The PASS panel survey after six waves

Abstract

The aim of the Panel Study “Labour Market and Social Security” (PASS) is to provide a database which allows analysing the dynamics of welfare benefits receipt after the introduction of the Unemployment Benefits II in Germany in 2005. This entails the take up and ending of benefits receipt as well as the social situation of households and individuals receiving benefits, their subjective ways of coping with the situation and the contact to institutions providing the basic income support. PASS is set up as a household panel study with a sample of approx. 10,000 households interviewed in each wave. In addition to household interviews with the heads of the households about 15,000 interviews with individual household members aged 15 and older are carried out. This article provides an overview of the first six waves of PASS. It focuses on the survey’s main goals, the questionnaire, the sampling and study design, the number of interviews, data access and methodological research on PASS. The article closes by describing the outlook for future developments.

Zusammenfassung

Ziel des Haushaltspanels “Arbeitsmarkt und soziale Sicherung“ (PASS) ist es, eine Datenbasis für die Analyse des Bezugs von wohlfahrtsstaatlichen Leistungen nach der Einführung des Arbeitslosengeld II in Deutschland im Jahr 2005 zur Verfügung zu stellen. Dies umfasst die Aufnahme und die Beendigung des Leistungsbezugs, sowie die soziale Lage von Haushalten und Individuen im Bezug, ihre Wege mit der Situation umzugehen und den Kontakt zu Trägern der Grundsicherung. PASS ist eine Haushaltspanelbefragung, in deren Rahmen in jeder Welle ca. 10.000 Haushalte befragt werden. Zusätzlich zu Haushaltsinterviews mit den Haushaltsvorständen werden ungefähr 15.000 Interviews mit Haushaltsmitgliedern ab 15 Jahren durchgeführt. Dieser Aufsatz gibt einen Überblick über die ersten sechs Wellen von PASS. Beschrieben werden Ziele, Erhebungsinhalte, Studien- und insbesondere Stichprobendesign, Fallzahlen, Datenzugang und methodische Begleitforschung. Der Aufsatz schließt mit einem Ausblick auf zukünftige Entwicklungen.

1 Introduction

Since the late 1990s, the German labour market and social policies have undergone major restructuring with the introduction of the Job-AQTIV and subsequent Hartz reforms. These reforms emphasised active labour market policies (ALMPs) and reorganised the welfare and unemployment benefits system. With the integration of welfare benefits (i.e., Sozialhilfe, social assistance, SA) and unemployment benefits for long-term unemployed (i.e., Arbeitslosenhilfe, unemployment assistance, UA), the fourth and final portion of the Hartz reforms was implemented in 2005. This resulted in the creation of a new benefits scheme called Unemployment Benefits II (Arbeitslosengeld II, UB II), codified in Social Code II (Zweites Buch Sozialgesetzbuch, SGB II).¹

The new UB II includes means-tested minimal benefits for individuals aged 15 to 64 who are in need and able to work. UB II can only be claimed when no other means of subsistence, such as savings, assets, or other state transfers (e.g., insurance based unemployment benefits), are available. Another feature of the new benefit scheme is the adoption of the benefit unit concept, which means that if a person in need has a partner able to support him or her financially, he or she cannot claim UB II. Partners can receive benefits only when they are both in need, in which case they serve as the benefit unit. The principal aim of the reforms was to end or reduce neediness through quickly reintegrating individuals into the labour market. Although the benefits should assure economic subsistence of the household, ALMPs were employed to strengthen the individual’s own responsibility for his or her employment prospects.

2 Goals of the study

In 2006, PASS was initiated by the Institute for Employment Research (Institut für Arbeitsmarkt- und Berufsforschung, IAB) to facilitate longitudinal research on the new benefits. The focal research questions were defined as follows (cf. Achatz et al., 2007):

• (A) The pathways out of UB II receipt

  On the one hand, this covers the question of how important employment is compared to alternative pathways out of poverty, such as retirement, finding a new partner or transferring to paramount support systems, and by which pathways different groups of recipients (e.g., by age or gender) overcome benefit receipt. On the other hand, PASS should illuminate which resources facilitate to exit receipt and which impediments keep people from gaining employment with sufficient income to overcome receipt. The first results on these topics have been published by Achatz and Trappmann (2011), Bruckmeier et al. (2013) and in a study for the Federal Government Report on Poverty and Wealth (Pollak et al., 2013).

• (B) The dynamics of the material and social situation of households in receipt

  To provide information about the material situation of households and changes over time, three different concepts to measure welfare are integrated in PASS. In addition to several individual income components and the overall household income, which are used for a resource-based approach, deprivation in living standard and receipt of social benefits are requested. The different measurement concepts have been used in studies by Andreß et al. (2010), Andreß and Hörstermann (2012), Christoph (2010), Christoph and Lietzmann (2013) and Lietzmann et al. (2011).

