Skating on Thin Ice: New Evidence on Financial Fragility

1 Introduction

Ice skating is a popular traditional winter activity in the Netherlands. However, the first days when frost sets in, skating is a precarious undertaking. If the ice is too thin, skaters may fall through the ice and into the freezing water. The expression “skating on thin ice” serves as a metaphor for describing a financially fragile situation faced by individuals or households.¹ This metaphor aptly illustrates situations where individuals are unable to afford an unexpected expense—such as the sudden breakdown of the car or of an essential household appliance, such as the refrigerator—and covering such costs drives them into acute debt. In these cases, individuals find themselves in a financially fragile situation that can plunge them into difficulties.

This study investigates the incidence of financial fragility in the Netherlands, delves into the channels that are used to deal with unexpected financial needs and its link to personal characteristics and circumstances. We use the term financial fragility to indicate that a household cannot cope with an unexpected expense—which is also referred to as a financially vulnerable household.

This is an important topic: not being able to face emergency expenses is worrisome because it could drive individuals into debt or being unable to meet necessities. The consequences stretch far beyond financial concerns. Baker et al. (2023) link financial fragility to lower happiness and life satisfaction. A financially fragile situation comes with financial stress and negatively affects mental health (Simonse et al. 2022) as well as physical health (Bialowolski et al. 2021). In turn, financial stress and negative health may lead to additional financial hardship, which can contribute to a vicious circle or a poverty trap.

Sergeyev et al. (2024) document that a median worker in the US is distracted from work for 5 hours per week as he/she worries about and deals with issues related to household finances. This distraction reduces the productive working hours and earning capacity of workers under financial stress. They argue that more financially sophisticated individuals understand the negative impact of financial stress, which incentivizes them to accumulate savings and stay away from being financially fragile. Naïve households, on the other hand, do not take this negative impact into account, save less, and are more prone to find themselves in a trap of financial stress, have less productive working hours, and have difficulties escaping financial hardship. Yakoboski et al. (2020) also show a strong negative association between financial sophistication and hours spent worrying and dealing with personal financial issues.

Bucher-Koenen et al. (2023) highlight another personal characteristic that may lead to financial fragility: limited understanding of probabilities. They show that individuals on average assign a lower probability to negative shocks hitting their own financial situation than to negative shocks hitting otherwise identical individuals. In our study, we will investigate the association of financial fragility with both financial literacy and probability literacy.

A broad range of indicators has been used in studies focusing on financially fragile households, including objective as well as subjective measures of financial fragility. Objective indicators use quantitative measures, such as debt to income ratios, debt service ratios, debt to assets ratios, loan to value ratios or illiquidity due “non-optimal” portfolio allocation (a portfolio that is overly exposed to liquidity risk) to define financial fragility (Jappelli et al. 2013; Albacete and Lindner 2013; Ampudia et al. 2016; Brunetti et al. 2016). Some of these measures recognize that financially fragile households may not necessarily be asset poor; they may hold illiquid assets that are costly or difficult to liquidate. Subjective indicators use measures such as difficulties in making ends meet or individual's perceptions of the housing cost burden (McCarthy 2011; Deidda 2015; Hoff et al. 2016). Naturally, all these measures have their strengths and weaknesses, and the choice of an appropriate measure depends on the research question at stake.

In our study, we measure financial fragility by asking whether respondents are (probably) able to come up with €2000 within a month in case of a financial emergency—a measure that was pioneered by Lusardi et al. (2011) in the wake of the financial crisis of 2007–2008. Similar measures have been used in other research studies to gauge household financial fragility. An advantage of this measure is that it has been widely used in the literature and it is possible to compare findings with the evidence in other countries (see the recent evidence in Clark et al. 2021). Another advantage is that this measure well reflects the state of households' balance sheet. A potential disadvantage is that some respondents who answer that they are unable to come up with €2000 could, in fact, come up with this amount if they would, for example, take out a loan, but are reluctant to do so or are unaware of this option. Another potential disadvantage is that our measure refers to a fixed amount, €2000, which can have a different meaning to a wealthy individual than to an individual with a low income.² A way to circumvent this problem is by referring to a percentage of their income. For example, the OECD (2022) uses the following question to measure financial fragility “If you, personally, faced a major expense today—equivalent to your own monthly income—would you be able to pay it without borrowing the money or asking family or friends to help”. Such a question measures whether respondents have sufficient liquid assets to face an immediate expense.³ Our financial fragility measure takes a broader perspective, allowing respondents to use a multitude of channels to come up with €2000, including borrowing funds, selling assets, or leveraging their social network. Moreover, by further asking which channels respondents use to come up with €2000, we can differentiate between those who have enough liquid savings to pay for an emergency expense and those who intend to use different coping methods. This provides useful insights into the various ways households can cope with a financially fragile situation. This information may have implications for policymakers designing measures to increase financial resilience. Since respondents have 30 days to come up with the amount and may use a multitude of channels, our measure represents a broader definition of financial fragility that aligns closely with real-life decision-making. By referring to a fixed amount, it acknowledges that high-income individuals can be less likely to experience financial fragility compared to low-income individuals.

In our work, we use a representative sample of households living in the Netherlands and delve into the many factors and personal characteristics that can be linked to financial fragility. The survey was conducted in 2016; financial fragility was, therefore, measured in normal times, several years after the great financial crisis and before the pandemic hit the economy worldwide. We examine both the prevalence of financial fragility among Dutch households and the different coping methods they would use to come up with the funds needed to finance an unexpected expense. Furthermore, we examine the importance of both financial literacy and probability literacy as predictors of financial fragility and the coping methods that individuals use.

