/ EUROPEAN COMMISSION
EUROSTAT
Directorate F: Social Statistics

Doc. Eurostat/F/13/DSS/01/4.2 EN

Item 4.2

SOCIAL TRANSFERS IN KIND:

PROBLEM STATEMENT

MEETING OF THE EUROPEAN DIRECTORS OF SOCIAL STATISTICS

luxembourg, 4 – 5 March 2013

BECH building, room AMPERE

Eurostat project on valuation of social transfers in kind in household surveys

Introduction

One of the five targets in the Europe 2020 strategy is to reduce the number of people in or at risk of poverty and social exclusion by 20 million. Monitoring performance against this target requires good quality information on, inter alia, all of the types of income received by households.Social transfers in kind, or STIK for short, are not included in the current definition ofincome used for monitoring the Europe 2020 targets. However, they may have a redistributive impact. Eurostat's two year project is investigating how STIK might be valued in household surveys in a harmonised way for inclusion in comparative income distribution statistics for, and across, the EU.

Action for Directors of Social Statistics

This paper is for information only.

Overview of project plan and summary of progress

The project lasts from December 2011 till November 2013, and has four stages:

Stage / Progress
(1)information gathering from December 2011 through to spring 2012 /
  • External studies reviewed
  • Introduction of project to the Social Protection Committee’s Indicators Sub-Group (ISG)
  • Meeting with Eurostat experts on data sources
  • Meeting of external experts on setting up and taking project forward

(2)reviewing sources, methods and assumptions and making first estimates using in-house information sources to autumn 2012; /
  • Discussion with Eurostat Working Group on Statistics on Living Conditions: suggestions received on how to make the scope of the project manageable and on tackling issues
  • Development of first estimates
  • Presentation of first results to the ISG

(3)iterative revisions according to expert commentary to spring 2013; /
  • N/A

(4)consolidation of conclusions and dissemination to November 2013. /
  • N/A

What are the issues?

Four broad issues have been identified:

(i)Which services should be covered by the project?

(ii)How should economy-wide social transfers in kind (the base valuation') be valued?

(iii)How should the base valuation be distributed across the population?

(iv)How should the differing needs of households associated with social transfers in kind be accounted for?

Dealing with each of these in turn:

(i)Which services should be covered by the project?

Discussions with experts in the field, as well as with key stakeholders including the Social Protection Committee’s Indicators Sub-Group and the Eurostat Working Group on Statistics on Living Conditions, have led to the conclusion that, given their importance (in terms of the size of government expenditure) and the fact that these are the services that have been covered in the existing literature, health care and education are the services which should be covered first. Early childhood education, child care, elderly care and housing are the next most importantservices.

(ii)How should economy-wide social transfers in kind (the ‘base valuation') be valued?

There are a number of different sources of information on the amounts governments spend on the services covered by social transfers in kind. No single source provides all of the information needed, and each source has been designed with a different aim in mind. The sources being looked into include National Accounts, European System of integrated Social Protection Statistics and the System of Health Accounts.

(iii)How should the base valuation be distributed across the population?

There are a number of different approaches that could be used for distributing social transfers in kind across the population using household surveys. All studies that have an EU-member-state comparison perspective published thus far have a common theme: they have all used EU-SILC data. In general, household surveys do not capture the exact information that is needed on the social transfers in kind that households receive. Instead, a value needs to be imputed for social transfers in kind for each household. This imputation needs to be modelled based on a set of rules which link the information that is available in EU-SILC to external sources. There are a number of normative assumptions that need to be made in specifying the model, the most important include:

a) Is an insurance or a consumption approach appropriate? The insurance approach is one where the amount of income imputed to a person is based on an estimate of what the equivalent (notional) insurance premium might be. The actual consumption approach is one where the amount of income imputed to a person is based on that person's use of the service. There is as yet no consensus on which is the most appropriate approach for all services.

b) What variables ought to be used to distinguish between different households?

Whether the insurance approach or the actual consumption approach is used, there needs to be a set of variables chosen which exist in EU-SILC and for which there are external sources that provide information on how social transfers are distributed across the population. In practice, and the simplicity of this method is acknowledged, published studies rely on age and sex, as well as information on hours of education and care received.

c) Should other sources beyond EU-SILC be used? Sources with much richer information on individuals’ use of services, for example health care and social services, could be used that would allow for a much better model than the simple use of age and sex. However, the trade-off would be that there would then be a need to link these more specific pieces of information with the income information in EU-SILC. It is not clear whether using extra sources would improve the overall quality of estimates and it is not clear how we might establish what are the relative levels of quality of the different models.

(iv)How should the differing needs of households associated with social transfers in kind be accounted for?

The differing relative needs of households with different compositions (for example households with different numbers of people and/or members with different mixes of sexes and ages) means that care needs to be taken when making comparisons of those households' relative income needs. It is generally accepted that a household with more members, or a household with a greater proportion of children compared with adults, requires a lower per-capita level of income to acquire a given standard of living.

