Transition from work to retirement. Evaluation of the 2012 labour force survey ad hoc module

Transcription

1 Transition from work to retirement Evaluation of the 2012 labour force survey ad hoc module

2 Preface This report evaluates the 2012 labour force survey (LFS) ad hoc module (AHM), which examined the transition from work to retirement. The main objective of the report is to assess the way in which the module was conducted, by providing information on the quality of the data set and presenting preliminary results. Recommendations relating to a possible repeat of the module as part of a future survey are also included. The EU LFS is a large-sample survey of private households, which provides detailed quarterly and annual data on employment, unemployment and economic inactivity. The LFS was established by Council Regulation (EC) No 577/98 of 9 March 1998 on the organisation of a labour force sample survey in the European Union. This Regulation and its amendments set out provisions for the design, characteristics and decision-making process of the survey. The transition from work to retirement was the subject of the 2006 LFS ad hoc module, and the same topic was chosen again for 2012 (Regulation No 365/ ). The 2012 proposal was prepared in the light of lessons learnt during the course of the 2006 LFS ad hoc module and, where possible, appropriate changes were made. The involvement of a large number of labour market specialists from national statistical offices, Eurostat and other Commission Directorate- Generals also played an important role in the planning of the 2012 module. The national statistical offices all contributed to a documentation exercise, for which each of them drafted a mapping between the possible responses for the PENSTYPE variable (the type of pension the person is currently receiving) and their national pension system. The evaluation of the 2006 module and the documentation produced in preparing the 2012 module are publicly available. 2 Both are designed to make it easier for researchers and the public to understand and use AHM data. Administrative differences between the pension systems in different countries are highlighted, and the areas in which, as a result, comparison between countries is not possible, are made clear. The first chapter of this document gives general information on AHM Subsequent chapters then provide a detailed description of each variable, together with information as to the comparability of this variable both across countries and between 2006 and 2012, and other information on data collection. The annexes to the document include country abbreviations, the list of tables proposed for online publication and the text of Regulation No 365/2008 with the list of variables. This document is based on data sent to Eurostat before the end of Although minor revisions of the data set may have happened after this date, the data was considered stable enough for analysis and interpretation. The quality reports provided by participating countries were particularly useful in helping Eurostat to interpret certain values and have also contributed to ideas for a potential repeat of the module. Colleagues from many national statistical offices provided Eurostat with insight into the national circumstances, explaining specific results that did not fit patterns seen in other countries. Eurostat would like to thank all contributors. This report was prepared by Diana Ivan and Håvard Lien of Eurostat s unit working on labour market statistics (F3). Luxembourg, March Transition from work to retirement. Evaluation of the 2012 module Page 2

3 Table of contents Preface... 2 Table of contents... 3 Chapter 1: General information on the module... 5 Executive summary for researchers... 5 Recommendations relating to a repeat of the module... 6 Description of the module... 6 Aims of the module and main findings... 6 Participating countries... 7 Target population... 7 Main findings... 9 Description of the variables... 9 Links with AHM Links with the core LFS General issues relating to data collection Sample size Non-response rates Other measurement issues Chapter 2: Quality analysis by variable PENSION: Person receives or does not receive a pension Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations PENSTYPE: Type of pension(s) currently received Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations EARLYRET: Incidence of early retirement Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations AGEPENS: Age at which a person first received an old-age pension Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations REASNOT: Main reason for not remaining in employment longer Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations WORKLONG: Wish to remain in employment longer Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations Transition from work to retirement. Evaluation of the 2012 module Page 3

4 7. REDUCHRS: Reduced working hours as a step towards full retirement Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations STAYWORK: Main reason for remaining in employment Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations PLANSTOP: Plans to stop working in the future Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations BUILDPEN: Information on pension rights built up to date Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations CONTWORK: Expectations of continuing working or looking for a job after starting to receive an old-age pension Short description Filter conditions and codes Analysis of the questionnaires Analysis of the results Conclusions and recommendations Annexes Annex 1: Abbreviations Annex 2: Main AHM 2012 tables Annex 3: Commission Regulation (EU) No 249/ Transition from work to retirement. Evaluation of the 2012 module Page 4

5 Chapter 1: General information on the module Executive summary for researchers The EU LFS sample size is about 1.5 million people and surveys are carried out every quarter. Only private households are included. The survey is conducted by means of interviews with each individual in the sample. The interview method varies across countries. In most countries, proxy interviews with another person in the household are allowed. Interviews are generally conducted in person, at least for the first wave, but subsequent follow-up interviews can be conducted by phone. Participation in the survey 3 is compulsory in seven EU countries and in two of the participating European Free Trade Association countries. The variables on which the LFS collects data will be referred to in this document as core LFS variables, to distinguish them from the AHM variables. Their list is available as an annex to Regulation 377/ on codification and filters. Explanatory notes on each of the variables are also available 5. Regulations on multi-annual programmes of ad hoc modules and Regulations defining the list of variables to be collected in a specific year provide further legal basis 6 for the LFS AHMs. Commission Regulation (EU) No 249/2011 adopting the specifications of the 2012 ad hoc module defines the eleven variables on which data was collected in AHM 2012 and describes the target population of the module and of each variable. A task force was commissioned to define a proposed list of variables to be collected and to provide explanatory notes to accompany them. A document has been prepared giving answers to frequently asked questions on the concepts covered by the variables. These are all publicly available. This document summarises their main elements, and adds further information on data comparability between countries and between surveys. The first chapter explains in detail the structure of the target population of AHM It sets out which populations are included and which are excluded from the module. Since the module focuses on people involved in the transition from work to retirement, not all potential respondents aged were interviewed. The main side effect of the choice of target population for this survey is that analysis by gender is limited, because the labour market participation rate is different for men and women. The AHM 2012 database does not include a non-applicable field (which applies to those not in the AHM target population) for all countries for which data was collected. For this report, the size of the non-applicable category for Germany, France, Austria, Sweden and Switzerland was estimated by crossing the AHM data with the core LFS data. Non-response rates by variable and country are also included in the first chapter. Any non-response rate higher than 15 % will be systematically flagged in this document, as this is the level considered to make the remaining data for that question and population unreliable. The reader is therefore made aware of the limits of the data quality for specific countries and/or specific variables of the module. The second chapter presents each variable in detail, from a data collection perspective as well as with a view to a possible repeat of the module. At a glance, the quality of each variable can be summarised as follows: a good level of comparability: PENSION and AGEPENS; comparison is possible, but analysis should take into account the specific differences existing at national level: PENSTYPE, EARLYRET, WORKLONG, REASNOT, REDUCHRS and STAYWORK; and 3 See table 3 for information on participation at AHM 2012 by country Transition from work to retirement. Evaluation of the 2012 module Page 5

6 lowest quality variables whose use should be confined to national analysis (provided response rates are sufficient): PLANSTOP, BUILDPEN and CONTWORK. Please note that names of variables will always be given in capital letters. Definitions and code lists of the variables are available in the subchapters of chapter 2. With the exception of table 2 on sample size, table 4.1 and related graphs on percentile and mode, all data is weighted. Recommendations relating to a repeat of the module Were a module with the same (or a similar) topic to be run again, a number of recommendations made in the light of the 2012 experience should be taken into account. These recommendations were formulated by experts from the national statistical institutes based on their experience of the data collection and analysis conducted at national level, and by Eurostat based on their experience of analysing the dataset for all countries. (1) Consider changing the topic of the module from the transition from work to retirement to an analysis of the whole population within a certain age group. In general, it is considered difficult to collect detailed information on a complex phenomenon such as the transition from work to retirement by means of a small number of questions that assume a linear progression from work to retirement, because in practice the transition may be longer and atypical. A module focussing on a transition of any kind will inevitably involve retrospective questions and questions about future intentions. Experience, including from AHM 2012, has shown that these questions are most difficult to deal with in an AHM. (2) Choose simpler filters, both for the module and for each of the variables. Complex filters are difficult to manage during the preparation of the survey, interviewing, data processing and data analysis. For example, it would have been easier to collect comparable data across all countries for an AHM whose only filter was age. Moreover, users would have benefited from a richer dataset, allowing detailed analysis by gender and by country. A broader filter also reduces the risk of the target population being too small. The drawback of this approach would be a slight increase in the response burden but the effect of this could be mitigated throughout the module if simpler variables were used. (3) Reconsider the definition of a pension in the context of the module. Due to specific national legal provisions, it is not always easy to distinguish between pensions and other social benefits and this can often lead to a situation where similar social schemes are considered as a pension in one country and as a different kind of benefit in another. The 2012 AHM included a proposal to supplement the data collection with mappings of the pension systems in each country. This approach has the advantage of offering greater transparency of data collection. Little can be done however to improve the comparability at European level of any PENSTYPE data collected for this module. Moreover, the documentation and updating of administrative information involved in this exercise placed a considerable burden on participating countries. Description of the module Aims of the module and main findings The aim of this AHM was to analyse: how people leave the labour market; why they leave the labour market; why they do not remain in the labour market longer; and for how long the active population aged 50 to 69 expects to be in the labour market. The results were intended to be used to prepare the joint report on social protection and social inclusion, and in the areas covered by the open method of coordination in the field of pensions and the Europe 2020 Transition from work to retirement. Evaluation of the 2012 module Page 6

7 Strategy, and in particular for monitoring the guideline on increasing labour market participation. Participating countries The module was carried out in all EU countries (27 in 2012), Croatia, Iceland, Norway and Switzerland. The aggregated EU totals include all 28 current EU Member States (i.e. the 27 Member States as of 2012 and Croatia). Annex 1 gives the country abbreviations used in this report. Target population The target population of AHM 2012 is people aged either currently working or having worked beyond age 50. The target population is the same as for the AHM 2006 on the transition from work to retirement. The target group excludes people who have not worked beyond age 50 on the basis of their reduced proximity to the labour market and in order to reduce the volume of responses to be handled. It is however important to verify whether excluding these respondents had any negative effects on data analysis. It was often difficult to identify the exact target population of the module at the point of collecting data. In particular, some of the countries using paper questionnaires (Bulgaria, Greece and Hungary) reported significant difficulties in ensuring that respondents respected the conditions imposed by each of the filters. Many countries (including Malta and Switzerland, and a number of the countries using paper questionnaires) had difficulties in extracting data on experiences of working beyond age 50 from the core LFS, and had to collect the relevant information on age, work history and working life beyond age 50 again. In these cases, three extra questions were usually added to the interview as a check. There were also instances of filters being applied when coding the data, rather than at the point of collecting the data. This affected the data in the non-answering category and in some cases resulted in the data not being able to be used in the module. As noted in the Austrian report, having additional filters reduced the number of observations of more narrowly defined variables. Filtering for a very specific target population makes both understanding and communication of the data more difficult. It also limits the analysis of the data, by precluding certain breakdowns, including by gender. The following graph shows the target population of AHM 2012 compared to the total population aged 50-69, by gender, for EU-28. While at EU level there are more women than men in the age group (52 % of the total population aged are women), the target population of AHM 2012 (people aged 50-69, and currently working or having worked beyond age 50) contained more men than women. This is because over three-quarters of the group that was not part of the survey was female. Graph 1: Population aged by gender, employment status and participation in AHM 2012, EU-28 (number of people, in millions) Women, not in employment and not in the survey Men, not in employment and not in the survey Women, not in employment but in the survey Men, not in employment but in the survey Women, in employment Men, in employment Note: not applicable values are estimated for Germany, France, Austria and Sweden. Transition from work to retirement. Evaluation of the 2012 module Page 7

