1 Introduction

The Great Recession brought a sudden halt to the growth path that the Spanish economy had been experiencing since the mid-1990s. In the five years that followed, GDP per capita contracted by 8.6% and unemployment soared to 26.9%. Part of the devastating effects of the 2008–2013 crisis can be attributed to the high level of debt that Spanish households had accumulated during the expansionary phase of the economic cycle, which placed many of them in a particularly vulnerable situation in the face of an economic downturn.

Although the economic recovery that followed the 2008–2013 crisis put the country back on the path of sustained growth and reduced its staggering levels of unemployment, it might not have had an impact on the financial situation of households since the employment created in these years was biased towards short (or very short) term employment contracts, with an increase in involuntary part-time employment, and stagnant wages (Felgueroso et al. 2017; Simón et al. 2017; Cuadrado and Tagliati 2018; Conde-Ruiz and García, 2019).

Whilst there is a plethora of research on household indebtedness and financial vulnerability (Costa and Farinha 2012; Albacete and Lidner, 2013; Fasianos et al. 2014; Kim et al. 2016; Terraneo 2018; Daud et al 2019; Scott and Pressman 2019), the main studies on the Spanish case (Casado García and Villanueva López 2018; Aller and Grant 2018) focuses on the ex-post analysis of the determinants of household default after its occurrence, but not on the estimation of the number of potentially unstable households. Our study thus fills this gap in the literature by conducting an ex-ante analysis aimed at determining the share of households in a potentially vulnerable situation that could trigger severe financial stress—which could even lead them to default on their outstanding debt.

Given the far-reaching effects for Spain of the economic slowdown caused by the global COVID-19 pandemic and the subsequent 2021–2022 inflation surge, we consider it necessary to perform the aforementioned ex-ante analysis to establish how prepared were Spanish households for a crisis of this scale and what can be expected in the future in light of this scenario.

The purpose of this research, therefore, is twofold: (i) on the one hand, it seeks to determine the financial situation of Spanish households before and after the financial crisis of 2008, in order to establish their capacity to cope with the stresses caused by the looming recession; and, (ii) on the other hand, after classifying households according to their level of default-risk, it aims to identify the economic and socio-demographic determinants of the financial instability of the most vulnerable households.

To achieve these objectives, we will rely on microdata from the Household Finance and Consumption Survey (HFCS). This survey, designed as an equivalent for the US Survey of Consumer Finances, provides income, wealth, debt and consumption data and a set of relevant socio-demographic variables for a representative sample of Spanish households, thereby ensuring the reliability and comparability of the results obtained.

This paper is organized as follows. Section 2 presents a review of the literature discussing the classification of households by their level of default-risk and also outlines the main previous studies on household financial vulnerability and their main drivers. In Sect. 3, we describe the data used for the study, its source, and its most relevant characteristics, and we define the variables to be used in our study. In Sect. 4, we perform a descriptive analysis of the indebtedness of Spanish households and identify vulnerable and potentially vulnerable households using the aforementioned classification. In Sect. 5, we estimate the determinants of the financial vulnerability of the households identified as such in the previous section and discuss the results obtained. Finally, Sect. 6 summarises the main findings of our research.

2 Literature review

There is a large body of scientific literature on household indebtedness, its determinants and the factors that lead indebted households to suffer financial distress (Anderloni and Vandone 2008). These studies are generally based on the life cycle hypothesis (LCH) and the permanent income hypothesis (PIH), although more recent research have questioned the validity of several of their key assumptions and their effectiveness to explain the actual spending, saving and borrowing behaviour of households (Deaton 1992).

The LCH suggests that households plan their expenditure and savings throughout the life cycle, trying to maintain more or less stable consumption levels, so they will be inclined to borrow in circumstances where their income is unexpectedly cut (Modigliani and Brumberg 1954, 1979; Ando and Modigliani 1963). The PIH states that the individuals' consumption is not only determined by their present income level, which is transitory, but by their expected income over their lifetime, which is permanent, so that borrowing and spending will ultimately depend on this permanent income estimated by the individuals themselves (Friedman 1957).

