Financial Literacy and Pension Expectations:
Adjustments to Reforms in the Netherlands
a, bRemko Struik*
13th of January, 2016
Abstract
The Dutch pension system was reformed between 2010 and 2015. By fielding a new survey on financial literacy and pension expectations in October 2015, repeating the surveys from 2010 by Alessie et al. (2011), we are able to study the changes in the level of financial literacy of the Dutch population and the adjustment in pension expectations of Dutch employees. We find that the level of financial literacy of the Dutch population did not significantly increase between 2010 and 2015. Women, the young and the least educated still show the lowest levels of financial literacy. Of the Dutch employees, the less financially literate and women are less likely to form pension expectations. Compared to 2010, the youngest born after 1975 and the least educated have a higher likelihood of reporting ‘do not know’ in 2015. This implies these groups have become more uncertain. Dutch employees expect to retire later in 2015 compared to 2010. Findings suggest that the link of life expectancy to the retirement age is only partly incorporated in retirement expectations in 2015. Dutch employees expect a lower replacement rate in 2015 compared to 2010. The adjustment is the strongest for the less financially literate. While the more financially literate tend to report a wider range between the minimum and maximum expected replacement rate, the less financially literate have a higher probability to report ‘do not know’. The latter effect is strengthened from 2010 to 2015.
Keywords: Financial literacy, Retirement Expectations, Uncertainty JEL Classifications: D80, H55
a This paper is submitted as a thesis for the MSc Economics and for the MSc Finance for the Faculty of Economics
and Business of the University of Groningen (RuG)
b I am grateful to Maarten van Rooij of the Economic and Research Division of De Nederlandsche Bank and Rob
Alessie of the University of Groningen for their input and support during the process of writing this paper.
* The author can be contacted by email: r.struik@gmail.com; student number: s1984411. Supervisors of this thesis
1
1. Introduction
Up until the late ‘90s the profits made by pension funds on their investments were substantial due to the good performance of the stock markets around the world. During those times, pension premiums were declining, and participants could expect indexation of their pension income year after year. The Dutch pension system used to be generous, but appeared to be unsustainable. One of the reasons for the unsustainability is the aging of the population and the (underestimated)1 increasing life expectancy. Additionally, the structural decline in the interest rate has increased the liabilities of the pension funds considerably. Furthermore, throughout the 2000s, the stock market endured large fluctuations during the dot-com bubble and the 2008 credit crunch, followed by the 2010 sovereign debt crisis. Dutch pension funds have been applying incomplete indexation and pension benefit cuts depending on the coverage ratio of pension funds. Retirement schemes transformed from collective defined benefit (CDB) schemes into collective defined contributions (CDC) schemes. Therefore, the employees instead of employers are exposed to the risk of financial downturns that cause low or negative investment returns affecting pension wealth. The combination of this caused uncertainty whether the obligations of pension funds could be met. In 2010, the Dutch pension system was found to be insufficiently future-proof, according to Commissie Goudswaard (2010). Therefore, a reform of the pension system was necessary.
The Dutch pension system was reformed between 2010 and 2015. The reforms consisted of a gradual increase of the statutory retirement age and a reduction in the pension accrual rates. Both the retirement age and accrual rate are determinants in the replacement rate, which is defined as gross pension income as a percentage of final working salary. The question is whether Dutch employees possess the financial knowledge to incorporate the implications of the reforms in their expectations on the retirement age and the replacement rate. Therefore, this paper studies whether the level of financial literacy in the Netherlands changed between 2010 and 2015. Consecutively, this research investigates to what extent Dutch employees adjusted their retirement expectations after the reforms. Additionally, it relates the retirement expectations of Dutch employees to their level of financial literacy. Is financial literacy essential in forming and adjusting pension expectations? Will the more financially literate adjust their expectations to a different extent? And what other demographic factor play an important role in adjusting pension expectation? This paper
2
is a follow-up of Alessie et al. (2011a). By repeating their 2010 surveys on financial literacy and pension expectation in 2015, special modules of the Dutch Central Bank Household Survey (DHS), we are able to study the adjustments in pension expectations.
The main findings of this paper are the following: the level of financial literacy in the Netherlands did not significantly increase between 2010 and 2015. Consistent with other research (Lusardi and Mitchell, 2008, 2011a, 2014), the young, women, and the ones with low level of education and income are the least financially literate. Compared to 2010, the youngest born after 1975 and the least educated have an increased likelihood of reporting ‘do not know’ in 2015 implying that they have become more uncertain. Overall, Dutch employees expect to retire later and expect to obtain a lower replacement rate in 2015 compared to 2010. The adjustment of the replacement rate is the strongest for the less financially literate. While the more financially literate tend to report a wider range between the minimum and maximum expected replacement rate, the less financially literate have a higher probability to report ‘do not know’. The latter effect is strengthened from 2010 to 2015.
This study contributes to the literature in various ways. First, this research provides a clear description of the pension system and reforms in the Netherlands between 2010 and 2015 and the subsequent effects on the replacement rate. Second, in collecting the same data in both 2010 and 2015, adjustments in pension expectation due to recent reforms is investigated. Third, organisation with the goal of increasing financial literacy can use this research to identify which demographic groups to address.
3
2. Literature review
There is an extensive amount of research in the field of financial literacy in the last decade. According to several studies (Huston, 2009; Remund, 2010), a variety of conceptual definitions has been used. Huston (2009) reviews a wide range of financial literacy measures used in research, and argues that in order to do comparison studies and/or meta-analyses, consistency in the concept and measurement of financial literacy is required. Collaborating teams from various countries introduced the same questions to measure financial literacy that was used in eight different countries. Lusardi and Mitchell (2011a) present an overview of the results of these studies. In order to provide consistency, this study follows the definition of financial literacy by Lusardi and Mitchell (2014):“people’s ability to process economic information and make informed decisions
about financial planning, wealth accumulation, debt, and pensions”. Lusardi and Mitchell (2011a,
2014) provide an overview of the level of financial literacy around the world. According to Lusardi and Mitchell (2011a), financial illiteracy is prevalent in both well-developed and rapidly changing markets. Women are less financially literate than men. The middle-aged are more financially literate than the young and the old. Finally, education is positively related to financial literacy. Van Rooij, Lusardi and Alessie (2011a) argue that education is an imperfect proxy for financial literacy, since education cannot fully account for the effect of financial knowledge. Lusardi and Mitchell (2014) argue that financial literacy has important implications for welfare, since increased financial literacy helps individuals “to maintain a budget, to understand credit, to understand investment vehicles, or to take advantage of our banking system”.
4
(DHS) and find a strong positive relation between financial literacy and household wealth. They examine two channels through which financial literacy might enable wealth accumulation, namely the likelihood of investing in the stock market and the likelihood of planning for retirement and developing a savings plan. Findings show that through both channels financial literacy has strong positive effects in boosting household wealth.
Before individuals act to provide themselves with an adequate pension level by saving more through a savings plan or by investing in the stock market, they need to form expectations. Gustman and Steinmeier (2001) used data from the US Health and Retirement Study (HRS) on older Americans and find that individuals who overestimate their social security and pension benefits are likely to retire later than they planned. Clark, Morril and Allen (2010) used data on 1,500 workers nearing retirement at three large companies and find that workers who believe to be allowed to access Social Security benefits at earlier ages than the eligibility age, expect to retire earlier. Chan and Stevens (2008) combine self-reported and administrative pension data and find that self-reported pension data are significant drivers of retirement behaviour, even after controlling for employer-reported administrative pension data. Having misconception on retirement, such as retirement age and generosity of pension benefits, might have implications for retirement well-being. Research of the AFM (2010) shows that there exists a gap between the actual level of benefits pension funds can provide future retirees with and the level Dutch employees expect to receive. Therefore, it is of great importance to investigate what drives the formation of retirement expectations.