8 MT EL IT BE LU IE ES HR NL CY PL EU-28 PT RO AT FR DE SI HU UK LT SK SE BG LV DK EE FI CZ NO IS CH In each EU country, there were more women than men who: (i) were not in employment; and (ii) have not worked beyond age 50. There are however significant differences between countries, as shown in graph 2. People who are not in employment and have not worked beyond age 50 are less likely to have access to pension rights, although those who have worked in the past may still have pension rights. Unfortunately, the data set provides no information at all on people who stopped working before age 50. The Hungarian quality report points out that the choice of target population was more problematic for countries with a lower retirement age, because people who do not work beyond age 50 are in general very likely to be receiving a pension at the time of the interview or to be due a pension in the future, as a result of their previous employment. Moreover, although the module included a special variable for past contributions to several pension schemes (BUILDPEN), the results are unfortunately not available for those who have worked in the past, but not beyond age 50. This has therefore detracted somewhat from the usefulness of these results. Graph 2: Non-applicable AHM 2012 population aged by gender (% in the population aged of the same gender) 80% 70% 60% 50% 40% 30% 20% 10% 0% Men Women Note: values for Germany, France, Austria, Sweden and Switzerland are estimated. Excluding those who have not worked beyond age 50 from the data also means that the results from AHM 2012 cannot be generalised to apply to the full population aged and its transition to retirement. It is only in the case of the group in employment at the time of the interview that the analysis has not been affected by the choice of target population, neither with respect to gender nor country. This particular group represents only half of the total population aged in the EU however, and the percentage represented varies significantly by country, with employment rates ranging from 63 % in Sweden to 34 % in Malta. Transition from work to retirement. Evaluation of the 2012 module Page 8

9 Main findings Amongst the target population at EU level: the average age at which respondents receive an old-age pension for the first time is 59 years, for both women and men; 42 % are currently receiving a pension; the statutory old-age pension is by far the most common type of pension, with 81 % of those receiving some form of pension receiving a statutory old-age pension; approximately four in 10 people receiving an old-age pension had taken early retirement; and the two most frequently cited main reasons for not remaining in employment for longer were reaching eligibility for a pension and the respondent s own health problems or disability (reasons given by 37 % and 21 %, respectively, of economically inactive people receiving a pension). Description of the variables A full description of the variables, filters and coding, as defined in Commission Regulation (EU) No 249/2011, is available in Annex 3. The module contains the following 11 variables: PENSION PENSTYPE EARLYRET AGEPENS REASNOT WORKLONG REDUCHRS STAYWORK PLANSTOP BUILDPEN CONTWORK Links with AHM 2006 Person receives a pension Type of pension(s) Early retirement Age at which person first received an old-age pension Main reason for not remaining in employment longer Wish to remain in employment longer Reduced working hours as a step towards full retirement Main reason for remaining in employment (for respondents already receiving a pension) Plans to stop working in the future (for respondents already receiving a pension) Pension rights built-up to date Expects to continue working or looking for a job after starting to receive an old-age pension. The policy background for AHM 2012 is similar to that of AHM 2006, namely monitoring participation in the labour market. There are however no variables which are completely and directly comparable between the two surveys. The evaluation of AHM made the following recommendations for changes to be made to the variables were the module to be repeated. We have included only the six variables that, in principle, cover the same content in 2012 as in (1) REDUCHRS caused problems for respondents in 2006 because it asked about their future plans, when in many cases they had not made such plans at the time of the interview, due in particular to changes in the pensions regulation. The response item progressive retirement schemes 8 was not used in several countries because the concept itself was unfamiliar. The 2006 task force s recommendation to simplify 7 The publication is available at 8 Progressive retirement scheme/part-time pension was defined in the Transition from work into retirement publication, on page 48, as follows: 'this measure concerns older employees in some countries. This is part of measures to keep older employees in employment (incentives to stay at work). To avoid the exit from work, in case the employee wishes to decrease his/her working time before retiring, he/she could take a part-time job for example. It could be a classic part-time or what is called a progressive retirement scheme/part-time pension. The latter ensures a better remuneration than the classic part-time (e.g. 50% work paid 80%). In other words, it corresponds to a reduction of the number of hours worked with a less than proportional reduction in salary (e.g. 50% work paid 80%)' Transition from work to retirement. Evaluation of the 2012 module Page 9

10 this variable was followed in (2) PLAGESTP (corresponding to AHM 2012 variable PLANSTOP) caused problems for the same reason as REDUCHRS. Unfortunately, the fine-tunings proposed for 2012 did not solve the original problem of people finding it difficult to answer questions about their future plans. (3) REASRET (corresponding to AHM 2012 REASNOT) was reported to have too many response items each of which was too detailed. This made it difficult to distinguish between several of the items. Changes to this variable were proposed but not such as would necessarily simplify it. (4) AGEPENS caused problems due to difficulties in understanding the expression individual retirement pensions. Accordingly, this was changed to old-age pension for (5) OTHBENF (corresponding to AHM 2012 PENSTYPE) caused problems due to the interpretation of the term individual benefits in the explanatory notes. The 2012 survey asks for broad categories of pension types. (6) FININCTV (corresponding to AHM 2012 STAYWORK) had many missing values in its data set due to the filter conditions being too restrictive in The advice relating to this variable was followed in Links with the core LFS The target population of the module is in part based on the International Labour Organisation (ILO) employment status, which has three main subgroups: employed, unemployed, and economically inactive. While the ILO status is obtained from a combination of several core variables (WSTATOR, SEEKWORK, AVAILBLE, METHODA to METHODM), the AHM 2012 status is based on a simplified version, derived from only two variables in the core LFS: WSTATOR (labour status during the reference week) and SEEKWORK (seeking employment during the previous four weeks), in the following way: Employed: WSTATOR (1, 2); Unemployed (simplified): WSTATOR (3, 5) and SEEKWORK (1, 2, 4); and Inactive (simplified): WSTATOR (3, 5) and SEEKWORK (3). In the schematic overview of the relation between the variables, given on the last page of the explanatory notes, the basic ILO terminology is used, without a clear indication of the intended simplification (in respect of SEEKWORK). Whilst this is not ideal, the Regulation takes precedence over the addendum to the explanatory notes, so the omission is not of material consequence. The fact that the scheme does not include all WSTATOR categories does not have any impact on the module, as those not included refer to people who are either younger than 15 years, older than 75 years, or engaged in compulsory military service. None of these people are in the years target group. YEARPR (year in which person last worked) and YEARBIR (year of birth) are used in combination to define the target population of the module. General issues relating to data collection This section gives detailed information on sample sizes and non-response rates. It also includes information on differences in interviewing methods and experiences between countries. Sample size The sample size of AHM 2012 is determined by the number of people interviewed during the labour force survey in the specific 2012 quarter(s). 9 Only people aged 50 to 69 and meeting specific requirements relating to labour market status (see the target population chapter for more details) were interviewed. The following table shows, for each country, the number of interviews conducted and the percentage of the target group population that this number of interviews represents. 9 See the reference period of the module (in table 3) for the specific quarter(s) by country. Transition from work to retirement. Evaluation of the 2012 module Page 10

11 Table 1: Sample sizes of AHM 2012, by country AHM sample (persons interviewed) AHM sample as a percentage of the corresponding population in each country (%) EU % BE % BG % CZ % DK % DE % EE % IE % GR % ES % FR % HR % IT % CY % LV % LT % LU % HU % MT % NL % AT % PL % PT % RO % SI % SK % FI % SE % UK % IS % NO % CH % Note: the corresponding AHM 2012 population is the weighted total of yes, no and blank answers from the PENSION variable. Non-response rates If the non-response rate is higher than 15 %, the data for that country for that variable was considered for the purpose of this report to be of very limited use. Only nine of the 31 countries involved in the survey have acceptable response rates (i.e. above 85 %) for all variables. The two variables PLANSTOP and CONTWORK are of minimal use for analytical purposes due to the very high non-response rate at EU level. Transition from work to retirement. Evaluation of the 2012 module Page 11

12 PENSION EARLYRET AGEPENS REASNOT WORKLONG REDUCHRS STAYWORK PLANSTOP CONTWORK Table 2: Non-response rates in AHM 2012, by variable EU-28 2 % 4 % 5 % 4 % 6 % 11 % 12 % 35 % 19 % BE 0 % 0 % 2 % 19 % 21 % 12 % 1 % 2 % 0 % BG 3 % 0 % 0 % 0 % 0 % 0 % 0 % 60 % 52 % CZ 0 % 0 % 0 % 0 % 0 % 0 % 1 % 5 % 6 % DK 0 % 0 % 0 % 0 % 0 % 0 % 4 % 0 % 29 % DE 8 % 27 % 30 % 26 % 33 % 21 % 42 % 45 % 19 % EE 0 % 0 % 0 % 0 % 0 % 0 % 0 % 19 % 13 % IE 5 % 0 % 2 % 0 % 1 % 57 % 2 % 16 % 8 % GR 9 % 0 % 0 % 0 % 20 % 19 % 0 % 40 % 26 % ES 0 % 0 % 2 % 0 % 5 % 2 % 1 % 67 % 17 % FR 0 % 0 % 1 % 0 % 0 % 1 % 33 % 26 % 16 % HR 0 % 0 % 0 % 0 % 0 % 2 % 18 % 89 % 0 % IT 0 % 0 % 0 % 0 % 1 % 0 % 0 % 63 % 24 % CY 0 % 0 % 0 % 0 % 0 % 0 % 0 % 0 % 0 % LV 0 % 0 % 1 % 1 % 1 % 0 % 0 % 11 % 0 % LT 0 % 0 % 1 % 0 % 0 % 11 % 0 % 0 % 46 % LU 0 % 1 % 2 % 0 % 2 % 3 % 8 % 14 % 13 % HU 3 % 0 % 0 % 0 % 0 % 1 % 0 % 48 % 0 % MT 0 % 0 % 0 % 0 % 0 % 0 % 0 % 0 % 0 % NL 1 % 0 % 1 % 0 % 1 % 1 % 4 % 19 % 15 % AT 0 % 0 % 0 % 0 % 0 % 0 % 0 % 16 % 0 % PL 0 % 0 % 0 % 0 % 0 % 0 % 0 % 0 % 0 % PT 2 % 0 % 0 % 0 % 1 % 1 % 0 % 22 % 8 % RO 6 % 0 % 0 % 0 % 0 % 3 % 0 % 83 % 61 % SI 0 % 0 % 0 % 0 % 0 % 0 % 8 % 0 % 0 % SK 0 % 0 % 0 % 0 % 0 % 0 % 0 % 5 % 3 % FI 1 % 1 % 3 % 0 % 1 % 2 % 3 % 8 % 4 % SE 0 % 0 % 0 % 1 % 1 % 3 % 1 % 52 % 17 % UK 0 % 0 % 1 % 0 % 1 % 43 % 1 % 29 % 22 % IS 2 % 1 % 3 % 35 % 37 % 15 % 13 % 82 % 32 % NO 1 % 62 % 19 % 0 % 2 % 11 % 4 % 25 % 36 % CH 2 % 1 % 9 % 2 % 1 % 1 % 17 % 25 % 31 % Note: highlighted cells are those where the non-response rate is above the critical level of 15 % Other measurement issues The following table gives information on various aspects of how the survey was conducted in each country; these are liable to influence the quality and comparability of the results: the reference period(s), i.e. during which quarter(s) of 2012 the data was collected; whether participation was compulsory or voluntary; whether a proxy could answer on behalf of the respondent; whether pilot surveys or other testing was carried out before the survey itself; and the order in which questions were asked when collecting LFS and AHM data, i.e. whether AHM questions were either integrated into the survey by topic, or asked all together after the LFS questions. Transition from work to retirement. Evaluation of the 2012 module Page 12