While there is no single definition of financial vulnerability, most of the studies cited later in this section use a limited set of indicators that link outstanding debt or debt payments to income or gross wealth to determine whether a household is under financial stress (ECB 2013). When a household exceeds a certain threshold on these indicators, it is considered to be at risk of suffering difficulties in the near future, even though it may not be in default at the present moment (Leika and Marchettini 2017).

The implementation of the Household Finance and Consumption Survey (HFCS, henceforth) at European level has created the opportunity to produce research on the financial situation of European households that can be compared over time and between countries. These surveys, carried out at national level by 18 euro area countries every three years, collect harmonised data on assets, liabilities, income and consumption from a representative sample of households within each country.

One of the first precedents in the study of household financial vulnerability using this type of data is the report by Costa and Farinha (2012) on Portuguese household indebtedness in 2010. First, while the level of indebtedness of Portuguese households is one of the highest in the euro area, the upward trend of recent decades has been interrupted by the credit crunch triggered by the Great Recession. Moreover, the results obtained indicate that low-income and younger households with outstanding mortgage debts are the most financially vulnerable groups of the population. However, the authors also suggest that the limited participation of lower-income households in debt markets reduces the potential impact of their higher probability of default on the financial situation of creditors.

In a similar vein, Albacete and Lindner (2013) conducted a study using data from the Austrian HFCS. According to the authors' research, Austrian households have a generally high debt burden, and low-income and low-wealth families are found to be particularly vulnerable—especially, the ones with an unemployed head of household. However, the contribution of indebtedness to the financial vulnerability of these households appears to be low.

Exploiting the cross-country comparability of HFCS data, Fasianos et al. (2014) study household indebtedness in five European countries in 2010. The authors draw conclusions consistent with the research presented earlier: an overall increase in household indebtedness, and a relationship between financial fragility and household income, as well as the age and educational level of the household head.

Later, Terraneo (2018) surveys the financial vulnerability of households in four southern European countries (Greece, Italy, Portugal and Spain) using data from the first wave of the Eurosystem Household Finance and Consumption Survey conducted between 2009 and 2010. The author notes that, in the period analysed, Spain and Portugal were the countries with the highest debt participation and the highest proportion of vulnerable households of the four analysed, followed by Greece in an intermediate position and Italy rather further behind. Having taken on more debt than they could repay put many Spanish households in a delicate situation in the event of loss of income due to unemployment, a particularly pervasive phenomenon during the years analysed in the paper.

Regarding the Spanish economy separately, Casado García and Villanueva López (2018) studied the indebtedness of its households based on data from the HFCS, focusing on households that declared their bankruptcy in the twelve months prior to the survey. Thus, the authors highlight the significant increase in households unable to pay their debts during the period under analysis. The estimates attribute this increase to the fall in income resulting from the Great Recession and the loss of employment of the head of household. Meanwhile, Aller and Grant (2018), who study the determinants of the arrears rate using HFCS data from 2002 to 2011, find that lower-income and unemployed households are more likely to default, as are younger households and households with lower educational attainment.

Similar studies using household microdata for countries outside the European Union include the cases of South Korea (Kim et al. 2016), where the authors construct indicators of Korean household over-indebtedness and estimate the possibility of future defaults; Malaysia (Daud et al 2019), where the authors detect that both household indebtedness and potential vulnerability to crisis scenarios are increasing, and that both economic and sociodemographic variables are among the determinants of this vulnerability, but also point at the importance of attitudes towards risk; or the United States (Scott and Pressman 2019), where the authors, using a Minsky-inspired framework, observe that the proportion of financially vulnerable households has been rising for more than three decades, a fact that is alarming in the face of possible economic downturns.

In summary, the studies discussed in this section point not only to purely economic factors as an explanation for financial vulnerability, but also to socio-demographic variables (such as age, education level, gender, household size, etc.) to explain household over-indebtedness. However, some studies also hint at the importance of attitudes towards risk, which are difficult to capture in standardised surveys such as those used in most of the literature, to explain whether a household becomes financially distressed (Husniyah and Fazilah 2011; Daud et al. 2019).

3 Data source and variables

3.1 Data source

As stated earlier, the data source used to conduct this analysis is the Household Finance and Consumption Survey (Encuesta Financiera de las Familias, in Spanish—henceforth, HFCS). The HFCS is the only statistical source available in Spain that allows linking the income, assets, debts and expenditure of each family unit. This survey provides information on income, assets and liabilities, and consumption, as well as a wide range of socio-demographic variables, for a representative sample of households.