We contribute to the literature by building on the earlier work of Lusardi et al. (2011). They examined the ability of American households to cope with an unexpected shock and compared the evidence in the United States with that of households in seven other countries, including the Netherlands. Similar measures have been used to gauge household financial fragility (see, e.g., Anderloni et al. 2012, and Gathergood and Wylie 2018). More recent studies using the financial fragility metric proposed in Lusardi et al. (2011) include Demertzis et al. (2020) in the context of the EU, Cziriak (2022) for Germany, Bucher-Koenen et al. (2023) for Austria, and Clark et al. (2021), Hasler et al. (2023), and Baker et al. (2023) for the United States.

Our study also builds upon the literature that links financial literacy to sound financial decision-making, financial well-being and financial fragility (see Stolper and Walter 2017, and Lusardi 2012, for an overview). Babiarz and Robb (2014), Hasler et al. (2018), and Bucher-Koenen et al. (2023) find that households who are more financially knowledgeable are more likely to have emergency savings and are less likely to be financially fragile. More recent studies during and after COVID-19 pandemic have also documented a negative association between financial literacy and fragility in Germany (Cziriak 2022) and the United States (Hasler et al. 2023).

While numeracy is an important element of financial literacy, in our study, we go a step further and include a separate probability literacy measure—a measure taken from Hudomiet et al. (2018) and link this measure to financial fragility. Our motivation is that the understanding of probabilities seems particularly relevant for how households prepare for unexpected shocks, especially in view of the findings of Bucher-Koenen et al. (2023) that individuals may underestimate the probability that they are hit by negative shocks. While the earlier study by Lusardi et al. (2011) already included a measure of risk literacy, linking the ability to cope with an unexpected expense to probability literacy is a relatively unexplored area in the literature on financial fragility.

The main findings of our empirical investigation can be summarized as follows: First, about 14% of Dutch households report that they are probably or certainly unable to come up with €2000 within a month in case of a financial emergency. This is lower than what was found in 2009, when a quarter of Dutch households was classified as being financially fragile (Lusardi et al. 2011). Second, relatively higher levels of financial fragility are observed among the young, households with children, renters, low-income households, those with lower education levels, and the unemployed. Third, turning to methods of coping, we find that most Dutch households would rely on cash, checking, or savings accounts in case of a financial emergency. However, some households would resort to less formal methods, such as seeking help from their network of family and friends. Others would resort to credit card borrowing. Fourth, we find that the ability to cope is associated with financial knowledge and probability literacy, even after controlling for many determinants of financial fragility, including education and income. Specifically, we find probability literacy to be a strong predictor of financial fragility.

These findings not only increase our understanding of financial fragility but also have important policy implications. Using data from 2016, several years after the financial crisis, allows us to observe if and how households' balance sheets were able to recover. Moreover, compared to earlier studies, we have new data that allow us to take a deeper dive into the factors influencing financial fragility and the choice of coping methods of Dutch households. Specifically, we can identify those who are financially fragile and which demographic groups are best to target. Finally, while many previous studies on financial security and financial well-being have paid attention to retirement savings, our findings highlight that it may be useful to also pay attention to short-term savings.

The remainder of the paper is organized as follows: in the next section, we describe our data. In Section 3, we present the findings on financial fragility. In Section 4, we examine the ways Dutch households foresee coping with financial shocks and the underlying characteristics influencing the choice of coping methods. In Section 5, we perform some robustness checks, and in Section 6, we summarize our main findings and discuss their implications for future research as well as for policymakers and financial institutions.

2 Data

In July 2016, we fielded a new module designed to measure financial fragility and the methods Dutch households use to cope with an emergency expense in the Centerpanel, which is an internet panel managed by Centerdata, a research institute at Tilburg University specializing in internet surveys. The panel was originally designed to be representative of the Dutch population. Participation in the panel is not dependent on the use of and access to the internet. If necessary, equipment is provided for households without a computer or internet connection (e.g., a device that enables members to participate through their television). Nonetheless, some groups are over-represented (e.g., respondents older than 65) or underrepresented (respondents younger than 40). When we present summary statistics (i.e., the percentages presented in Tables 1–3 and 5) we use the sample weights supplied by Centerdata, which are based on gender, age, education level, and net household income. Panel attrition is dealt with by drawing refreshment samples biannually (see Teppa and Vis 2012, for information about the Centerpanel). Notably, there could be more than one respondent (a person who is at least 16 years old) within the household. In total, 2893 members of the Centerdata panel aged 16 and older were approached to fill out the financial fragility module and 2550 members did so, for a response rate of 88%. We removed 113 respondents from the sample because they were not the head of the household, that is, the household member who reports on household finances, nor the partner of the head of household, and are left with 2437 respondents.

Following the measure of financial fragility designed by Lusardi et al. (2011), respondents were asked: “How confident are you that you could come up with €2000 if an unexpected need arose within the next month?” Respondents could answer: “I am certain I can come up with €2000”; “I could probably come up with €2000”; “I could probably not come up with €2000”; or “I am certain I cannot come up with €2000”. Out of the 2437 respondents, 76 stated they “don't know” answer to this question. We removed these respondents from the sample. Respondents stating that they probably could not or certainly could not come up with €2000 are classified as being financially fragile. In a follow-up question, these respondents were asked to provide an estimate of the amount they could come up with within a month. For this question, respondents had to provide an amount or could state they did not know the answer to this question.