13 Table 3: General measurement issues in AHM 2012 Reference period Participation Use of proxies Pilot survey or other testing AHM questions after core LFS BE Q2 Compulsory No Yes BG Q1-Q4 Voluntary Yes Yes, field testing Yes CZ Q1-Q4 Voluntary Yes No DK Q2 Voluntary Yes Yes DE Q1-Q4 Voluntary Yes Yes EE Q2, Q4 Voluntary Yes Yes IE Q2 Voluntary Yes Yes GR Q2 Compulsory Yes Yes ES Q1-Q4 Compulsory Yes Yes FR Q1-Q4 Compulsory No Yes, field testing (185 and 283 respondents) Yes HR Q2 Voluntary Yes Yes IT Q2 Compulsory Yes No CY Q2 Compulsory Yes Yes LV Q2 Voluntary Yes Yes LT Q2 Voluntary Yes Yes LU Q1-Q4 Voluntary Yes Yes, field testing Yes HU Q2 Voluntary Yes Yes MT Q1-Q2 Voluntary Yes No NL Q1-Q4 Voluntary Yes No AT Q1-Q4 Compulsory Yes Yes, field testing (195 respondents) Yes PL Q2 Voluntary Yes Yes PT Q2 Voluntary Yes Yes RO Q2 Voluntary Yes Yes, field testing (593 respondents) Yes SI Q2 Voluntary Yes Yes SK Q2 Voluntary Yes Yes FI Q1-Q4 Voluntary Yes Yes, cognitive testing (22 respondents) Yes SE Q1-Q4 Voluntary Yes Yes, cognitive testing Yes UK Q1-Q4 Voluntary Yes No IS Q2 Voluntary Yes No NO Q1-Q4 Compulsory No Yes CH Q1-Q4 Compulsory Yes Yes The national reports on the interviewing phase of the survey paint a positive overall picture of the experience. There are, however, several issues that would need to be addressed were the module to be repeated, such as the method of collecting information on pension(s) (spontaneous answer of the user versus administrative definition with examples) and the use of filters and the routing of the questionnaire in general. More extensive testing of questionnaires, better mapping of national pension systems, and more training for interviewers could have helped to avoid some of the problems which resulted in the low quality of some of the data collected. In other cases, the complexity and changing nature of pension systems, as well as respondents lack of understanding of them, created difficulties when collecting the data for AHM. The interviewing method sometimes limited the options open to interviewers, as did some of the variables themselves, which were too rigid, e.g. where I do not know was not allowed as a spontaneous answer. Other issues encountered included: questions being asked twice (for LFS and again for AHM), difficult administrative concepts, and problems in recalling past events or in predicting future decisions on retirement. These are likely to have had a negative influence on the quality of the data. All of the points mentioned above should be re-assessed before any new data is collected on the same topic. The various issues will be explored in more detail in the next chapter, under the relevant variables. Transition from work to retirement. Evaluation of the 2012 module Page 13

14 Chapter 2: Quality analysis by variable This chapter assesses AHM 2012 in more detail, with analysis of each variable. The eleven variables included in the module are presented in the same order as in the Regulation. This is the order of columns in the database, 10 but it does not imply that variables were collected in this order in all countries. The questionnaires used in each country to collect the AHM 2012 data are available, 11 often in several languages. 1. PENSION: Person receives or does not receive a pension Short description The purpose of the variable is to split the target population into two groups: those currently receiving a pension and those not currently receiving a pension. Respondents were expected to decide, based on their own judgment, whether the type of benefit they receive is a pension. The general condition for a payment to be considered as a pension was that it had to be a regular and periodic benefit received in cash, other than a salary or wage. Lump-sum payments and benefits in kind were excluded from the definition. 12 A list of benefits which are not considered as pensions was also drawn up 13. Symbolic payments, even if labelled as pensions, were excluded. 14 Filter conditions and codes This question was asked to all respondents, i.e. people aged who are either currently working (WSTATOR = 1, 2) or have worked beyond the age of 50 (WSTATOR = 3, 5 and (YEARPR - YEARBIR) > 49)). See the target population section for more information. Code 1 Yes 2 No Analysis of the questionnaires Description 9 Not applicable (not included in the filter) Blank No answ er or does not know This variable is of particular importance: firstly, because it distinguishes between those receiving and not receiving a pension; and secondly, because it acts as a filter for the rest of the module. Ideally, the same question would have been used in all countries, to minimise the risk of introducing national differences in a variable whose output was crucial for the whole AHM. In general, the recommendation of asking respondents to judge for themselves whether they receive a pension was followed. In some questionnaires however (the Belgian and Swiss questionnaires, for example), this question was repeated for several types of pensions mainly the ones from the PENSTYPE variable while in others (Spain, Portugal and Romania) additional types of pensions were already given as examples in the opening question. In Bulgaria and Hungary, the interviewer first checked whether respondents were receiving an old-age pension, and then asked about other types of pensions. It can be assumed that these variations would not have fundamentally affected the comparability of the data but the actual impact cannot be measured. This variable played an important role in the routing of the questionnaire. Those answering no (code 2) 10 AHM 2012 columns ranged from 197 to 218, with one or more digits per variable. 11 See description of the 2012 AHM at 12 Explanatory notes, page Explanatory notes, page Frequently asked questions, page 4. Transition from work to retirement. Evaluation of the 2012 module Page 14

15 MT EL HR IT BE IE PL ES LU NL RO CY EU-28 HU PT SI UK FR LT AT DE BG LV SK EE SE DK FI CZ NO CH IS were routed outside the variables PENSTYPE, EARLYRET, AGEPENS, REASNOT, WORKLONG, STAYWORK, and PLANSTOP. They were routed instead to REDUCHRS (if in employment), BUILDPEN and CONTWORK. In most cases, those who wrongly answered no to the first open-ended question did not receive follow-up questions on the same topic. Those who answered yes (code 1) were routed towards the PENSTYPE question. The combination of an open-ended variable (PENSION) and a yes/no variable (PENSTYPE), the latter based on a pre-determined list of administrative schemes and not allowing do not know answers, was not ideal in all cases. When faced with inconsistencies in a respondent s answers to the two questions, interviewers are likely to have made an on the spot decision and, as a result, the routing of the questionnaire was not always as consistent as might have been hoped. Data was not collected for the variable PENSION in 2006, as a result of which we do not have any time series data with which to compare the 2012 data. The results were, however, compared to the LFS variables MAINSTAT (respondents own view of their main labour status) 15 and LEAVREAS (main reason for leaving last job or closing business). The testing of the model questionnaire had predicted that data obtained for this variable would be of high quality and would be useful as a starting point for the module. Results from the full survey generally confirm this. Analysis of the results The response rate was good in all participating countries. Univariate analysis of a categorical variable offers limited options in terms of methods, but we can see that this variable has the highest response rate out of all variables included in the module, at 98 % (table 2). The impact of the target population on any potential interpretation of the data is also evident. Analysing the existence of pension rights based on a subgroup of the population aged (moreover a subgroup with an unequal gender distribution) is not without risks. This issue was discussed in the target population section, where a breakdown by country and gender is given. Graph 1.1 provides an overview of the distribution of this variable for the population aged 50-69, by country. Graph 1.1: Distribution of the AHM 2012 PENSION variable (% of the population aged 50-69) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Yes No No answer Not applicable Considering only the target group of the AHM (i.e. excluding the respondents answering not applicable from the graph above), 42 % of respondents in the EU receive a pension (see graph 1.2). A total of 19 of the 31 countries in the survey are within ±5 % of this. The percentage of the target group receiving a 15 MAINSTAT is an optional variable in the LFS, measuring the (perceived) main labour status. Transition from work to retirement. Evaluation of the 2012 module Page 15

16 CZ SI SK RO PL HR EE BG FR LT AT MT UK LV LU SE FI EL PT EU-28 IT DK HU BE DE ES CY NL IE CH NO IS pension is equal to the EU average in three countries (Finland, Greece and Portugal), and lower than it in nine EU member states and in Switzerland, Norway and Iceland. The yes-no curve in graph 1.2 is fairly smooth and there are therefore no outliers. The highest value for yes, i.e. the percentage of the target group receiving a pension (53 % in the Czech Republic) is two and a half times larger than the lowest value (in Iceland). From a geographical perspective, the countries with the highest percentage of the target group receiving a pension are eastern European (the Czech Republic, Slovenia, Slovakia, Romania and Poland), whereas the countries with lower percentages of the target group receiving a pension are more mixed, and include Mediterranean as well as north-west European countries (Iceland, Ireland, the Netherlands, Cyprus, Spain and Norway), and also Hungary. Graph 1.2: Distribution of the AHM 2012 PENSION variable (% of the target population aged 50-69) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Yes No No answer Graph 1.3 shows the split of people in employment receiving and not receiving a pension. This subgroup is not affected by the choice of target population. The height of each bar represents the overall employment rate for people aged in each country. In all countries, only a small proportion of the population is in employment and simultaneously receiving a pension. In general, countries with a relatively high proportion of employed people receiving pensions also have a high employment rate, e.g. Sweden, Estonia, the United Kingdom and Norway. There are also however several countries where employment rates are high and the proportion of employed people receiving pensions is relatively low, e.g. Germany, Denmark and the Netherlands. Transition from work to retirement. Evaluation of the 2012 module Page 16

17 SI CZ SK HR EL BG FR AT PL LU RO MT IT HU LT EU-28 BE LV FI DK PT ES EE UK DE SE CY NL IE CH NO IS SE EE DE DK UK NL FI LT LV CY CZ EU-28 AT PT IE FR SK RO LU BG BE PL IT ES HU SI HR EL MT IS NO CH Graph 1.3: Distribution of the AHM 2012 PENSION variable for people aged in employment (% of the entire population) 80% 70% 60% 50% 40% 30% 20% 10% 0% Yes No No answer Graph 1.4 shows that the majority of people receiving a pension are no longer in employment. At EU level, 34 % of the AHM 2012 population receive a pension and are not in employment, while only 7 % receive a pension and are in employment. Graph 1.4: Distribution of the AHM 2012 PENSION variable for people aged 50-69, by employment status (% of the AHM 2012 population) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% In employment and PENSION=no Not in employment and PENSION=yes No answer In employment and PENSION=yes Not in employment and PENSION=no On the basis of the graphs above, we can conclude that there is a strong link between employment status and the PENSION variable, but that there are also many other aspects of the labour market that play a role in determining the distribution of this variable. Some countries (such as Malta, Greece, Italy and Transition from work to retirement. Evaluation of the 2012 module Page 17

18 Belgium see graph 2) have a significant population who have not worked beyond age 50, and this population was not interviewed. In other countries (such as Sweden, Estonia and the United Kingdom see graph 1.3) a high proportion of the population are in employment and receiving a pension, but in general, those receiving a pension are no longer in employment, as seen in graph 1.4. Among those receiving a pension, there are significant variations in their employment status at country level: over a third of those receiving a pension are still in employment in Sweden, Estonia and the United Kingdom, while only 3 % of those receiving a pension are still in employment in Greece. Finally (see graph 1.2), more than half of respondents in the Czech Republic and Slovenia do receive a pension, while in Ireland and the Netherlands fewer than 30 % do. The next section discusses the aggregated PENSION data and analyses links between the PENSION variable and the age of respondents at EU level. Graph 1.5: Distribution of the AHM 2012 PENSION variable, by age groups and employment status, EU-28 (million persons) No answer Not in employment and PENSION=no Not in employment and PENSION=yes In employment and PENSION=yes In employment and PENSION=no Unsurprisingly, the distribution of the PENSION variable changes significantly with the age of the respondents. The majority of respondents below age 60 are still in employment and do not receive a pension. The majority of respondents aged 60 or above are not in employment and do receive a pension. Population groups who are in employment and receiving a pension, or not in employment and without a pension are less prominent. While age 60 could be assumed to be a natural turning point for the EU population in its transition from work to retirement, analysis based on data relating to pensions needs to take into account the significant demographic differences at country level. The following graphs examine in more detail the population aged 55-64, in order to describe the actual transition from the labour market into retirement by analysing the population at the ages at which, in many countries, this transition takes place. In the age group, only a small proportion of people receive pensions (6 % at EU level, with some country variations), and in the age group, a very high proportion do so (95 % at EU-level), thus illustrating that most transitions take place between the ages of 55 and 64. Ireland and the Netherlands are the only EU countries where the percentage of respondents aged 65 to 69 receiving a pension was under 90 %. In Greece, Romania and Spain, 90 % of respondents in this age group were receiving a pension. In some countries 100 % of the AHM population aged were receiving a pension (Estonia, Malta and Latvia), indicating that the right to a pension becomes universal at a certain age (at least) for those having worked beyond age 50. Transition from work to retirement. Evaluation of the 2012 module Page 18

