Pension profile preferences: the influence of trust and expected expenses

ABSTRACT

This paper studies the influence of people’s expectations about expenses during retirement and trust in pension funds on preferences for different pension arrangements. Although most workers prefer a flat-rate annuity, many workers want to deviate from it. The most popular option is a high/low, annuity-based profile, followed by a partial lump sum payment. Workers who expect declining expenses during retirement are more likely to opt for a high/low annuity-based pension and/or a lump sum payment at retirement. Furthermore, workers and pensioners who do not trust their pension fund are more likely to prefer a lump sum over annuity-based arrangements.

KEYWORDS:

• Consumer choice
• consumption
• trust
• pension annuities
• lump sum

JEL CLASSIFICATION:

• D14
• D91
• G20
• J26

Acknowledgement

We are grateful for comments by an anonymous referee, Leo de Haan, Lieke Kools, Joris Knoben, Marike Knoef, Jante Parlevliet, Maarten van Rooij and Mathijs van Zaal, and for discussions with seminar participants at the AFM, Netspar and De Nederlandsche Bank. We thank Miquelle Marchand (CentERdata) for arranging the survey and Gareth Budden for linguistic services. The views expressed in this paper do not necessarily reflect the views of De Nederlandsche Bank or those of the Eurosystem.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Existing literature shows that expectations of earnings before retirement affect pension choices (e.g. Cocco and Lopes 2011), we take a different approach by researching the effect of expected expenses after retirement on pension choices.

² Bikker, Knaap, and Romp (2014) show that Dutch pension funds limit pension rights by 4 percentage points when the funding ratio falls below the minimum threshold level of 105%.

³ E.g. Van Rooij, Lusardi, and Alessie (2012) show that financial literacy contributes to retirement planning and retirement saving.

⁴ According to the literature it is difficult to indicate a priori how financial literacy influences participants’ preferences. Financially literate participants may be aware of the longevity risk and prefer the default full annuity pension plan, whereas financially illiterate participants may underestimate the longevity risk (Brown 2007). Yet, financially literate participants may also opt for non-default pension plans as they may think to be able to achieve higher returns than their pension funds (investment option) (Van Rooij et al. 2007; or Banks, Crawford, and Tetlow 2015). Van Rooij and Teppa (2014) use survey data on Dutch consumers and show that when it comes to economic decision-making financially literate individuals are more likely to opt out the default and to set apart additional savings via third pillar retirement savings products.

⁵ Health care expenses include for example health care insurance, medical out-of-pocket costs, and payments to service providers. Since the introduction of the mandatory base health care insurance in 2006, which covers a wide range of medical treatments, the Dutch have been confronted with rising health care costs for insurance contributions due to a rising excess of the Dutch base health care insurance, and some cuts in its coverage. In addition, the Dutch administration has economised on state-subsidised domestic help for people with health problems.

⁶ Since the start of the 21^st century the Dutch pension system has undergone several reforms, see Parlevliet (2017).

⁷ However, only a minority of Dutch consumers value individual freedom of choice for most pension contract issues (Van Dalen and Henkens forthcoming).

⁸ Arts and Ponds (2016) show that both young and older generations in the Netherlands may benefit from the exchange of home equity and pension wealth by the use of reverse mortgages and lump sum take-ups of accrued pension wealth.

⁹ In general, there is little information about members’ usage of different pension options in the Netherlands. However, Dellaert and Ponds (2014) show the choices of the members of the civil servant pension fund (ABP) who are born between 1940 and 1950. It turns out that in 2010 20.5% of these people was retired earlier or later than the default retirement age, 7% retired part-time and 1.5% made use of the high/low option. The popularity of the high/low profile is relatively low compared to other options. However, there is some evidence that between 2010 and 2015 the high/low profile has gained in popularity (Van Den Bleeken et al. 2017).

¹⁰ The questionnaire is available on request. The ordering was the same for all individuals surveyed.