5
their retirement age more directly, and individuals tend to retire at the minimum statutory retirement age. Additionally, both the effects of retirement age reforms and of pension accrual reforms are reflected in the replacement rate. Van Santen, Alessie and Kalwij (2012) use self-reported retirement expectations, such as expected retirement age and expected replacement rate, and relate it to financial literacy while controlling for a wide range of background characteristics. They find that less-educated individuals are, on average, biased towards more optimistic replacement rates. Van Santen (2012) finds that households tend to save more when they are uncertain about their pension income. Mastrogiacomo and Van Ooijen (2014) focus on a possible policy reform in the Dutch housing market and find that policy uncertainty is the sole driver of precautionary saving, which leads to a welfare loss. These results implicate that it is relevant to form pension expectations in order to save adequately. Apart from retirement age expectations, this study examines replacement rate expectations as well.
Besides investigating expectations, it is crucial to examine to what extent individuals adjust their expectations and adapt to pension reforms, since a large number of pension systems are being reformed throughout the world at the moment. Baldini, Mazzaferro and Onofri (2015) investigate the evolution in retirement expectations of Italian workers from 2000 to 2012 using 6 waves of the Survey on Household Income and Wealth (SHIW). They find that workers adapt their expectation to reforms with delays and in an incomplete and confusing fashion. A shift from general overestimation towards underestimation of pension benefits is observed, predominantly since many are unaware that the retirement age will be linked to an increasing life expectancy. They argue that many workers are still not able to form correct pension expectation that due to the recent economic crisis.
6
3. Dutch pension system and recent reforms
3.1 Dutch pension system
The Dutch pension system compiles three pillars. The first pillar is the old-age state pension, the AOW (Algemene Ouderdomswet), in which a person is eligible for social security when reaching the statutory retirement age. Although the retirement age was fixed at 65 from 1957 onwards, it is increasing gradually due to the reforms installed between 2010 and 2015 (see section 3.2.1). The first pillar is characterised by a flat-rate pay-as-you-go (PAYG) monthly social security benefit. AOW entitlements are accrued at a 2 percent accrual rate for each year one resides in the Netherlands, thus full entitlement is obtained after 50 years. The benefit is independent of past contributions or labour earnings. The benefit depends on marital status. A single retiree receives 70 percent of the minimum wage, whereas a married retiree receives 50 percent of the minimum wage. The social security benefits are partly financed by premiums paid by workers and partly by general taxes. The AOW benefit is indexed to the minimum wage growth.
The second pillar is a fully-funded occupational pension system facilitated by employers in which workers are obliged to accrue pension rights based on their labour earnings. Premiums are paid partly by the employer and partly by the employee. The pension benefit from the second pillar depends on various factors, namely on contributions, investment returns, and indexation. Whether the pension rights are indexed depends on the pension funds’ coverage ratio. One can receive pension income from the second pillar at an age different to the statutory retirement age. However, that situation might result in a substantial drop in pension income, as the pillar two pension income bridges the age gap to the statutory retirement age. Furthermore, it is mandatory to convert the pension wealth into an annuity, and consecutively an actuarially fair pension income is computed. Additionally, in the pension ambition formulated by pension funds, the pillar one social security benefit is taken into account. Therefore, the pension age in the second pillar usually coincides with the statutory retirement age.
7
In the third pillar, people can voluntarily participate in individual pension products that are subject to limited fiscal facilitation. However, Mastrogiacomo and Alessie (2011) show that the third pillar is of little importance: only one-third of the Dutch population engages in the third pillar, and the additional pension income obtained is relatively small compared to pension benefits from pillar 1 and 2. For an extensive description of the Dutch pension system, see Reichert (2014).
3.2 Pension reforms 3.2.1 Pension age
8
3.2.2 Pension accrual rate
The so-called Witteveenkader is a Dutch law that is complementary to the law on income tax (Wet op loonbelasting) from 1964, which prescribes how much pension wealth is allowed to be accumulated in a tax-exempted fashion as a percentage of the pension base in pillar 2.2 This Witteveenkader was adjusted twice: the accrual rate was adjusted from 2.25 percent to 2.15 percent in January 2014 and from 2.15 percent to 1.875 percent in January 2015 for the average-salary schemes. Furthermore, from January 2015 onwards, a maximum salary over which an employee can accrue pension rights was set at 100.000 euro. A salary exceeding this limit does not apply to the Witteveenkader.
The increase of the retirement age (see section 3.2.1) and the reduction of the accrual rate have important implications. Firstly, less pension wealth can be accumulated in a tax-exempted manner. Secondly, people have to work longer, and thus, will retire later. One can compensate for the fact that less pension wealth is accumulated because of the reduction of the accrual rate by working longer. The combined effect of these reforms can be captured by the replacement rate. To show this effect, an example is provided. Assume a young employee that starts working at the age of 24 earning an initial gross salary of 24000 euro, and follows a lifecycle income path in which this employee experiences a wage growth in real terms of 3 percent before age 35, 2 percent between age 35 and 44, 1 percent between age 45 and 54, and finally no wage growth from age 55 until retirement (see Figure 4).3 The employee accrues at a rate of 1.875 percent in the 2015 situation and at a rate of 2.25 percent in the 2010 situation. The example includes euros in real terms in order to simplify the computations4. Whereas in the 2010 situation, the employee obtained a gross replacement rate just over 73 percent at age 65, in the 2015 situation this replacement rate is obtained just over age 70 (see Figure 5). Figure 5 shows that a worker can compensate for the reduced accrual rate by working 5 years longer.
2 Tax exemption is two-fold. First, pension wealth is accumulated before income tax is paid. Second, tax on wealth is not applicable on pension wealth (no ‘vermogensrendement heffing’). When receiving the monthly pension benefits, income tax is paid, which is usually lower than the income tax on working salary. 3 Wage profile based on Dutch legislation for defined contribution plans: Artikel 18a, derde lid, onderdeel b, Wet op de Loonbelasting 1964.
9
4. Data
The DNB Household Survey (DHS) is an annual survey fielded among over 2000 households in a panel administered by CentERdata. For this research, we fielded a new module of the DHS in September 2015 on financial literacy and pension expectations. The questions in this module are similar to the questions of two separate modules from 2010 fielded by Alessie et al. (2011a).
The 2010 module on financial literacy was fielded in July in which the household head and the partner were questioned who were 25 years and older. 1665 completed the survey and the response rate was 65.4 percent. The 2010 module on pension expectations was fielded in November in which household members of 16 years and older were questioned, including other than the household head and the partner, such as children, parents (in law) and other relatives. 1854 completed the survey and the response rate was 76.9 percent. As for the retirement expectation questions, people younger than 65 were questioned, conditional on the fact that the respondent did not retire early. Focusing on employees only, 887 filled in the retirement age question and 876 filled in the replacement rate question.
In the module of September 2015, household members of 18 years and older were questioned. The full survey was answered by 2182 respondents and the response rate was 89.6 percent. As the 2010 module on financial literacy only contained household heads and their partners, we take the same selection for 2015 to avoid a selection bias. Hence, 2108 that fall within the selection criteria filled in all three financial literacy questions. As for the retirement expectation questions, people that did not retire were questioned. Focusing on employees younger than 66 only, 950 answered both the retirement age question and the replacement rate question. In this research we focus on respondents with the age of younger than 66, since in 2015 some employees are still required to work while being 65 and will retire when being 66.
10
CentERdata pursues representativeness for the Dutch-speaking population in the panel. Although the panel is representative along various dimensions, exceptions exist with respect to some demographic variables. Therefore they use sample weights in order to make the sample statistics representative in accordance with CBS statistics on, in the case of the modules used in this paper, age, gender, income quartiles and education level. For an extended description of the methodological aspects and representativeness of the DHS, see Teppa and Vis (2012).
