4.1 How much do individuals know?

We measure financial literacy by using the three questions that were first proposed by Lusardi and Mitchell (Reference Lusardi, Mitchell, Lusardi and Mitchell2011) and added to the 2004 U.S. Health and Retirement Study. The first two questions are rather basic and measure respondents' ability to perform simple calculations and understand the effect of inflation. To be able to classify respondents according to different levels of financial sophistication, a third and more complicated question was added to the module. This question measures understanding of risk diversification. The precise wording of the questions is as follows:

1. 1. Understanding of Interest Rate (Numeracy). Suppose you had €100 in a savings account and the interest rate was 2% per year. After 5 years, how much do you think you would have in the account if you left the money to grow? (i) More than €102; (ii) Exactly €102; (iii) Less than €102; (iv) Do not know; (v) Refusal.

2. 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) Do not know; (v) Refusal.

3. 3. Understanding of Risk Diversification. Do you think that the following statement is true or false? Buying a company stock usually provides a safer return than a stock mutual fund. (i) True; (ii) False; (iii) Do not know; (iv) Refusal.

Van Rooij et al. (Reference Van Rooij, Lusardi and Alessie2011a) designed a financial literacy module for the DHS in 2005, which contained the three questions presented above, in addition to other questions. To assess the relevance of the wording of the questions, a randomly chosen group was exposed to the risk diversification question with inverted wording. The pattern of responses changed dramatically when the order of the wording was inverted. The number of correct answers was very low when respondents were asked whether ‘buying a stock mutual fund usually provides a safer return than a company stock’ but doubled when respondents were asked whether ‘buying a company stock usually provides a safer return than a stock mutual fund’. For the current study, in order to keep the data for the Netherlands as comparable as possible with the data collected by the other countries participating in the international comparison, we did not do any randomization. However, the results of Van Rooij et al. (Reference Van Rooij, Lusardi and Alessie2011a) show that there are considerable measurement (classification) errors in our financial literacy variables. We will address the problem of measurement error when we assess the effect of financial literacy on retirement planning.

Responses to the three financial literacy questions asked of participants in the 2010 survey are reported in Tables 1a–c. Most respondents (84.8%) have at least some idea about interest rate calculations, with the percentage of incorrect answers at only 5.2% (Table 1a). About 10% of individuals refuse or do not know how to respond to this question. About 77% of respondents answered the inflation question correctly and about 11% responded incorrectly (Table 1b). To answer the second question correctly, individuals need to have some basic understanding of the concept of inflation and its impact on purchasing power. Note that well before the start of the Economic and Monetary Union (EMU) in 1999, the Dutch Central Bank closely followed the strict monetary policy of the German Bundesbank. As a result, inflation has been quite low since the mid-1980s and the Dutch population has not experienced periods of high inflation.

Table 1a. Interest question (weighted percentages)

Table 1b. Inflation question (weighted percentages)

The proportion of correct answers decreases considerably, to a little more than 50%, when we consider the more complex question on risk diversification (Table 1c). A sizable fraction (35%) of respondents either refuses to or is not able to answer the risk diversification question. In order to interpret the low percentage of correct answers, one should realize that the questionnaire is representative of the Dutch population aged 25 years and older on an individual level, not on the household level. The sample contains respondents who are the primary household financial decision makers and respondents whose partner is in charge of the household finances. Moreover, about three-quarter of Dutch households neither holds company stocks nor stock mutual funds. Furthermore, concepts like ‘stock mutual funds’ are typically not covered in (lower secondary) high schools.

Table 1c. Risk question (weighted percentages)

Given the low percentage of correct answers to the risk diversification question, it is not surprising that only 45% of respondents answered all questions correctly (Table 1d). A considerable fraction of respondents (73%), however, provided the correct answer to both the interest rate and the inflation questions, and we thus find a strong positive association between the ability of respondents to correctly answer the first and second basic literacy questions.

