HOUSEHOLD BUDGET SURVEYS

Transcription

1 EUROPEAN COMMISSION EUROSTAT Directorate D: Single market, employment and social statistics Unit D-2: Living conditions and social protection Doc. HBS/157B/2005/EN Working Group HOUSEHOLD BUDGET SURVEYS JUN 2005 Eurostat-Luxembourg Amended with the comments of the national HBS delegates received during and after the meeting Quality report of the Household Budget Surveys round of 1999

2 CONTENTS 1. INTRODUCTION RELEVANCE HBS potential: scope and structure of the information collected Comparability Multi-dimensional subject-matter Tool for harmonisation Users and uses: projects by or in collaboration with Eurostat Users at national level EU perspective Steps taken to promote data use Consultation with users Dissemination of aggregated HBS data Promotion of micro-data use ACCURACY Response rates and overall achieved sample sizes Item non-response Sampling error Sampling error and related measures for household final consumption expenditure Sampling error for expenditure components Sampling errors for population subgroups (subclasses) TIMELINESS AND PUNCTUALITY Timeliness Reducing the impact of poor timeliness Improving timeliness ACCESSIBILITY AND CLARITY Accessibility Forms of dissemination Reference database (the 'Production Data Base') Dissemination of aggregated tables on NewCronos Publications by Eurostat: Technical Documentation Clarity 27 2

3 Dissemination on NewCronos Publications COMPARABILITY The concept Comparability across countries Recommendations for survey structure and sampling Common concepts, definitions and classifications Household, household members and reference person Household Final Consumption Expenditure Data processing and checking Comparability in time Synchronisation with reference year Comparability between rounds COHERENCE EU sources Comparison of household final consumption expenditure with National Accounts (NA) Comparison of household monetary consumption expenditure with Harmonised Index of Consumer Prices (HICP) COST AND BURDEN Cost for de NSI Cost for Eurostat and other commission services Respondent burden

4 1. INTRODUCTION The Working Group (WG) on Household Budget Surveys (HBS) was created by a decision of the DGINS conference on The aim was to compile the existing information in the Member States (MS) of the European Union on household budgets in order to make all this information available at European level as well as to improve harmonisation of surveys, in terms of concepts used, classification of variables, data collection and data processing methods. This project filled a gap in the area of social statistics, allowing completing the social portrait of Europe. Contrarily to ECHP 1 and EU-SILC 2, which focus on household income, HBS relies on the concept of household consumption expenditure. The HBS is a unique source of information on household final consumption expenditure and living conditions in the European Union because of the comparability of the data generated as well as the multi-dimensional coverage. These specific features made it possible to respond to the increasing demand for comparable information on household consumption expenditure. Numerous HBS data requests originating from the Commission (SANCO, EMPL, MARKT, ), the European Parliament, other EU institutions, European governmental bodies, EU research institutions, EU enterprises and finally public in general. Eurostat publications drawing on HBS results of the round of 1999 include so far 1 Statistics in Focus, 2 Panoramas of the European Union (in collaboration with DG SANCO), 2 methodological volumes, and over 50 technical and methodological documents. HBS data has been also used in the context of some major Commission reports, e.g. the annual The Social Situation in the European Union. The Household Budget surveys (HBS) in the European Union are sample surveys of private households carried out regularly under the responsibility of the National Statistical Offices (NSIs) in each of the twenty five Member States (European Statistical System). Essentially, they provide information about household final consumption expenditures on goods and services, with considerable detail in the categories used; information on income, possession of consumer durable goods and cars; basic information on housing and many demographic and socio-economic characteristics. Conversely to other European statistical domains, HBS is voluntary and no EU regulation exists. Therefore there is a great freedom for each Member State to decide on the objectives, methodology, programming and resource assignment for their respective HBS. Since 1989, Eurostat has collected three data rounds for the following reference years: 1988, with a participation of 10 Member States, 1994, with a participation of 15 Member States (EU-15) and two EFTA countries, 1 European Community Household Panel 2 European Union - Statistics on Income and Living Conditions 4