5. Financial literacy
5.1 The measure of financial literacy
The 2010 financial literacy module fielded by Alessie et al. (2011a) contained, among others, three financial literacy questions that were developed by Lusardi and Mitchell (2008, 2001a, 2011c) for the 2004 U.S. Health and Retirement Study. The three financial literacy questions are displayed below:
1) Understanding of Interest Rate (Numeracy)
Suppose you had €100 in a savings account the interest rate is 2% per year. After 5 years, how much do you have in the account if you the money to grow? (i) More than €102; (ii) Exactly €102; (iii) Less than €102; (iv) Refusal; (v) Do not know.
2) Understanding of Inflation
Imagine that the interest rate on your savings account was 1% per year and inflation was 2% per year. After 1 year, how much would you be able to buy with the money in this account? (i) More than today; (ii) Exactly the same; (iii) Less than today; (iv) Refusal; (v) Do not know.
3) Understanding of Risk Diversification
11
Based on the questions above, a measure of financial literacy is constructed which is denominated as the total number of correct answers on the financial literacy questions. Thus, the measure can take the values from zero to three.
5.2 Financial literacy across years
The responses to the three financial literacy questions collected from both 2010 and 2015 are shown in table 1 through 4. In 2015, 87 percent of the respondents that answers the interest question correctly, which is slightly more than the percentage of 84.8 that answered this question correctly in 2010 (see Table 1). The percentage that answers incorrectly rose slightly by 1.7 percentage point to 6.9. Whereas in 2010, about 10 percent answered ‘do not know’ or refused to answer, in 2015 this is only 6.2 percent. The percentage of respondents who answers the inflation question correctly rose with 0.2 percentage point to 77.1 percent (see Table 2). The percentage that answers ‘do not know’ or refuses to answer declined from 14.7 to 12 percent. The percentage that answers question incorrectly increased with 2.5 percentage point to 10.9. As for the risk question, the percentage of respondents that answers the risk question correctly in 2015 was 55.7 percent, while this was 51.9 percent in 2010 (see Table 3). Less respondents (5.7 percentage point) answer ‘do not know’ or refuse to answer in 2015 compared to 2010. The percentage of incorrect answers increases with 1.9 percentage point to 15.2 percent. The percentage of respondents that is able to answer the risk diversification question correctly is substantially lower compared to the interest and the inflation question. This phenomenon holds for both 2010 and 2015. Several reasons might cause the low percentage of correct answers on the risk question. Few household do actually hold stocks (see e.g. Alessie, Hochguertel and van Soest, 2002). Additionally, respondents not in charge of finances are included in the sample. Moreover, Alessie et al. (2011a) note that concepts such as “stock mutual funds” are typically not covered in (lower secondary) high school.
12
is maintained between 2010 and 2015, though slightly weakened. Although in 2015, more respondents have been able to answer the individual financial literacy questions (less reported ‘do not know’ and less refusals), both the percentage of respondents that were correct and the percentage of respondents that were incorrect has increased. Despite the increases in the percentage of people that have answered the individual financial literacy questions correctly, Table 5 shows that the level of financial literacy of the Dutch population has not significantly changed between 2010 and 2015 at the 5 percent significance level. Overall, the percentage of individuals with a specific number of questions answered correctly does not significantly differ between 2010 and 2015.
Since we deal with an unbalanced panel, the data is tested whether attrition of the respondents in the 2010 survey is random (see Table 6). As for the interest question and risk question, attrition is random, since the percentages do not significantly differ between years. The inflation question shows to be significantly different for in and out of panel observations between 2010 and 2015, though only at the 10 percent level. The percentages of the total number correct, which includes the inflation question, are not significantly different between 2010 and 2015. Overall, this suggests that attrition is random and not related to the level of financial literacy of the respondent.
Another concern when using a panel, is that people could have learned from the fact that they have had a similar questionnaire before. Table 7 presents a test of this so-called learning effect. As for the interest question and risk question, there is no learning, since the percentages do not significantly differ between 2010 and 2015. In the inflation question the percentages of in and out of panel observations is significantly different at the 1 percent level between 2010 and 2015. However, the percentages of the total number correct, which includes the inflation question, are not significantly different between 2010 and 2015. Overall, this suggests that there is no learning-effect observed.
13
5.3 Financial literacy across demographics
There is quite some variation in the level of financial literacy across demographic variables for the Dutch population in 2015. The following demographic variables are displayed in Table 8, and are discussed below: age, gender, education, socioeconomic status, marital status, income and home ownership.
Financial literacy is significantly related to age. Less young respondents relative to older respondents answers the financial literacy questions correctly. Additionally, the young tend to answer ‘don’t know’ more often. This holds for all financial literacy questions. Consistent with previous research (Bucher-Koenen, Lusardi, Alessie, & Van Rooij, 2012), the gender gap remains large and significantly related to financial literacy. Women are less financially literate compared to men. In every of the three financial literacy questions, the percentage of men that answers the question correctly is significantly higher compared to women. However, interestingly, women answer much more often ‘do not know’. The level of education is positively correlated with financial literacy. Whereas, for instance, about 38 percent of the respondents with lower secondary education answers all three questions correctly, this percentage is more than 77 for respondents with a university degree. Financial literacy is significantly related to socioeconomic status. The self-employed and the employed are most financially literate. The unemployed, housewives and househusbands, and the (partially) disabled are least financially literate. The level of financial literacy increases with income. A significantly higher percentage of individuals answers all financial literacy questions correctly when in the highest income quartile (62.8 percent versus 37.3 percent for the lowest income quartile). Home ownership is positively related to financial literacy. Additionally, the percentage of home owners reporting ‘do not know’ is much lower. For the inflation question, the non-home owners report 2.7 times more ‘do not know’.
14
6. Pension expectations
In the retirement expectation module of 2010, people younger than 65 were questioned, conditional on the fact that the respondent did not retire early. In 2015, people that did not retire were questioned. This includes both people that retired early and retired according to the statutory retirement age, since the statutory retirement age differs across age groups which implies that some are still required to work during the age of 65. The description as well as the further analysis of retirement expectation are based on the two question displayed below:
1) Retirement age expectation At what age do you expect to retire? (i) At age [RET_AGE];
(ii) Do not know.
2) Replacement rate expectation
Imagine you will retire at age [RET_AGE]. Take in mind your total pension income. What do you think your minimum gross retirement income will be, as a percentage of gross final (working) salary? And maximum?
(ia) I expect gross retirement income to be at least [MIN] as a percentage of final salary; (ib) I expect gross retirement income to be at most [MAX] as a percentage of final salary; (ii) Do not know.
15
Although the questionnaire was submitted to employees, unemployed, housewives and househusband, students and self-employed, this research focusses on employees under 66 only. While all residents of the Netherlands accrue pension rights in the first pillar, only employees accrue pension wealth in the second pillar. Additionally, employees will have to make a decision when to retire from their company. The descriptive statistics of the retirement expectations are shown in Table 9.
6.1 Expected retirement age
Table 10 shows that the expected retirement age increased significantly during the period from 2010 to 2015. Whereas 56.7 percent of the employees expected to retire at 65 or earlier in 2010, only 18.3 percent of the employees expects to do so in 2015. Whereas in 2010, a bimodal distribution is observed in which 33 percent of employees expects to retire at age 65 and 28.6 percent at age 67, in 2015 a unimodal distribution in which 47.7 percent of employees expects to retire at age 67. Additionally, in 2015 it is more common for employees to expect to retire later than age 67, namely close to 30 percent, while this percentage was less than 5 percent in 2010.
Also note that a large number of employees have become uncertain at what age they will retire. This is reflected in the increased percentage of employees that answer ‘do not know’ when they are requested to report their expected retirement age. Although in 2010, 11.8 percent was unable to report their expected retirement age, in 2015 this percentage rose to 18.3 percent. Figure 6 is a graphical representation of Table 10. The median of the expected retirement age was 65 for 2010 and 67 for 2015.
6.2 Expected replacement rates
16
Furthermore, in 2015, substantially more employees have become uncertain compared to 2010. In 2015, almost 70 percentage point more employees were unable to report their expected replacement rate. The percentage of employees reporting ‘do not know’ in 2015 is over 45 percent compared to only 27.2 percent in 2010. The fact that relatively more employees are able to report their expected retirement age rather than their replacement rate, is not surprising. In order to come up with a replacement rate, one should make assumption on both future final salary and future pension income. Figure 7 is a graphical representation of Table 11.