Table 1d. Answers across questions (weighted percentages)

4.2 Who knows the least?

Financial literacy varies widely across demographic variables such as age, gender, education and socioeconomic status (Table 2). Interestingly, the 65 plus cohort – who experienced the period of high inflation in the Netherlands during the 1970s – scores somewhat better on the inflation question than younger respondents, while the reverse is true for the other two questions and the overall score. Overall, however, differences across age are not statistically significant. These results differ from those found in a number of other studies, which find a hump-shaped age profile for financial literacy.

Table 2. Distribution of financial literacy across demographics (weighted percentages)

In line with the results of other studies, we find large gender differences in financial literacy in the Netherlands: women display much lower knowledge than men, and differences are statistically significant. Notice, however, that women do not give many more incorrect answers than men. Instead they state ‘do not know’ much more often. As expected, financial literacy increases strongly and significantly with the level of education. About a third of respondents with primary or lower secondary education answered all literacy questions correctly. Around half of those individuals answered at least one of the three literacy questions with ‘do not know’ or ‘refuse to answer’. Conversely, the majority (70%) of respondents with a university degree gave correct answers to all three literacy questions.

Financial literacy is significantly correlated with socioeconomic status. Self-employed respondents not only have higher literacy scores overall but also provide a correct answer to each literacy question more often than those who are employed, retired or unemployed (other than retirees). This is important, since the self-employed in the Netherlands are in charge of their own retirement savings, while employees typically participate in their employer's mandatory pension plan. Employees seem more financially literate than retirees (including those who have retired early), who, in turn, score much better than do unemployed respondents. The latter group includes those who are unemployed but looking for a job, those who are not able to work and receive a disability benefit, and housewives and househusbands.

4.3 Financial literacy and religious affiliation

Renneboog and Spaenjers (Reference Renneboog and Spaenjers2011) have used the DHS panel to investigate the differences in economic attitudes and financial decisions between religious and non-religious Dutch households. They find the Netherlands to be an important case study for examining the effect of religion on individual financial decision-making for two reasons. First, there is a considerable variety in religious beliefs in the Netherlands: Catholicism, different types of Protestantism, and several other beliefs (e.g., Islam). Nevertheless, Christian religions dominate. An essential difference between Catholics and Protestants is that ‘the former rely on salvation by works with enforcement by the Church and the latter on salvation from divine grace with enforcement from social interaction’ (Arruñada, Reference Arruñada2010). Second, the distinction between religious and non-religious individuals is probably easier to make in the Netherlands than in other countries. Generally, those who claim affiliation with a specific religious denomination also practice their religion, which is not the case in all countries.

Renneboog and Spaenjers (Reference Renneboog and Spaenjers2011) find that religious households (those practicing a religion) consider themselves more trusting, have a stronger bequest motive and – most important for our work – a longer planning horizon. (Given this last finding, one would expect religious individuals to put more emphasis on retirement planning than those who are not religious. We will return to this issue in Section 5 of the paper.) Furthermore, they find that Catholics invest less in the stock market. At the same time, in line with the differences between the two Christian religions, Protestants seem to have a weaker internal locus of control than Catholics (i.e., they feel less able to influence the course of their life), but a higher awareness of individual financial responsibility. Given this last result, one may expect differences in the level of financial literacy across religious groups. As our survey contains information on religion, we are able to explore this relationship. While our religion variable measures affiliation and does not necessarily capture upbringing and/or active practice, we feel we can assume that claiming of affiliation indicates a likelihood of religious practice. In our empirical work, we make a distinction between no religion (including humanists), Catholic, Protestant (including Evangelicals) and ‘other’ religions. The last category consists of Muslims and other smaller religious groups. Table 3a shows that individuals with religion designated ‘other’ display the least financial literacy. However, we do not find that Protestants are more financially literate than Catholics or non-religious individuals. Individuals of ‘other’ religion report that they do not know the answer to the financial literacy questions more often than other groups.