5 1999, with full participation of 15 Member States (EU-15) and one EFTA country and a limited participation (partial supply of data) of the 10 New Member States and 2 Candidate Countries. The HBS Working Group has decided to program the next round for the reference year In the first round, the methodologies used by the MS to carry out the HBS were very far from being harmonised. Since then, all the countries participating in this project and Eurostat have made big efforts in order to harmonise their HBS and to improve data comparability. However, there is still room for improvement. In order to allow Eurostat to process received data to perform an ex-post harmonisation and to answer specific requests of the users, countries deliver micro-data to Eurostat. However, the implemented gentlemen s agreement only allows Eurostat to disclose aggregated tables or indicators. In co-operation with the National Statistical Offices of the Member States, Eurostat has for many years worked on the quality - mainly the comparability of HBS statistics within the EU. In spite of the important progress already done, there is still big room for improvement regarding quality and harmonisation of HBS data. This document addresses the quality aspects of the last HBS round already collected, that is for the reference year Since this round has been done in a different way for the EU-15 and for 12 candidate countries (the 10 new Member States which acceded to the EU on May 1 st 2004 plus Bulgaria and Romania), many sections of this document will contain separate tables and analyses for both groups of countries. In general we will refer to the first group of countries as EU-15 and to the second as CC (Candidate Countries). The methodology and the structure of this analysis is consistent with the recommendations of Qualistat, the Eurostat s approach to statistical quality, which is based in the ISO definition of quality: the totality of features and characteristics of a product or service that bear on its ability to satisfy stated or implied needs. Quality of statistics is defined by Eurostat with reference to the following six criteria: relevance accuracy timeliness and punctuality accessibility and clarity comparability, and coherence. Each of these components will be covered by chapters 2 to 7 of this document. Although not a measure of quality, the costs involved in the production of statistics as well as the burden on respondents act as a constraint on quality. This topic will also be covered in Chapter 8. Below please find a short definition of the seven dimensions of statistical quality. Relevance 5

6 Relevance is the degree to which statistics meet current and potential users needs. It refers to whether all statistics that are needed are produced and the extent to which concepts used (definitions, classifications etc.) reflects user needs. Accuracy Accuracy in the general statistical sense denotes the closeness of computations or estimates to the exact or true values. Timeliness and punctuality Punctuality refers to the time lag between the release date of data and the target date when it should have been delivered, for instance, with reference to dates announced in some official release calendar, laid down by Regulations or previously agreed among partners. Timeliness of information reflects the length of time between its availability and the event or phenomenon it describes Accessibility and clarity Accessibility refers to the physical conditions in which users can obtain data. Clarity refers to the data s information environment whether data are accompanied with appropriate metadata, illustrations such as graphs and maps, whether information on their quality also available (including limitation in use ) and the extent to which additional assistance is provided. Comparability Comparability aims at measuring the impact of differences in applied statistical concepts and measurement tools/procedures when statistics are compared between geographical areas, non geographical domains, or over time. We can say it is the extent to which differences between statistics are attributed to differences between the true values of the statistical characteristic. There are three main approaches under which comparability of statistics is normally addressed: comparability over time, between geographical areas, and between domains. Comparability over time refers to comparison of results, derived normally from the same statistical operation, at different times. The geographical component of comparability emphasises the comparison of statistics between countries and/or regions in order to ascertain, for instance, the meaning of aggregated statistics at European level. Geographic comparability is not of course limited to the comparability within EU. The EU statistics can be compared with other international statistics, for example with Japan and USA. In the EU context, it can exist a European reference, to which each Member State statistic could be benchmarked Comparability between domains refers to non-geographical domains, for instance between industrial sectors, between different types of households, etc. Coherence Coherence of statistics is therefore their adequacy to be reliably combined in different ways and for various uses. It is, however, generally easier to show cases of incoherence than to prove coherence. When originating from different sources, as HBS are, and in particular from statistical surveys of different nature and/or frequencies, statistics may not be completely coherent 6

7 in the sense that they may be based on different approaches, classifications and methodological standards. Conveying neighbouring results, they may also convey not completely coherent messages, the possible effects of which, users should be clearly informed of. 7

8 2. RELEVANCE Relevance is the degree to which statistics meet current and potential users needs. It refers to whether all statistics that are needed are produced and the extent to which concepts used (definitions, classifications etc.) reflect user needs. The aim of this chapter is to describe the extent to which the Household Budget Surveys are useful to, and used by, the broadest array of users HBS potential: scope and structure of the information collected Several characteristics make the HBS a unique source of information on household income and living conditions in the European Union. These include comparability of the data generated, multi-dimensional subject matter, a framework for co-operation and harmonisation for the generation of comparable social statistics in the EU Comparability. The need for genuinely comparable data arises not only because it is important in itself but because such data give Member States and the EU the possibility of bench-marking and defining best practices in terms of social and economic policy. Apart from its obvious relevance at the EU level, comparable information is also very valuable for policy at the national level, as it helps each country to judge its place relative to others in the EU. The HBS is designed to achieve data comparability in a two-step process: Before each round of actual data collection process, Eurostat proposes a set of methodological recommendations which are approved by the HBS Working Group. Once field surveys are completed and data are transmission to Eurostat, data are processed in Eurostat for additional harmonisation, basing on the methodological differences between countries found from the metadata communicated to Eurostat. Though the survey organisation and execution is left to individual countries, the HBS design has a number of features which enhance cross-national comparability. These include the publication of a methodological guide as a reference, common concepts and classifications and a common format for the production data-base where Eurostat stores the harmonised HBS micro-data for exploitation. 8

