Psychological Factors of Voluntary Retirement Saving (August 2015) Extended Abstract 1 Psychological Factors of Voluntary Retirement Saving Andreas Pedroni & Jörg Rieskamp University of Basel Correspondence should be addressed to: Andreas Pedroni University of Basel, Department for Psychology Missionsstrasse 62a 4055 Basel, Switzerland Phone: (+41) 61-267-3905 E-mail: andreas.pedroni@unibas.ch Introduction The funding of the future pension system is one of the most pressing global challenges many societies are facing today. Governments throughout the industrialized world have anticipated this challenge and have begun to increasingly bank on people s responsibility to save for their sunset years, mainly by adopting policies, in which voluntary retirement saving programs are favorably taxed. There is mixed evidence whether this trend has helped to increase retirement savings (1), however, it undoubtedly places a large share of the asset allocation responsibility and investment risk directly on individuals. Naturally, there are various factors that potentially affect people's decision to voluntarily save for their pension. Most importantly, the monetary resources need to exist. Another precondition for retirement savings and in particular whether a specific savings program is followed, is sufficient knowledge that the program exists and how it works. These factors can be summarizes as necessary conditions for voluntary retirement savings. Even if these conditions are met, people may not necessary enroll in a voluntary retirement savings plan. We argue that the decision to enroll in such a program and how investments are allocated within the program are decisively affected by psychological factors, in particular self-control capacity. Studies have shown that people vary to a great extent in their self-control capacity, commonly defined as the ability to resist immediate consumption in favor of a later reward. Given that voluntary retirement saving requires resisting immediate consumption, there is a good case to believe that it is influenced by individuals self-control capacity. Previous research has mainly focused on socio-demographic influences on retirement saving, such as age, income and gender (2), as well as the influence of peoples financial literacy (reflecting a persons ability to perform financial calculations and knowledge of some fundamental financial concepts) (3, 4). Surprisingly, research focusing on other psychological factors is sparse. The goal of the present study was to fill this gap, by examining to what extent psychological factors, such as people's self-control capacity, their risk attitude as well as their financial literacy and knowledge about the pension system affect if and how much they save by choice. Furthermore, we examined whether these factors affect how much risk people accept in allocating money (i.e. in funds or savings). For this purpose, we collected the data in a national survey conducted in Switzerland. The survey consisted of a sample of 1008 participants in the age range of 19 45 years, which is representative for the Internet using population (LINK, Institut für Markt- und Sozialforschung). In this survey participants were asked about their voluntary retirement saving activity, focusing on tax-exempted savings in the so-called Pillar 3a of the Swiss pension system. Most importantly, we assessed participants knowledge about the Swiss pension system, their self-control and their risk attitudes (see methods for further information about the assessment).
Psychological Factors of Voluntary Retirement Saving (August 2015) Extended Abstract 2 Results Who is prepared to save? 79 % of the respondents reported having a Pillar 3a account. In line with previous studies (e.g. 2) we find a pronounced effect of household income and age on the incidence of having a 3a account (Fig. 1). Figure 1. Incidence of having a Pillar 3a account in relation to A) household income and B) age. While controlling for socio-demographic variables (see methods for a complete list of variables), our analyses indicate that knowledge about the Swiss pension system is strongly related to having a Pillar 3a account (Fig. 2a). In contrast to previous studies (e.g. 3, 4) we find no significant influence of financial literacy on voluntary retirement saving. Interestingly, our analysis shows that that individuals with low self-control capacity compared to individuals with high-self control 10 % less often possess a Pillar 3a account (Fig. 2b). Figure 2. Incidence of having a Pillar 3a account in relation to A) pension system knowledge and B) self-control capacity
Psychological Factors of Voluntary Retirement Saving (August 2015) Extended Abstract 3 How much do people save? As a measure of how actively individuals voluntarily save for their retirement, we examined their reported deposits in the Pillar 3a of the last year (2014). 66.3 % of those respondents who possess a Pillar 3a account reported that they have deposited on average CHF 5014 in the last year (2014), 23 % did not know how much they deposited, while 9 % did not deposit any money into their Pillar 3a account. Considering sociodemographic factors, we find significant effects of age, gender, household income, and education level on the magnitude of deposits made in 2014. Specifically, older male individuals with high income and education reported to have deposited significantly more money in their Pillar 3a account. Our analyses reveal that knowledge about the pension system significantly predicts the magnitude of deposits made in the last year, indicating that those individuals with higher knowledge about the pension system not only more frequently possess a Pillar 3a account, but they also save more. In line with our hypothesis, individuals with higher self-control deposited 22 % (CHF 977) more in their Pillar 3a account in the last year, compare to those with low selfcontrol. Figure 3. Effects of pension system knowledge and self-control capacity on the deposit made in 2014. How do individuals