There are four elements of this financial fragility indicator that are important to highlight. First, it is essential to pose the question in terms of the respondent's confidence about his/her ability to come up with €2000 rather than a yes or no answer, because we are dealing with an unanticipated event in the future. This subjective measure captures the confidence levels and perceptions of individuals regarding their financial situation, which are crucial for understanding financial behavior and decision-making under uncertainty. Second, the question measures the capacity to come up with funds, not whether households have those funds, as there may be many methods one could use to deal with shocks, for example relying on family and friends or borrowing. Third, the amount of €2000 is meant to represent not a specific amount of money but, rather a midsize shock, measuring, for example, the amount needed to cover an unanticipated expense such as a car repair, a legal expense, or replacing a broken fridge.⁴ Such unexpected expenses are shown to be a source of financial distress in the Netherlands. Other triggers are sudden loss of income (e.g., due to illness, work disability, job loss, caregiver tasks that lead to less work) and divorce (Money Wise Platform 2019).⁵ Fourth, the question asks respondents whether they can come up with €2000 within a month rather than immediately. In this way, respondents have a relatively long timeframe to assess and utilize all the resources at their disposal, thereby extending the scope of our fragility measure beyond short-term liquidity constraints.

We are also interested in assessing how households intend to cope with a financial emergency. Respondents who are able or probably able to come up with €2000 were asked the following question: “If you were to face a €2000 equivalent unexpected expense in the next month, how would you come up with the funds you need?” Financially fragile respondents (those who are not able or probably not able to come up with €2000) were asked to provide an estimate of the amount they could come up with within a month. For them, the question mentioned above was changed and showed their estimated amount rather than the €2000 figure. For the financially fragile respondents who did not provide an estimate, “an unexpected small expense” was used in place of the €2000 figure. Respondents were presented with a list of 15 options (together with the options “Other” and “I do not know”) and were instructed to select one or more coping methods with no limitation on the number of methods that could be chosen. With this question, we tried to elicit the preferred methods of coping (without indicating an order of preference). The list was randomized on the screen to avoid response-order bias.

The list consisted of the following coping methods: (1) draw from cash or checking accounts, (2) draw from savings accounts, (3) borrow or ask for help from my family, (4) borrow or ask for help from my friends (not members of my family), (5) liquidate or sell investments, (6) draw from annuity or single premium insurance, even if I have to pay a penalty or taxes, (7) sell my home, (8) use my credit card, (9) open or use a home equity line of credit or a second mortgage (using the house as collateral), (10) take out a personal loan or revolving credit (without collateral), (11) use a short-term mini loan,⁶ (12) ask for a payroll advance⁷, (13) work overtime, get a second job, or another member of my household would go to work (longer), (14) pawn an asset I own at a pawnshop (e.g., the “Stadsbank van lening” or pawnshop “Used products”), (15) sell things I own (except my home), for example via marktplaats.nl (an online platform for individuals to buy and sell products).

The data from the financial fragility module are enriched with demographic information from the Centerpanel, which includes information on gender, age, education, household size, household income, housing, and labor market status. Net monthly household income is categorized into quartiles as follows: income up to €2000, between €2000 and €2750, between €2750 and €3600, and income above €3600. We categorized labor market status into employed and self-employed, unemployed, retired, and economically inactive. Unemployed people consist of respondents who are looking for a job, while the economically inactive are students, those who are occupationally disabled, housekeepers, and those who perform unpaid work while retaining benefits or do volunteer work.

The financial fragility module did not have questions about financial literacy. However, we can use data from a separate module on financial literacy that was fielded in September 2015 to 2208 Centerpanel respondents aged 18 years and older. In this module, financial literacy was measured using the Big Three financial literacy questions developed by Lusardi and Mitchell (2011). These questions measure an individual's knowledge of basic concepts at the basis of financial decision-making, that is, knowledge of interest compounding, inflation, and risk diversification. We construct a financial literacy index that goes from zero (lowest financial literacy) to three (highest financial literacy) based on the number of correct answers to the Big Three financial literacy questions. Do not know responses to the financial literacy questions are grouped together with incorrect answers. Appendix A provides the exact wording of the financial literacy questions. Out of the 2208 respondents, 2191 provided an answer to all Big Three questions. We removed 17 respondents from the sample who did not answer the Big Three questions.

To perform the empirical analysis, we merged the data from our financial fragility module (with 2361 observations) with data from the 2015 financial literacy module (with 2191 observations). Only 1766 respondents completed both modules, for an attrition rate of about 25%. Following Verbeek and Nijman (1992), we tested for attrition bias by using data from the financial fragility module and running linear probability regression models. We consider several dependent variables. The first one, used in Section 3 below, is the financial resilience indicator which is equal to 1 if the respondent reports being certainly or probably able to raise €2000 and zero if not. We also consider indicators indicating the coping methods as dependent variables. We use three empirical specifications. The first specification uses only one regressor, which indicates whether the respondent is in the matched sample, henceforth referred to as the matching indicator. The second specification extends the first one with the inclusion of the sample weights and background variables (Model 1 of Table 4). The third specification extends the second one with interaction terms between the matching indicator and the background characteristics. The full results of this analysis are in Table A1. For the first specification, the matching indicator is statistically significant but only at the 10% level (p value 0.0786). However, this significance is no longer present for the second specification, when background variables are controlled for (p value 0.5307). Adding interaction terms doesn't significantly improve the fit of the model. Thus, on the basis of specification 2, we conclude that the results presented in columns 1 and 2 (including a financial literacy measure) of Table 4 are not affected by attrition bias. Attrition also seems to be exogenous (and the interaction terms insignificant) for all coping methods, except “borrow or ask help from family, friends” and “liquidate or sell financial or real assets.”