It is also generally accepted that the relative need for income of households with differing numbers of adults and children does not change in direct proportion to the actual number of household members. Rather, the relative need for income takes into account the fact that households benefit from economies of scale as the number of household members increases: for example, a household with four members, say two adults and two children, does not (on average) have four times the heating needs of a household with only one member.

Economies of scale act differently for each of the wide variety of items required by households in acquiring a given standard of living. Heating is an example of one such item which does not increase very quickly as the size of the household, in terms of number of members, increases. Average household expenditure on heating is relatively inelastic to household size.Food is perhaps an example of an item of household consumption at the opposite end of the scale: expenditure on food does increase probably fairly closely in direct proportion to household size. Its elasticity to household size will be close to unity, on average.

In concept, and with perfect information on the household compositional factors that have an impact on the amount of expenditure needed to acquire a given standard of living and on expenditure patterns, it would be possible to calculate exactly what income levels are required to acquire a particular standard of living for all types of household given their relative needs for all consumption items.In practice, such a wealth of information is not typically available. Instead, a consensual approach to the problem of comparing income for households with differing compositions has been reached. This consensual approach involves the use of relatively simple 'equivalence scales' which can be thought of as average ratios that take into account the needs of households across all different possible consumption items (including the two examples given above – heating and food - as well as all other household consumption items).

The equivalence scale adopted for use in practice by Eurostat is the modified OECD equivalence scale. This simple scale is based on applying one of three weights to different types of household member, as follows:

1.0 to the first adult;

0.5 to the second and each subsequent person aged 14 and over;

0.3 to each child aged under 14.

The equivalence scale for each individual household is simply the sum of the weights for all of the household members. So, returning to the two different households mentioned above, the weight for the single adult household is 1, whereas for the two adults-two-children household, it is 1.0 + 0.5 + 0.3 + 0.3 = 2.1. The interpretation of these (relative) weights is that the four-member household requires 2.1 times the level of income compared with the single adult household to acquire the given standard of living.

Most experts[1] agree that as this scale has been designed solely with cash income in mind, it may not be suitable for cash + non-cash income combined. Eurostat is currently looking into the effects of using alternative equivalence scales, which have been estimated by other organisations.

Other issues

Besides these four broad issues, the project continues to identify many other issues. To provide Directors with a flavour of the work of the project, a selection of these issues, along with their solutions, is below.

  • Is it income? Consider a low-income individual living on their own who has received expensive medical care, for example, triple heart-bypass surgery. In cash terms, this is a low income household. Adding the value of the health care received would almost certainly mean the household was considered high income. But the economic circumstances of the individual have not really changed (they have no more income at their disposal) and the welfare circumstances of the individual are worse (triple heart bypass surgery is highly traumatic). One part of the answer to this is the use of an appropriate equivalence scale (which would take into account the extra need for health care of such an individual, thus higher health care needs requires higher income to achieve the same standard of living). Another part of the answer is that valuation of STIK is not meant to provide meaningful results at the micro (for example, individual or household) level, but at the macro (income quintile) level.
  • Public versus private It is not possible to distinguish in EU-SILC whether individuals are using public or private services. This may be a large source of bias, due to the fact that many government-funded or government-provided services are means-tested or directed to lower income households. It is also not possible to distinguish between those who contribute to the cost of the service (by making a co-payment) and those that do not. Eurostat is currently considering how extra information might be collected which would help with these connected issues.
  • Validation of estimates As there are so many issues to be resolved in calculating estimates of STIK in household surveys for the purpose of improving our understanding of how income is distributed across the population, Eurostat is considering how the different methods and accompanying datasets might be validated.
  • Exclusion of people living in communal establishments EU-SILC does not cover those people living in communal establishments, such as care homes. This will reduce the quality of the estimates to an unknown extent, especially for estimates of health care and elderly care STIK, where many live individuals benefitting from these services will be living in care homes and retirement homes.
  • Missing data For the year chosen, 2008, there are a small number of missing values in EU-SILC, which are being excluded from the analysis. For a limited number of countries, the information on the distribution of expenditure on health care by age and sex is based on EU averages. These latter estimates are probably not appropriate for use in estimating STIK.
  • Calculate estimates for a single year, or for a time series? First estimates are being calculated for a single year with further years to follow. The year chosen is 2008, on the grounds that this would be the most recent year available that would be unaffected by recession.
  • Valuation of non-market output Eurostat's STIK project is taking the conventional approach to valuing non-market output, which involves summing the costs of production. This means that any valued added by the health care and education services is not reflected in the valuation.

Page 1 of 6

[1] See, for example, Radner, Daniel B., Noncash Income, Equivalence Scales, and the Measurement of Economic Well-Being. Review of Income and Wealth, Vol. 43, Issue 1, pp. 71-88, 1997. Available at SSRN: or for an early exposition of the issue