Do you have an occupational pension?

Transcription

1 Do you have an occupational pension? Michela Coppola 1 Bettina Lamla 2 This version: April 2013 Abstract In response to an aging society, public pensions are being reduced in almost all developed countries. Recognizing the increasing importance of supplementary pension, this study investigates the determinants for participation in an occupational pension scheme. Most importantly, we account for the fact that employers offering an occupational pension and employees demanding one are not randomly matched on the labor market. Our analysis is based on the German SAVE study, a household panel survey with the main focus on saving and old age provision, which has been linked to administrative records containing information on the workers and establishment level. Occupational pensions are typically found in larger firms, paying higher wages and hiring more high skilled workers. The degree of information about the supply of an occupational scheme varies by firm size, with employees in medium-sized firms being less informed. In line with existing literature we find that low educated individuals are less to participate in an occupational pension scheme. From a policy perspective raising awareness about the need provide for the old age and informing about the ways to do so seems of utmost priority. Keywords: occupational pensions, linked employer-employee data, saving, SAVE study JEL: J26, J32, J33 1 Munich Center for the Economics of Aging at the Max Planck Instiute for Social Law and Social Policy, Amalienstr. 33, Munich, Germany. coppola@mea.mpisoc.mpg.de 2 Munich Center for the Economics of Aging at the Max Planck Instiute for Social Law and Social Policy, Amalienstr. 33, Munich, Germany. coppola@mea.mpisoc.mpg.de

2 1 Extended abstract Understanding how households form their long-term saving and investment decisions to shoulder risks not covered by social security systems has been of primary importance in all countries which introduced major reforms to face the challenges of an aging population. These reforms have shifted longevity and income risks from the state to the individual and therefore require individuals to make their own provisions. At the same time, also employers tried to shift risks by reducing their defined benefit occupational pension plans and increasing defined contribution plans. While data availability and quality has vastly improved, we still face crucial gaps in achieving a comprehensive picture of old-age provision. One aspect of this is our little knowledge about the second pillar of occupational pensions leaving important questions unanswered, e.g., which individuals are covered by occupational pension plans, and what determines their decision to participate in a particular plan? Several explanations have been put forward to why workers want pensions and why employers offer occupational pensions. Demand side theories focus on employees preferences, while accounting for plan features. Employer motivations for supplying pensions rely on the fact that they can be used as a human resource tool. They are seen as a device for attracting workers with certain characteristics, for eliciting work effort, turn-over rates, and retirement. Accordingly, a number of stylized facts have been established for the U.S. (Gustman and Mitchell, 1990): Workers covered by an occupational pension retire earlier, exhibit lower job mobility and work for companies offering higher compensation. With both sides interacting on the labor market pension coverage outcomes arise from matching employers supply and workers demand for pensions (Andrietti, 2000). Most empirical work has not been able to incorporate the influence of the employer-employee match which is mainly due to the heavy data requirements (Gustman and Mitchell, 1990). Moreover, while most of the empirical evidence relates to the U.S., incentives for employers and employees differ depending on cultural differences and the institutional framework that need to be taken into account. For instance, occupational pension coverage outcomes can vary widely depending on the generosity of the public pension system and the tradition to save privately for retirement. Kortmann (2008) documents the recent dynamics of the distribution of occupational pension plans in Germany. From December 2001 to December 2007 the coverage with occupational pension plans increased from 38% to 52% for all employees subject to social security in the

3 2 private sector. His data permit only analyses on the company level, no individual covariates are available. This paper provides an empirical analysis of what determines participation in an occupational pension scheme, grounded on the insight that both sides of the labor market need to be considered. Pension coverage is modeled as a binary outcome explained by a vector of personal and firm characteristics taking into account selection on the labor market, i.e. we observe the choice of each individual only, if he works for a company offering an occupational pension. Facing a two-equation model, our results will be biased if the error terms are correlated. For instance, workers who have a preference to switch jobs ( movers, see Allen et al., 1993) might be less likely to work for a company that uses delayed vesting rules as a sorting device. At the same time an unstable employment history should be negatively correlated with the decision to subscribe such a pension. Our analysis is based on the German SAVE study, a longitudinal survey of private households initiated in 2001 by the Mannheim Research Institute (since 2011 Munich Center) for the Economics of Aging (MEA) and run since 2005 on an annual base. The main goal of the study was to create a sound empirical base to understand households saving behavior. In 2011 respondents (and their partners) were asked for written consent to link their survey data with their administrative records stored at the Federal Employment Agency. The records contain individual information such as daily wages and are further enhanced with establishment level information about the employer (such as economic sector of activity or qualification and age structure of its labor force). All individuals who have worked at least one day as an employee paying social security contributions in Germany are included in these administrative records. While the survey collects a rich set of psychological indicators and includes information on occupational pensions, the administrative records allow linking employers and employees. Our analysis is therefore restricted to individuals who consented to data linkage. As such, the reader has to keep in mind that the findings might be subject to selection bias that we do not control for. In total, 956 households signed and sent back at least one consent form, leading to a consent rate at the household level of about 58%. In a further paper we find only mild evidence for consent bias (Coppola and Lamla, 2012). Apart from income and education, privacy concerns in