According to Hudomiet et al. (2018), individuals who have a better understanding of probabilities (how likely an event is to affect them) can be more successful in planning for the future and in developing strategies to deal with uncertainties.⁸ Since we examine how households deal with unexpected shocks, it is important to measure respondents' knowledge of probability and risk, in addition to their basic financial literacy. The annual DNB Household Survey (DHS) from 2017 includes four questions that measure knowledge of probabilities or probability literacy (as suggested by Hudomiet et al. 2018). This survey data are again collected by Centerpanel. Probability literacy is defined as the ability of individuals to think in probabilistic terms and use probabilities effectively in everyday life (Hudomiet et al. 2018). We view probability literacy to be part of a broader financial literacy measure. The questions in the DHS survey are ordered from less to more complex and respondents do not have the option to choose “do not know” or refuse to answer. Similar to the financial literacy index, we construct a probability literacy index based on the number of questions a respondent answered correctly. Appendix A reports the exact wording of the probability literacy questions. Column 3 of Table 4 below presents the “baseline” model which also includes a probability literacy measure as control variable. Notice that additional observations are lost when one adds data on probability literacy: the matched sample now consists of 1498 observations (instead of 1766). The findings in Table A2 about attrition show that this reduced sample does not lead to biased estimation results. In Table A2, we also investigate the issue of attrition bias for the binary coping method indicators considered in Table 6 of the paper. The overall conclusion is that attrition seems to be exogenous (and the interaction terms insignificant) for all coping methods, except “borrow or ask help from family, friends” and “work more, other.” This result implies that we should interpret the regression results for those coping methods with caution.

Finally, respondents in the Centerpanel provide information on their assets and liabilities in the annual DHS. In Section 5 (Further analysis), we use data from the DHS 2016 questionnaire to get information on financial wealth. Yet again observations are lost when we add both the risk literacy data and the data on household wealth to our baseline dataset. The matched sample consists of 1386 observations (instead of 1766). In Table A3, we investigate further whether attrition is endogenous. The results presented in Tables A2 and A3 are similar: specifically, we cannot reject (at the 5% level) the null hypothesis that attrition is exogenous in the case of the financial resilience indicator variable. However, attrition becomes endogenous for the coping method indicator “draw from checking or saving account,” which warrants caution when interpreting our empirical findings.

3 Financial Fragility of Dutch Households

In this section, we present our findings about Dutch respondents' capacity to cope with a financial emergency. We first document financial fragility and its correlation with different socioeconomic and demographic characteristics, financial literacy, and probability literacy. Next, we estimate probit models to study the determinants of financial fragility.⁹

Table 1 provides descriptive statistics of financial fragility across a set of demographic and economic characteristics. All frequencies are weighted using population weights. The first row reports respondents' capacity to raise €2000 within a month. About 14% of respondents state they probably could not (4.5%; see Appendix D, Table A5) or certainly could not (9.0%; Table A5) come up with €2000 within a month. Comparing our findings in 2016 to those of Lusardi et al. (2011), who documented that about 27% of Dutch households were financially fragile in 2009 (a period of economic crisis), we observe that financial fragility among Dutch households has decreased over time but is still sizable.¹⁰ In 2009, nearly half of American respondents were financially fragile, while in European countries such as France, the United Kingdom, and Germany, more than 35% of households were financially fragile.¹¹ Using data from the 2015 National Financial Capability Study, Hasler et al. (2018) find that financial fragility is still prevalent among the US population (36%), even several years after the financial crisis.

One explanation for the differences in financial fragility may be related to differences in social welfare systems. The Netherlands has a basic state pension which covers everyone who lives in the Netherlands and, contrary to the United States, it has a universal healthcare system. Out-of-pocket health care spending per capita in the Netherlands ($605) is relatively low compared to the US ($1122) (OECD 2019).¹² Furthermore, employees in the Netherlands are covered by disability insurance and unemployment insurance, and households with little or no income and limited savings are entitled to social assistance benefits.

Respondents who indicated that they probably or certainly could not come up with the €2000 were asked to provide an estimate of how much money they could come up with within a month. About 70% of the financially fragile respondents indicate that they could come up with amounts less than or equal to €500, 25% believe that they could come up with amounts between €500 and €1000, and 5% between €1000 and €2000.

Table 1 further illustrates how financial fragility varies with demographic and economic characteristics. For all characteristics, except for age, cross-sectional differences in coping ability are statistically significant. Women are far more likely to be financially fragile than men (17.5% vs. 10%). Single-headed households are significantly more fragile, with more than 36% of those with children reporting that they probably or certainly would not be able to cope with a financial emergency. A possible explanation is that, in these households, a sudden increase in household expenses must be dealt with by a single income.¹³ Spouses responsible for household finances exhibit similar levels of financial fragility to other spouses (9.4% vs. 8.9%). Furthermore, a large difference in the ability to cope is observed between renters and homeowners, a simple proxy for wealth. Renters report far more often than homeowners (30.4% as opposed to 6.7%) that they are probably or certainly unable to come up with €2000.