7. Econometric analyses pooled regression
7.1 Expected retirement age
The percentage of respondents that is unable to answer their expected retirement age is considerable, namely 11.8 percent in 2010 and 18.3 percent in 2015 (see Table 10). Therefore, we employ a two-step approach in our empirical analysis. First, a linear probability model (LPM) is estimated in which the binary dependent variable takes the value of 1 for the respondent who reported ‘do not know’ on the retirement age question, and zero otherwise. Results of this model are displayed in column 1 through 3 in Table 12. Second, we estimate a model for the expected retirement age, conditional on knowing the retirement age. Results can be found in column 4 through 6 in Table 12.
17
one additional financial literacy question correctly, reduces the probability of being able to report a retirement age with 5.8 percentage point. In the third column, the time interaction effects are added which causes the time dummy to become insignificant, while coefficient remains similar in sign and in magnitude. Note, however, that in 2015 in particular the youngest cohort is significantly more likely to report ‘do not know’ on the retirement age question. Compared to the oldest cohort, the youngest cohort is 15 percentage point (10.4+4.6) more likely to report ‘do not know’ in 2015, while the difference in likelihood between these cohorts was insignificant in 2010. Interestingly, whereas in 2010 the effect of education disappeared by adding financial literacy, in 2015 the effect of tertiary education has become significant. Since the effect of all levels of education relative to lower education show negative signs, it can be interpreted that the least educated have an increased likelihood of being unable to answer the retirement age question in 2015. Compared to respondents with a university degree, the least educated respondents have a 10.6 higher likelihood of reporting ‘do not know’ in 2015, whereas there were no significant differences among different levels of education in 2010. Thus, the data suggests that the lower educated and the young have become more uncertain about their retirement age due to the reform. A probit regression confirms the results of the LPM in the likelihood of answering the retirement age. The signs and magnitudes of coefficients of the LPM are similar to the marginal effects obtained from the probit model.
18
but this does not alter the signs and significance of the coefficients of the various variables. This suggests that, apart from an overall increase in the retirement age expectation as a result of the reforms, no adjustment of expectations is observed among different demographic groups between 2010 and 2015.
In 2010, the expectation of the youngest cohort is to retire 1.5 years later relative to the oldest cohort, and this expectation does not significantly differ in 2015. The difference between the youngest and the oldest cohort in 2010 reflected the retirement plans at the time, while not incorporating the link to life expectancy. So ex ante, we might have expected an effect in 2015 due to the link of the retirement age to the life expectancy. To illustrate, a person from the youngest cohort who expects to retire the latest has the following characteristics: having a university degree, being in the second income quartile, and displaying the highest level of financial literacy (3 financial literacy question correct). An individual with the abovementioned characteristics expects to retire at an age of 68.34 in 2015. The realistically expected statutory retirement age in 2015 for a person born after 1975 is at least 69.75 according to CBS life expectancy projection (see Figure 3). This implies that even the ones with the highest expected retirement age from the youngest cohort expect to retire earlier than the realistically expected statutory retirement age, namely at least 1.36 years. Since this cohort has a relatively large likelihood of being unable to answer the retirement age question, this result needs to be interpreted with caution. Although the link of the retirement age to the life expectancy was already heavily debated in 2010 (see section 3.2.1), it was not incorporated in the expectations in 2010 and only partially incorporated in the expectations in 2015.
7.2 Expected replacement rates
19
for the range of the replacement rate. The results of the model for the expected replacement rate, conditional on knowing the replacement rate, are displayed in column 4 through 7 in Table 13. The results of the model for the range of the expected replacement rate, conditional on knowing the replacement rate, are displayed in column 8 through 11 in Table 13.
20
question. The signs and magnitudes of coefficients of the LPM are similar to the marginal effects obtained from the probit model.
21
point lower replacement rate than the least financially literate in 2010. In 2015, there is no significant effect of financially literacy on replacement rate expectation (-1.511+2.214=0.703 which is not significantly different from zero). Thus, while in 2010 the more financially literate were more pessimistic on their expected replacement rate, the less financially literate have reduced their replacement rate expectations stronger than the more financially literate in 2015. Turning back to the year of birth categories, interestingly, both in 2010 and 2015 respondents born between 1966 and 1975 expect a lower replacement rate relative to the oldest cohort. There are no significant adjustments of the expected replacement rates from 2010 to 2015 observed among cohorts. Overall, in 2015 respondents have reduced their replacement rate expectations and the adjustment is the strongest for the less financially literate, while there is no significant difference in adjustment in expectations among age cohorts.
22
Note, however, the likelihood of being unable to answer the replacement rate question, and consequently having no value for the range, is higher for a less financially literate person. In the model with interaction effects (see column 11), the effect of education disappears. Finally, the younger the respondent, the higher the uncertainty in terms of a wider reported range.
8. Conclusion
In the period between 2010 and 2015, the statutory retirement age has been gradually increased and from 2021 onwards is linked to life expectancy. Additionally, the pension accrual rate has been reduced. This paper is a follow-up of Alessie et al. (2011a). By repeating their 2010 surveys on financial literacy and pension expectation in 2015, we are able to study the adjustments in pension expectations. Our empirical results show that Dutch employees find it substantially more difficult in 2015 to form retirement expectations. Moreover, financial literacy shows to be a significant determinant in forming retirement expectations. Our findings shows that the level of financial literacy of the Dutch-speaking population did not significantly increase between 2010 and 2015. Consistent with previous research (Lusardi and Mitchell, 2011, 2014, and Alessie et al., 2011a), the young, the women, the lower educated, the housewives and househusbands, the (partially) disabled, the low income, and the non-home owners remain the least financially literate.
23
relatively large likelihood of reporting ‘do not know’ on retirement age question. However, it suggests that the link of the life expectancy to the retirement age due to the reforms is only partly incorporated in their pension expectation. An overall increase is observed in the retirement age expectation after the reforms were installed.
A larger share of the Dutch employees has reported ‘do not know’ on their replacement rate in 2015 compared to 2010. The more financially literate have a reduced probability of being unable to report an expected replacement rate. This effect is strengthened in 2015. While in 2010 there was no difference in the likelihood of reporting a replacement rate among year of birth categories, the youngest cohort (born after 1975) has become significantly more likely to report ‘do not know’ in 2015 compared to the oldest cohort. Thus uncertainty increased especially for the less financially literate and the youngest in 2015. Finding from Alessie et al. (2011a) suggested that many Dutch employees held overly optimistic replacement rate expectation in 2010, especially the less financially literate. Overall, many employees have reduced their replacement rate expectations in 2015 compared to 2010. There is no significant adjustment in expectations among age cohorts. However, the adjustment is the strongest for the less financially literate, whereas the more financially literate already held lower replacement rate expectations. Another measure of uncertainty is the range of the expected replacement rate. The more financially literate or the younger the respondent, the wider the range that is reported. Thus young and the more financially literate are more uncertain. This holds in both 2010 and 2015. We find that respondents in general have become less uncertain in 2015, indicated by a decreased range, irrespectively of demographic background.
Lusardi and Mitchell (2014) argue that financial literacy has far-reaching economic consequences. The more financial literate are better equipped to complex financial decisions and are more likely to plan for retirement. Organisation with the goal of increasing financial literacy can use this research to identify which demographic groups to address.
24
the formation of replacement rate expectations, since Guise et al. argues that replacement rate uncertainty is more important retirement age uncertainty.
25
9. References
Alessie, R., Van Rooij, M., & Lusardi, A. (2011a). Financial literacy, retirement preparation and pension expectations in the Netherlands. NBER Working Paper, 17109.
Arrondel, L., Debbich, M., & Davignac, F. (2013). Financial Literacy and Financial Planning in France. 6(2 Art. 8).