9 Multi-dimensional subject-matter The need for multi-dimensional coverage results from the fact that many of the social issues and social processes are clearly inter-linked. Policy makers need, on the one hand, simple and well-defined measures which can serve as a focus for action (for instance, how much people expend in a specific good or service, in which countries ). However, for policy measures to be effective, there also needs to be detailed research on specific measures and processes. These two conflicting demands cannot be pursued with separate methods and data sources: a high degree of consistency between the basic measures and the conceptual research is required. It is, therefore, necessary that our data be comprehensive in coverage. In response to this concern, the HBS has been designed to cover, with household consumption expenditure as the core topic, diverse aspects of living conditions simultaneously, for the same set of households and persons. This permits linkage between different aspects and greatly enhances the potential for policy-relevant research Tool for harmonisation The HBS also serves as a tool for the development of new modes of operational cooperation between Eurostat and NSI and other national institutions in the design and implementation of a common methodological framework, and in the development of common approaches and arrangements for data analysis. Big efforts have been done in using ESA-95 as a fundamental reference for HBS methodological framework, t The most visible outcome of this feature is all the methodological documents and manuals produced by the HBS Working Group leaded by Eurostat Users and uses: projects by or in collaboration with Eurostat Users at national level Historically, the prime objective of conducting HBS in all the Member States was to collect information on household consumption expenditures for use in updating the weights for the basket of goods used in the Consumer Price Indices (CPI). The weights measure expenditures on specific goods and services items as a proportion of total expenditures. Over the years, the range of uses has grown, as the surveys also had to meet the requirement to give a picture of living conditions of private households in certain areas and periods of time. To this end, the surveys provide detailed descriptions of a private household s total consumption and expenditures by household characteristics such as income, possession of consumer durable goods, housing and many demographic and socio-economic characteristics. The surveys also provide information on levels of living in terms of income and expenditure. Hence HBS are multi-purpose surveys which cater for a large number of uses and users. In terms of the scope and detail of information supplied, the surveys are an invaluable source on economic and social living conditions of households and individuals in the EU Member States. The HBS are not only used by the NSIs to produce the weights in the CPI, but also for determining the weights in the harmonised index of consumer prices (HICP) at the EU 9

10 level. On 7th March 1997 the European Commission (Eurostat) published the first set of HICP, which since then have appeared regularly each month with a time-lag of one month. The NSIs are obliged by the EU Council Regulations to submit the weights (together with the HICP sub-indices) using the four digits categories and sub-categories, derived from the COICOP international classification, known as COICOP/HICP. 3 Although, it is not required for the weightings in the HICP that HBS are carried out more frequently than once every five years, Member States are obliged to up-date the weightings with a frequency sufficient to meet the comparability requirement. Accordingly, Member States experiencing changes in consumption patterns should increase the frequency of the survey (Commission Regulation (EC), No 2454/97). The HBS is also a source for household consumption expenditures for the purpose of measuring consumption expenditures in the National Accounts in many countries (see Table 6). Therefore data must comply the ESA 95 definitions and classifications, implying for instance that the COICOP 1999 must be used. Finally, the survey information is used widely in the MS as the basis for studies of living conditions of private households. These might be studies focusing on certain subjects for example patterns of consumption expenditures or income, or relations between different subjects such as the influence of the level of income on consumption patterns, or patterns of consumption expenditure in relation to different types of households. Other applications are the: assessment of the impact on household living conditions of existing or proposed economic or social measures, particularly changes in the structure of household expenditures and in household consumption; estimation of the redistributive effects of direct and indirect taxation, and of a wide range of social benefits, on the situations of the various types of households; analysis of the variations in levels of living over a period of years and the disparities among households in the different socio-economic groups, geographical areas, rural and urban zones, etc.; Data derived from the HBS (in connection with income surveys) may also be used in connection with the determination of minimum wage levels, assessment of the need for revision of minimum wages and other questions relating to wage determination. Specific sub-populations might be studied such as the elderly, the young, manual workers or private sector employees. A number of countries specifically use the survey information to study the distribution of income/consumption with the purpose of looking at poverty issues. Moreover, there is a wide range of customers for the data outside the statistical offices. The different groups of users are: Ministries and public administrations, which use the data for economic and social policy planning purposes; Universities and research organisations, which use the data for research on living conditions of private households; 3 From January 2000, the COICOP/HICP is based on the COICOP 1999 (Commission Regulation (EC), No 1749/1999). The COICOP/HBS has the same basis and has been applied for the 1999 data. The most recent ESA 95 data are also according to the COICOP 1999 (period ). 10