The Bank of Spain, which has conducted this survey every three years since 2002, guarantees the representativeness of the HFCS by using stratification techniques and oversampling of the most affluent households in order to include a large number of households with high net worth and a wide variety of assets in their portfolio. Subsequently, each household is assigned a weight, which must be used for any calculations made with the microdata.

In addition, in the case of non-response to certain items, five imputations are made for the missing data based on other information provided in the survey by the respondents. These five possible versions of the database must be used together to produce any estimates.

Both the descriptive and econometric analysis presented in later sections will use data from the seven available waves of HFCS in Spain (2002, 2005, 2008, 2011, 2014, 2017 and 2020), in order to track the evolution of Spanish household indebtedness for almost two decades.

3.2 Definition of the variables

According to the HFCS, outstanding household debt is defined as the sum of the outstanding debt of all household members aged 16 and over. As identified in the survey questionnaire, this includes debts from loans with mortgage guarantee used for the purchase of real estate properties, outstanding debts from mortgages and other secured loans not related to the purchase of real estate assets, outstanding debts from personal loans, outstanding credit card balances, and other debts not contained in the preceding categories.

Gross household income is defined as the sum of the pre-tax income of all household members aged 16 and over, both in cash and in kind; while gross household wealth is defined as the sum of the real and financial wealth of all household members older than 16.

Regarding the socio-demographic variables of the households, we will use the number of household members (from one to five or more) and the following data of the head of the household: age, gender, employment status (employed, self-employed, retired, or other type of economic inactivity), and highest educational level attained (primary, secondary, or tertiary education).

Finally, to account for changes in purchasing power over time, all amounts of monetary variables have been converted to euros in 2017 using the factors provided by the Bank of Spain.

4 Descriptive and econometric analysis

4.1 Descriptive analysis

Aiming to analyse how debt is distributed among Spanish households and how this distribution has changed throughout the twenty-first century, we will split the population according to economic (gross income and gross wealth) and socio-demographic variables (gender, age, household size, educational level and labour status) described in the previous section.

As shown in Fig. 1, more than half of Spanish households have some form of outstanding debt. This percentage increased during the central years of the 2000s, when access to credit was easier, stagnated during the years of the Great Recession and rose sharply again when the country returned to growth in the second half of the 2010s.

Fig. 1
figure 1

Percentage of households with outstanding debts (total and by income quintile), 2002–2020

Full size image

Table 4 in the Appendix 1 shows the debt ratio for Spanish households according to different classification criteria (wealth quintile, number of household members, gender, age, educational level and employment status of the head of household).

Regarding the gender of the reference person, households where the main breadwinner is female significantly increased their debt participation during the expansion phases of the economic cycle. Meanwhile, households headed by men, which already started from a significantly higher debt participation, are also becoming increasingly indebted over time. In this regard, we can state that the convergence in debt participation between these two groups of households came to a halt during the 2008–2013 economic crisis, but has resumed during the post-crisis economic recovery.

In terms of the age distribution, as might be expected, younger households tend to be more indebted and debt participation declines as the age of the household head increases, which is consistent with the assumptions of the life cycle hypothesis presented in a previous section. However, in households where the main breadwinner is under the age of 35, a differing trend can be observed: while in the other age groups indebtedness generally grew since the early 2000s, in younger households we detected a break in this increase and a drop in the percentage of households in debt with the end of the 2008–2013 recession, and an expansion of the debt participation since then. This finding might be related to the work instability experienced by Spanish youth during the post-crisis years (Sanz-de-Galdeano and Terskaya 2020), which, coupled with tighter access to credit, might have hindered the accumulation of debt in younger households until the economic recovery that unfolded during the second half of the 2010s.

Concerning the number of persons in each family unit, as expected, the indebtedness grows in parallel with the household size, reaching almost two thirds of the households of 3 or more persons. Nonetheless, the survey registers a quite substantial increase in the debt participation of single-person households, still the least indebted, which could be a source of vulnerability in the event of a financial shock that negatively affects the income stream of the sole member of the household.