Table 3a. Distribution of financial literacy across religion (weighted percentages)

5.1 Descriptive evidence

Our main interest is explaining why some households prepare for retirement more effectively than others. To that end, we included the following question in our 2010 survey: How much have you thought about retirement: A lot, some, little, or hardly at all? This question was also included in the 2005 DHS questionnaire on financial literacy. As we stated in the introduction, the recent financial crisis has shown that the Dutch pension system is vulnerable to financial-market shocks. As a result, Dutch policy-makers have proposed additional pension reforms, such as an increase in the statutory retirement age. A key element of the proposed reforms is more individual choice concerning the timing of retirement. Furthermore, the market risk of pension investments will be deferred away from employers, meaning individuals will face more uncertainty with their second pillar pension. In this respect, one would expect that in the years since the 2005 survey, individuals would have taken more responsibility in preparing for retirement.

Table 3b indeed suggests that in 2010 respondents think significantly more about their retirement than they did in 2005 (see the results of the χ²-tests). Note that there are some differences in the composition of the 2005 and 2010 samples. In 2005, the financial literacy questionnaire was only filled out by the household member in charge of household finances, whereas in 2010 we selected all household members aged 25 and older, including both the household head and partner, if present. In order to have comparable results, we consider in Table 3b the same group of individuals in 2005 and 2010, i.e., individuals aged 25 or over in charge of household finances.

Table 3b. Retirement planning across years (weighted percentages)

Note: Respondents are in charge of household finances and at least 25 years old.

While the observed increase in retirement preparation is a comforting result, it should be made clear that about a third of respondents acknowledged having thought ‘a little’ (28.1% in 2005; 21.9% in 2010) or ‘hardly at all’ (7% in 2005 and 2010) about retirement. Only a small group of respondents (12.9% in 2005; 17.1% in 2010) have thought ‘a lot’ about retirement. The majority (about 50%) took an intermediate position and reported having thought ‘some’ about retirement. If we consider the subsample of non-retirees – the group of individuals who should be preparing for retirement – compared with the whole sample, we find that relatively more respondents in this subgroup have thought ‘hardly at all’ or ‘little’ about retirement.

Overall, most respondents seem to prepare only to a limited extent for retirement. Moreover, one may debate whether respondents who reported thinking ‘a lot’ or ‘some’ about retirement are actually thinking about the sufficiency of retirement savings rather than of ways to enjoy life after retirement. Psychological research, however, has shown that having a concrete picture in mind induces action and has a positive effect on the likelihood of taking concrete steps (McCrea et al., Reference McCrea, Liberman, Trope and Sherman2008). Indeed, the Dutch data show that respondents who think more about retirement not only find it more important to save but also manage to save more (Van Rooij et al., Reference Van Rooij, Lusardi and Alessie2011b). Lusardi and Mitchell (Reference Lusardi and Mitchell2007) report evidence from the US showing households who think a little, somewhat or a lot about retirement accumulate substantially more wealth than those who do not think about retirement at all. For the median household, older planners (51–56 years old) hold twice as much wealth as non-planners.

Table 3c summarizes changes in financial literacy levels in the Netherlands between 2005 and 2010. As mentioned in Section 4, we randomized the risk diversification question in 2005 so that half of the sample answered the same question but with an inverted order. Hence, for an appropriate comparison, we restrict the 2005 sample to respondents who got exactly the same question on risk diversification as the members of the 2010 sample. Table 3c shows that the 2010 respondents had somewhat more trouble answering the interest rate and risk diversification questions correctly than did the 2005 respondents. This is a worrisome result as, given the events in the years 2005–2010 and the changes at the policy level, individuals should be taking on more responsibility for retirement preparation.Footnote ¹

Table 3c. Financial literacy across years (weighted percentages)

Note: Respondents are in charge of household finances and at least 25 years old. The 2005 sample is restricted to respondents who got the same risk question as 2010 sample.

Exploring the correlation between financial literacy and thinking about retirement, we find that respondents who reported thinking ‘a lot’ or ‘some’ about retirement have, on average, a similar level of financial literacy.Footnote ² Moreover, these respondents are more financially literate than individuals who reported thinking ‘little’ or ‘hardly at all’ about retirement. Based on this evidence, in our multivariate analysis, we construct a dummy variable for retirement planning that takes on the value 1 if the respondent reported thinking ‘a lot’ or ‘some’ about retirement and 0 otherwise. Table 4 reports the relationship between this dummy variable and financial literacy, showing a strong positive correlation between financial literacy and thinking about retirement.