save? Of the individuals, who possess a Pillar 3a account, 67 % reported to save in a savings account, 11 % in a fund and 13 % indicated to save in a savings account and fund. 9.5 % did not know whether they save in a savings account or fund. Thus, a majority prefers the security of a savings account while accepting the downside of comparably low expected return rates. The analyses reveal that gender, the self-assessed expertise in the Swiss pension system as well as risk preference significantly influence whether an individual saves in a savings account or in a fund. Specifically, we find that female respondents significantly more often invested their money in a virtually risk-free savings account. Individuals, indicating that they were proficient in pension matters were more likely to invest retirement money in funds. However, we find no influence of the actual knowledge about the pension system. Interestingly, individuals, who were reluctant to take any investment risk in the risk attitude measure (i.e. would not invest any money) more often reported to save for their retirement in a savings account and hence preferred a virtually risk-free asset. Controlling for other ways to save. As some individuals may rather invest in more liquid assets than the Pillar 3a, which is bonded (with few exceptions) to retirement, we asked whether or not individuals save outside the Swiss pension system (e.g. in a non tax-favored saving account, stocks or real estate, etc.). We included this information in the analyses and found that the above reported results hold, also after controlling for alternative savings.
Psychological Factors of Voluntary Retirement Saving (August 2015) Extended Abstract 4 Discussion Our results show whether and how much people voluntarily save for their retirement is largely determined by necessary conditions, in particular sufficient income and knowledge about the system, in which one saves. Yet, even if these conditions are met, not everyone takes action and saves for retirement. Individuals with low self-control capacity less often possess a taxexempt voluntary retirement account (Pillar 3a account) and deposit less money. Specifically, individuals with low self-control (i.e. categorized in the first tertile) deposited CHF 977 less than the individuals with high self-control (i.e. categorized in the third tertile), suggesting that many people may have difficulties to forego immediate consumption in favor of saving for their future. This finding is in line with research, suggesting that low self-control capacity is one of the major predictors for suboptimal life outcomes in the financial, academic, and health domain and extends this prediction to the retirement domain (5). Our results furthermore support previous work regarding socio-demographic influences on retirement saving (2). The incidence of having a Pillar 3a account as well as the volume of the deposit increases significantly up to an age of approximately 30 years, where it reaches a stable level (examined up to the age of 45 years). We also find that being female and less educated results in fewer deposits, which is consistent with the opinion of (6), indicating that retirement planning is least pursued by those who need it the most, particularly young, female and economically disadvantaged individuals. In contrast to previous work, highlighting the influence of general financial literacy on retirement saving (4), our study suggests that it may be more important to be knowledgeable about the specifics of the pension system, in which one saves. The effect of pension knowledge on the magnitude of deposits made in 2014 is surprisingly high: The difference in deposits of individuals with a low score in pension knowledge to a high score is comparable to the difference in deposits made, related to an increase in two household income categories. Despite this strong effect, it is difficult to ascertain direct causality of pension system knowledge. It is likely that those who save more have more thoroughly concerned themselves with the pension system and have gained proficiency in pension matters. Besides investigating the factors, influencing if and how much people save on a voluntary basis we wanted to gain insights on how retirement investments are allocated. A majority of 67 % of the respondents saved in a tax-exempt savings account, suggesting that people generally invest conservatively and thus try to avoid investment losses at the cost of rather low expected return rates. Interestingly, this finding seems to reflect their preference for investment risk, as also in the risk attitude measure most people would not invest any money in a risky asset, even if investing the maximum would yield the highest expected value. Thus, in line with a wealth of empirical findings, individuals seem on average risk-averse in the pension domain. Interestingly, we find that women are less likely to invest in funds than men. This difference may be partly explained by more pronounced risk aversion in women (7). In addition to this, women rated themselves on average less proficient in pension matters compared to men, suggesting that women may feel less competent in making retirement decisions and hence opt for the safer investment strategy. In summary, our results show that voluntary retirement saving is strongly influenced by psychological factors, which may open new ways for interventions. For instance, augmented educational efforts and accessibility to information about a specific pension system may help increasing voluntary retirement saving. Furthermore, self-control problems may be outsmarted by introducing automatic enrollment schemes in the Pillar 3a for instance at the beginning of the first job in ones professional life and the promotion of adaptive saving plans, in which for example participants can decide to commit themselves to a series of contribution increases, which coincide with pay raises. Enrollment schemes akin have been implemented with great success in the USA (8) and UK in the recent years.