4 3 combination with a general resistance towards the survey are found to be the main determinants of the consent decision. Restricting our sample further to respondents that have been working in 2011/12 we are left with 252 observations. Despite the small sample size, German workers and establishments seem to be well represented. Comparing employees in our sample with official statistics from the Federal Employment Agency, we find that employees in the sample are representative for the working population regarding several aspects of their employment history and the composition of the labor force. Similarly, establishments the respondents work for constitute a good picture of the German landscape in terms of firm size, business sector and (median) wages. Respondents were asked whether the establishment they are currently working for is offering an occupational scheme and if yes, whether they participate themselves in the scheme. More than half of the respondents (55%) indicate working in an establishment that offers an occupational pension, while 34% say their establishment does not. A large string of the literature has shown that knowledge of employees regarding occupational pensions is limited (e.g. Luchak and Gunderson, 2000; Gustman and Steinmeier, 2005; Gustman et al., 2007). Gustman et al. (2007) investigate the reason for differences between the pension plan type reported by respondents to the Health and Retirement Study and information obtained from employers. Discrepancies are found to be mainly due to misreporting by respondents. In line with these findings a fraction of 11% of the SAVE respondents claim not to know or refuse to answer the question whether occupational pension schemes are on offer at their establishment. The question about the availability of an occupational pension is fundamental as workers can only respond to incentives they are aware of. Several contributions aim at explaining such a lack of awareness about saving possibilities (e.g. the seminal works on financial literacy). However, as outlined throughout the paper, both sides of the labor market need to be considered when trying to understand the formidable lack of knowledge among certain employees. Pensions and retirement programs are often complex areas of HR administration (Greer et al., 1999), and therefore costly. Against this background the low level of knowledge of employees is frustrating from the employers point of view as this suggests that workers may not appreciate their occupational pension (Luchak and Gunderson, 2000). While pension provision can be seen as a wider package in human resource management, its present role in this package appears therefore less clear (Hales and Gough, 2002, p. 320). Many employees do not even seek information about

5 4 pensions during recruitment process (Clark and Pitts, 1999) limiting further the power of occupational pension as an HR tool. Yet, in Germany employers are obliged to set up some sort of occupational pension scheme if actively requested by the employees. Depending on the beliefs of the employers about the function of occupational pensions the incentive to inform employees about their right and possibilities might differ. In particular, individuals working for medium size firms are less aware of the availability of an occupational pension which might reflect different information policies. At the same time we find a positive correlation between firm size and the availability of occupational pensions. {Graph 1 about here} Comparing further firm characteristics we find that the share of marginally employed workers (so called mini-jobbers ) is significantly lower in establishment offering occupational pensions while the share of high skilled workers is higher. Moreover, they pay higher median wages on average and experience less turn-over. In conclusion U.S. stylized facts can be replicated for Germany, also confirming that employees as well as employers are systematically different depending on the availability of occupational pensions. In order to answer our research question we therefore correct for the selection using a Heckman model (Heckman,1979). Our exclusion restriction is the share of marginally employed workers at the establishment level. Data limitations do not allow us to control for whether the subscription is mandatory or voluntary. However, the model might be misspecified if e.g. firms with less minijobbers are more likely to introduce mandatory pension schemes. Further missing variables are the concrete eligibility rules or the pension plan type.. {Table 1 about here} Due to the many data limitations we will present results for a Probit model explaining the determinants of owning an occupational pension conditional on the reported availability at the establishment, and the Heckman model presented above. In both specifications not much is significant. We find that lower educated individuals are less likely to participate in an occupational pension scheme. This is in accordance with the common finding that lower educated

6 5 individuals accumulate lower wealth for retirement and are less likely to participate in voluntary pension schemes (for Germany: Börsch-Supan et al., 2012). Moreover, employees working for medium-sized firms are less likely to opt for an occupational pension. As already indicated by the results above this might reflect different information policies by firm size. Somewhat surprisingly our exclusion restriction is not significant indicating that the error terms are not correlated and selection does not play a role. However, the results need to be interpreted with caution: {Table 2 about here} An essential requirement to answer the research question is a rich data set linking employers and employees, including quantitative information on pensions as well as qualitative aspects on motives and expectations on the workers and the establishment level. No currently available data set meets the necessary requirements as described in Gustman and Mitchell (1990). We were therefore forced to deal with a series of data limitations and the results are likely to be sensitive to the sample and model specifications. Most of the time there is insufficient information in the data to remedy the identification problem, and all we can do is make the reader aware of its existence. On the supply side of occupational pensions, we describe a number of establishment level outcomes (such as turnover, share of skilled workers, compensation) that seem to differ between firms offering occupational pension and those who do not. The findings are in line with previous evidence for the U.S. labor market, replicating stylized facts for Germany. While many incentives for employers to offer an occupational pension were already discussed in the literature (e.g. Gustman et al., 1993), we lack evidence on how effective pensions are in managing firm outcomes and to what extent there are differences between pension plan type. On the demand side, we want to highlight two findings: First, many workers do not know whether the establishment they are working for offers an occupational pension scheme. We suggest that while the well-documented lack of knowledge (see e.g. Gustman et al., 2007) might be due to the lack of interest of workers, the fault may also lie with the establishments. From a policy perspective standardized information requirements might help to overcome poor information policies in some firms. Especially as employer provided information can be regarded as to general and is not easily understood by many employees (Hales and Gough, 2002; Gustman et al., 2007).