As expected, unemployed individuals exhibit significantly higher levels of financial fragility compared to the (self)-employed (32.7% vs. 10.7%). Retirees are the least financially fragile, a finding in line with the low poverty rates observed in this demographic group in the Netherlands. Respondents with household income in the bottom quartile are significantly more financially fragile than those in the highest income quartile (22.9% vs. 4.4%). Further, compared to other respondents, respondents with low financial wealth (< €5000) more frequently report being financially fragile (33.2%).

The capacity to cope with financial shocks also increases with educational attainment. Notably, fewer than 8% of respondents with a college or graduate degree are classified as being financially fragile. This contrasts with the findings of Lusardi et al. (2011), who report that American respondents with a college degree exhibited relatively limited coping capacity. They attribute this finding to a combination of high student debt, increased cost of living, and greater income inequality. Student loan debt in the United States reached unprecedented levels, with many college graduates entering the workforce burdened by substantial financial obligations (Dynarski and Kreisman 2013). Conversely, Dutch students benefit from a more generous system of higher education funding, with lower tuition fees and more generous student loan terms (Knoester 2017) than the US system. This disparity in debt levels can partly explain why Dutch college graduates experience lower financial fragility. Moreover, the Dutch social security system is more comprehensive, offering stronger protection against financial hardship compared to the US system (see, e.g., Van der Linden et al. 2020).

To summarize, the large majority of Dutch households is capable of dealing with financial shocks. Nevertheless, our analysis shows that financial fragility is more prevalent among specific groups, including women, younger individuals, the unemployed, those with lower education levels, lower-income households, and households with children. Furthermore, our results show higher financial fragility among renters and those with low amounts of financial assets.

Table 2 tabulates financial fragility and its association with various measures of financial literacy and probability literacy (see also Table A6). We find statistically significant differences in financial fragility across different levels of financial literacy and probability literacy. One noticeable finding is that those who incorrectly answer or could not answer the two easiest financial literacy questions (interest compounding and inflation) are more likely to be financially fragile compared to those who incorrectly or could not answer the more difficult question (risk diversification).

About 14% of the respondents answered at most one of the financial literacy questions correctly. Even though individuals in the Netherlands are taught about interest rates and inflation from an early age in school, we still observe individuals lacking basic knowledge of these two topics. This is important because financial fragility decreases sharply with the number of correctly answered financial literacy questions.

We observe a sharp gradient in financial fragility across the number of correct answers and the number of don't know answers to the Big Three financial literacy questions. In particular, the percentage of respondents unable to come up with €2000 within a month decreases from 32% among those who could not answer any of the financial literacy questions correctly to 8% among those who answered all three questions correctly. Also, the fraction of financially fragile respondents is considerably higher for respondents who gave at least two don't know answers to the financial literacy questions (about 30%).

Table 2 further shows that the ability to raise €2000 within a month increases with higher levels of probability literacy. Among those who can answer up to two probability questions correctly, 17% are financially fragile as compared to 10% of respondents with correct answers to at least three probability literacy questions. In summary, the data show that higher levels of financial literacy and probability literacy are linked to higher coping capacity.

Given the link between financial fragility and financial literacy and probability literacy, in Table 3, we present a breakdown of the individual financial literacy questions and the count of correct responses on the probability literacy questions. We found a strong correlation between these two measures: individuals with increased levels of probability numeracy are also more likely to provide correct answers to the financial literacy questions.

Recall that the findings in Table 1 showed that women are significantly more financially fragile than men. Building upon the work of Bucher-Koenen et al. (2017) and Hudomiet et al. (2018), the results in Table A4 show gender disparities in financial literacy and probability literacy in our dataset as well. Thus, compared to men, women in our sample are not only significantly more financially fragile but also score lower on financial literacy and probability literacy.

To gain further insights, we now turn to a multivariate analysis of the determinants of financial fragility. We use probit regressions linking financial fragility to a set of demographic and economic characteristics.¹⁴ The dependent variable is equal to one if the respondent reports being probably or certainly able to raise €2000 within a month, hence reports being financially resilient, and is equal to zero if the respondent reports being probably or certainly not able to raise €2000 within a month. Table 4 displays the average marginal effects from these probit regressions. Model 1 includes only demographic and economic characteristics. In Model 2, we add the financial literacy index. In the last column, we include a dummy variable that takes value 1 if a respondent can answer three to four probability questions correctly (high probability literacy) and value 0 otherwise.

The estimates reported in the first column of Table 4 largely confirm the findings of the descriptive statistics. Financial resilience increases with age and educational attainment. However, gender is no longer statistically significant. After accounting for other background characteristics, the difference in the capacity to cope between men and women found in the descriptive statistics disappears.¹⁵ An increase in the number of children in the household significantly reduces the likelihood of being able to cope, and singles (with children) are more financially fragile than respondents with a partner. Spouses responsible for household finances exhibit similar levels of financial fragility to other spouses. Compared to renters, homeowners are 13% points more likely to be able to cope with a financial shock.¹⁶ This result might be due to being wealthier. In Section 5, we discuss how the relationship between home ownership and the ability to cope is affected by extending the baseline model (Model 3 of Table 4) and adding financial wealth. Moreover, the results show that economic characteristics, such as labor market status and household income, are significantly linked to financial fragility. The capacity to cope increases with household income, while those who are unemployed are more likely to be financially fragile.