Table 4. Financial literacy by retirement planning (weighted percentages)

5.2 Multivariate analysis of retirement preparation

In this section, we perform a multivariate analysis of the relationship between retirement planning and financial literacy. We use several measures for financial literacy: (1) a dummy variable that equals one if a respondent correctly answered all three financial literacy questions and (2) a variable counting the number of correct answers to these three questions. To compare results with other countries, we also use three dummies for answering each financial literacy question correctly. We include dummy variables that control for age, education, gender, marital status, net monthly household income quartiles, home ownership and religion to take into account individual heterogeneity that might affect the relationship between retirement planning and financial literacy.

In Table 5, we first report the results of a simple Ordinary Least Squares (OLS) regression of retirement planning on socioeconomic controls and financial literacy. In this analysis, we consider only the 2010 sample. Moreover, we select all respondents who are 65 or younger and not yet retired. The OLS estimates in Table 5 show that there is a strong positive relationship between retirement planning and financial literacy. The size of the estimated coefficient for the number of correct answers (0.101, see column 3) suggests that one extra correct answer is associated with an increased probability (by 10 percentage points) of having thought (some or a lot) about retirement. Answering all three questions correctly raises the probability of planning by 13 percentage points, and the last column of Table 5 (column 5) indicates that it is the understanding of interest rate and risk diversification that matters most for retirement planning. The OLS results also indicate that respondents do not tend to think much about retirement when they are young and retirement is a distant concept. After controlling for financial literacy, there is no role for education in explaining retirement planning once we control for other individual characteristics. While in the raw data, men think more often about retirement than women, the effect of gender disappears in the multivariate setting. Interestingly, Catholics think more about retirement than respondents of other religious affiliations (or no affiliation). This finding is consistent with the results of Renneboog and Spaenjers (Reference Renneboog and Spaenjers2011) who find that, compared with respondents with ‘other’ religions or without religion, Catholics have a longer planning horizon.

Table 5. Multivariate analysis of retirement planning (1,166 observations)

Note: Standard errors reported in parentheses are clustered at the household level;

*** P<0.01, **P<0.05, *P<0.1.

Based on these simple estimates, we cannot yet interpret the relationship between financial literacy and planning as causal. The literacy variable might itself be endogenous (by planning more for retirement, one becomes more literate) or could proxy for unobservable variables (e.g., ability). On the basis of these arguments, one might state that the estimated literacy coefficient is biased upward. On the other hand, Van Rooij et al. (Reference Van Rooij, Lusardi and Alessie2011a) show that financial literacy is rather difficult to measure. It is likely that there are considerable measurement (classification) errors in our discrete financial literacy variables, which might lead to a downward (attenuation) bias in the estimated financial literacy coefficient.Footnote ³ In either case, we cannot simply rely on the OLS estimates reported in Table 5 to assess the effect of literacy on retirement planning. To remedy this problem, we have collected additional information that can serve as instruments for advanced financial literacy. We asked respondents about the financial experiences of their siblings and parents.Footnote ⁴ Specifically, we collected information on whether the financial situation of the oldest sibling is ‘better’, ‘the same’, or ‘worse’ than the financial situation of the respondent. The experience of siblings is not under the control of the respondent, but respondents can learn from those around them and increase their own financial literacy. One may argue that the experience of siblings can proxy for a common set of preferences or for a family fixed effect. While plausible in theory, the first stage regressions indicate that when siblings are in worse financial condition than the respondents, respondents are more likely to have higher financial literacy.Footnote ⁵ In addition to the financial situation of siblings, we also consider parents' understanding of financial matters as perceived by the respondent.

The estimates in the second stage, reported in columns 2 and 4 of Table 5, show that the relationship between literacy and retirement planning remains positive and statistically significant in the Generalized Method of Moments (GMM) regression. The results of the exogeneity test indicate that the OLS estimates differ significantly from the GMM estimates and that therefore the OLS estimates are inconsistent. Moreover, the Hansen's J test does not indicate rejection of the over-identifying restrictions. Overall, our GMM estimates show that financial literacy is an important determinant of retirement planning: Those who have low financial knowledge are less likely to plan for retirement.