19 SI SK MT CZ AT PL LT FR LV RO EE LU IT HR BG UK HU EU-28 BE FI EL PT DK DE CY ES SE NL IE NO CH IS The graph below shows the breakdown of the PENSION variable by 5-year age groups, and In both age groups, there are significant differences between countries in the proportion of respondents receiving a pension. In some countries, such as Denmark, Hungary, and Malta, there is a significant increase in the proportion of respondents receiving a pension just after age 60. It is not surprising that countries with a relatively low proportion of respondents receiving pensions in these age groups (Ireland, the Netherlands, Spain and Cyprus) will tend to be in the lowest part of the overall PENSION distribution (see graph 1.2). The overall proportion of Swedish respondents receiving a pension is however higher than the EU average, even though a relatively low proportion of people in the and age groups receive pensions. This is due the high proportion of respondents in the age group receiving a pension (98 %). Graph 1.6: Proportion of yes responses to the AHM 2012 PENSION variable, by respondents age group (% in the corresponding age group) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% We conclude the section dedicated to the age group 55 to 64 with a graph that shows the gender differences among the year-old AHM population. In most countries, women complete their transition towards retirement before men. Countries where women retire earlier than men are also likely to have a high overall proportion of respondents receiving a pension. Transition from work to retirement. Evaluation of the 2012 module Page 19

20 SI CZ PL SK RO AT HR LT EE BG UK LV FR MT LU EU-28 EL IT FI HU PT DE DK BE ES CY SE NL IE NO CH IS Graph 1.7: Proportion of yes responses to the AHM 2012 PENSION variable for the AHM population aged 55-64, by gender (%) 80% 70% 60% 50% 40% 30% 20% 10% 0% Men Women As this variable was not included in the 2006 module. the best way of assessing the quality of the data is therefore to compare it to the LFS MAINSTAT variable, as recommended in the explanatory notes. The two variables are related, as PENSION classifies respondents according to whether they receive a pension, and MAINSTAT classifies them by their main activity status, as judged by the respondents themselves, with one option (code 4) being in retirement, in early retirement or having given up business. By definition, the relevant values of the variables are not identical and so will not obtain exactly overlapping population groups, but the two variables would be expected to show the same picture to a large extent. Graph 1.8: Proportion of yes responses to the AHM 2012 PENSION variable for selected categories of the LFS MAINSTAT variable, EU-28 (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Retired Permanently disabled Other inactive person Unemployed Employed Note: EU-28 average excluding Germany and the United Kingdom. Transition from work to retirement. Evaluation of the 2012 module Page 20

21 AHM LFS 2012, PENSION = 1 Graph 1.8 confirms these expectations, with most of those who define themselves as retired also receiving a pension, while people in employment or those unemployed are less likely to receive a pension. It also shows that, at EU level, there are more respondents in PENSION = 1 than in MAINSTAT = 4 (i.e. more people receiving a pension than who would describe themselves as in retirement, in early retirement or having ceased business activities). The next graph shows the correlation between those coded 4 for the core LFS variable MAINSTAT (in retirement, in early retirement or having ceased business activities) and those coded 1 in the LFS AHM PENSION variable (is receiving a pension), for the age group 50 69, by country. The clusters of countries are familiar from graph 1.1, with Iceland and Ireland at the lower end (i.e. with a smaller proportion of the population receiving a pension), and the Czech Republic and Slovenia at the higher end (i.e. with a larger proportion of the population receiving a pension). Data for Germany and the United Kingdom is not shown in graph 1.8 because they did not provide data for MAINSTAT in The data points are all reasonably close to the trend line. At country level, the proportion of respondents answering yes for the PENSION variable is also higher than the proportion answering code 4 for the MAINSTAT variable, as was noted from graph 1.8. This can now be seen to be a systematic pattern that holds for all countries, with the exception of Hungary and Croatia. In the case of Hungary, this is due to the recent pension reform 16 that has transformed several of the previous types of pension into social benefits. This illustrates the sensitivity of this variable to national administrative definitions, and serves as a warning of the errors that can occur if data is generalised and interpreted outside the AHM 2012 context. Graph 1.9: Correspondence between the LFS MAINSTAT variable and the AHM 2012 PENSION variable (%) 60% 50% 40% 30% IE CZ SK EE PL RO BG FR MT LT HR LV AT SE LU EU-28 PT FI EL IT DK BE HU CH ES CY NL SI 20% IS 10% 0% 0% 10% 20% 30% 40% 50% 60% Core LFS, 2012, annual, MAINSTAT = 4 It would also be possible to compare the PENSION variable to the core LFS LEAVREAS variable. The LEAVREAS variable (main reason for leaving last job or ceasing business activities) is collected annually from all respondents not in employment but who have previous employment history and have stopped work within the last eight years. The overlap between this variable and the PENSION variable is represented by the group of respondents aged not in employment but having previous employment history beyond age 50 and within the last eight years. 16 More information is available in the document mapping the national pension systems and the AHM 2012 PENSTYPE variable. Transition from work to retirement. Evaluation of the 2012 module Page 21

22 Graph 1.10: Proportion of yes responses to the AHM 2012 PENSION variable for selected categories of the LFS LEAVREAS variable, EU-28 (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Normal retirement Early retirement Illness or disability All other reasons As shown by graph 1.10, the two variables are correlated, as expected. A large majority of people who left their last job due to retirement (whether normal or early) declare receiving a pension. This also holds for people leaving the labour market for health reasons. The grouping all other reasons on graph 1.10 includes the following codes from the LEAVREAS variable: dismissal (code 00), end of a contract of limited duration (code 01), and other personal reasons. For all values of the LEAVREAS variable, the distribution of the PENSION variable is in line with expectations. Conclusions and recommendations The form of this variable suggests that data can be collected using a very simple yes/no question, without complicated routing or coding. Deciding whether or not one receives a pension (as defined by AHM 2012) might not however always be entirely straightforward. Difficulties in answering this were reported to have occurred most often in situations where respondents received some benefits (in particular unemployment benefits or disability benefits) that might be called pensions in everyday language, but which are not considered as pensions for the purpose of this module. The opposite situation was also reported to have occurred, where people receiving pensions (usually not old-age pensions) spontaneously answered no to the question. Countries having recently undergone changes to their pension systems (in particular Bulgaria, Hungary and Poland) had to ensure consistency between the old and new systems in terms of what is classed as a pension. In order to prevent errors resulting from these situations, national mappings were prepared and linked the existing national terminology with the corresponding PENSTYPE types of pension. The PENSION variable is by definition the sum of all the possible values for PENSTYPE (i.e. receiving any of the possible types of pension included in the PENSTYPE variable equals a yes answer to the PENSION variable). It should however be noted that the conceptual complexity of what is and what is not a pension was not fully accommodated by the yes/no question which was supposed to be asked in the questionnaire. In several countries (Bulgaria and Hungary, for example), data for this variable was not in fact collected using a yes/no question, but was deduced from answers to other questions on the type of pensions received. The complexity of the opening question also varied, ranging from the simple question Do you receive a pension? to questions listing several types of pensions received in that country. Depending on the order in which questions were asked, the interview method, the training received by interviewers for working on this module and their level of experience, answers to the PENSION variable may have Transition from work to retirement. Evaluation of the 2012 module Page 22

23 sometimes been re-coded in light of the information collected for the PENSTYPE variable. In other cases however, it may not have been possible to ensure fully consistent coding. Very few of the participating countries reported any problems in collecting data for this variable and the response rate was good. The Netherlands report explained that their data would underestimate the number of pensioners because it was difficult in some cases to make the respondents understand the module s broad definition of pensions. The report from Greece highlighted a situation which can arise in their system where an individual is formally entitled to a pension, but the actual payment is delayed for more than a year, thus making it difficult to know whether to answer yes or no to the question of receiving a pension. Moreover, no country reported that their definition of the variable in the module differed from the definition included in the Commission Regulation (249/2011). The response rate was good, there were no outliers, and there was also strong correlation with MAINSTAT and LEAVREAS. We can therefore conclude that the quality of data for the variable is good. The variable is however based on national definitions and classifications of pensions and other social benefits. Conceptually its comparability is unfortunately limited at any level higher than country level. Should the module be repeated, consideration should be given to the question of whether one single variable is sufficient for the purpose of dividing the target population into two groups: those receiving and those not receiving a pension. Were a similar variable to be used again as a filter for the rest of the module, the risk of inconsistencies or errors and their impact on the overall data set should also be assessed. Moreover, the concept of a pension was defined in 2012 at national level, as a specific provision entitling an individual to selected social benefits. Possible ways of improving its comparability at European level could be analysed in more detail. 2. PENSTYPE: Type of pension(s) currently received Short description The purpose of the variable is to determine which type of pension(s) the respondents who answered yes to the previous question (PENSION) are receiving. However, not all participating countries asked the questions in this order in their questionnaires. As shown in the code list provided, PENSTYPE has eight possible answers: four types of old-age pension scheme, unemployment pension, disability pension, survivor s pension and other pension or type of pension unknown. More than one answer could be given to this question, as one person can receive several types of pension at the same time. The answer do not know was not included as an option for this question. Lists of the different pensions and benefits available in each country were prepared at national level and each assigned to the relevant code. 17 Filter conditions and codes This question was asked to all respondents who answered yes (code 1) to the question relating to the PENSION variable. Code Description PENSTYP1 1: Yes; 0: No Old-age pension. Statutory scheme PENSTYP2 1: Yes; 0: No Old-age pension. Occupational scheme PENSTYP3 1: Yes; 0: No Old-age pension. Personal scheme PENSTYP4 1: Yes; 0: No Old-age pension. Scheme unknow n PENSTYP5 1: Yes; 0: No Unemployment pension PENSTYP6 1: Yes; 0: No Disability pension PENSTYP7 1: Yes; 0: No Survivor's pension PENSTYP8 1: Yes; 0: No Other pension(s) or type of pension unknow n Not applicable (not included in the filter) 17 A detailed document presenting the mapping of the types of pension available in each country onto the pension types from this variable is available online at: Transition from work to retirement. Evaluation of the 2012 module Page 23

24 Analysis of the questionnaires Countries were advised to ask the PENSTYPE question only after the PENSION question. Most countries followed this advice and collected information on whether pension benefits were received at all first, and then asked about the type of pensions received. Some countries, including Belgium, Lithuania and Switzerland, used a slightly different questionnaire structure. The risk of inconsistencies between the two variables and the implications of this for the overall questionnaire were discussed in the PENSION section. For the PENSTYPE variable, mappings of pensions available in each country were prepared at national level. These were helpful but required substantial documentation. Where national systems are extremely complex (e.g. in Denmark and Italy) or undergoing changes (e.g. in Hungary and Poland), countries reported difficulties in ensuring that all eight components of the variable were well defined and distinct. Bulgaria and Sweden reported minor deviations in the implementation of this variable. In the case of Bulgaria, this is due to a change made to the social security code in 2012, which separated social disability pensions from disability pensions (PENSTYP6), with an estimated impact on the results of PENSTYP6 of 12 %. Sweden reported that it did not use PENSTYP4, and that it included the data which would have taken this value in PENSTYP8 instead, i.e. it combined the two codes as PENSTYP8. In some countries (Finland and the Netherlands), the terminology of this AHM was not always familiar and respondents had difficulty answering the question. For example, in Finnish, the concept of an old-age pension is not always understood correctly. In some situations (reported in Finland and Italy and likely to have occurred in other countries as well), the respondent received a benefit whose label had changed at some point in time, e.g. when a disability pension was automatically converted into an old-age pension at a certain age. In these cases, respondents were not always able to indicate accurately the correct type of pension they received. Errors in the PENSTYPE variable had an impact on the overall routing of the questionnaire, as well as on the accuracy of the AGEPENS variable. It can however be assumed that the impact of these cases on the overall data set was not significant. A somewhat similar variable was collected in 2006, as discussed in the chapter on links with the AHM In Hungary, there was a major overhaul of the pension system in 2012, making any attempt at comparison to the 2006 module meaningless. Analysis of the results This variable is defined as eight separate variables collated together. Each of the eight sub-variables allowed yes and no answers, and, as a result, multiple types of pensions could be selected by each respondent. Respondents receiving two (or more) types of pension are therefore counted as many times as they receive pension types. Unfortunately, the data for this variable is not easy to interpret at EU level, as very similar types of pension could be classified in different countries as being under different schemes. Moreover, for some types of pension (PENSTYP 5 or 6), similar social rights are classified as pension schemes in some countries and social benefits (not included in the module) in others. The table below shows the percentage of yes answers out of the total number of answers for each subcategory of the PENSTYPE variable, at EU level and by country. Transition from work to retirement. Evaluation of the 2012 module Page 24