¹¹ More information on the CentERpanel is available at www.centerdata.nl/en/projects-by-centerdata/the-center-panel. URL last accessed on 3 August 2018. Teppa and Vis (2012) also give a good overview.

¹² See www.centerdata.nl/en/publications for all publications. URL last accessed on 3 August 2018.

¹³ More information on the DHS is available at www.centerdata.nl/en/projects-by-centerdata/dnb-household-survey-dhs. URL last accessed on 3 August 2018.

¹⁴ There were 54 incomplete responses.

¹⁵ The response rate is high compared to offline surveys, but not uncommon for internet-based surveys using panels. Moreover, it is in line with response rates reported by other scientific articles based on surveys among the CentERpanel. For example, Van der Cruijsen et al. (2013) report a response rate of 85% and Van der Cruijsen and Diepstraten (2017) document a response rate of 81.5%. For an overview of the response rates to the separate DHS modules over time, we refer to Teppa and Vis (2012).

¹⁶ Other factors with the response shares between brackets are: ‘the default option’ (16%), ‘too far away in time’ (8%), ‘not thinking about pensions’ (7%), and ‘minimisation income tax’ (4%).

¹⁷ Other factors with the response shares between brackets are: ‘receive extra savings after a while’ (10%), ‘large expense in first years of retirement’ (8%), and ‘additional return by investing’ (3%).

¹⁸ Other factors with the response shares between brackets are: ‘high life expectancy’ (18%), ‘labour income during first years’ (15%), ‘labour income partner during first years’ (8%), ‘minimising capital tax’ (3%), and ‘large expenses after a while’ (2%).

¹⁹ With respect to the choice between a flat profile and a high/low profile, we focus on workers only because this choice is made before retirement.

²⁰ Note that the group of respondents wanting a low/high profile is too small to include in the analysis.

²¹ It may very well be the case that some pensioners want to spend more but do not have enough income to do so. We find a significant positive correlation of 0.23 between the extent to which one thinks that one’s pension income has been disappointing and the extent to which one spends less than expected beforehand.

²² Additional analyses (available upon request) show that the overall expected expenditure pattern of employees with pension rights is not significantly related to age. It is also unrelated to most other background characteristics. Exceptions are gender and financial ability. Males are less likely to expect an increasing pattern than females and more likely to expect a decreasing pattern, whereas people who consider themselves knowledgeable with respect to financial matters are more likely to expect an increasing expenditure pattern and less likely to expect a decreasing expenditure pattern than people who consider themselves unknowledgeable.

²³ In our survey we included a question to measure whether one has joined a company’s pension scheme.

²⁴ Little is known about the drivers of trust in pension funds. An exception is the recent study by Van Dalen and Henkens (2018), who find that pension participants’ trust depends on perceived integrity, competence, stability and benevolence of pension providers.

²⁵ These self-reported measures for financial ability are akin to variables used to proxy financial literacy (e.g. Lusardi and Mitchell 2007; Van Rooij, Lusardi, and Alessie 2011).

²⁶ Please note that pension funds in the Netherlands are obliged to provide their members with information about their pension every year. So all respondents with pension rights have received information from their pension funds. However, many of them are not aware of that, indicating that the answers to this question should be considered as self-assessed.

²⁷ Previous research shows that attitudes in the context of pension depend on the age. For example, Mello et al. (2017) find a strong positive correlation between a person’s age and support for public spending on pensions.

²⁸ To test for spurious correlations and the importance of cohort effects, we ran regressions that exclude the control variables. Our findings are robust to the exclusion of control variables: the relationships barely change. The results are available upon request.

²⁹ Note also that multicollinearity is not a problem. For the regression in Table 4 column 1 the mean Variance Inflation Factor (VIF) of the explanatory variables is 1.44, the maximum VIF is 2.35, the minimum is 1.06. For the regressions in Table 4 column 2 these figures are 1.50, 2.76 and 1.09 and for the regressions reported in column 3 the values are 1.40, 1.09, and 2.56.