11 Private firms and consultants. Their use is often directed towards analyses of consumption patterns of households in relation to the marketing of private consumer products; The general public that often gets the information via mass media in general or publications published by the statistical offices. The NSIs also produce statistical paper publications presenting the results of the surveys. Most countries have a system of producing short, focused reports emphasising particular aspects of households living conditions; these are often published within a relatively short time of finishing the survey. Most offices also disseminate data via electronic means, for example on diskettes or CD-ROMs - and, in some countries, via the Internet EU perspective EU objectives overlap national objectives to a large extent, but the focus is different. The three main users at the national level also appear at the European level. From the EU perspective, the goal is to contribute to the higher quality of HBS consumption expenditure required for the ESA, for the weights in the HICP and for studies on living conditions for all the by now 15 EU countries. This would indirectly serve the most important European users, i.e. the ECB and the European Commission. Other users of Household Budget Surveys data at European level have been the Commission s Directorates-Generals of Economic and Financial Affairs, Employment and Social Affairs, Health and Consumer Protection, Research and Regional Policy. The Commission uses the data mainly for policy purposes in connection with social and economic planning, and to help identify needs or establish targets. Social policy purposes include transport and environmental issues; economic purposes relate to the analysis of poverty and social exclusion and also to consumer (protection) policies. Eurostat also produces printed statistical publications presenting the results of the survey. An example is the Statistics in Focus series of short, focused reports emphasising particular aspects of household living conditions (e.g. on expenditure patterns of elderly people). A main dissemination line of the HBS results is the Eurostat s network of data shops. The results are also available on CD-ROM. Eurostat takes the position that quality improvements should focus on the comparability of certain items as well as reducing periodicity and improving timeliness. To have frequent and timely data, available maybe even on an annual basis, is likely to activate a large latent demand for HBS household expenditure data, in a world where consumption patterns could change drastically in short time. Moreover, the link to ESA and HICP will make more extensive and integrative statistical descriptions possible. The following list gives a selection of uses of HBS data at EU level: Eurostat dissemination network: A regular and extensive use of HBS data is done by means of the access to the HBS domain on NewCronos. Until October 2004, this access was limited to a rather small number of users: personnel of the EU institutions, personnel of the NSI and other national governmental bodies. However, in October 2004, NewCronos has become accessible for the general public too through the web page of Eurostat. This fact has dramatically widened the number of potential users of the European HBS data. 11

12 Besides this regular use of HBS information, a large number of specific requests from public bodies, private companies, journalists, academic researchers, received continuously since the first HBS data became available, have been answered. Other Eurostat teams and units: HBS is often requested as a complementary source of methodological information and data for other statistics managed by Eurostat. In particular the HBS team has received questions from the teams in charge of statistics on economical matters, food, education, income and living conditions, demography and energy among others. DG Health and Consumer Protection (Sanco): In the beginning of 2002, DG Sanco published a document titled Consumer Policy Strategy This document sets out the Commission s strategy for consumer policy at European level over the period One key pillar for the development and assessment of this strategy was the development of the Commission s knowledge base on information and data on consumers and the market. Eurostat, through the HBS team, has an essential role in this task. The HBS team of Eurostat collaborates regularly with DG Sanco giving support on statistical matters. DG Sanco has not only benefited from HBS data but also from methodological support in statistical issues connected with consumer policy and from the preparation and edition of some publications. Special relevance has the first two editions of the publication Consumers in Europe. Facts and figures of the series Panorama of the EU, which are based to a great extent on HBS data. The first edition, which was issued at the beginning 2002, was the best seller of the Panorama series during that year. The second edition will be issued by July The third editions is already planned for the end of DG Employment (Empl): DG Empl often uses HBS data as a complementary source of statistical information in support of the Employment policy. The scope of this use includes some specific requests sent DG Empl as well as some contributions to various reports and publications of DG Empl. As an example, we may mention the last issues of the yearly publication The social situation of the European Union, which is prepared jointly by DG Empl and Eurostat and which contains a statistical annex including some data form HBS. DG Internal Market (Markt): In a similar way to DG Sanco and DG Empl, DG Markt has also addressed numerous requests of statistical information on the HBS domain in support of the information needs for their political activities. These requests have been connected with several areas of household consumption expenditure: e.g. services of general interest, transport European Parliament: Occasionally, the HBS team has also received some request from the European Parliament regarding statistical information on this domain, usually connected with particular topics such as housing costs, expenditure on life-long training and so forth. 12