Autoriteit Financiële Markten (AFM). (2010). Geef Nederlanders pensioeninzicht, werken aan
vertrouwen door dichten van de verwachtingskloof. Den Haag.
Baldini, M., Mazzaferro, C., & Onofri, P. (2015). Pension Expectations and Reality. What do Italian Workers Know About Their Future Public Pension Benefits? Working Paper DSE
No. 1007.
Beer, J. d. (2013). Een levensduur van meer dan 100 jaar: van uitzondering naar regel? NIDI-Webartikel 2013-02, September 2013. Retrieved from www.nidi.nl/nidi-webart-2013-02
Bucher-Koenen, T., Lusardi, A., Alessie, R., & Van Rooij, M. (2012). How Financially Literate are Women? Some New Perspectives on Gender Gap. Netspar Panel Papers, 31.
Chan, S., & Stevens, A. H. (2008). What You Don't Know Can't Help You: Pension Knowledge and Retirement Decision-Making. The Review of Economics and Statistics, 90(2), 253-266.
Clark, R., Morril, M. S., & Allen, S. G. (2010). Th Role of Financial Literacy in Determining Retirement Plans. NBER Working Paper No. 16612.
Commissie Goudswaard. (2010). Een sterkte tweede pijler: Naar een toekomstbestendig stelsel
van aanvullende pensioenen. Den Haag.
Guiso, L., Jappelli, T., & Padula, M. (2013). Pension Wealth Uncertainty. The Journal of Risk
and Insurance, 80(4), 1057-1085.
Gustman, A. L., & Steinmeier, T. L. (2001). Imperfect Knowledge, Retirement and Saving.
26
Lusardi, A., & Mitchell, O. S. (2008). Planning and Financial Literacy: How Do Women fare?
American Economic Review, 98(2), 413-17.
Lusardi, A., & Mitchell, O. S. (2011a). Financial literacy around the world: an overview. Journal
of Pension Economics and Finance, 10(4), 497-508.
Lusardi, A., & Mitchell, O. S. (2011b). Financial Literacy and Planning: Implications for Retirement Wellbeing. Financial literacy: Implications for Retirement Security and the
Financial Marketplace, 1, 17-39. Oxford and New York: Oxford University Press.
Lusardi, A., & Mitchell, O. S. (2014). The Economic Importance of Financial Literacy: Theory and Evidence. Journal of Economic Literature, 52(1), 5-44.
Mastrogiacomo, M., & Alessie, R. (2011). Did you really Save so little for your retirement? An analysis of retirement savings and unconventional retirement accounts. Netspar
Discussion Paper, No. 12/2011-94.
Mastrogiacomo, M., & Van Ooijen, R. (2014). Policy Uncertainty and Precautionary Savings: Does a Possible Reduction of the Mortgage Interest Deduction Increase Savings in the Netherlands? Netspar Academic Series, DP 10/2014-91.
Parlevliet, J. (2015). What Drives Public Acceptance of Reforms? Longitudinal Evidence from the Run-Up of the Increase of the Dutch Retirement Age. DNB Working Paper, No. 492.
Reichert, S. J. (2014). The Dutch Pension System, an overview of the key aspects. Brussels,
Belgium: the Dutch Association of Industry-wide Pension Funds and the Dutch association of Company Pension Funds.
Remund, D. L. (2010). Financial Literacy Explicated: The Case for a Clearer Definition in an Increasingly Complex Economy. The Journal of Consumer Affairs, 44(2), 276-295.
Teppa, F., & Vis, C. (2012). The CentERpanel and the DNB Household Survey: Methodological Aspects. DNB Occasional Studies, Vol.10/No.4.
Van der Wiel, K. (2009). Essays on Expectations, Power and Social Security. Dissertation,
27
Van Duin, C., & Stoeldraijer, L. (2012). Bevolkingsprognose 2012-2060: langer leven, langer
werken. Statistics Netherlands (CBS), December, The Hague/Heerlen.
Van Rooij, M., Lusardi, A., & Alessie, R. (2011a). Financial Literacy and Stock Market Participation. Journal of Financial Economics, 101, 449-472.
Van Rooij, M., Lusardi, A., & Alessie, R. (2011b). Financial Literacy, Retirement Planning, and Household Wealth. CFS Working Paper, No. 2011/21.
Van Santen, P. (2012). Uncertain Pension Income and Household Saving. Netspar Discussion
Paper Series, DP 10/2012-034.
Van Santen, P., Alessie, R., & Kalwij, A. (2012). Probabilistic Survey Questions and Incorrect Answers: Retirement Income Replacement Rates. Journal of Economic Behavior &
28
10. Appendix A.
Computation of pension income
1. The so-called AOW franchise is computed from the level of annual AOW benefit, which is multiplied by the franchise multiplication factor (see Panel A). By using this AOW franchise, the fact that pension benefits consist of both AOW benefit and pillar 2 accrual is taken into account. 2. In the computation of the accrual of pension rights, the AOW franchise is
subtracted from pensionable income, which is defined by gross annual salary including 13th month and/or holiday pay (see Panel B). This results in the pension base.
3. A pension accrual rate is applied on the pension base which results in an annual pension accrual in pillar 2. All annual pension accruals during working life amount to the pillar 2 annual benefit after being retired.* 4. The AOW benefit plus the pension accrual in pillar 2 form the total pension
income of a retiree. Pillar 3 pension products can increase the pension income of a retiree.
29 Panel A. Franchise computation
in 2015 euros 2015 situation 2010 situation Annual AOW* € 9,482 € 9,482
Franchise multiplication factor 100/75 x 100/70 x
AOW Franchise € 12,642
€ 13,546 * In the case of a married person for which the annual AOW is 50% of minimum wage, equal to €9,482.