Moreover, if we analyse the distribution of the debt burden related to the educational level of the head of the household, we find that the group of households with the highest debt participation are the ones where the main respondent has only completed secondary education, while households with a university graduate head fall slightly below. Given that the average income level increases with the educational level of the head of household, it is not entirely unexpected that the most indebted households are those with the greatest repayment capacity. Nonetheless, the fact that both groups have similar levels of indebtedness suggests that households in the second group might need to complement their income with additional debt to meet their consumption needs.

Regarding the classification by employment status, as in the two previous cases, all the groups analysed see an increase in the percentage of households with some type of outstanding debt. Households with an economically active main respondent have a much higher debt participation than the population average and, depending on the type of employment of the head of household, ranges between 60 and 75 percent. Besides, it is particularly striking the case of inactive non-retired households, who despite the obvious difficulties in accessing credit without a formal employment, must resort to furthering their indebtedness in order to achieve their consumption goals.

Finally, in terms of the strictly economic variables, the percentage of households in debt tends to grow with their gross income, reaching nearly 70% in the most affluent ones. Yet we must point out that it is the households at both ends of the distribution where the debt participation has grown the most, while the percentage of households in debt in the second and third quintiles decreased temporarily since the beginning of the 2008 crisis, only to experience further expansion during the next expansionary phase of the economic cycle. Particularly noteworthy is the case of the bottom 20%, where the proportion of households with outstanding debts has almost doubled despite their glaring difficulties in accessing credit.

In terms of gross household wealth, the ratio of indebted households increased significantly for households in the bottom 80%, compared to a much more moderate increase for the top 20%. Unlike the case of gross income, the most indebted households are not at the top of the distribution, but around the median where the proportion of households with outstanding debt is well above 50 per cent in all years of the sample. On the other hand, the persistent growth of household debt in the two lower quintiles of the distribution is again noteworthy, and contrasts with the mild drop experience by the remaining three groups after the 2008 crisis—from which they have, however, recovered since 2014.

Furthermore, if we examine in Fig. 2 the volume of debt accumulated by Spanish households over the last two decades, it is clear that there has been an extraordinarily large increase, which, nonetheless, reached its peak during the fourth wave of HCFS conducted between 2010 and 2011. Although Spanish households as a whole began to deleverage from that point on, a breakdown by group shows that for certain groups this increase in debt never stopped.

Fig. 2
figure 2

Average outstanding debt of indebted households in euros (total and by income quintile), 2002–2020

Full size image

As shown in Table 5 in Appendix 1, particularly remarkable is the case of households in the upper four quintiles of gross wealth, which, in addition to having increased their debt participation, experienced increases of around 50–60% in their outstanding debt between 2002 and 2020. This accumulation of debt in households in the lower half of the distribution—although not in the bottom 20%—could be a source of financial vulnerability in the face of an economic downturn, due to the strong pro-cyclicality of employment and wages of low-income households in Spain (De la Roca 2014).

Meanwhile, other groups that experienced strong relative increases in debt were families with a retired head of household—especially if the head of household is between 65 and 74 years-old. Although they are still the least indebted age group and the segment with the most stable income source, this behaviour seems to contradict the results predicted by the life-cycle hypothesis and calls for greater attention. In contrast, there are no major gender differences in the volume of debt accumulated during the period under analysis.

Households with more than three members, the groups with the highest debt participation, also experienced above-average increases in their debt stock, which places them as the most indebted family units by far, whereas at the beginning of the period they had the least outstanding debt.

Finally, households led by an economically inactive individual (mainly unemployed) are another group that show an increase in indebtedness above the average. Although this increase seems to have slowed down markedly during the central years of the last economic crisis, the fact that this group has increased its indebtedness this much could be a potential source of economic instability.

In short, although most debt remains in the hands of middle-aged households in the upper half of the income and wealth distribution, with more than two members, higher education and economically active, our analysis suggests that this bias has gradually been fading in favour of an extension of indebtedness to traditionally low-debt households.

4.2 Financial vulnerability classification

Once the general situation of the debt in Spanish households has been contextualized, we will proceed to classify them into groups according to their level of financial fragility.

