A Eurosystem Household Finance and Consumption Survey: Developing the questionnaire for the planned

1
A Eurosystem Household Finance and Consumption
Survey Developing the questionnaire for the
planned survey

• Panagiota Tzamourani and Carlos Sánchez Muñoz
• Q2008, Rome, July 2008

2
The Household Finances and Consumption Survey
(HFCS)

• Micro data essential to study household financial
  behaviour, impact of monetary policy, effect of
  shocks
• Similar surveys exist in some euro area countries
  only lack of data or data non-harmonised
• EU surveys either do not cover wealth and
  liabilities (eg EU-SILC) or target only parts of
  the population (eg SHARE)

3
The Household Finances and Consumption Survey a
proposal

• Eurosystem Household Finance and Consumption
  Network (HFCN)
• Proposal for a survey on HFC for the euro area
• Proposal should include
• a common questionnaire
• modalities for the survey implementation
• an estimate of the associated costs
• The survey has not yet been approved

4
This presentation on the HFCS questionnaire

• Topics
• Questionnaire challenges
• Due to the survey topic
• Due to the international nature of the survey
• Development process
• pretests
• Resolutions
• Content and definitions
• Structure and wording

5
Questionnaire to cover

• Real assets and financial wealth
• Liabilities and credit constraints
• Income
• Consumption
• Pensions and insurance policies
• Intergenerational transfers
• Demographics and employment
• Payment habits

6
Data requirements

• Data adequate to address key issues for monetary
  policy analysis, financial stability analysis and
  research
• Consistency with macro data (National Accounts,
  banking statistics)
• Consistency with other European survey data (eg.
  SILC)

7
Challenges posed by the survey topic

• Detail required, but questions sensitive,
  questionnaire can become too long
• Information required difficult, eg portfolio
  composition find alternative ways of getting
  information, or, simply resign (!)
• Translate economic concepts and statistical
  classifications into everyday language
• Ensure a natural flow of the questionnaire,
  relax respondents after demanding sections

8
Challenges posed by the international nature of
the survey

• Different structure of household assets
  (real/financial) level of detail required
  different in each country
• Different prevalence and use of loan types
  (mortgage collaterals credit cards)
• Understanding and knowledge of common concepts
  varies, eg
• Salaries Gross or net, monthly or annual
• Reference period for income and wealth (more
  general connected with the use of tax records,
  time period of fieldwork)
• Differences in the characteristics and labelling
  of financial products, eg
• Sight accounts (current accounts, saving
  accounts)
• Individual retirement accounts or life insurance
• Differences in the pension systems across
  countries
• Social security /occupational pensions
• Comparability with existing time series

9
The process of developing the questionnaire

• May-July 2007 drafting questionnaire, written
  procedure
• In parallel Drafting of the handbook of
  definitions
• September 2007 Irish pretest (60 cases, spread
  over 3 phases)
• November 2007 Revision based on pretest results
  and written procedure
• January 2008 Greek pretest (30 Cases) of new
  version
• February 2008 revision and streamlining from
  Questionnaire subgroup
• Spring 2008
• German pretest (20180 cases)
• Belgian pretest (40 cases)
• Portuguese pretest (80 NCB staff)
• Summer 2008 final revision

10
Questionnaire pretests aims

• Ascertain appropriateness of questionnaire
  content
• Test understanding of the questions and ease with
  which questions can be answered
• Time each phase of interview
• Core and non-core (optional) components
• Test impact of sectioning ordering
• Test appropriateness of reference person for
  household questions // feasibility of getting
  responses from all adults
• Phrasing in own language
• Implementation issues, e.g. survey letter,
  non-response, interviewers training,
  understanding and performance, oversampling the
  wealthy

11
Questionnaire pretests some lessons (1)

• Ireland (September 2007, 60 cases, PAPI)
• July version comprising core and non-core
  questions unworkably long
• Perceived duplication of income questions (last
  years annual income and current monthly income)
• Lack of knowledge about pensions held
• Interest rates not known
• Self-employed often found income difficult to
  report
• Popular questions consumption, attitudes,
  payment habits (non-core)
• Resented questions parents occupation, credit
  constraints, cash in the house

12
Questionnaire pretests some lessons (2)

• Germany (Spring 2008, 200 achieved interviews,
  CAPI self-completion)
• Questionnaire overall well received
• Fine-tuning needed modify some questions and
  lists of answer categories
• Belgium (Spring 2008, 40 cases, PAPI)
• Questions overall well received - very few
  criticism on the questions, except
• Income and pension sections inappropriate for
  many children above 16
• Questions on mutual funds considered intrusive

13
The HFCS approach

• Overall output oriented approach, i.e. countries
  bound to common definitions rather than
  questionnaire
• Streamlined core questionnaire /output
  variables, to be complemented by non-core and
  country specific variables, so that
• average interview length about 1 hour
• basic requirements of data provided covered
• concepts applicable and relevant to all countries
• Blue-print common questionnaire to be used in
  countries with new surveys

14
HFCS current questionnaire structure
Pre-interview selection of respondent/ Household
listing
Employment
Income
Demographics
Pensions and insurance policies
Real assets and their financing
Intergenerational transfers / gifts
Other liabilities / Credit constraints
Consumption
Private businesses / Financial assets
Post-interview interviewer debriefing
Pre/post interview
Individual questions
Household questions
15
HFCS data issues comparability

• Comparability with National Accounts
• Countries to make the distinction between
  producer households and quasi corporations
• Calculation of some ESA95 wealth components
  further comparability possible by adding
  non-core questions
• Comparability with other survey data
• Income in line with Canberra report
  recommendations
• Definitions to the extent possible consistent
  with EU-SILC and Eurostat's Concepts and
  Definitions Database, the OECD glossaries and
  other standards

16
HFCS data issues reference period

• Reference period for income and wealth
• optimal for wealth current status
• for (annual) income
• Last calendar year
• Accurate respondents could consult records, e.g.
  tax files
• Could be out-of-date, particularly if fieldwork
  after the middle of the year
• Inconsistent with wealth
• Last twelve months (chosen)
• Closer to fieldwork
• Consistent with wealth

17
HFCS data issues pensions, consumption

• Pensions
• Only indicator variables social insurance/
  private plans, contributions
• Consumption
• Only indicator variables Money spent on food
  outside home, money spent on food at home

18
HFCS ease respondents burden

• Use a communication strategy that creates trust
  and stresses confidentiality and anonymity
• Streamlined questionnaire
• Add some soft questions to relax respondents
• Allow households to report what they know best
  (eg monthly amounts for some income types)

19
Conclusions and next steps

• Overall current questionnaire well received
• Countries with existing surveys will be using
  their own questionnaire, with amendments so that
  they can provide the core variables
• New surveys will use Eurosystem questionnaire
• Governing Council will decide on the proposal
  this autumn
• First countries to implement survey in 2009

20
Thank you