Panel B. Pension income computation in 2015 euros 2015 situation 2010 situation Monthly Salary € 2,500 € 2,500 12 x 12 x
Gross annual salary € 30,000 € 30,000 13th month € 2,500 + € 2,500 + Pensionable Income € 32,500 € 32,500 AOW Franchise € 12,642 - € 13,546 - Pension Base € 19,858 € 18,954
Annual pension accrual € 372 € 426
30
11. Appendix B. Graphs
Figure 1. Type of DB scheme (as % of total DB schemes)
Figure 2. Statutory retirement age by date of birth (2013-2025)
0 20 40 60 80 100
Final Salary Scheme Average Salary Scheme Combination Other
31
Figure 3. Statutory retirement age by date of birth (2013-2060)
Figure 4. Lifecycle income path in 2015 euros
32
Figure 5. Replacement rate lifetime development average salary scheme (in terms of final salary including full AOW)
Figure 6. Expected retirement age or earlier (in cumulative percentages)
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 P er cen tag e Age Difference
Replacement rate 2015 sitation Replacement rate 2010 situation Age 65
Break even replacement
0 10 20 30 40 50 60 70 80 90 100 P er ce n teg ae
Expected retirement age
33
Figure 7. Expected replacement rate or less (in cumulative percentages)
0 10 20 30 40 50 60 70 80 90 100 P er ce n tag e
Expected replacement rate
34
12. Appendix C. Tables
Table 1. Interest question
Unweighted frequencies, weighted percentages
2015 2010
Whole Sample Age 25-65 Whole Sample Age 25-65
Frequency % Frequency % Frequency % Frequency %
More than 102 euro 1907 86.97 1291 86.93 1465 84.83 1121 85.54
Exactly 102 euro 80 4.64 55 4.65 45 3.44 29 3.01
Less than 102 euro 38 2.23 22 2.09 22 1.74 20 2.11
Refuse 8 0.54 7 0.63 15 1.08 12 1.08
Do not know 75 5.61 43 5.70 118 8.90 85 8.26
Total 2108 100 1418 100 1665 100 1267 100
Table 2. Inflation question
Unweighted frequencies, weighted percentages
2015 2010
Whole Sample Age 25-65 Whole Sample Age 25-65
Frequency % Frequency % Frequency % Frequency %
More 58 3.78 43 4.25 36 2.74 32 3.09
Exactly the same 124 7.12 83 6.55 86 5.65 68 5.77
Less 1761 77.06 1168 75.66 1361 76.86 1026 75.90
Refuse 9 0.63 8 0.74 16 1.20 13 1.24
Do not know 156 11.40 116 12.80 166 13.54 128 14.01
Total 2108 100 1418 100 1665 100 1267 100
Table 3. Risk question
Unweighted frequencies, weighted percentages
2015 2010
Whole Sample Age 25-65 Whole Sample Age 25-65
Frequency % Frequency % Frequency % Frequency %
Incorrect 'true' 319 15.23 191 13.76 201 13.32 131 11.98 Correct 'false' 1266 55.69 885 56.73 939 51.91 737 53.29
Refuse 7 0.56 5 0.63 21 1.57 20 1.87
Do not know 516 28.51 337 28.88 504 33.20 379 32.86
Total 2108 100 1418 100 1665 100 1267 100
Table 4. Answers across questions
Unweighted frequencies, weighted percentages
2015 2010
Whole Sample Age 25-65 Whole Sample Age 25-65
Frequency. % Frequency. % Frequency. % Frequency. % Interest and inflation correct 1671 72.24 1115 71.20 1313 73.36 994 73.11 All correct 1118 47.17 780 48.41 836 44.83 655 46.18
No correct 88 6.59 60 7.11 136 10.46 102 10.45
At least one answer ‘dk/refuse' 563 31.75 370 32.40 560 37.60 422 37.25
35 Table 5. Financial literacy across years
Weighted percentages
Whole sample
Interest Inflation Risk Number correct
wave N correct incorrect correct incorrect correct incorrect 0 1 2 3 2010 1665 84.83 15.17 76.86 23.14 51.91 48.09 10.46 10.30 34.41 44.83 2015 2108 86.97 13.03 77.06 22.94 55.69 44.31 6.59 14.25 31.99 47.17 F-statistic 1.566 0.0102 3.214 1.873 P-value 0.211 0.920 0.073 0.171 Table 6. Attrition in 2010 Weighted percentages
Interest Inflation Risk Number correct
N correct incorrect correct incorrect correct incorrect 0 1 2 3 respondent in 2010 only 752 84.87 15.13 74.26 25.74 51.00 49 10.83 11.92 33.53 43.72 respondent in 2010 & 2015 913 84.80 15.2 79.26 20.74 52.75 47.25 10.12 8.80 35.22 45.86 Total in 2010 1665 84.83 15.17 76.86 23.14 51.91 48.09 10.46 10.3 34.41 44.83 F-statistic 0.000636 3.19 0.331 1.081 p-value 0.980 0.0743 0.565 0.299 Table 7. Learning in 2015 Weighted percentages
Interest Inflation Risk Number correct
N correct incorrect correct incorrect correct incorrect 0 1 2 3 respondent in 2015 only 1195 86.79 13.21 74.43 25.57 55.13 44.87 6.68 16.07 31.47 45.78 respondent in 2010 & 2015 913 87.29 12.71 81.72 18.28 56.69 43.31 6.45 11.03 32.91 49.62 Total in 2015 2108 86.97 13.03 77.06 22.94 55.69 44.31 6.59 14.25 31.99 47.17
F-statistic 0.0469 7.251 0.298 2.432
36
Table 8. Distribution of financial literacy across demographics in 2015
Unweighted frequencies, weighted percentages
Interest Inflation Risk Number Correct
Frequency correct dk correct dk correct dk 0 1 2 3
Age categories <=35 159 83.77 9.71 67.57 19.57 52.46 35.06 8.33 23.95 23.32 44.40 36-50 596 86.94 5.66 74.79 14.16 58.43 27.06 7.76 13.46 29.65 49.13 51-65 663 89.02 4.73 81.92 8.93 57.86 28.28 5.65 10.23 33.78 50.34 >65 690 87.10 5.58 81.84 6.87 52.16 27.56 4.84 12.13 40.10 42.93 Gender Male 1114 89.99 3.70 85.30 5.88 65.90 20.16 4.46 8.98 27.49 59.08 Female 994 84.15 8.44 69.37 17.78 46.16 37.40 8.59 19.17 36.20 36.04 Education
Primary education (basisonderwijs) 84 77.22 10.79 63.01 21.00 41.23 37.82 14.35 17.99 39.51 28.15 Lower secondary (VMBO) 548 84.86 7.19 70.54 13.60 49.59 32.95 7.06 18.80 36.21 37.92 Upper secondary (mbo) 470 83.31 7.65 71.29 15.44 51.57 32.75 8.87 18.63 29.97 42.53 Upper secondary (havo/vwo) 213 94.63 1.51 88.85 6.56 57.85 27.70 1.82 6.10 41.01 51.07 Tertiary (hbo) 539 91.82 4.59 87.70 7.00 65.83 21.48 2.85 8.95 28.19 60.00 Tertiary (university) 254 99.54 0.24 97.58 1.10 78.23 13.93 0.24 1.32 21.28 77.16 Socioeconomic status Employee 924 91.64 3.96 79.69 10.64 60.16 27.14 4.73 12.20 29.93 53.15 Self-employed 119 86.52 2.91 83.98 3.65 68.82 18.88 3.39 9.66 31.18 55.77 Unemployed 133 78.92 14.24 72.21 17.69 57.42 30.46 16.30 8.37 25.81 49.52 Housework 192 79.99 8.66 65.57 16.07 36.15 43.35 7.16 30.32 36.16 26.36 (Early) retiree 645 86.65 5.70 82.46 6.79 56.43 24.09 5.25 10.75 37.22 46.78 (Partially) disabled 95 77.30 11.69 62.32 29.69 42.50 40.28 13.65 22.71 31.51 32.13 Marital status Single 405 86.44 6.89 75.96 17.77 50.20 35.16 8.46 16.22 29.57 45.75 Married, no child 959 85.85 5.17 80.01 8.60 54.85 28.94 5.52 14.48 33.77 46.23 Married, child 649 91.97 4.50 79.20 8.74 60.87 24.14 4.22 12.33 30.63 52.81 Single parent, other 95 71.97 16.03 56.37 24.69 55.26 31.34 16.33 15.03 37.35 31.29
Quartiles monthly net household income