25 PENSTYPE1 PENSTYPE2 PENSTYPE3 PENSTYPE4 PENSTYPE5 PENSTYPE6 PENSTYPE7 PENSTYPE8 Table 2.1: Yes answers as a percentage of total answers for each of the AHM 2012 PENSTYP1 to 8 variables, EU-28 (% of those with PENSION = 1) EU % 16 % 8 % 2 % 1 % 9 % 8 % 5 % BE 73 % 7 % 7 % 7 % 16 % 9 % 5 % 0 % BG 90 % 1 % 0 % 0 % 0 % 11 % 3 % 0 % CZ 90 % 0 % 4 % 0 % 0 % 9 % 12 % 0 % DK 83 % 37 % 14 % 1 % 0 % 13 % 2 % 0 % DE 86 % 19 % 17 % 12 % 0 % 11 % 16 % 15 % EE 81 % 0 % 6 % 0 % 0 % 19 % 0 % 0 % IE 46 % 49 % 5 % 2 % 0 % 6 % 7 % 2 % EL 95 % 1 % 1 % 0 % 1 % 3 % 1 % 0 % ES 67 % 3 % 2 % 0 % 10 % 15 % 11 % 4 % FR 88 % 3 % 5 % 0 % 1 % 12 % 10 % 0 % HR 67 % 0 % 0 % 0 % 0 % 27 % 3 % 7 % IT 92 % 3 % 1 % 0 % 0 % 7 % 0 % 4 % CY 89 % 3 % 2 % 0 % 0 % 6 % 7 % 0 % LV 88 % 0 % 0 % 0 % 0 % 12 % 0 % 0 % LT 85 % 0 % 0 % 0 % 0 % 12 % 12 % 5 % LU 80 % 12 % 7 % 6 % 0 % 13 % 8 % 4 % HU 97 % 0 % 0 % 0 % 0 % 0 % 14 % 0 % MT 83 % 18 % 0 % 1 % 1 % 6 % 3 % 2 % NL 57 % 54 % 10 % 2 % 0 % 17 % 12 % 2 % AT 87 % 7 % 2 % 0 % 2 % 14 % 10 % 1 % PL 81 % 0 % 0 % 0 % 0 % 0 % 3 % 10 % PT 74 % 2 % 1 % 2 % 5 % 14 % 14 % 1 % RO 85 % 0 % 0 % 0 % 0 % 14 % 1 % 0 % SI 88 % 0 % 0 % 0 % 0 % 8 % 5 % 5 % SK 84 % 0 % 1 % 1 % 5 % 8 % 20 % 0 % FI 78 % 5 % 9 % 0 % 1 % 15 % 7 % 1 % SE 84 % 74 % 47 % 0 % 0 % 6 % 6 % 6 % UK 65 % 60 % 20 % 0 % 0 % 0 % 4 % 3 % IS 18 % 49 % 5 % 0 % 0 % 37 % 9 % 0 % NO 50 % 31 % 4 % 0 % 0 % 26 % 5 % 11 % CH 73 % 58 % 10 % 6 % 0 % 9 % 7 % 4 % The main conclusion to be drawn from this data is that statutory old-age pensions (PENSTYP1) are by far the most commonly received type of pension. In the EU as a whole, 81 % of those who declared receiving some form of pension receive a statutory old-age pension. The only countries where this pension type is not the most common are Ireland and Iceland, where more people receive occupational old-age pensions. In the majority of countries, more people are receiving statutory old age pensions than are receiving all the other types of pension combined. The only exceptions to this are Germany, Ireland, the Netherlands, Sweden, the United Kingdom and the European Free Trade Association countries, whose data does not fit the same pattern. Seven countries (Bulgaria, Greece, Italy, Hungary, Poland, Romania, and Croatia) show a very low use of all types of pension other than statutory old-age pension schemes. Transition from work to retirement. Evaluation of the 2012 module Page 25

26 HU EL IT CZ BG CY LV FR SI AT LT RO DE SK SE MT DK PL EU-28 EE LU FI PT BE HR ES UK NL IE CH NO IS The following graph shows the frequency of different combinations of pensions received by the same respondent. Unlike for the PENSION variable, where each respondent receiving a pension is counted only once, this variable counts the number of pensions, thus allowing the number of each of the eight types of pensions being paid to be quantified. As can be seen on the graph below, in many countries the stacked bar combining different types of pensions is only slightly above 100 %, indicating that in these countries it is rare for one person to receive several pensions. Graph 2.1: Distribution of the PENSTYP1 to PENSTYP8 AHM 2012 variable (% of yes answers in total answers given for each PENSTYPE) 250% 200% 150% 100% 50% 0% PENSTYP1 PENSTYP2 PENSTYP3 PENSTYP4 PENSTYP5 PENSTYP6 PENSTYP7 PENSTYP8 Note: the maximum height of the stacked bar would be 800 % if each respondent received all eight types of pension at the same time. The data for PENSTYP5, the unemployment pension, reflects the national mappings, as the majority of countries reported that this specific type of scheme did not exist. Ireland, Italy and Lithuania confirmed that unemployment pensions do exist in their countries, but they are not reflected in the data collection (and are likely to feature only on a small scale). As PENSTYP1 is the most frequently received type of pension, the other categories of the PENSTYPE variable will be assessed in relation to PENSTYP1, i.e. by comparing the proportion of respondents receiving: 1) only PENSTYP1; 2) PENSTYP1 in combination with other types of pensions; and 3) one (or more) of PENSTYP2 to 8 but not PENSTYP1. Transition from work to retirement. Evaluation of the 2012 module Page 26

27 EL LV HU BG IT RO CY SI PL CZ EE FR LT MT AT SK HR FI PT EU-28 LU ES BE DK IE DE UK NL SE CH NO IS Graph 2.2: PENSTYP1 considered in relation to PENSTYP2 to PENSTYP8 (% of respondents with PENSION = 1) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% PENSTYP1=yes, PENSTYP2 to 8=no PENSTYP1=no PENSTYP1=yes, at least another PENSTYP=yes The previous graph shows that, in the EU as a whole, the majority of people who receive a pension are receiving only a statutory old-age pension. There are however significant differences at country level, as the percentage of people receiving only a statutory old-age pension (out of those in receipt of some sort of pension) ranges from over 90 % in Greece to less than 5 % in Iceland. With the exception of certain countries (Sweden, Germany, the Netherlands and the United Kingdom), the phenomenon of combining pensions from different types of schemes is not very common. The next section analyses the population group not receiving any of the old-age pensions (PENSTYP1 to 4 = no). This group represents 12 % of those receiving a pension. Considering data for this group only, the total of PENSTYP5 to 8 as a percentage of this population, (i.e. the total number of these four types of pension being paid, per person not receiving an old age pension), is fairly constant across the EU. Furthermore, the figure is near to 100 % for most countries, which shows that, in these countries, when a person does not receive an old-age pension, it is unlikely that that s/he will receive more than one of types PENSTYP5 to 8. Germany is the only exception to this general trend. In the graph below, data is ordered by PENSTYP6. A full analysis on each of PENSTYP5 to 8 could also be carried out, but would need to include people receiving old-age pensions as well. Analysis of combinations of pensions does however have limited potential, because the patterns seen in each country s data generally say more about the organisation of the pension system in that country than about the respondents themselves. Transition from work to retirement. Evaluation of the 2012 module Page 27

28 EE LV CZ RO DK BG HR FI LT AT SI FR NL IT CY PL EL SE LU SK MT EU-28 PT IE ES BE DE HU UK IS NO CH Graph 2.3: Distribution of the PENSTYP5 to PENSTYP8 AHM 2012 variable if PENSTYP1 to PENSTYP4 = no (% of yes answers in total answers given for each PENSTYP) 140% 120% 100% 80% 60% 40% 20% 0% PENSTYP5 PENSTYP6 PENSTYP7 PENSTYP8 Note: in the data for Sweden, PENSTYP8 also includes PENSTYP4. The maximum height of the stacked bars would be 400 % if each respondent received all four types of pension at the same time. The possibilities for comparison to the core data or to AHM 2006 are quite limited. Although the 2012 PENSTYP1 to 4 variables (different types of old-age pension) do overlap to some extent with the AGEPENS variable from 2006, the 2006 variable covers a wider range of pensions (not just old-age pensions), and its main purpose was to determine the age of the recipients, not the type of pension. PENSTYP6 (disability pension) has some overlap with OTHBENF = 1 from 2006, but again, the variables are far from being exactly equivalent, as the 2006 variable also covers sickness benefits and disability pensions. The 2006 module did not include any variables parallel to unemployment pensions and survivor s pensions (PENSTYP5 and PENSTYP7 respectively). Conclusions and recommendations The PENSTYPE variable is constructed as a combination of eight different variables. The information captured by this variable is by its very nature difficult to compare across countries, due to the differences in legal frameworks and pension systems. Furthermore, some of the types of pension do not exist in all of the countries in the survey, which makes total results from across countries difficult to interpret in some cases. In some countries, it was difficult to code this variable due to the complexity of the pension system. Nonetheless, national statistical institutes generally reported positive experiences. The absence of an LFS variable to compare to means that it is not possible to carry out a data-based assessment of the quality of the data for this variable. As noted in the Austrian quality report, comparisons with administrative data sources would also be difficult, given the particular conventions followed in collecting this data. Were the module to be repeated, it would be important to consider whether data on all the types of pensions included in 2012 can reasonably be collected by an AHM and, for each type of pension, whether the data can provide comparable and meaningful information at EU level. This variable created a relatively heavy burden in terms of data collection, as eight yes/no variables had to be merged into one, and for each of these variables a list of possible corresponding pension schemes had to be prepared. In addition, in the absence of the code no answer/do not know, respondents and interviewers sometimes had to take difficult yes/no decisions. Unfortunately, this variable offers limited potential for analysis at Transition from work to retirement. Evaluation of the 2012 module Page 28