13 2.3. Steps taken to promote data use It is clearly in the interest of the HBS project to be widely used and be visible through relevant analyses and publications. The above list shows that, from the beginning of the project, users have been a key focus of the project Consultation with users In pursuing this goal, Eurostat has been collecting formal and informal feedback from the main HBS users. The contacts between Eurostat and other DGs with regard to the statistical information support of their political areas have been frequent and some times formal. We may highlight the following formal contacts: DG Sanco and Eurostat have held several formal meetings and hearings during the last years in order to manage and to coordinate their activities in this area. Some of these meetings have been held at the highest level. Eurostat participates as a member of the Inter-Service Group (ISG) on Consumer Policy organised and leaded by DG DG Sanco. The person representing Eurostat in this ISG is the head of the HBS project. Eurostat has answered numerous Inter-Service Consultations (ISC) launched by DG Sanco regarding questions about consumer policy. The answers to these consultations have been usually prepared by the HBS team members. Representatives of DG Sanco and DG Empl have participated in several HBS working group meetings in order to explain the statistical needs of their DGs. Representatives of the HBS teams have been invited to participate in various meetings and seminars organised by other teams of Eurostat in order to interchange information about the use of HBS data in support of other statistical domains. Besides these formal contacts, there have been quite frequent informal contacts for specific purposes, ad-hoc requests and regular collaboration with other teams of Eurostat, DG Sanco, DG Empl and DG Markt. The amount and variety of the information interchanged in these contacts has been huge Dissemination of aggregated HBS data Dissemination of aggregated HBS data (tables and indicators) has been one of the most important means of promoting the use of HBS data. Dissemination has been done by several ways: Electronic publication of HBS tables and metadata on NewCronos, the statistical data base of Eurostat. Since October 2004, the access to this data base is no longer limited to a relative small number of users but it is also available for the general public through the web site of Eurostat. Paper publication with statistical and methodological contents about HBS: Statistics in Focus, Panorama of the EU, Methodological manuals Answers to ad-hoc requests received by the HBS team from numerous users. 13

14 Promotion of micro-data use Following a request sent by numerous European research institutions, Eurostat has tried to make HBS micro-data base accessible for research purposes. After the HBS WG meeting of October 2001 in Luxembourg, Eurostat sent a consultation to all the delegates of the Member States in the HBS WG about the possibility of delivering HBS to third parties for scientific or research purposes. Thirteen countries out of fifteen answered this question with the following results: Only two countries gave a positive answer without any kind of limitation. Seven countries gave a positive answer with some kind of constraint (the most usual one was the requirement of asking a particular permission for each delivery). Four countries were fully opposed to this initiative. In consequence, this initiative was not accepted. At the moment, only a few statutory people from Eurostat are allowed to directly work on the HBS micro-data sets delivered by the countries. However, Eurostat is allowed to produce and to disclose any aggregate table or indicator from the above mentioned micro-data base, provided that these tables and indicators do not contain any information on individual households or household members. 14

15 3. ACCURACY The aim of this section is to assess the quality of the HBS data in relation to the sampling process. As noted in Section countries are almost fully responsible for the sampling design of HBS. Although HBS surveys in most countries are based on probability sampling by design, there also are some exceptions. This lack of uniformity in sampling methods and methodology has made impossible the analysis at European level. This chapter collects all available information at county level in order to give a picture about the overall reliability of the data, reflecting the combined effect of sampling, nonresponse and response errors. Due to the limited scope of data collection for CC in 1999, this analysis has only been carried out for the EU Response rates and overall achieved sample sizes A critical aspect of the quality of survey data is the extent to which households and persons are successfully interviewed. Table 1 presents some information on response rates in terms of the sample of households: Response rates for the EU round Achieved net sample (n) Response rate (R%) (1) (3) BE 3,745 10% DK 2,727 62% DE 62,150 (d) EL 6,258 68% ES 9,891 65% FR % IR 7,644 55% IT 20,930 79% LU 3,012 50% NL 1,851 83% AT 7,098 25% PT 9,643 73% FI 4,359 63% SE 3,031 53% UK 7,100 63% Explanatory notes columns: (1) Households finally completed, along with diaries and all interviews. (3) Completed households, as % of the number originally selected plus substitutes if any. 15

16 3.2. Item non-response By item non-response is meant the failure to obtain the required information on some particular items or questions in an otherwise completed interview. This failure can arise from two types of sources. Firstly, no response, or no usable response may have been obtained to a question included in the interview. Secondly, in a comparative multicountry survey such as the HBS some items may not have been included in the required data supplied by every country to Eurostat. The first source of item non-response is closely link to the design and implementation of the sampling procedures. Thus it has been controlled directly by each country without any participation of Eurostat. HBS data delivered to Eurostat is supposed to be free of this type of problem. As regards to the second source, not all the countries have supplied to Eurostat all the requested variables. Practically all countries have omitted the transmission of some variable, particularly among those classified as not essential on the methodological recommendations of Eurostat. Below please find a few examples of missing items, particularly chosen for their relevance: Breakdowns for the first division of COICOP-HBS cp01 - Food and not alcoholic beverages are not available for German and Swedish data. Various countries did not supply information at the latest level of detail of COICOP- HBS (5 digits). For these reason Eurostat has decided to remove this level from data tables published on the HBS domain of NewCronos. Therefore, COICOP classification used for breaking down household consumption expenditure is no longer COICOP-HBS but the standard COICOP as developed by the OECD. Almost one half of the countries have not given information about the variable: HA09 population density domain Sampling error Again, the freedom of each country to choose sample sizes and methods prevents from doing a systematic analysis for the whole European data set. The information presented below is based on the metadata supplied by the countries in a non-harmonised way Sampling error and related measures for household final consumption expenditure All countries have supplied detailed information about the samples sizes (see the table at the beginning of this chapter) and sample design: Main sampling characteristics and response rates for the EU-15 16