Lowest income quartile 520 79.50 11.28 68.45 19.40 47.98 37.10 11.81 17.69 33.24 37.25 Second income quartile 518 90.33 3.53 80.70 7.37 55.95 26.98 3.70 14.61 32.71 48.99 Third income quartile 519 90.78 3.31 81.18 8.22 56.96 27.88 3.94 12.29 34.68 49.09 Fourth income quartile 519 94.48 1.46 87.64 6.19 68.99 17.18 1.65 8.35 27.25 62.76 Refuse/dk 32 88.91 5.27 66.87 9.19 57.74 25.92 7.57 18.37 27.03 47.03
Dummy: Home ownership
Home owner 501 78.09 10.13 67.92 20.93 44.82 38.94 12.19 19.72 33.18 34.91 Other 1607 91.31 4.21 81.53 7.69 61.01 24.25 3.86 11.57 31.41 53.15
37 Table 9. Descriptive statistics pension expectations of employees under 66
Variable Retirement age Expected replacement rate Expected minimum replacement rate Expected maximum replacement rate Range replacement rate
Mean Median SD Mean Median SD Mean Median SD Mean Median SD Mean Median SD 2010 64.9 65.0 2.8 72.0 72.5 16.0 65.6 70.0 16.4 78.4 80.0 17.4 12.8 10.0 10.9 2015 66.6 67.0 3.2 70.0 70.0 15.5 63.6 67.5 16.1 76.5 75.0 16.5 12.9 10.0 10.0
Table 10. Expected retirement age of employees under 66 Unweighted frequencies, weighted percentages
2015 2010
Age Frequency Percentages Cumulative Frequency Percentages Cumulative
<=60 31 4.13 4.13 58 8.48 8.48 61 4 0.30 4.43 7 0.75 9.23 62 16 2.32 6.75 82 7.87 17.10 63 19 1.86 8.61 50 4.57 21.67 64 12 0.82 9.43 20 1.94 23.61 65 83 8.86 18.29 298 33.04 56.65 66 51 4.24 22.53 87 10.28 66.93 67 398 47.65 70.18 191 28.56 95.49 68 74 10.56 80.74 16 2.25 97.74 69 20 2.89 83.63 0 0.00 97.74 70 85 14.36 97.99 14 1.96 99.70 71 3 0.39 98.38 0 0.00 99.70 72 5 0.62 99.00 0 0.00 99.70 >=75 7 1.00 100.00 2 0.30 100.00 Total 808 100.00 825 100.00 Do know 808 81.73 825 88.20 Do not know 142 18.27 62 11.80 Total 950 100 887 100
Table 11. Replacement rates in categories of employees under 66 Unweighted frequencies, weighted percentages
2015 2010
Replacement rates Frequency Percentages Cumulative Frequency Percentages Cumulative
38
Table 12. Financial literacy and retirement age expectations Employees under 66
Expected retirement age unknown Expected retirement age
Time dummy (1 for 2015) 0.0656*** 0.0665*** 0.0728 1.701*** 1.671*** 1.198*
(0.0141) (0.0145) (0.0650) (0.135) (0.148) (0.674)
Number correct -0.0576*** -0.0628*** 0.132 0.0859
(0.0130) (0.0163) (0.112) (0.139)
Number correct × time dummy 0.00987 0.0962
(0.0225) (0.221)
Year of birth (base: year of birth < 1955)
1956-1965 0.0466** 0.0478** 0.0453* 0.367** 0.332* 0.465** (0.0193) (0.0216) (0.0235) (0.157) (0.169) (0.221) 1966-1975 0.0592*** 0.0585*** 0.0381 0.848*** 0.834*** 0.723** (0.0197) (0.0219) (0.0232) (0.226) (0.247) (0.322) 1976<= 0.134*** 0.111*** 0.0464 1.506*** 1.456*** 1.490*** (0.0258) (0.0272) (0.0371) (0.271) (0.324) (0.460)
Year of birth × time dummy (base: year of birth <= 1955 × time dummy)
1956-1965 × time dummy 0.0139 -0.226 (0.0363) (0.264) 1966-1975 × time dummy 0.0450 0.196 (0.0370) (0.374) 1976<= × time dummy 0.104** -0.0332 (0.0506) (0.565) Female 0.0575*** 0.0481*** 0.0489*** -0.167 -0.0653 -0.0640 (0.0155) (0.0174) (0.0175) (0.168) (0.182) (0.178)
Education (base: lower education (base: primary education + VMBO))
Upper secondary (mbo) -0.0133 0.00245 0.0444 -0.0395 -0.0875 -0.318
(0.0263) (0.0271) (0.0324) (0.254) (0.259) (0.311)
Upper secondary (havo/vwo) 0.0205 0.0492 0.0734* 0.519* 0.341 0.0169
(0.0351) (0.0381) (0.0420) (0.269) (0.269) (0.356)
Tertiary (hbo) -0.0664*** -0.0311 0.0116 0.434* 0.369 0.354
(0.0228) (0.0246) (0.0242) (0.252) (0.247) (0.277)
Tertiary (university) -0.0993*** -0.0514** 0.00627 0.587** 0.390 0.179
(0.0242) (0.0251) (0.0250) (0.296) (0.309) (0.407)
Education × time dummy (base: lower education× time dummy)
Upper secondary (mbo) × time dummy -0.0782 0.438
(0.0500) (0.479)
Upper secondary (havo/vwo) × time dummy -0.0448 0.654
(0.0593) (0.496)
Tertiary (hbo) × time dummy -0.0818* 0.0780
(0.0440) (0.422)
Tertiary (university) × time dummy -0.106** 0.411
(0.0452) (0.554)
Quartile dummies monthly net household income (base: first quartile)
Second income quartile -0.0367 -0.0278 -0.0298 0.204 0.209 0.225
(0.0261) (0.0279) (0.0278) (0.214) (0.241) (0.240)
Third income quartile -0.0451 -0.0320 -0.0343 -0.140 -0.131 -0.120
(0.0276) (0.0293) (0.0291) (0.238) (0.266) (0.261)
Fourth income quartile -0.0711** -0.0708** -0.0733** -0.580** -0.541* -0.542*
(0.0277) (0.0294) (0.0294) (0.265) (0.310) (0.307)
Refuse/dk 0.0216 0.0481 0.0349 -0.613 -0.434 -0.370
(0.0816) (0.0863) (0.0870) (0.836) (0.865) (0.884)
Constant 0.0574* 0.163*** 0.156*** 64.37*** 64.15*** 64.37***
(0.0334) (0.0477) (0.0535) (0.282) (0.414) (0.418)
Marital status Yes Yes Yes Yes Yes Yes
Number of children Yes Yes Yes Yes Yes Yes
Home ownership dummy (1 for home owner) Yes Yes Yes Yes Yes Yes
Observations 1837 1572 1572 1633 1395 1395
R-squared 0.072 0.088 0.092 0.142 0.138 0.141
p-value number correct 0.000 0.000 0.240 0.500
p-value year of birth 0.000 0.000 0.003 0.000 0.000 0.000
p-value year of birth year interaction 0.177 0.710
p-value education 0.000 0.0190 0.054 0.0312 0.169 0.254
p-value education interaction 0.213 0.576
p-value marital status 0.508 0.328 0.274 0.342 0.450 0.494
p-value income 0.103 0.0720 0.068 0.017 0.076 0.073
39
Table 13. Financial literacy and replacement rate expectations Employees under 66
Do not know expected replacement rate Expected replacement rate Range expected replacement rate Time dummy (1 for 2015) 0.172*** 0.200*** 0.362*** -2.931*** -3.673*** -4.277*** -9.818** -1.027** -1.584*** -1.790*** -4.771**
(0.0202) (0.0204) (0.0847) (0.802) (0.910) (0.950) (3.958) (0.512) (0.527) (0.538) (2.137)
Expected retirement age 0.449** 0.577** 0.557** 0.337*** 0.330*** 0.327***
(0.212) (0.226) (0.230) (0.116) (0.123) (0.125)
Number correct -0.149*** -0.107*** -0.717 -1.511** 1.621*** 1.389***
(0.0151) (0.0206) (0.615) (0.682) (0.357) (0.419)
Number correct × time dummy -0.0778*** 2.214* 0.726
(0.0269) (1.341) (0.717)
Year of birth (base: year of birth <= 1955)
1956-1965 0.0585* 0.0375 0.0477 -2.869** -3.034** -2.518* -2.228 2.670*** 2.546*** 2.173*** 1.632** (0.0310) (0.0322) (0.0391) (1.352) (1.354) (1.482) (1.578) (0.601) (0.607) (0.640) (0.726) 1966-1975 0.0615* 0.0586* 0.0338 -3.634*** -4.025*** -3.568*** -3.841*** 5.972*** 5.678*** 5.199*** 5.403*** (0.0317) (0.0332) (0.0391) (1.220) (1.236) (1.317) (1.424) (0.669) (0.692) (0.743) (0.997) 1976<= 0.131*** 0.0702* -0.0210 -1.013 -1.818 -1.325 1.120 8.204*** 7.598*** 7.268*** 7.190*** (0.0358) (0.0375) (0.0487) (1.421) (1.497) (1.673) (2.091) (0.969) (0.955) (1.031) (1.534)