29 CZ SI SK RO PL HR EE BG FR AT LT MT UK LV SE LU FI EL PT EU-28 IT DK HU BE DE ES CY NL IE CH NO IS EU level. Some countries (Austria, Slovenia and Norway) suggested that less time should be spent classifying pension types, given that the resulting classification will differ greatly between countries and is inconsistent with administrative data. Should it be decided to simplify variables for a possible repeat of the module, one question to consider would be whether analysing only old-age pensions would be sufficient, as this would still provide scope for extensive analysis but without placing an unnecessary burden on respondents. The current breakdown by scheme (statutory, occupational, personal, unknown) shows how much the organisation of pension systems varies from country to country, even when only considering old-age pensions. In addition, nonold-age pensions are often very similar to social benefits, and their analysis at any level higher than country level is hindered by the divergent administrative classifications and terminology. The following graph is one suggestion for how this variable could be simplified. It shows separately: (i) those who receive only an old-age pension (irrespective of the type of old-age pension); (ii) those who receive an old-age pension and a non-old-age pension (i.e. those receiving at least one of the types of old-age pension and at least one of unemployment pension, disability pension, survivor s pension, or other pension of unknown type); and (iii) those who exclusively have a non-old-age pension. As these three groups are mutually exclusive and jointly exhaustive, the sum of the three (the total height of the bar) is the overall level of pension coverage in each country. Graph 2.4 shows that old-age pensions are the dominant subset. They represent 87 % of all pensions in the EU. Graph 2.4: Distribution of those receiving only an old-age pension, an old-age pension and a non-old-age pension, and only a non-old-age pension (% of respondents with PENSION = 1) 60% 50% 40% 30% 20% 10% 0% Some type of pension, but no old-age pension Old-age pension in combination with another pension type Only old-age pension 3. EARLYRET: Incidence of early retirement Short description The purpose of this variable is to determine whether those currently in receipt of an old-age pension took early retirement or stopped working at normal retirement age. Early retirement includes the following: anticipated old-age pensions, disability pensions, early retirement pensions in the case of reduced ability to work, early retirement pensions for labour market reasons, early retirement pensions due to seniority (long career or long contribution period) and early retirement pensions for family reasons. In specific situations, full unemployment benefits and partial retirement pensions can be considered as early retirement. Early retirement is defined in relation to the standard Transition from work to retirement. Evaluation of the 2012 module Page 29

30 retirement age for a given gender, occupational group, etc. People with low retirement ages should not be systematically coded as receiving an early retirement pension, unless they were subject to specific early retirement measures which go beyond what would otherwise be allowed in their profession. 18 In cases where a respondent is receiving multiple pensions, it is the one which started being paid first that determines early retirement status. 19 This variable does not have the same definition as the indicator from the European system of integrated social protection statistics 20 anticipated old-age pension beneficiaries. There are also differences between this variable and the definition of early retirement in the LFS LEAVREAS set of variables, as the focus of the LEAVREAS variable is on economic factors, such as difficulties in specific sectors of the economy, while EARLYRET includes personal and labour market reasons. Filter conditions and codes This question was asked to all respondents receiving an old-age pension (PENSTYP1 = 1 or PENSTYP2 = 1 or PENSTYP3 = 1 or PENSTYP4 = 1). Code 1 Yes 2 No Analysis of the questionnaires This is a yes/no variable based on a concrete measure (early retirement). Respondents would be assumed to remember the relevant information and be able to answer the question easily. This question was not asked in In Italy in particular, and also in a number of other countries, the complexity of the pension system created uncertainty as to whether, in a given period, for individuals of a given age and given profession, particular cases of retirement would count as early retirement or not. In the same countries, AGEPENS was collected after EARLYRET, and interviewers could not assist respondents or check the consistency of responses (this is why a suggestion is made in the report from Italy to consider collecting AGEPENS before EARLYRET). In some countries, respondents receiving benefits from both statutory and occupational schemes struggled to make the distinction between the two types, especially where there was a misconception that employers make contributions to both pension systems. In Bulgaria, it was not easy to communicate to respondents what measures going beyond the normal rule for their profession are, in particular for professions with a lower retirement age (e.g. police officers). Finland reported difficulties in interpreting early retirement in relation to part-time pensions. Analysis of the results Description 9 Not applicable (not included in the filter) Blank No answ er or does not know Due to high non-response rates, the data from Germany and Norway must be used with caution. In the case of Norway, this is because the question was not asked to people older than 66 years of age. The nonresponse rate at EU level was 4 %. Graph 3.1 shows how common it is to take early retirement in different countries, based on the proportion of respondents answering yes and no. 18 Explanatory notes, pages Frequently asked questions, page The European system of integrated social protection statistics is built on the concept of social protection, or the coverage of precisely defined risks and needs including health, disability, old age, family and unemployment. It records the receipts and expenditure of the organisations or schemes involved in social protection interventions. See for more information. Transition from work to retirement. Evaluation of the 2012 module Page 30

31 IT IE ES AT PT HR NL HU SE BE PL FR LU SI EU-28 LV MT DE SK FI UK RO CY LT DK EL EE CZ BG IS CH NO Graph 3.1: Distribution of the AHM 2012 EARLYRET variable (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Yes No No answer The graph shows a very varied picture, with early retirement being close to non-existent in Bulgaria and the Czech Republic, whilst applying to almost three in four pensioners in Italy and Ireland. Both Bulgaria and Italy reported some difficulties in collecting information on this variable, as detailed in the previous section. At EU level, around four out of ten people receiving an old-age pension had taken early retirement. The proportion of people taking early retirement varies gradually across the graph, and there are no striking outliers. No clear geographical pattern is evident. Graph 3.2 analyses early retirement by gender. It shows that at EU level there are more men than women taking early retirement (at EU level, around 60 % of those taking early retirement are male). The high percentage of males amongst those taking early retirement in Malta is a result of the definition of the target population for the module, which included gender. In general, the fact that more men than women take early retirement is a direct consequence of the retirement age being higher for men than women in many countries. This means that a man retiring before a certain retirement age would be counted as having taken early retirement, whilst a woman retiring at the same age might not be. The result can also be explained by specific measures taken for certain occupations, in which the proportion of men and women might not be equal. In 13 EU countries, early retirement is however more common among women, with women representing over 60 % of those taking early retirement in Hungary, Latvia and Estonia. Transition from work to retirement. Evaluation of the 2012 module Page 31

32 AHM LFS 2012, EARLYRET, proportion of yes in total, per cent MT IT LU NL ES UK BE CY IE PT EU-28 AT CZ FR SK DE SE BG RO FI SI PL EL HR LT DK EE LV HU NO IS CH Graph 3.2: Proportion of men among EARLYRET = 1 (%) 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% It would be reasonable to expect a clear correlation between the rate of early retirement and the age at which an old-age pension is first received. Graph 3.3 shows however that this is not the case there is only very weak correlation between these two variables. Graph 3.3: The AGEPENS and EARLYRET variables plotted against each other (% and age in years) 80% 70% IT IE 60% 50% 40% 30% 20% PL SI SK RO HR HU UK AT LU FR EU-28 MT PT LV LT BE DE FI ES CY NL CH SE IS NO EL DK 10% EE BG CZ 0% AHM LFS 2012, AGEPENS, average, in years The LFS LEAVREAS variable provides information on the main reasons for leaving the last job or ceasing business activities. Even if the definitions and the target of the two variables are different, the comparison allows an assessment of the comparability of the data at EU level. Transition from work to retirement. Evaluation of the 2012 module Page 32

33 LT SK ES HR AT PT IE LV SI SE CY LU EE IT RO HU PL BE NL FR EL MT EU-28 FI UK CZ DK CH Graph 3.4: AHM 2012 EARLYRET = 1 among respondents who took early retirement from their last job or from business activities (LEAVREAS = 06), (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Note: data from Germany and Norway is excluded from the analysis because of high non-response rates for the EARLYRET variable. Data from Bulgaria and Iceland is not available for LEAVREAS = 6. This graph confirms that the data should be interpreted with caution, as the correspondence between the two variables (EARLYRET and LEAVREAS) differs significantly between countries, with Denmark, the Czech Republic, UK and Finland showing the smallest overlap in respondents answers to the seemingly related questions. In the case of Denmark, the reduced overlap is due to a problem in filtering for this question, as people with PENSTYP1 = yes were not asked the question on early retirement. In Finland, the low level of correspondence is likely to be a result of the different time horizon and precision of the two variables: EARLYRET was asked after having collected precise information on several types of pensions, while LEAVREAS was a more general variable, and the question was therefore probably easier for respondents to answer. Conclusions and recommendations The main challenge when using this variable is to provide meaningful information at aggregate level, given that the variable itself is defined in relative terms. Retirement provisions which are very similar in terms of what they offer and their conditions could be considered regular policy measures in one country and early retirement in another. As a result, the label early retirement can only be meaningful within a group defined by occupation, age and gender, from a specific country. Then however, the sample from this type of subgroup would be too small to allow detailed analysis. The results for this variable show that patterns of early retirement vary greatly between countries, adding a further dimension to the already divergent national situations and practices. This variable cannot easily be benchmarked against other data sources (administrative or otherwise). As a result, it cannot be interpreted in isolation, but only in the broader context of national data on pensions, pension age, occupation, gender, etc. Were the module to be repeated, further analysis should be conducted in order to establish whether this variable could provide comparable and meaningful information at EU level. Transition from work to retirement. Evaluation of the 2012 module Page 33

34 SI RO PL SK BG HR EL IT UK HU AT CZ FR LU EU-28 MT EE LT LV PT BE IE DE FI CY ES DK NL SE CH IS NO 4. AGEPENS: Age at which a person first received an old-age pension Short description The purpose of this variable is to determine the age at which the respondent started to receive his/her first old-age pension. Filter conditions and codes This question was asked to all respondents receiving an old-age pension (PENSTYP1 = 1 or PENSTYP2 = 1 or PENSTYP3 = 1 or PENSTYP4 = 1). Code Analysis of the questionnaires This is the only numerical variable in the module, and as such allows more extensive analysis than is possible for the other variables. There were some slight variations in the way in which data was collected: in some countries, respondents were asked for the age at which they started to receive their first old-age pension; in other countries, interviewers asked respondents for the year and month in which they started to receive their first old-age pension and then calculated their age at that time from the year and month of birth. These differences are not expected to have had a significant impact on the quality of the variable. The variable is identical in content, but not in coding, to the equivalent variable included in the 2006 AHM module. Analysis of the results 2 digits Description 99 Not applicable (not included in the filter) Blank No answ er or does not know Due to the high non-response rate, the data from Germany and Norway must be used with caution. With the exception of Germany, the response rate at EU level is very good, at 95 %. No major problems were reported in collecting the data for this variable but a small number of respondents did have difficulties recalling the date when they first received an old-age pension, especially in cases where an old-age pension was first received after an unemployment pension. Graph 4.1: Distribution of the AHM 2012 AGEPENS variable (average age in years) Transition from work to retirement. Evaluation of the 2012 module Page 34

35 HR SK SI CZ PL HU RO EE LT LV AT BG EU-28 EL DK MT BE DE IE NL SE FI ES LU IT UK FR PT CY IS CH NO The EU average for the age at which an old-age pension is first received is 59 years. The countries where old-age pensions are received earliest are Slovenia, Romania, Poland, Slovakia, and Bulgaria. Among EU Member States, the average age at which an old-age pension is first received is highest in Sweden, at almost 64 years. Among all participating countries, Iceland and Norway have the highest average ages, with the first old-age pension typically being received after 64 years. The variation across countries is quite high for this variable, with eight years separating the highest (almost 65 in Norway) and lowest (less than 57 in Slovenia) average ages. The geographical pattern in the data appears much like that seen in graph 1.1 (PENSION). The correlation between these two variables is not surprising when the average age for starting to receive a pension is lower, the probability of receiving a pension and being in the AHM target group will be higher. Graph 4.2: Distribution of the AHM 2012 AGEPENS variable (average age in years), by gender Men Women Graph 4.2 displays the average ages at which respondents start receiving a pension, by gender. Countries are ordered by the difference in years between the age of men and the age of women. In most EU countries and two European Free Trade Association countries the average age at which respondents first received an old age pension is higher for men than for women. A very clear east-west divide emerges, with the difference between the average age for receiving a first old-age pension for men and women being most pronounced (more than three years of difference) in Croatia, Slovakia, Slovenia and the Czech Republic, and with smaller but still clearly visible differences in Poland, Hungary, Romania, Estonia, Lithuania and Latvia. The countries where there is no difference at all between men s and women s pension ages, or where women on average receive their first old-age pension at a higher age than men, are western European or Mediterranean (Cyprus, Portugal, France, the United Kingdom, and Italy). One possible way of assessing the quality of data for the AGEPENS variable is to compare it to the expected duration of working life. The duration of working life indicator 21 measures the number of years a person aged 15 is expected to be active in the labour market throughout his or her life. 21 This indicator is derived from demographic data and labour market data. It is published online on the Eurostat database page as table reference lfsi_dwl_a. Transition from work to retirement. Evaluation of the 2012 module Page 35