17 Survey population: main Over-sampling of Number of Probability Ultimate 1999 round Sampling frame exclusions special domains stages sampling sampling unit (5) (6) (7) (8) (9) (10) BE Brussels (a) National population register 2 X Household DK Greenland and Faroe Islands Central Population register 1 (XX) (c) Address DE (e) Microcensus 1997 and Household EL Census of Population XX Dwelling ES 1991 Population Census with the inclusion of new dwellings 2 XX Dwelling FR 1990 Census updated by new constructions 2 XX Dwelling IE Oversampling of households with more than Census of population enumeration agricultural three lodgers areas households 2 XX Household IT Population register (on commune basis) 2 XX Household LU Central Population register NS X Household NL (11) Geographic Base Register, selfemployed from the General Business Register 2 X Address AT Autrian micorcensus 1994 and 1995 (new building/dwellings were added) 2 XX Dwelling Master sample based on 1991 PT 1 XX Dwelling Population Census Sparsely populated FI Central Population register 2 XX Individual regions Individuals of more than 74 SE Population Register 1 XX Individual years of age (h) 2 (NS in N. UK Postcode Address File X Household (i) Ireland) Explanatory notes: (5) National areas or categories of households not included in the survey. (6) Population groups for which higher sampling rates and/or special procedures are used. (7) The main source(s) for sample selection. (8) Number of stages in which the sample is selected. (9) Indicates departures from probability sampling. XX: probability sampling. X: Sample selected randomly but low response rates indicate possible departures from representativity of sample. -: No probability sampling. (XX) for DK: the sample is concretely decided upon taken into account the place of residence of the interviewer. (10) The unit selected at the last stage of the sampling procedure. (11) About half the sample of the Dutch HBS 1999 included households that participated in HBS However, only two countries from EU-15 have supplied accurate figures on sampling errors Sampling error by expenditure components In this case only three countries have communicated concrete standard errors and coefficients of variations for household final consumption at the level of COICOP division (2 digits). In both countries, as it could be expected, standard errors where smaller than for total population (because they are referred to a single component of consumption expenditure), but coefficients of variation are generally greater. 17

18 Estimated variable (for the whole population) BELGIUM GREECE NETHERLANDS Standard Coefficient Standard Coefficient Coefficient of error of variation error of variation variation Mean household final consumption expenditure 4935,5 0, ,16 0,68 Mean household final consumption expenditure on food an non-alcolic beverages (COICOP-01) 656,38 0, ,92 1,16 Mean household final consumption expenditure on alcolic beverages (COICOP-02) 267,37 0, ,73 3,32 Mean household final consumption expenditure on clothing and fooware (COICOP- 03) 573,97 0, ,2 1,76 Mean household final consumption expenditure on housing, water, electricity, gas and other fuels (COICOP-04) 1291,11 0, ,12 0,78 Mean household final consumption expenditure on furnishing, household equipment and routine maintenance of the house (COICOP-05) 1598,16 0, ,19 2,49 Mean household final consumption expenditure on health (COICOP-06) 800,29 0, ,41 3,54 Mean household final consumption expenditure on transport(coicop-07) 2139,3 0, ,49 2,96 Mean household final consumption expenditure on communications (COICOP-08) 324,3 0, ,44 1,54 Mean household final consumption expenditure on recreation and culture(coicop- 09) 1440,21 0, ,55 1,69 Mean household final consumption expenditure on education (COICOP-10) 184,12 0, ,96 5,18 Mean household final consumption expenditure on hotels, cafes and restaurants(coicop-11) 773,84 0, ,26 2,23 Mean household final consumption expenditure on miscellaneous goods and services(coicop-12) 1394,63 0, ,53 0, Sampling errors by population subgroups (subclasses) Most analyses require separate estimates for subgroups in the population, such as subclasses defined in terms of household size, household type and other household and personal characteristics. On the one hand, the magnitude of the sampling error increases in inverse proportion to square-root of the subclass sample size. On the other hand, this 18