Turning to the additional variables in models 2 and 3, we find a strong link between financial resilience and both financial and probability literacy. While Lusardi et al. (2011) did not find a strong relationship between risk literacy and the ability to cope, our results show that financial literacy and probability literacy are both linked to financial resilience. It is noticeable that the effects of financial literacy and probability literacy are statistically significant, even after controlling for education. These results align with the findings of Hasler and Lusardi (2019), who observe that financially literate US households are significantly less likely to be financially fragile, even after controlling for socioeconomic factors such as education and income. These findings are also supported by Brunetti et al. (2016), who use data from the Bank of Italy Survey on Household Income and Wealth and show that a higher level of financial literacy is linked to a reduced likelihood of financial fragility.

The estimation results in Table 4 further show that the significance of financial literacy diminishes when accounting for probability literacy. This suggests that knowledge of probabilities plays an important role when it comes to financial fragility. A possible explanation is that holding buffer stock savings is a financial decision that involves a good understanding of risk. Consequently, a grasp of probabilities becomes of great relevance.

To summarize, our estimates from the probit regression analysis are in line with the results from the bivariate analysis with the exception of gender, which does not contribute to explaining the ability to cope, after accounting for sociodemographic background information. Lastly, probability literacy emerges as a strong predictor of the ability to cope with a financial emergency and its effect goes above and beyond that of education.

4 Methods of Coping With A Financial Emergency

In this section, we examine the strategies respondents intend to use in case of a financial emergency, along with the socio-demographic and economic characteristics influencing their choice of coping methods. As outlined in Section 2, respondents were presented with a list of 15 potential coping methods (plus the option “other” and “don't know”) and were instructed to select one or more methods that applied to them. To mitigate response-order bias, the list of 15 options was presented in a randomized order.

We first examine whether this randomization impacts the answer patterns by creating a table that indicates which coping method was mentioned first alongside each coping method. Subsequently, we perform a χ² test to explore whether the choice of coping method is independent of the order in which coping methods were presented.

For most coping methods, we do not reject the null hypothesis (at the 5% level) that randomization does not impact the response pattern. For one coping method, “withdrawing from a savings account,” however, we reject the null hypothesis at the 1% level. For the response options “drawing from cash or checking accounts” and “taking out a personal loan or revolving credit (without collateral)” randomization seems to play a role as well. These findings can possibly be explained by the fact that drawing from cash or checking accounts and savings accounts are the most frequently selected methods (see Table 5). It is important to note that these coping methods involve very liquid assets, as funds in these accounts are readily accessible. By mentioning these methods first, we may affect the likelihood that respondents choose them, as they could perceive them as the simplest and quickest ways of managing a financial emergency. In contrast, opting for a personal loan or revolving credit is only chosen by 2.5% of respondents. These types of unsecured loans tend to be less attractive due to their typically high interest rates. We expect that showing an unappealing method, like unsecured loans, as the first option does not usually lead respondents to select this method unless they have a strong preference for it.

Table 5 reports the coping methods chosen by respondents. In the table, the coping methods are combined into seven categories as follows: (1) Draw from checking or saving accounts; (2) borrow or ask for help from family and friends; (3) take out a loan with or without collateral; (4) use a credit card; (5) liquidate or sell financial or real assets; (6) Work more; (7) Do not know other.¹⁷ Table 5 also shows the distribution of coping methods (categories) among those who rely on a single coping category versus those who use multiple coping categories. The last column of Table 5 shows that most of the respondents (86%) opted for using funds from their checking or savings accounts as a way of dealing with a financial emergency. As previously discussed, these accounts typically involve liquid assets, which may be the easiest or cheapest way to acquire the necessary funds. Moreover, using funds from savings accounts is the most common individual coping method. However, it is worth noting that respondents who opt for withdrawing funds from their accounts as a coping strategy might not necessarily have sufficient funds available in those accounts. In fact, 6% of respondents who selected using funds from their accounts as a coping method belong to households that have at least one checking account with a negative balance. At first glance, this result seems to be counterintuitive. However, several comments are in order. First, the financial fragility module was fielded in weeks 26, 27, and 29 of 2016, whereas the information on (financial) wealth stems from the wealth module of the 2016 wave of the DNB household panel, which elicits information on the checking account balances as of December 31, 2015. It is possible that households have a negative balance in their checking accounts on December 31, 2015, and a positive balance when the financial fragility questionnaire was in the field. Second, a respondent could have a checking account with a positive balance whereas another household member has an overdrawn account. Such a respondent could plausibly use the funds in a checking account as a coping method.

Turning back to the last column of Table 5, we observe that approximately 15% of respondents would turn to their family or friends and 8.4% would opt for using a credit card in the event of a financial emergency. This percentage is somewhat unexpected given the relatively limited usage of credit cards in the Netherlands.¹⁸ It is also apparent from the table that the remaining coping methods were less popular among respondents. Notably, Dutch households are unlikely to address financial emergencies by taking out a loan. Notice also that less than 0.1% of the respondents mentioned “sell my home” as coping method.

Summarizing our key findings, we observe that Dutch households intend to rely on a range of coping methods when faced with a financial emergency. Furthermore, our results demonstrate that households not only rely on formal coping methods like withdrawing funds from their bank accounts, but they also rely on informal methods, such as seeking support from their family and friends.