36 AHM LFS 2012, PENSION, proportion of yes to no, per cent LFS 2012 indicator, Duration of working life, in years lfsi_dwl_a Graph 4.3: AHM AGEPENS variable average by country, plotted against LFS 2012 working life indicator (duration of working life), age expressed in years SI SK PL RO BG UK EL IT HR HU AT EE EU-28 FR CZ LU MT PT LV LT DE FI IE BE DK CY ES CH NL SE IS NO AHM LFS 2012 AGEPENS, average age in years Graph 4.3 shows that there is a fairly strong correlation between these two variables, which provides good grounds for believing AGEPENS to be measured accurately. As the length of working life increases, so does the age at which respondents first receive an old-age pension. Graph 4.4: AHM AGEPENS average by country, plotted against AHM PENSION (years and %) 120% 100% 80% 60% 40% RO SI SK PL BG HR EL UK IT HU CZ EE FR AT MT LU EU-28 LT LV PT FI DE BE IE DK ES CY CH NL SE NO IS 20% 0% AHM LFS 2012, AGEPENS, average, in years Graph 4.4 shows that, as the average pension age in a country increases, the number of respondents in the target group of the AHM that receive a pension decreases, as would be expected. The fact that the correlation between these two variables follows the expected pattern provides assurance that the quality Transition from work to retirement. Evaluation of the 2012 module Page 36

37 DE DK BG FI HU SK SI RO CZ EE EU-28 FR LV LU AT BE PL NL LT SE EL HR MT CY ES PT UK IE IT NO CH IS of the AGEPENS variable can safely be assumed to be good. Graph 4.5. Difference in average AGEPENS between answer categories of EARLYRET (EARLYRET = yes minus EARLYRET = no), by gender, difference expressed in years Men Women Graph 4.5 shows the difference (in number of years) between the average age for AGEPENS for those who did and those who did not use early retirement schemes. In Germany, Denmark, Belgium, Finland and Norway, the average age at which respondents who did not use an early retirement scheme receive their first old-age pension is lower than the average age for those who did use such a scheme. The difficulties encountered in Germany, Denmark, Finland and Norway when collecting the EARLYRET variable were mentioned in the previous section. The fact that for other countries the difference is a positive value (as would logically be expected) is only an indirect argument to support the quality of the data collected for the AGEPENS variable. The data should nonetheless be used and interpreted with caution. Transition from work to retirement. Evaluation of the 2012 module Page 37

38 P25 P50 P75 Mean Mode Shape P25 P50 P75 Mean Mode Shape P25 P50 P75 Mean Mode Shape The following section, relating to the results in table 4.1 and graphs 4.6 and 4.7, uses unweighted data. Table 4.1: AHM AGEPENS unweighted quartiles, mode, and mean, in years and curve shape, by gender All Men Women EU Neg BE BG CZ Norm DK Pos DE EE Norm IE Neg Neg Neg EL ES FR HR Norm IT CY LV Neg LT Norm LU Pos HU Norm MT NL Neg Neg AT PL Pos PT RO Norm Neg SI Norm Norm SK Norm FI SE UK Neg IS Neg NO CH The placement of the 25th, 50th (median), and 75th percentiles, together with the mode and the mean, show the shape of the distribution of the age at which respondents start receiving their first old-age pension. If the mean, median and mode are equal, the distribution is normal, as in the case of Lithuania and Romania. Where the mean is the highest estimate of central tendency and the mode the lowest estimate, the distribution is positively skewed, as in the case of Poland. Where the opposite is true (i.e. the mean is the lowest estimate and the mode the highest), the data creates a negatively skewed curve. This type of distribution is more common among the country data sets in the table above, with Ireland, Latvia, and the Netherlands showing this shape curve. Only for two countries is the curve shape the same for men and women (Ireland and Slovenia). Transition from work to retirement. Evaluation of the 2012 module Page 38

39 Graph 4.6: Box-and-whisker plot for the AHM 2012 AGEPENS variable A box-and-whisker plot is a powerful tool for illustrating both the central tendency of a data set as well as the spread of the values. The length of the box represents the interquartile range (from the 25th to the 75th percentile). The square symbol inside the box marks the mean value. The horizontal line inside the box indicates the median, and the vertical lines extending from the box show the minimum and maximum values. The most obvious difference between the countries highlighted by the box-and-whisker plots is that some of the lines are very long, and some are quite short. The data range is almost four times larger for the country with the largest range than for the country with the smallest range. In Malta, the highest age at which any respondents started to receive a pension was 65 years, whereas in twelve other countries the highest age was 69 years, which is the highest possible maximum point, given that the data was collected from a population aged 50 to 69. In Spain, France, Italy, Sweden, and the Transition from work to retirement. Evaluation of the 2012 module Page 39

40 United Kingdom, some respondents were recorded as receiving their first old-age pension in their early thirties. An age of 30 years was set as a lower limit when editing the data for this variable, as any ages below this would be very implausible and are highly likely to result from an error in data coding. It should be noted that the spread of the data ranges does not show notable correlation with the average pension age in each country. At the other end of the scale to the countries mentioned above, with very low youngest ages and wide ranges, Iceland has the smallest range of ages over which respondents first received an old-age pension, and the youngest age recorded was in the late fifties. The interquartile range also differs between countries. As can be seen from table 4.1, Portugal has the largest interquartile range, at 8 years. The countries with the shortest interquartile range, three years, are Cyprus, Finland, France, Lithuania, Luxembourg, Malta and Switzerland. Relationships between the various aspects of the data plots can also be analysed, e.g. it can be noted that France has a very wide range but a very short interquartile range, whereas Greece has the exact opposite. This indicates that most of the data from France is concentrated in the middle of the age range, with a few but very extreme outliers, whereas Greece has no real outliers, but the age at which an old-age pension is first received is much less concentrated in the middle of the age range. Graph 4.7: Histogram with imposed Gauss curve for the AGEPENS variable, for all participating countries Graph 4.7 shows that the AGEPENS variable is not normally distributed. The far end of the tails are close to a Gauss curve, but there are also two peaks, one in each of the two middle sections, on either side of the centre of the distribution, as well as a very pronounced spike in the data at around 60 years of age. The graph illustrates that, at European level, there is a high likelihood of receiving a pension for the first time at the age of 60, with the other two ages where there is a concentration of pensions going into payment being at around 55 and 65 years of age. Transition from work to retirement. Evaluation of the 2012 module Page 40

41 LFS AHM 2012, AGEPENS, average in years, per country Graph 4.8: Distribution of the AHM 2012 AGEPENS variable, for all participating countries, by gender (% in the male/female population receiving an old-age pension for the first time) Men Women Graph 4.8 shows that, although the clear peak in the data for both men and women is at 60 years, there are proportionally more women than men first receiving an old-age pension at this age. The age at which men most frequently start receiving an old-age pension is 60 followed by 65, whereas women most often first receive an old-age pension at 60, and next most often at 55. These three main data points together account for almost half of the total volume of data. Almost four times as many women start receiving their first pension at the age of 60 than at the age of 65, whereas for men the equivalent figure is only 1.4 times. There is a similar pattern between the ages 55 and 65, in that women more often start receiving a pension at 55 than at 65, whereas men are three times more likely to start receiving their first old-age pension at 65 than at 55 years of age. Graph 4.9: AHM AGEPENS variable in 2006 and in 2012 (average age in years) NO IS SE 63 ES NL SK LV IT HU LT EE AT CZ UK PT FR EL BE LU CY DK DE FI IE MT 57 RO BG PL SI LFS AHM 2006, AGEPENS, average in years, per country Transition from work to retirement. Evaluation of the 2012 module Page 41

42 This variable was also included in the 2006 AHM, although the question asked then was about the age at which the respondent started receiving an individual retirement pension, and not an old-age pension. Graph 4.9 provides a scatter plot of the two results against one another. A deviation of plus or minus one year must be allowed for. This means that the results are consistent for most countries. Slovenia explained in their quality report for 2012 that the pension system has been under revision since 2006, and comparability has therefore been compromised, albeit unavoidably. Table 4.2: Confidence limits for the AHM 2006 and AHM 2012 AGEPENS variable, in 2006 and 2012, for the outliers in graph 4.9 (average age in years) Mean estimate 95 % Confidence limits Mean estimate 95 % Confidence limits Latvia Slovenia Norway As can be seen from the overview of the confidence intervals, the changes observed for Latvia and Norway cannot be explained by wider confidence intervals. Comparative analysis between 2006 and 2012 should therefore be avoided for these countries. Conclusions and recommendations Response rates were very good overall for this variable. None of the national quality reports indicate any major deviations from the question or problems that would have an impact on comparability. Suggestions for improvement came from Austria, whose report suggests that personal and occupational pensions should have been excluded in the filter for this question, or that respondents should at least have been asked about these types of pension separately from the other old-age pensions, as this would facilitate comparison with administrative data. France and Greece report that some respondents had difficulty determining or recalling the exact age at which they started to receive an old-age pension. These ideas could be taken into consideration were the variable to be used again in a future module, but the general assessment of this variable remains that the data is of good quality and few problems were reported in collecting it. 5. REASNOT: Main reason for not remaining in employment longer Short description The purpose of this variable is to identify the main factor that caused the respondent to leave the labour market. The moment of leaving the labour market is defined as the moment at which the respondent left their last job. The list of possible reasons for not remaining in employment for longer included the following: favourable financial arrangements, inability to find another job, reaching the maximum retirement age, reaching pension eligibility, job-related reasons, and personal reasons. Only the main reason was recorded. Where a respondent spontaneously answered early retirement, the question was repeated in order to elicit the reason for taking early retirement. 22 Filter conditions and codes This question was asked to all inactive respondents (simplified ILO status: WSTATOR = 3, 5 and SEEKWORK = 3) receiving a pension (PENSION = 1). 22 Explanatory notes, page 15. Transition from work to retirement. Evaluation of the 2012 module Page 42

43 Code Description 1 Favourable financial arrangements to leave 2 Lost job and/or could not find a job 3 Had reached the maximum retirement age 4 Had reached eligibility for a pension 5 Other job-related reasons 6 Ow n health or disability 7 Family or care-related reasons 8 Other 9 Not applicable (not included in the filter) Blank No answ er or does not know Analysis of the questionnaires Data on a similar question was collected in 2006, under the AHM 2006 REASRET variable. Despite the difference in name ( Main reason for not remaining longer in employment for REASNOT in 2012 and Main reason for retirement or early retirement for REASRET in 2006), the objective of the 2006 variable 23 was the same as that of the 2012 variable, namely to establish the main factor that caused the respondent to stop working, using a choice of responses better suited to older people. In 2006 however, exit from the labour market was understood as being the time of retirement, which could affect the comparability of the two sets of data. The fact that this variable refers to the moment of leaving the last job detaches it to a certain extent from the analysis of pensions, and could theoretically render the data less sensitive to variations resulting from national definitions of pensions. For example, respondents who continued working after starting to receive a pension from their main employment, e.g. taking a small part-time job or casual work, should have referred to the moment of leaving their last job in their answer, even if the moment of retirement could have been defined as the moment of starting to receive a pension. This nuance limits the comparability of the 2006 and 2012 modules to those respondents who stopped work at the same age as when they first received a pension. As seen at the end of the section on the AGEPENS variable, only seven of the 31 participating countries show no difference between the average age of leaving the last job (core 2012 data) and the age of first receiving a pension (AHM 2012 data). In practice however, the overlap between the 2006 and 2012 variables might have been greater than intended given that, in most countries, questions on the age at which respondents first received a pension and on the reason for leaving their last job were asked consecutively, and because several of the possible responses referred to pensions and retirement, it is likely that some respondents were actually answering in reference to the moment of leaving the job that provided them access to a pension. The purpose of the 2012 variable is therefore identical to the LFS LEAVREAS variable on the main reason for leaving the last job or ceasing business activities. The only slight difference is in the populations being asked this question. For those respondents answering both the LFS LEAVREAS variable and the AHM REASNOT variable, in is understandable that there may have been some confusion. A number of questions were raised during the planning 24 of the module on the difference between codes 3 and 4. The intention of codes 3 and 4 was to distinguish between those who are forced to leave the labour market because of their age and those who leave the labour market as soon as their age allows it. In several countries, for example the Czech Republic, code 3 was not relevant. Most countries have however included the response category in their questionnaire, because it could in principle have been applicable to a small minority of respondents, or in specific sectors of activity or occupations, for example as a result Frequently asked questions, page 11. Transition from work to retirement. Evaluation of the 2012 module Page 43