19 tends to be partially compensated by some reduction in population variance and design effects to the extent the subclasses represented homogenous partitions of the population. Unfortunately only two countries have communicated these data and, therefore, we have little numeric evidence of this. Population subset (estimated variable: mean household final consumption expenditure) Population size Sample size BELGIUM Standard error Coefficient of variation Population size Sample size GREECE Standard error Coefficient of variation Total population ,5 0, ,17 Employment status of the reference person = Manual ,6 0, ,97 workers in industry an services Employment status of the reference person = Non Manual ,32 0, ,82 1,72 workers in industry an services Employment status of the reference person = Selfemployed ,21 0, ,04 Employment status of the reference person = Unemployed Employment status of the reference person = Retired Employment status of the reference person = Inactivesothers Households with zero active persons ,98 0, , ,7 0, , ,36 0, , ,68 0, ,99 Households with 1 active persons ,63 0, ,74 Households with 2 active persons ,57 0, ,4 Households with 3 or more active persons ,23 0, ,12 Type of household = single person ,12 0, ,83 Type of household = single parent with dependent children ,35 0, ,53 Type of household = two adults ,37 0, ,85 Type of household = two adults with dependent children ,7 0, ,51 Type of household = three or more adults ,84 0, ,57 Type of household = three or more adults with dependent ,73 0, ,54 children Age of teh reference person = less than 30 years ,93 0, ,12 Age of teh reference person = between 30 and 44 years ,64 0, ,52 Age of teh reference person = between 45 and 59 years ,2 0, ,82 Age of teh reference person = 60 years and over ,4 0, ,79 19

20 4. TIMELINESS AND PUNCTUALITY In addition to relevance and statistical accuracy, timeliness is the third important aspect of data quality. The main concern here is the time lag between data collection and data release. Related issues include the timing and duration of the data collection phase, and the format and quality of the data released Timeliness The character multi annual of the European HBS and the requirement of ex-post harmonisation in Eurostat after the reception of each data package make HBS different from other statistics. This causes the extension of the production period over several years. The following table gives information about the actual periods of data collection in each country: Table: Survey frequency and data collection years for the HBS round of 1999 EU-15 Frequency Collection year CC Frequency Collection year Belgium Annual 1999 Bulgaria Annual 1999 Denmark Annual 1998 Czech Republic Annual 1999 Germany Annual 1998 Estonia Annual 1999 Greece 5 years Cyprus 5 years Spain Annual 1998 Latvia Annual 1999 France 5 years Lithuania Annual 1999 Ireland 5 years Hungary Annual 1999 Italy Annual 1999 Malta 5 years Luxembourg Annual 1998 Poland Annual 1999 Netherlands Annual 1999 Romania Annual 1999 Austria 5 years Slovenia Annual 1999 Portugal 5 years Slovakia Annual 1999 Finland 5 years 1998 Sweden 5 years United Kingdom Annual Data of most EU-15 countries were received by Eurostat during Then they were fully processed and uploaded onto NewCronos by July The main exceptions were: Ireland. Although provisional data from Ireland was available in the dates indicated for the bulk of countries, certain difficulties in the calculation of Irish imputed rent which required a long technical discussion between Eurostat and the Irish NSI delayed the availability of this component of household final consumption expenditure until February France and Portugal. These countries were the last ones in completing their surveys. For this reasons they sent their data much later than the others and they were not fully process and uploaded onto NewCronos until June The operation NewCronos loading with HBS data of 1999 was completed in September with the following operations: Overall revision of all data and correction of a limited number of small errors found for a few countries. 20

21 Calculation of all the aggregated indicators for the Eurozone, EU-15 and EU-25 Attachment of metadata files using the standard format recommended by Eurostat (SDDS) The following gantt chart gives a global overview of the round timeliness: ID Task Name 1 Reference year 2 Surveys 3 BE 4 CZ 5 DK 6 DE 7 EE 8 EL 9 ES 10 FR 11 IE 12 IT 13 CY 14 LV 15 LT 16 LU 17 HU 18 MT 19 NL 20 AT 21 PL 22 PT 23 SI 24 SK 25 FI 26 SE 27 UK 28 Dissemination on NC 29 1 (EU15 - FR & PT) 30 2 (CC) 31 3 (IE FR PT) 32 Final revision Qtr 1 Qtr 3 Qtr 1 Qtr 3 Qtr 1 Qtr 3 Qtr 1 Qtr 3 Qtr 1 Qtr 3 Qtr 1 Qtr 3 Qtr 1 Qtr 3 Qtr 1 Qtr 3 Qtr 1 Qtr 3 Qtr 1 02/07 01/06 01/06 15/ Reducing the impact of poor timeliness As we have just explained before, the timeliness of HBS figures for the round 1999 was dramatically affected by the lack of availability of a few countries. However the bulk of data was available in a quite reasonable period. The solution devised to overcome this problem was to start the exploitation of HBS data much before than all data were available. In certain cases missing data of 1999 was replaced by data of Quite often this type of replacements did not produce serious distortions because the structures of household consumption expenditure in many EU-15 countries use to be rather stable over the time, but in other opportunities analyses could not be as accurate as wished because of the lack of complete data sets. This has been actually done for the issue of a first Panorama publication basing on HBS data of 1999 by the end of Nevertheless, although these measures are possible and, in some cases, may be acceptable, for future rounds improving timeliness will be a must. The following paragraphs propose various initiatives to achieve it. 21