The second, third, and fourth column of Table 5 present the coping methods chosen by respondents who selected one, two, three, or more coping methods (broadly classified). Note that 76% would rely on one coping method, 14% on two, and 6% on at least three coping methods. Four percent of the respondents gave a don't know answer to the coping method question. These percentages are quite different if we only consider the financially fragile households (as defined before): 40% of them selected one method, 30% two methods, 11% at least three methods and 19% stated they don't know (see the bottom of Table 5). In other words, financially fragile respondents more frequently use two or more coping methods or do not know how to cope with an emergency.

The findings reported in the second column of Table 5 show that using funds from bank accounts is by far the most common standalone coping method, while other methods are rarely used in isolation. Using bank accounts is also the dominant strategy for respondents who opt for more than one coping method. However, some other coping methods are important as well when considering multiple coping methods. Approximately 48% of the respondents who selected a combination of two coping methods would partly rely on their networks of family and friends, 30% would use their credit card, and 22% would liquidate or sell their assets. Those methods are rarely used in isolation but are used in combination with other methods of coping.

Overall, the findings suggest that, in case of a financial emergency, most respondents would rely on their cash, checking, or savings accounts. This method of coping is used by almost 9 in 10 respondents who uses only one coping method. Apart from relying on bank accounts, resorting to family or friends and using a credit card seem to be the next most popular methods of coping among our respondents, and even more so among respondents using a combination of coping methods, and among the financially fragile respondents.

We turn now to estimating probit regressions to explore the relationship between the choice of one of the seven (aggregated) categories of coping methods and various demographic and economic characteristics. We use the same variables as in Table 4. Table 6 presents the average marginal effects from these probit estimates.^19,20 While we found that gender is not related to the coping capacity, it is related to the coping strategies. Specifically, women are less likely than men to use a credit card and sell their assets. On the other hand, women are more likely to use money on their accounts. Many of these findings are generally aligned with those by Lusardi et al. (2011) for the US, despite their lack of information regarding those unable to cope with an emergency. Other findings also are consistent with the study of Lusardi et al. (2011). Older individuals are less likely to resort to family or friends compared to respondents younger than 40. Specifically, individuals aged 65 and older are 17 percentage points less likely to select family or friends as a coping method compared to individuals who are younger than 40. One plausible explanation is that children more often rely on their parents than the other way around. Although we do not find an association between the number of children in a household and the choice of coping methods, we do find that singles are more likely to select “family and friends,” sell assets' and “use a credit card” as coping strategies compared to respondents with a partner. Respondents with monthly household income of less than €2000 are less likely to use a credit card as a coping strategy. We observe few significant relationships between economic attributes, such as labor market status, education level or household income, and the selection of coping strategies. However, we find that homeowners are more likely to use formal coping methods, such as relying on funds from bank accounts.

The probit regressions also show the relationships between financial literacy and probability literacy and the coping strategies chosen by respondents. Financial literacy is positively related with a credit card and negatively related with working more. Probability literacy is negatively related with the choice of working more as a coping strategy and positively related with using bank accounts and selling assets. The positive relationship between financial literacy and the use of credit cards as a coping strategy, along with the negative relationship to working more, may be explained by better access to financial products and an understanding of opportunity costs. Financially literate individuals tend to use credit strategically, managing liquidity without increasing labor supply, which may be seen as less efficient (Lusardi and Mitchell 2014; Lusardi and Tufano 2015). There is room for using credit cards strategically, because most credit cards issued in the Netherlands are basically debit cards and outstanding credit card debt is cleared each month without paying any interest, if the checking account balance is positive. The market of revolving credit cards is rather small in the Netherlands. Probability literacy, on the other hand, is positively linked with selling assets and negatively with working more, arguably because these individuals are better at assessing risks and returns and may prefer adjusting their portfolios rather than increasing labor supply, consistent with risk management theory.

Some respondents answer don't know (DK) to the coping method question. Interestingly, the probability of a DK answer is strongly negatively associated with both financial and probability literacy.

In summary, the empirical findings show that the choice of coping strategies is related to various demographic characteristics and, to a lesser extent, to economic attributes. Liquid assets (using funds from cash, savings and checking accounts) and the networks of family or friends emerge as important coping strategies. Individuals also turn to credit cards to manage financial emergencies. Notably, both financial literacy and probability literacy have an impact not only on the capability to cope with a financial emergency but also on the choice of coping strategies. Financially literate individuals more often report being able to cope with a financial emergency and are more likely to use credit cards as a coping mechanism. Likewise, individuals with higher probability numeracy are less financially fragile, and they are less likely to increase work hours or sell assets to deal with a financial emergency. These results highlight the importance of financial literacy and probability literacy in influencing financial decision-making.

5 Further Analysis

6 Conclusions and Implications

In this study, we evaluate the financial fragility of Dutch households by examining their ability to come up with €2000 within a month in case of an unexpected need, as well as their intended coping strategies for addressing such a need. We examined different predictors of financial fragility, including financial literacy and probability literacy.