7 6 Secondly, we find that low educated individuals are less likely to participate in an occupational pension scheme. Evidence has broadly accepted that low educated individuals save less and therefore might accumulate insufficient resources for retirement. Taken together the results are even more alarming: Low educated individuals are known to be less likely to plan for retirement (Lusardi, 2003) and hence, to actively seek information on occupational pensions. At the same time more financial knowledge implies a higher propensity to plan and succeed in planning, in particular for households with low income and low education (Lusardi and Mitchell, 2011). Raising awareness of the need to supplement decreasing statutory pension and the ways to save for retirement in the general population seems of utmost priority. We believe that our results provide a good starting point for more research on occupational pensions outside the U.S. As many important questions remain untouched there is a clear and urgent need for more and richer data sets.

8 7 References Andrietti V (2000) Occupational Pension Coverage in The European Union. An Empirical Analysis. ISER Working Paper Series: Allen S, Clark R, McDermed A (1993) Pensions, Bonding, and Lifetime Jobs. Journal of Human Resources 28(3): Börsch-Supan A, Coppola M, Reil-Held A (2012) Riester Pensions in Germany: design, dynamics, targeting success and crowding In. NBER Working Paper Series Clark R, Pitts M (1999) Faculty Choice of a Pension Plan: Defined Benefit versus Defined Contribution. Industrial Relations 38(1): Coppola M, Lamla B (2012) Empirical Research on Households Saving and Retirement Security: First Steps towards an Innovative Triple-Linked-Dataset. MEA Discussion Paper 258 Gustman A, Mitchell O (1990) Pensions and the U.S. Labor Market. NBER Working Paper Series 3331 Gustman A, Steinmeier T (2005) Imperfect Knowledge of Social Security and Pensions. Industrial Relations 44(2): Gustman A, Steinmeier T, Tabatabai N (2007) Imperfect Knowledge of Pension Plan Type. NBER Working Paper Series Greer C, Youngblood S, Gray D (1999) Human resource management outsourcing: The make or buy decision. Academy of Management Executive 13(3):85 96 Hales C, Gough O (2003) Employee evaluations of company occupational pensions: HR implications. Personnel Review 32(3): Heckman J (1979) Sample Selection Bias as a Specification Error. Econometrica 47(1): Kortmann K, Halbherr V (2008) Alterssicherung in Deutschland 2007 Methodenbericht. Forschungsbericht F391/M BMAS Luchak A, Gunderson M (2000) What do employees know about their pension plan? Industrial Relations 39(4): Lusardi A (2003) Planning and saving for retirement. Dartmouth College Working Paper Lusardi A, Mitchell O (2011) Financial Literacy and Planning: Implications for Retirement Wellbeing. NBER Working Paper Series 17078

9 8 Appendix Graph 1: Share of employees working in an establishment that offers an occupation pension by establishment size 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Does the establishment where you are currently working offer an occupational pension? Yes DK/Missing < > 500 Establishment size (number of employees)

10 9 Table 1: Descriptive statistics Does the establishment where you are currently working offer an occupational pension? West test East Yes Yes vs. No average # of empl. 991 *** 668 *** % of minijobs 11% *** 5% *** % of high skilled workers 11% *** 18% *** average median wage 3,293 *** 2,713 *** Turnover 12% ** 12% % of discharges by seniority < 4 years 51% *** 58% *** 4 to 9 y 23% *** 20% > 9 years 26% *** 22% *** No average # of empl % of minijobs 37% 25% % of high skilled workers 7% 3% average median wage 2,262 1,717 turnover 22% 19% % of discharges by seniority < 4 years 74% 74% 4 to 9 y 13% 18% > 9 years 13% 8% DK/ No Answer DK vs. No average # of empl % of minijobs 43% 21% % of high skilled workers 5% 3% average median wage 2,243 1,727 turnover 18% 17% % of discharges by seniority < 4 years 55% * 72% 4 to 9 y 26% ** 17% > 9 years 19% 12% Mann-Whitney test

11 10 Table 2: Estimation results Have an occupational pension? LPM Heckman Demographics Coef. Coef. female HH East Age Education Hauptschule ** ** Realschule Ref. Ref. Abitur University 0.19 * 0.17 Gross Income Log (Income) Saving Attitudes Motive old-age: important Saving regularly Employer changes No changes Ref. Ref. Moderate changes Many changes Firm characteristics Turnover rate Establishment size <50 Ref ** ** > Constant 0.65 * 0.73 R-squared N mills Coeff. S.E. lambda