Data Sources and Quality Improvements for Statistics on Agricultural Household Income in 27 EU Count

1
Data Sources and Quality Improvements for
Statistics on Agricultural Household Income in 27
EU Countries

• Berkeley Hill
• Emeritus Professor of Policy Analysis
• University of London (Imperial College)

2
Introduction

• Rising awareness of multiple incomes
• 1985 Commission Green Paper Annex
• Eurostat IAHS statistics 1988-2002
• Based in system of national accounts
• Harmonised methodology (Definitions)
• Sector-level results
• Declining EU and national priority
• Some political and institutional hostility
• Some MS used micro data as source of results

3
IAHS hiatus of early 2000s

• IAHS results increasingly out-of-date
• Importance of distributional information
• EU enlargement new types of agricultural
  household and business
• Court of Auditors 2003 review of IAHS
• Met central objective of CAP but
• Statistics of poor quality (NL evidence)
• Recommended a feasibility study of a uniform
  micro-approach across all MS endorsed by Council

4
Interim research work

• Gradual accumulation of information on data
  sources Eurostat, OECD etc
• ISTAT 2002 TAPAS Initiative reviewed income data
  sources and alternative calculations in Italy
• Statistics Sweden 2006 study on feasibility of
  adding questions to the FADN farm form
• No significant technical barriers

5
The 2007 feasibility study - AgraCEAS

• Template of uniform key definitions (income,
  household etc.) developed from
• 2005 UNECE Handbook
• Survey of users
• Visits to MS and surveys to find data sources
• Feasibility of using template assessed
• Method of filling data gaps proposed and costed
• Recommendation made to Eurostat

6
Key definitions

• Household single budget unit
• Household classification
• narrow agriculture main income source of
  reference person
• broad range of possibilities (any farm
  income, holding characteristics - FSS, SFP)
• Net disposable income as in Handbook
• Detailed breakdown Imputed items shown separately

7
Inventory of data sources (25 MS)

• Farm accounts surveys
• Household income not part of EU-FADN
• Only some MS collect household data
• EU-SILC
• Generally few agricultural cases
• Income data often of poor quality
• Household budget surveys (as above)
• Taxation records and registers
• In many MS farmers not taxed on actual income
• Tax income definitions pose problems / disclosure

8
Example Austria

• Farm accounts - sample of 2,500 holdings which
  also covers household income
• EU-SILC - 271 cases in 2005.
• HBS - carried out once every 5 years.
• Tax records - For a large proportion of farmers
  tax payment is not based on accounting income
  (farmers pay 'lump sum' taxes)

9
Example Poland

• Farm accounts - data on five types of
  non-agricultural income collected from about
  10,000 farmers in 2005.
• EU-SILC - agricultural cases not known they are
  combined with other self-employed.
• HBS - Some 2,000 agricultural households in 2005
  problems with income data quality.
• Tax records - assessment (mainly) uses a standard
  rate based on land and forest area, land quality
  and distance to market.

10
Example - Spain

• Farm accounts survey no household data 
• EU-SILC - only 253 agricultural cases in 2004
  (similar number in 2005).
• HBS - only about 120 agricultural cases, but
  there are difficulties with incomes from
  self-employment.
• Tax data - some farmers do not pay tax based on
  actual incomes, and incomes may be estimated.

11
Example - Luxembourg

• Farm accounts survey - questions covering
  household income were used for 1989 only.
• EU-SILC - only 78 agricultural cases in 2004 .
• HBS - few agricultural cases.
• Tax records - most farmer incomes are not on an
  accounts basis.
• Other - There is a poverty survey of households
  (CEPS).

12
Feasibility testing of definitions

• Each aspect of the template (household, agric
  household narrow and broad, income definition,
  comparison with others) was assessed as
• Currently in use
• Not in use but technically possible
• Requires development of existing data sources
• Requires a new data source

13
Example narrow agric household based on the
reference person

• Currently used 3 MS
• Technically possible 16 MS
• Required data source development 4 MS
• Requires new data source 3 MS

14
Example Can comparisons be drawn with other
socio-professional groups?

• Currently made 7 MS
• Technically possible 10 MS
• Requires data source development 4 MS
• Requires new data source 5 MS

15
Example use of Net Disposable Income?

• With existing data source 19 MS
• New data source needed 5 MS
• UK main data source does not collect tax paid
• Germany questionable reliability of existing
  sources
• (others were Slovakia, Hungary and Luxembourg)

16
Filling the data gaps

• MS fall into three broad groups
• Special survey needed to cover both narrow and
  broad definitions of an agricultural household
  hybrid of FADN and EU-SILC questions, collected
  by EU-SILC method
• Narrow covered, but special survey for broad,
  typically below FADN size threshold
• No new data collection needed only extraction

17
Costing MS costed individually

• Transparent calculations allow alternative
  figures to be used
• Survey costs based on existing national EU-SILC
  data costs, and commercial rates
• Case numbers
• narrow as for existing FADN samples
• broad below FADN threshold same sample rate
  as above, and at a 1 rate

18
Examples

• Denmark data extracted from existing registers
  (no additional survey)
• Germany additional special period survey for
  both the narrow and broad definitions
• Poland use of existing farm accounts survey for
  narrow additional survey below FADN threshold
  to cover the broad. Other work needed to
  faciliate comparisons with other spg.

19
Results

• Aggregate cost of collecting data to enable
  comparable and robust statistics (one-off
  surveys)
• 11.5 million survey costs 1 central costs for
  the narrow definition of an agricultural
  household
• Additional costs of extending coverage to the
  broad definition 9.1 13.3 (totalling
  22-26m)
• In comparison EU-SILC costs c.27m p.a.
• If the cost led to a 1 efficiency gain in Pillar
  1 spending, this would be 19 times greater

20
In conclusion

• Good quality data are essential to quality
• A useful inventory of data sources on incomes of
  farm households is now to hand
• A technical assessment of the feasibility of
  producing robust EU-wide IAHS statistics has been
  made and costed.
• Eurostat has not taken up the proposed actions
  some MS do not seem to be keen
• The Court of Auditors has expressed interest in
  why progress has not been made