We find that financial fragility is relatively low among Dutch households when compared to US statistics. In 2016, about 14% of households are probably or certainly unable to come up with €2000 within a month. Lusardi et al. (2011) reported that, in 2009, the Netherlands was one of the countries with the lowest percentage of financially fragile households (27%). Not surprisingly, this percentage has decreased since the financial crisis. Despite these findings, we still observe elevated levels of financial fragility among specific groups of the population, in particular the young, households with children, renters, low-income households, those with lower education levels, and the unemployed. Moreover, among the financially fragile respondents, a majority report that they could only come up with an amount equal to or below €500 within a month if an unexpected need were to arise.

Our dataset allows us to explore a wide range of methods households anticipate using in the event of a financial shock. We examined both formal methods, such as withdrawing funds from bank accounts, and more informal methods, such as seeking support from family or friends. These methods range from cheaper options to costlier ways of coping. The findings show that a majority of Dutch households would turn to formal and also cheaper methods; however, a sizeable fraction of households would also resort to less formal methods, such as their network of family and friends. Moreover, one out of three households would adopt a combination of two or more coping strategies in order to be able to cope with an unexpected need.²²

Lastly, we find that financial literacy and probability literacy are linked to financial fragility. These effects are present in both bivariate correlations and multivariate regressions, which account for a wide set of sociodemographic and economic characteristics, including the respondent's education level. Noticeably, probability literacy is a strong predictor of financial fragility. This finding highlights the importance of understanding probabilities (another facet of financial knowledge) when it comes to precautionary savings.

Considering that certain groups are more likely to be financially fragile, policymakers might focus efforts to enhance these groups' capacity to withstand financial shocks. For example, policy initiatives should focus not only on long-term savings strategies, such as retirement preparation, but also promote buffer stock savings. Our findings also underscore the importance of basic financial knowledge and probability literacy. A potentially promising approach to tackle financial fragility involves implementing financial education programs aimed at improving households' skills to manage their resources. Several studies demonstrate that improved financial literacy leads to better financial decision-making (Lusardi and Mitchell 2011; Van Rooij et al. 2011; Van Rooij et al. 2012).

Particular emphasis should be directed towards enhancing the understanding of probabilities. This, in turn, can help individuals gain a clearer grasp of the likelihood of events impacting their finances and develop strategies to deal with them. By doing so, they can be “more successful in planning their own futures, knowing when they should seek advice from experts, and helping them effectively use such advice” (Hudomiet et al. 2018, 1).

One approach is to integrate financial education programs in Dutch schools (as discussed by Kalwij et al. 2019). The levels of financial literacy in the Netherlands are still low and there is a need for enhancing knowledge of basic concepts. Exploring strategies to encourage Dutch households to save and build up financial reserves could also prove beneficial. A potential avenue is financial education initiatives in the workplace, with a particular focus on financially fragile subgroups of the population. In 2019, the Money Wise Platform launched a new mission, outlining actions to improve financial literacy and financial well-being within fragile groups (Money Wise Platform 2019). An additional policy option involves raising awareness among financially fragile households about the availability of special social assistance for covering “necessary” expenses.

Beyond household-level interventions, our findings also offer insights for policy. Widespread financial fragility can amplify economic shocks through reduced consumption, delayed debt repayments, or increased reliance on informal credit channels. These behaviors can contribute to wider financial stress in the banking sector and slow down macroeconomic recovery. Given the systemic nature of these risks, central banks and supervisory authorities may consider incorporating household resilience metrics into their financial stability monitoring frameworks. Enhancing household financial resilience is not only a matter of individual well-being, but also a pillar of broader economic stability.

In conclusion, our research has illuminated the financial fragility of Dutch households and the strategies they would resort to when faced with a financial shock. It has examined the interplay between financial fragility and financial and probability literacy, as well as various demographic and economic characteristics. Through this research, we were able to identify the groups most at risk and uncover some of the drivers of financial fragility.

Given the many changes due to the pandemic, additional data collection on financial fragility and the determinants of financial fragility is highly desirable. First, the probit models explaining the probability to cope and the choice of coping methods lack some important control variables. For instance, it might be important to control for the financial situation of family/relatives when seeking financial support from them. Unfortunately, the Centerpanel dataset does not contain variables, such as education of mother/father/siblings or inheritance received in the past that could act as a proxy for the financial status of relatives. Other important determinants such as intergenerational transfers, credit constraints, and economic shocks are not observed as well.

Second, one could collect additional information to examine the sequence in which individuals resort to various coping methods. We have seen that financially fragile respondents also use other coping methods than drawing on their bank resources (if they have any). An interesting question is what they do next—whether they go to work more or ask for help from family or friends and how this order is affected by financial literacy or probability literacy.

Third, a limitation of our analysis is its cross-sectional design, which provides only a snapshot of financial fragility at one point in time. Therefore, future research could benefit from a longitudinal analysis to examine how financial fragility evolves over time and how coping strategies shift in response to changing life circumstances or economic conditions. Such analysis would further enrich policy interventions.

Finally, future research should aim to study potential remedies for financial fragility. In short, several Dutch households find themselves skating on thin ice, prompting the need to devise policies and initiatives that can provide support to these households, were the ice going to break.

Author Contributions

Jasmira T. E. Wiersma: conceptualization, formal analysis, methodology, writing – original draft, writing – review and editing. Rob J. M. Alessie: conceptualization, formal analysis, funding acquisition, methodology, supervision, writing – review and editing. Adriaan S. Kalwij: methodology, supervision, writing – review and editing. Annamaria Lusardi: conceptualization, methodology, supervision, writing – review and editing. Maarten C. J. van Rooij: conceptualization, funding acquisition, investigation, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.