Considering the multidimensional nature of household financial vulnerability, we will use several indicators that connect the level of indebtedness and the yearly debt payments that the household has to face with their present income and wealth. This classification is based on the approach proposed by Minsky in his Financial Instability Hypothesis (FIH), which grouped firms into three categories of increasing default-risk (Minsky 1992). Our study targets what the author calls Ponzi borrowers, for whom indebtedness grows steadily and full repayment is not possible.

The first of these indicators to be used to perform the classification is the debt-to-assets ratio (DA), or leverage ratio, and can be calculated by dividing each household's total liabilities by its gross wealth. This ratio provides information on the leverage of households and, ultimately, on their ability to repay any outstanding debt with their assets at current market value and whether they might need to deleverage in the medium-term.

The second indicator is the debt payments-to-income ratio (DPI), or debt service ratio, and can be calculated as the ratio of debt payments made to gross monthly or yearly income for each household in the sample. This indicator provides information on the short-term ability of households to pay back the debt they contracted in the past and their resilience to unforeseen exogenous shocks—such as job and/or income losses, rising interest rates, etc.

The third and final indicator to be used is the debt-to-income ratio (DI), which is calculated by dividing the household's total liabilities by its gross annual income. This variable captures the future liquidity of the household by relating their ability to service their debt burden to their current income level.

To determine which households are in a potentially risky position, it is necessary to establish at what levels of the three aforementioned ratios a household can be considered to be in a financially vulnerable or unstable situation. Thus, based on the previous literature reviewed in a preceding section, the most commonly used vulnerability thresholds for each ratio are: 0.75 for the DA ratio, 0.4 for the DPI ratio, and 3 for the DI ratio (Bricker et al. 2011; Costa and Farinha 2012; Albacete and Lidner 2013; ECB 2013; Terraneo 2018).

Applying the above classification, it is possible to note how the percentage of households exceeding the thresholds set for each ratio has increased sharply since the first wave of the HCFS in 2002 (Fig. 3). The ratio with the largest proportion of vulnerable households, and where this proportion increased the most in the period under analysis, is the one that connects gross household income and outstanding debt (DI ratio). This increase reflects the fact that the median number of years it would take a Spanish household to pay off its current debt using its entire income doubled between 2002 and 2014 from 0.72 to 1.54. This rapid increase is the result, on the one hand, of the massive increase in the debt stock of Spanish households described in the previous subsection and, on the other, of the wage stagnation experienced by most workers during and after the last crisis.

Fig. 3
figure 3

Percentage of households above the threshold for the three vulnerability ratios, 2002–2020

Full size image

This increase, shown in Table 5 of Appendix 1, was the result of a sharp rise in financially vulnerable households in the central and upper sections of the income and wealth distributions. We also found that this increase is greater in younger segments of the population pyramid, which is also apparent in the classification by employment status, where all groups experience a dramatic increase in financial fragility except retirees.

Regarding the DA ratio, which measures the leverage of households, there is a steady increase in the percentage of vulnerable households, with a particularly sharp leap in the post-crisis years. These results suggest that, although the average outstanding debt held by Spanish households has reached its peak during the Great Recession, the number of over-leveraged households has not stopped growing—probably, as a result of the collapse of the housing bubble and the decline in the value of real estate assets held by households.

As shown in Table 6 in Appendix 1, the increase in vulnerable households according to the debt-to-asset ratio is concentrated in income- and wealth-poor households with a relatively young head of household and an above-average number of members. In terms of educational level and employment status, we do not identify clear linear trends, as over-leveraged households are more often those with a high school graduate head of household, who may be either economically inactive or active, working as employee.

Meanwhile, the DPI ratio, which focuses on the short-term financial position of households, also points to a persistent if somewhat milder increase in vulnerability during the Great Recession, and a return to pre-crisis levels during the last wave of the survey, that captures an expansionary phase of the economic cycle. Considering that this sharp drop took place when the European Central Bank started to apply extremely lax and non-conventional monetary policies, it is likely that this drop is due to the decrease in interest rates that families have to face on their debts and later to an improvement in their financial situation. In view of this, although this near-zero interest rate policy might benefit many households in the short term, it is likely that the significant increase in interest rates from 2022 onwards could have a devastating impact on the financial sustainability of a large number of Spanish families.