Psychological Factors of Voluntary Retirement Saving (August 2015) Extended Abstract 5 Methods / References / Acknowledgments Study sample. A sample of 1008 working individuals (49.4 % female, 50.6 % male, 19 45 years, mean age: 36.12 years, SD: 7.2 years, German and Swiss speaking part of Switzerland), which is representative for Swiss Internet users was randomly drawn from the online panel of the survey institute LINK, Institut für Markt- und Sozialforschung. Socio-demographic measures. Our survey assessed household income and employment status, gender, age, marital status, number of children, education level, the number of family members that contribute to household income and the employment status (self-employed vs. generally employed, part-time employed or full-time employed). Financial literacy. We used a widely used measure of financial literacy (9), determining the proficiency of a respondent for solving basic mathematical calculations. We extended this common measure of financial literacy that focuses on respondents numerical ability, by assessing respondents basic understanding of economic mechanisms using two questions from (10). Pension system knowledge / self-assessed expertise. To measure participants knowledge about the Swiss pension system we asked six increasingly difficult questions regarding the Swiss pension system. In advance to test the knowledge about the Swiss pension system, participants self-assessed their expertise in the Swiss pension system on a 10-grade scale from considering oneself as an absolute layperson to an expert. Self control measure. Similar to other experimental measures of temporal discounting (see, e.g., 11), individuals were asked to decide on a series of hypothetical monetary amounts that makes them indifferent between receiving the amount early versus waiting x months for a fixed larger monetary amount. The answers to these questions made it possible to calculate individual discount factors, reflecting the extent of preference for smaller sooner rewards against larger later rewards. This preference has been commonly used as an indicator for impulsivity and the inverse as a measure for self-control capacity (12). Risk attitude measure. We used an approach to assess risk attitudes introduced by (13). In this task, a risky choice is framed as an investment decision, similar as in asset allocation decisions in retirement saving. Subjects had to decide how to allocate a hypothetical endowment of CHF 100 000 between a safe account and a risky investment that, with 50% probability, yields 2.5 times the amount invested, zero otherwise. In other words, subjects had to choose an amount k to invest in a risky way. In this investment game the choice of a larger fraction to be invested implies a change in the amount of money at stake, while the probabilities are not affected. Since the expected value of the task, equals to 4 + 0.25k, is higher than one for any k different from zero, a risk-neutral subject should invest all the endowment. Statistical analyses. The analyses for the three main dependent variables (incidence of a Pillar 3a account, deposit made in 2014 and allocation type) followed a three-step procedure: First, we ran regressions to infer the influence of socio-demographic variables, including the hypothesized pre-conditional factor of household income on the variable of interest (logistic regressions for the binary variables of interest and a linear regression for the deposit in 2014 variable). We then controlled for these variables and estimated the effects of the other suggested pre-conditional factors, financial-literacy as well as knowledge and self-assessed expertise about the pension system on voluntary retirement saving. In a third step, while controlling for all pre-conditional variables and sociodemographic variables, we turned our analysis to examining the influence of self-control capacity and risk attitude. References (1) Antolin P, De Serres A, & De la Maisonneuve C (2004) OECD Economics Department Working Papers (2) Fernandéz-López S, Otero L, Rodeiro D, & Vivel M (2010) Czech Journal of Economics and Finance (3) Lusardi A & Mitchell OS (2011) Journal of pension economics and finance (4) Brown M & Graf R (2013) Numeracy (5) Moffitt TE, et al. (2011) Proceedings of the National Academy of Sciences (6) Hayes CL & Parker M (1993) Journal of Women & Aging (7) Jianakoplos NA & Bernasek A (1998) Economic inquiry (8) Benartzi S & Thaler R, (2013) Science (9) Banks J & Oldfield Z (2007) Fiscal studies (10) Lusardi A & Mitchell OS (2009) NBER Working Paper (11) Meier S & Sprenger C (2010) American Economic Journal: Applied Economics (12) Madden GJ, Begotka AM, Raiff BR, & Kastern LL (2003) Experimental and Clinical Psychopharmacology (13) Gneezy U & Potters J (1997) Quarterly Journal of Economics Acknowledgments This study was funded by UBS Switzerland AG. The authors declare no competing financial interests.