romania 55 20. Latvia 52 21. United States 52 22. Croatia 48 23. Turkey 47 24. Slovenia 46 25. Luxembourg 39
Information Centre - essnet.admindata.eu Advanced search options - can search within work package articles, documents, events, reference library articles, glossary items etc. Project-wide Glossary defines various terms related to the use of administrative data and accounts data (Wiki functionality). Database of existing practices - user-friendly application database with up-to-date data and user interface (for online queries). Reference library - repository of reference literature (about 600 items).
Thank you any questions? 25