  In addition to re-integration into the labour market and ensuring subsistence, Social Code II explicitly targets recipients’ social inclusion. Thus it is important to focus not only on material aspects of the recipients’ lives but also on a broader range of issues including social networks, health (Eggs 2013), happiness and social inclusion (Gundert and Hohendanner, forthcoming).

• (C) Behaviour and attitudes of recipients over time

  The stronger focus on ALMPs and the less generous provision of benefits that were introduced through UB II were directed towards changes in job search behaviour as well as the work attitudes of benefit recipients. This change was meant to encourage individuals receiving UB II to accept less than adequate jobs more readily. Because PASS contains detailed information on recipients’ behaviours and attitudes, it allows for analyses of activities and work motivation (Beste et al. 2010), willingness to make concessions to secure employment (Bender et al. 2009), reservation wages (Blien et al., 2012) and the role of work values on labour market behaviour (Hohenleitner and Tolciu, 2011) of UB II recipients. Bethmann and Berngruber (2012) examined decision-making by couples in UB II receipt, whereas Fehr (2012) studied how young people’s access to leisure activities depends on their social position.

• (D) Interactions between benefit recipients and the agencies for the provision of UB II (Jobcenters)

  Other micro datasets that can be used for labour market or poverty research usually lack information about interactions between UB II recipients and agencies responsible for the provision of UB II. PASS was designed to close this gap. In the past, PASS data have been used to analyse the perception (Christoph and Hohmeyer 2012) and the effects (Wulfgramm 2011a, 2011b) of ALMPs as well as the effects of benefit sanctions (Hohenleitner and Hillmann 2012). Tisch (2010) investigated satisfaction of benefit recipients with the Jobcenters.

• (E) Pathways into UB II receipt

  The take-up of UB II receipt can be initiated by changes in the household composition (e.g., separation from partner), by labour market related events (e.g., job loss) and by individual resources (e.g., education). Initial results on pathways into benefit receipt have been published by Fuchs (2012).

In cooperation with Rainer Schnell, the IAB developed a sampling design (see Schnell, 2007; Rudolph and Trappmann, 2007) that not only supports answering these focal research questions in the context of the Social Code II but also makes PASS a basis for a much wider range of questions in the areas of labour market, poverty and social security research in Germany.

Because UB II is paid to benefit units that are in most cases identical to households, the household context must be considered to adequately investigate individual and social consequences of the reform. Thus, PASS is designed as a household panel survey with yearly waves.

3 Questionnaires and topics

A household questionnaire is first administered to the head of each household. The questionnaire focuses on topics concerning the household as a whole (e.g., household composition, housing, deprivation, and child care). All members of the household aged 15 or older are then interviewed on their personal situations including employment status, income, health, social networks, attitudes (e.g., life satisfaction and gender role attitudes) and behaviour (e.g., when searching for a job).

For respondents with receipt of UB II, questions regarding their subjective perception of interactions with Jobcenters, participation in active labour market policy programmes and benefit sanctions play a central role.

The regular questionnaire programme is supplemented by yearly focal topics. Past focal topics have included retirement provisions (Wave 3), social networks (Waves 3 and 5), health and health related behaviour (Waves 3 and 6), willingness to accept job offers and mobility (Wave 5), personality traits (‘Big Five’) (Wave 5), experiences of injustice (Wave 6), social participation of children and youth, including the educational package of the Federal Government (Wave 6) and sports activities (Wave 6). Table 1 provides an overview of the questionnaires of first six waves.

Table 1 Overview of the questionnaire modules

4 Sampling design

The original PASS sample is composed of two subsamples. The first one is a sample of households containing at least one benefit unit on the reference date in July 2006. This sample is drawn from recipient registers at the Federal Employment Agency. The second subsample is a sample of the residential population in Germany, oversampling low status households, thus creating a reference group interviewed in the same survey with the same instrument. Because the inclusion probabilities for all elements of the target population are known, each subsample can be projected to its target population using cross-sectional weights supplied with PASS. In addition, integrated weights can be used to project the complete sample to the German residential population (see Rudolph and Trappmann, 2007).

The initial recipient sample loses relevance as time passes. After the first wave, it can no longer be projected to all current recipients because new entries to UB II are not included. Thus Subsample 1 is refreshed each year to include new entries into UB II (i.e., households with at least one benefit unit on the reference date of the current wave and no benefit unit on any of the previous reference dates). All respondents who (still or again) receive benefits on the current reference date can thus be projected to all current recipients in Germany. Furthermore, the refreshment samples can be used to analyse entry cohorts separately or to identify determinants of inflows into UB II (cf. Trappmann et al., 2010, 2011).