22 4.3. Improving timeliness Reducing the time lag between data collection and release is being given the highest priority. The following actions are intended to improve timeliness and punctuality for the round of 2005 compared to the previous rounds: Better synchronisation of national surveys in the round of 2005 compared to the round of Simplification of the list of variables to be delivered by each country. Improved ex-ante harmonisation by wider agreement on the main methodological recommendations, and hence reducing the needs of additional ex-post harmonisation More standardised and secure data transfer procedure. 22

23 5. ACCESSIBILITY AND CLARITY 5.1. Accessibility Forms of dissemination Reference database (the 'Production Data Base') The responsible services of the NSIs provide Eurostat with the results from their national HBS. This information is stored in the so-called Production Data Base (PDB). In order to simplify data exploitation, this PDB is organised in two separate data sets: A full joint harmonised data set (with full contents for all countries at the levels of households and household members) A simplified joint harmonised data set (with a selection of the most variables most frequently used for all countries but only at the household level) We must remark here that the New Member States, Bulgaria and Romania have participated in the round of 1999 in a very limited way. They have sent a selection of aggregated tables which have been directly uploaded onto NewCronos without being submitted to any ex-post harmonisation process. Therefore, the HBS PDB of the round 1999 only contains data from the EU-15 countries. The micro-data files collected nationally by the NSIs of the EU-15 have been sent to Eurostat in specific formats for each country. For all the countries, data was anonymised before their transmission to Eurostat. This means that, households and household members are only identified by alphanumeric codes, in such a way that it is not possible to link any data record with the interviewed people and households. Data delivered by each country is structured in two main pieces of information: A household members file A household file Each record of these files is qualified by a weight depending on the sampling procedure employed in each country. These weights are supplied by the countries as an additional attribute of each record. These files are then processed by Eurostat with a multiple purpose: To put all data into a common format To harmonise, concepts, definitions and classifications as much as possible. This may include the reclassification and aggregation of certain variables as well as imputing certain values (e.g. owners-occupiers imputed rent), whenever this task has not been done previously by the country supplying a particular set of data. To calculate the derived variables, that is, certain variables which are not directly supplied by the countries but calculated from the basic variables sent by them. 23

24 To store all these data in the HBS PDB. This data base is only accessible by a small team of Eurostat statutory personnel. Their contents are used to create all the products for dissemination, but it cannot be directly set available to other people because the lack of authorisation to Eurostat for disclosing any HBS micro-datum. This process is represented by the following data-flow-diagram (DFD): Simplified Data Flow Diagram of HBS data processing in Eurostat NOT HARMONISED MICRO DATA DATA DELIVERED BY COUNTRY 1 DATA DELIVERED BY COUNTRY 2 DATA DELIVERED BY COUNTRY N SPECIFIC DATA PROCESSING FOR COUNTRY 1 SPCECIFIC DATA PROCESSING FOR COUNTRY SPECIFIC DATA PROCESSING FOR COUNTRY N COMMON DATA PROCESSING FOR ALL COUNTRIES HARMONISED MICRO DATA JOINT HARMONISED MICRO DATA (Households and hh members) SIMPLIFICATION AND MONETARY ADJUSTMENT SPECIFIC PROGRAMS FOR COMPLEX REQUESTS SPECIFIC PROGRAMS FOR NORMAL REQUESTS STANDARD TOOL FOR SIMPLE REQUESTS AGGREGATED TABLES AND INDICATORS TABLES AND IND. FOR AD-HOC REQUESTS SIMPLIFIED HARMONISED MICRO DATA (Households only) DATA EXTRACTION FOR NEW CRONOS NEW CRONOS TABLES LEGEND: COMPUTER PROCESS PROGRAMS FOR THE PREPARATION OF PUBLICATIONS TABLES AND INDICATORS FOR PUBLICATIONS DATA BASE Data of the New Member States, Bulgaria and Romania have been collected directly as aggregate tables and they have been uploaded onto NewCronos without any kind of expost harmonisation by Eurostat. As a consequence, data of these countries are much less comparable than the data of the EU Dissemination of aggregated tables on NewCronos As soon as data of each country have been received and processed into de PDB, they have been also processed for extracting a set of selected aggregated tables for NewCronos, the statistical data base of Eurostat. Data of the New Member states, Bulgaria and Romania has been received directly in aggregated format and were not processed before sending them to NewCronos; except for data formatting. The structure of NewCronos data tables for HBS is the following: 1. Consumption expenditure of private households 1.1. Global consumption expenditure (mean and medium) 24