Year of birth × time dummy (base: year of birth < 1955 × time dummy)
1956-1965 × time dummy -0.00124 -0.934 1.196 (0.0567) (2.411) (1.142) 1966-1975 × time dummy 0.0610 -0.0330 -0.273 (0.0573) (2.194) (1.338) 1976<= × time dummy 0.145** -3.950 0.255 (0.0673) (2.849) (2.051) Female 0.155*** 0.114*** 0.108*** -1.756** -1.644* -2.179** -1.975** 0.437 0.521 0.657 0.715 (0.0223) (0.0237) (0.0239) (0.867) (0.861) (0.948) (0.941) (0.554) (0.548) (0.622) (0.615)
Education (base: lower education (primary education + VMBO))
Upper secondary (mbo) -0.0509 -0.0100 -0.0293 1.321 1.358 0.867 0.424 -0.327 -0.299 -0.275 -0.597
(0.0356) (0.0355) (0.0470) (1.394) (1.396) (1.500) (1.646) (0.769) (0.769) (0.793) (1.022)
Upper secondary (havo/vwo) -0.0391 0.0344 0.00325 0.125 -0.0521 0.485 -1.183 2.612** 2.479* 2.217 1.556
(0.0457) (0.0460) (0.0565) (1.946) (1.959) (2.140) (2.226) (1.298) (1.292) (1.374) (1.317)
Tertiary (hbo) -0.129*** -0.0488 -0.0177 0.635 0.474 0.363 -0.122 0.897 0.776 0.713 0.625
(0.0348) (0.0354) (0.0446) (1.360) (1.382) (1.494) (1.858) (0.759) (0.760) (0.813) (1.032)
Tertiary (university) -0.172*** -0.0619 -0.0291 -4.175** -4.338*** -4.471** -4.027** 2.718*** 2.595*** 2.267** 1.293
(0.0385) (0.0401) (0.0487) (1.636) (1.654) (1.752) (1.985) (0.961) (0.964) (1.043) (1.247)
Education × time dummy (base: lower education× time dummy)
Upper secondary (mbo) × time dummy 0.0189 0.968 0.786
(0.0645) (2.441) (1.483)
Upper secondary (havo/vwo) × time dummy 0.0490 3.784 1.577
(0.0801) (4.148) (2.750)
Tertiary (hbo) × time dummy -0.0571 1.065 0.304
(0.0601) (2.347) (1.504)
Tertiary (university) × time dummy -0.0557 -0.705 1.925
40
Table 13 Continued. Financial literacy and replacement rate expectations
Marital status (base: single)
Married, no child 0.0503 0.0626* 0.0670* -0.905 -0.670 -0.926 -1.033 -0.330 -0.154 -0.840 -0.875
(0.0372) (0.0369) (0.0368) (1.694) (1.669) (1.796) (1.813) (0.882) (0.871) (0.917) (0.921)
Married, child 0.0476 0.0615 0.0622 1.090 1.133 1.546 1.417 -1.770 -1.738 -1.580 -1.604
(0.0526) (0.0522) (0.0522) (2.170) (2.164) (2.377) (2.420) (1.185) (1.174) (1.219) (1.222)
Single parent, other 0.0907 0.0532 0.0519 1.921 1.746 3.653* 3.567 -2.883** -3.015** -3.241** -3.167**
(0.0569) (0.0635) (0.0634) (2.540) (2.549) (2.199) (2.208) (1.334) (1.318) (1.464) (1.479)
Number of children 0.00770 -0.00109 5.49e-05 -0.244 -0.251 -0.659 -0.581 0.387 0.382 0.338 0.365
(0.0186) (0.0196) (0.0196) (0.687) (0.687) (0.732) (0.742) (0.412) (0.406) (0.410) (0.411)
Quartile dummies monthly net household income (base: first quartile)
Second income quartile -0.108*** -0.0898** -0.0987** -0.293 -0.414 -1.254 -1.007 0.554 0.463 0.497 0.632
(0.0381) (0.0386) (0.0385) (1.556) (1.545) (1.741) (1.743) (0.851) (0.847) (0.887) (0.900)
Third income quartile -0.116*** -0.110*** -0.121*** -0.925 -0.944 -0.233 -0.0498 0.826 0.811 0.580 0.698
(0.0401) (0.0400) (0.0400) (1.809) (1.799) (1.930) (1.947) (0.910) (0.909) (0.918) (0.928)
Fourth income quartile -0.185*** -0.175*** -0.182*** -1.702 -1.543 -1.035 -0.907 1.328 1.448 1.427 1.530
(0.0416) (0.0423) (0.0422) (1.842) (1.841) (1.918) (1.928) (0.979) (0.973) (0.986) (1.005)
Refuse/dk 0.127 0.127 0.119 2.524 3.072 -0.309 0.0604 -0.446 -0.0338 0.682 0.870
(0.112) (0.124) (0.124) (4.411) (4.608) (4.568) (4.818) (2.187) (2.338) (2.482) (2.321)
Dummy for home ownership (base: no home ownership)
Home ownership -0.0402 -0.00514 -0.00132 -0.911 -0.958 -0.830 -0.933 0.154 0.119 -0.214 -0.291 (0.0325) (0.0326) (0.0324) (1.249) (1.250) (1.278) (1.294) (0.733) (0.738) (0.767) (0.766) Constant 0.260*** 0.530*** 0.443*** 77.67*** 48.85*** 42.66*** 45.84*** 7.895*** -13.77* -16.44** -15.30* (0.0465) (0.0564) (0.0674) (1.935) (13.45) (14.32) (14.83) (0.968) (7.480) (7.996) (8.083) Observations 1826 1565 1565 1269 1269 1090 1090 1269 1269 1090 1090 R-squared 0.128 0.199 0.208 0.050 0.056 0.063 0.069 0.127 0.135 0.150 0.154
p-value number correct 0.000 0.069 0.000
p-value year of birth 0.004 0.235 0.141 0.013 0.009 0.042 0.063 0.000 0.000 0.000 0.000
p-value year of birth year interaction 0.086 0.450 0.620
p-value education 0.000 0.186 0.422 0.001 0.001 0.004 0.035 0.003 0.006 0.040 0.198
p-value education interaction 0.450 0.790 0.797
p-value marital status 0.353 0.391 0.335 0.646 0.732 0.164 0.161 0.159 0.110 0.175 0.195
p-value income 0.000 0.000 0.000 0.679 0.740 0.878 0.896 0.652 0.600 0.644 0.628
41 \ Table of content 1. Introduction 1 2. Literature review 3
3. Dutch pension system and recent reforms 6
3.1 Dutch pension system 6
3.2 Pension reforms 7
3.2.1 Pension age 7
3.2.2 Pension accrual rate 8
4. Data 9
5. Financial literacy 10
5.1 The measure of financial literacy 10
5.2 Financial literacy across years 11
5.3 Financial literacy across demographics 13
6. Pension expectations 14
6.1 Expected retirement age 15
6.2 Expected replacement rates 15
7. Econometric analyses pooled regression 16
7.1 Expected retirement age 16
7.2 Expected replacement rates 18
8. Conclusion 22
9. References 25
10. Appendix A. 28
11. Appendix B. Graphs 30
42 List of figures:
Figure 1. Type of DB scheme (as % of total DB schemes) 30
Figure 2. Statutory retirement age by date of birth (2013-2025) 30 Figure 3. Statutory retirement age by date of birth (2013-2060) 31
Figure 4. Lifecycle income path in 2015 euros 31
Figure 5. Replacement rate lifetime development average salary scheme 32 Figure 6. Expected retirement age or earlier (in cumulative percentages) 32 Figure 7. Expected replacement rate or less (in cumulative percentages) 33
List of tables:
Table 1. Interest question 34
Table 2. Inflation question 34
Table 3. Risk question 34
Table 4. Answers across questions 34
Table 5. Financial literacy across years 35
Table 6. Attrition in 2010 35
Table 7. Learning in 2015 35
Table 8. Distribution of financial literacy across demographics in 2015 36 Table 9. Descriptive statistics pension expectations of employees under 66 37
Table 10. Expected retirement age of employees under 66 37
Table 11. Replacement rates in categories of employees under 66 37
Table 12. Financial literacy and retirement age expectations 38