The general population sample does not require regular refreshment samples. PASS has implemented wide following rules. Everyone who moves or is born into a PASS household is included in the sample and followed after moving out (again). This broad following rule guarantees reflection of demographic change in the panel survey and enables researchers to project results based on this sample to all households in Germany.² New households established by international migration can however not be reflected by this design. As this constitutes only a very small proportion of the German residential population, this can be neglected for a few years for most research questions. PASS introduced a replenishment sample for the general population sample in its fifth wave (for details, see below) which also includes migrants who moved to Germany between the first and the fifth panel wave.

5 Fieldwork and number of interviews

PASS is conducted in a sequential mixed-mode design of computer aided telephone interviews (CATI) and computer aided personal interviews (CAPI) mode. The mix of modes aims to optimise contactability and cooperation given a restricted budget. Given the rather difficult survey population, which features high proportions of target persons with low education or migration background, a series of measures was implemented to assure acceptable response rates and panel retention. These measures include refusal avoidance trainings for interviewers, monetary incentives for respondents, foreign language questionnaires (in Turkish and Russian) and intense maintenance and tracking activities (cf. Trappmann et al., 2011).

Data collection for PASS started in 2007 with more than 12,000 responding households in Wave 1. The first three waves were conducted by TNS Infratest Sozialforschung. Since 2010, the fieldwork agency infas Institut für angewandte Sozialwissenschaft collects the PASS data.

Panel surveys can never entirely avoid panel attrition. Each year, a share of the previous wave respondents is lost due to death, moving abroad, non-contactability or refusal. Despite considerable efforts to avoid it, the PASS panel loses between 20 (Wave 5) and 43 per cent (Wave 2) of households to attrition between subsequent waves. Although approximately 20 per cent of the dropouts are only temporary and return in the following wave, this attrition leads to a pronounced decrease in sample size in the long run.

To compensate for the loss of statistical power associated with attrition, a replenishment sample of both the population and the recipient sample was drawn before Wave 5 (cf. Berg et al., 2012, Chap. 6.1). The number of participating households was increased from 7,848 in Wave 4 to 10,235 in Wave 5. In Wave 6, the most recent, 9,513 households participated. Table 2 indicates the number of household interviews by wave.

Table 2 Number of household interviews by wave

In PASS, informed consent to link individual survey responses to administrative datasets of the Federal Employment Agency is requested. The consent rate after Wave 6 was 87.2 per cent. For a large proportion of respondents, the potential for analysis can be increased by enriching survey data with individual information from the administrative registers.

6 Data access and documentation

The first five waves of PASS are currently available as a scientific use file from the Research Data Centre (RDC) of the Federal Employment Agency at the Institute for Employment Research, and Wave 6 is expected to be released in November 2013. The RDC website (http://fdz.iab.de/en/FDZ_Individual_Data/PASS/Working_Tools.aspx) provides extensive documentation for users. Users are particularly recommended to consult the User’s Guide (Bethmann and Gebhardt 2011) and the wave-specific data reports (for Wave 5: Berg et al., 2012).

7 Methodological research into PASS data quality

The difficult survey population and the moderate response rate and panel retention create demand for extensive methodological research into PASS data quality. In recent years, this demand has resulted in numerous studies dealing with the increase in efficiency by stratification (Trappmann et al. 2009), the selectivity of linkage consent (Beste 2011) and its effect on total survey error (Sakshaug and Kreuter 2012), item nonresponse (Jaenichen and Sakshaugk, 2012) and unit nonresponse (Schnell et al. 2010; Müller 2011; West and Little 2012), the interaction of unit nonresponse and measurement error (Kreuter et al. 2010) and implications of the mixed mode design for total survey error (Levenstein 2010). Research attention has also been directed towards interviewer effects (West et al. 2013) and the assessment of interviewer performance (West and Groves 2013).

8 Current and future developments

PASS has currently secured funding³ until at least 2016. Wave 7 fieldwork ended in September 2013. Focal topics in Wave 7 are employment quality and stigma consciousness of and prejudices against unemployed. Focal topics on the participation of youth and sports activities are also being continued. Wave 8 will continue the social participation topic and contain a new module on subjective expectations of job search outcomes.

The RDC and the PASS team are currently preparing the release of a combined survey and administrative dataset (named PASS-ADIAB) comprising all PASS respondents and variables plus information available for respondents in the SIAB dataset (Dorner et al. 2010). This includes detailed employment histories dating back until 1975. This dataset will be available to researchers for onsite use at the RDC in Nuremberg or one of the RDC external locations at Berlin, Bremen, Dresden, Düsseldorf, Mannheim or Ann Arbor (see http://fdz.iab.de/en/FDZ_Scope_of_Services.aspx for details on locations).

Another IAB project that aims to augment PASS data has started recently. The IAB plans to create mixed-method datasets by conducting semi-structured interviews with PASS respondents who are theoretically sampled with respect to research questions that cannot be fully answered by quantitative data. As a first step, these interviews will focus on recipients who overcome benefit receipt despite low model-predicted chances due to a high number of labour market impediments.