44 MT CZ SI BG HU EL PL AT LU FR SK IT RO HR EU-28 FI DK CY BE LV UK SE ES IE LT NL PT DE EE NO CH IS of specific collective agreements between trade unions and employers. 25 Finland reported problems in translating codes 3 and 4. In general, response codes 3 and 4 posed some problems for respondents in terms of understanding. Austria and Italy reported that they added additional possible responses. In Italy, the response fear of pension age being raised was included and has been combined with the other reason category when coding the data. Even without additional responses being offered, this question was quite challenging for respondents. There is a clear risk of low data comparability. The inherent subjectivity of the question should also not be ignored where there is a mix of reasons for not remaining in employment longer, it is the respondent who decides which was the main reason. Analysis of the results Data from Germany, Belgium and Iceland includes a significant proportion of non-responses, and should therefore be interpreted with caution. In Belgium, this was due to the data collection methods used. A variable with ten possible values and a distribution that differs markedly between participating countries is challenging to interpret in any form. Graph 5.1: Distribution of answer categories for the AHM 2012 REASNOT variable (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Had reached eligibility for a pension Had reached the maximum retirement age Favourable financial arrangements to leave Other job-related reasons No answer Own health or disability Lost job and/or could not find a job Other Family or care-related reasons Graph 5.1 shows the possible responses, ordered according to the proportion of respondents choosing each at EU level. Two reasons emerge clearly as the most common main reasons for stopping work: had reached eligibility for a pension (37 % at EU level) and own health problems or disability (21 % at EU level); the former was also used as the basis for ordering the countries. Not counting the non-response category or the response own health problems or disability, the proportion of respondents choosing reached eligibility for a pension is equal to the sum of the remaining six answer categories. Nonetheless, the graph shows a range of values for the proportion of respondents choosing this response across the EU, from over 85 % in Malta and the Czech Republic to 12.9 % in Estonia. In 13 of the 31 countries, more people answered own health problems or disability than answered reached eligibility for a pension. Moreover, graph 5.1 shows considerable differences between countries for the responses related to age (codes 3 and 4). This could reflect an actual difference, but could also have resulted from the imperfect choice of possible responses, which respondents found difficult to understand. There is no comparable 2006 data to use for benchmarking. The existing LFS variables refer to the last job, but when a 25 Explanatory notes, page 14, provides a full explanation for code 3. Transition from work to retirement. Evaluation of the 2012 module Page 44

45 respondent actually retired from an earlier rather than from the most recent job, it is not easy to confirm the previous occupation or sector. It may be the case that this variable has failed to provide comparable information on the role that age plays in the decision to retire. Graph 5.2 provides an analysis of the frequency of main reasons for not remaining in employment longer, by gender. At EU level, answers were quite similar for men and women, with two exceptions: family reasons or care related reasons was a more frequent answer among women (and not only at EU level but also in each individual EU country), while favourable financial arrangements for leaving was more frequently chosen by men. These results match the distribution by gender seen in other core variables on reasons for reduced participation in the labour market. Graph 5.2: Distribution by gender of the responses for the AHM 2012 REASNOT variable, EU-28 (%) Favourable financial arrangements to leave Lost job and/or could not find a job Had reached the maximum retirement age Had reached eligibility for a pension Other job-related reasons Own health or disability Family or care-related reasons Other No answer 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Men Women Unfortunately, there is no possibility for comparison between the 2012 and 2006 data, primarily because the response had reached eligibility for a pension was not included in 2006, but was the most frequently chosen one at EU level in Conclusions and recommendations There was a good response rate for this variable overall, with the exceptions of Belgium, Germany and Iceland. None of the participating countries reported major problems in collecting data for this variable. Two responses emerge clearly as being the predominant reasons for not remaining in employment longer: reached eligibility for a pension and own health problems or disability. This may indicate that the question was asked in a similar way across the participating countries in Moreover, the analysis by gender provides the same main results as were obtained from similar core LFS variables. Country differences are significant however. Given that the aim of the variable was to collect data on eight main reasons which are similar to, but not identical to the responses offered for the LEAVREAS variable, full comparability between countries cannot be guaranteed. Comparison with 2006 is also not possible. Were the module to be repeated, consideration would need to be given to whether this variable can add significant relevant information to the information already collected under the LFS LEAVREAS variable. Guidelines issued on the basis of recent analysis performed as part of the overall exercise of evaluating the current system of AHMs advise against the practice of adding a variable to the AHM whose main purpose is to increase the target population of an LFS variable. Transition from work to retirement. Evaluation of the 2012 module Page 45

46 6. WORKLONG: Wish to remain in employment longer Short description The purpose of this variable is to establish whether the person would have preferred to remain in employment for longer (be it in their last or any other job) at the point of leaving the labour market. The variable is designed to reflect the preference of the respondent at the moment of leaving the last job, so spontaneous answers such as I would have liked to continue working but it was not possible for me to do so should be coded 26 as yes. Filter conditions and codes This question was asked to all inactive respondents (simplified ILO status: WSTATOR = 3, 5 and SEEKWORK = 3) receiving a pension (PENSION = 1). Code 1 Yes 2 No Analysis of the questionnaires Description 9 Not applicable (not included in the filter) Blank No answ er or does not know This variable is binomial as the question asks for a yes or no answer. The purpose of the variable was sufficiently self-explanatory and there were no questions about it while the survey was being carried out. The variable is subjective however and hypothetical, as it assumes that, in principle, it would have been possible for all respondents to remain in employment longer. The reference to any other job in the definition of the variable added an extra level of complexity. This was most likely included to cover cases where a last job was not easy to identify, e.g. if a respondent had had a succession of short contracts or spells of inactivity combined with periods of employment. When interpreting the data, users need to keep in mind that for most respondents the variable relates to the last job they had, and measures respondents desire to continue working in the same job. For those with a less well-defined moment of transition from a last job to retirement, the variable reflects the general desire to have a job, or to continue working. One country (Italy) reported some problems in answering the question for people who are economically inactive and only receive survivors pensions. Despite this, the non-response rate for Italy is only 1 %. Nonetheless, even though such cases represent only a small proportion of the population, they provide a useful example of illustrating the imperfect overlap between the labour market and those in receipt of pensions. This also applies to other variables, such as REDUCHRS or REASNOT, albeit to a limited extent. In Romania, data for this variable was not collected for those who answered had reached eligibility for a pension (code 4) to the question for the REASNOT variable. The reason for this is that code 4 in REASNOT, when translated as preference for leaving the labour market when reaching eligibility for a pension 27, can be seen as equivalent to WORKLONG code 2 does not wish to remain longer in employment. As the question was very sensitive to the exact wording chosen, the data should be interpreted with caution. It is plausible that, in some cases, choosing code 4 for the REASNOT question could have triggered a negative answer to the question on the wish to remain in employment longer. At EU level (see graph 6.2), only approximately 12 % of those leaving the labour market on reaching the age of eligibility for a pension state that they would have liked to continue working for longer, thus possibly reflecting this. As detailed in the previous section, it is the REASNOT variable that is likely to suffer from the lack of comparability of its response options, rather than the WORKLONG variable. 26 Explanatory notes, page The formulation used in the Romanian questionnaire. Transition from work to retirement. Evaluation of the 2012 module Page 46

47 PT EE CY ES UK DK LV FI IE AT MT HR BE FR LU SE NL EU-28 IT SK DE CZ RO HU BG EL LT SI PL IS CH NO Analysis of the results Due to the high non-response rate, the data from Belgium, Germany, Greece and Iceland must be used with caution. In Belgium, the high non-response rate was a result of the data collection methods used. In Greece, it was more related to respondents own lack of certainty about their answer, in cases where the situation at the time of leaving the labour market had been complicated, and possibly difficult. Graph 6.1 shows that a majority of retired people receiving a pension in Europe did not wish to remain in employment longer at the time of leaving their last job. The EU average for wanting to continue working is slightly above one quarter (28 %) of respondents, whereas two thirds said they had not wanted to continue working at the point of leaving the labour market. The remaining 6 % gave no response to the question. There are only two countries where the proportion answering yes to this question was over 50 %, Portugal and Estonia. In these countries therefore, the majority of those receiving a pension stated that they would have liked to continue working at the point when they left the labour market. At the other end of the spectrum, in Poland and Slovenia less than one in ten respondents answered yes, indicating that they would have liked to continue working for longer. There is no obvious geographical pattern in the results. Graph 6.1: Distribution of answer categories for the AHM 2012 WORKLONG variable (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Yes No No answer Even though this variable is not perfectly correlated with the REASNOT variable, it would be reasonable to expect there to be a relationship between a wish to stay longer in employment and the main reasons for having left. An analysis at EU level (see graph 6.2) shows that, in general terms, the two variables are related: those who left the labour market because they could not find a job are more likely to have wished to stay in employment than the ones leaving as soon as they could reach eligibility for a pension. There are no noticeable gender patterns. It should be noted however that the level of correlation is weaker than would be expected, and the pattern does not hold at country level in many cases. For example, among those who were forced by law to leave their job (REASNOT = 3), the proportion answering yes to wishing to continue working (WORKLONG = 1) was expected to be higher. The fact that this inconsistency arises when two variables are combined (i.e. the answers to one question are analysed for a specific group defined by their answer to the other), supports the conclusion that the REASNOT variable has limited comparability, for the reasons detailed in the previous section. Transition from work to retirement. Evaluation of the 2012 module Page 47

48 Graph 6.2: Percentage of AHM 2012 WORKLONG = 1 variable, by answering categories of AHM 2012 REASNOT variable and by gender, EU-28 (%) There was no parallel variable in the 2006 module, and there is no link to any of the core variables. There are therefore no alternative ways of making a comparative analysis of this variable. Conclusions and recommendations Data for this variable was relatively easy to collect. The question was, however, subjective, hypothetical and very sensitive to translation wordings, and for these reasons its comparability at EU level cannot be fully guaranteed. Further testing of the variable would be recommended if it is to be included in a repeat of the module. 7. REDUCHRS: Reduced working hours as a step towards full retirement Short description The purpose of this variable is to determine whether the respondent reduced his/her working hours in a move towards full retirement, and, in cases where an old-age pension is being received, whether working hours were reduced before or after the respondent started receiving his/her first old-age pension. Both voluntary and involuntary, and both formal and informal reduction of working hours are relevant for this question. A transition from a full-time to a part-time job is considered to constitute a reduction in working hours. In the case of respondents who are economically inactive, the period of reference for the question is the time before the respondent left their last job. Where respondents are employed, the reference period is the present. If respondents are in employment and not receiving a pension, code 2 cannot be applicable, and as a result, the question becomes a yes/no question for this category of respondent. Similarly, if respondents are inactive and receiving a pension (PENSION = 1) but not an old-age pension, the same applies and the question becomes a yes/no question. Transition from work to retirement. Evaluation of the 2012 module Page 48