In this case, the increase in the ratio, which targets short-term financial problems, was fuelled by households at the lower end of the income distribution and at the centre of the wealth distribution, which have sufficient wealth to access collateralized debt. Also, this increase is generally more relevant in older households, with limited educational attainment, an average household size, and working employed or retired (Table 7 in Appendix 1).

Finally, the number of households experiencing some kind of financial distress has almost tripled since the early 2000s, reaching almost one in every five Spanish households. Moreover, the percentage of households exceeding the threshold in two or even three of the ratios also points to an even greater increase in vulnerability than what each indicator captures separately. An indicative of this is the number of households simultaneously experiencing excessive leverage, short-term financial distress and future liquidity problems, which has increased fourfold in fifteen years. In any case, our results suggest that the income and wealth of a growing share of Spanish households have grown at a slower pace than their debt payments and liabilities, placing them in an increasingly fragile and vulnerable situation in the face of an exogenous shock.

For a more detailed picture of the evolution of the number of financially vulnerable households according to each ratio, see Tables 6, 7, 8 included in Appendix 1.

To better understand the situation of Spanish households in the European context, we turn to the Eurosystem's Household Finance and Consumption Surveys, which have been carried out regularly since 2010 in a harmonised manner for a group of fifteen to twenty-two euro area countries (Household Finance and Consumption Network 2013, 2016, 2020, 2023).

The results of successive waves of this HFCS show that Spanish households are slightly above the European median in terms of their debt-to-income ratio, albeit with an increase of more than a third in over ten years. However, they are still well behind the Netherlands, Finland and Portugal, whose households are consistently the most vulnerable in the euro area in terms of this ratio.

With regard to the debt-to-income ratio, Spanish households are well above the European average, although with a noticeable decrease over the period analysed, with a median value for this ratio that is only surpassed by the Netherlands and Portugal, but also by Cyprus, Luxembourg and Malta.

Finally, with regard to the ratio of debt service to income, Spanish households once again have one of the highest median values in the group of Eurosystem countries analysed, this time only behind France and Cyprus, although with a notable reduction over the decade.

These additional data complement the results of our analysis by showing a reduction in vulnerability in two of the three ratios over the 2010s, but still place Spanish households among the most vulnerable in the euro area, along with their counterparts in the Netherlands, Portugal and Cyprus. Moreover, the picture provided by these data is incomplete, as it is not possible to compare the situation before the Great Recession at the European level.

5 Econometric analysis and discussion of results

To explore the economic and socio-demographic determinants of financial vulnerability in Spanish households, we will use, as has been done in recent studies on this topic, regression models for categorical and limited dependent variables. In these models, the dependent variable takes only two possible values, 0 and 1, representing the occurrence or non-occurrence of a certain event.

In our case of study, this event is whether or not the threshold for at least one, two or all three of our indicators is exceeded, and therefore the model will estimate which variables have an impact on a particular family unit belonging to the group of vulnerable households.

For our estimation we will use the logit regression model and it can be written as follows:

$$\begin{aligned} {\text{logit}}\left( {{\text{y}}_{i} } \right) & = \beta_{0} + \beta_{1} {\text{ln}}\left( {gross\_income} \right) + \beta_{2} {\text{ln}}\left( {gross\_wealth} \right) + \beta_{3} gender \\ & \quad + \beta_{4} age + \beta_{5} hhsize + \beta_{6} educ\_att + \beta_{7} emp\_status + \varepsilon \\ \end{aligned}$$
(1)

where there are three possible dependent variables, each one taking the value 1 when a certain household falls into the vulnerable group according to the ratios and 0 if not, and where the independent variables are the natural logarithm of the household's gross income, the natural logarithm of the household's gross wealth, the size of the household, and the following characteristics of the head of household: age, gender, educational level and employment situation—the last three in the form of dummy variables.

Nevertheless, in order to avoid endogeneity problems, gross wealth and outstanding debt are not included as regressors in the debt-to-assets regression, gross income is not included as a regressor in the debt-payments-to-income regression, and gross income and outstanding debt are not included as regressors in the debt-to-income regression.

The outcomes of the regressions (Tables 1, 2 and 3) confirm, as can be inferred from the results of the previous section, the importance of both economic and socio-demographic variables when explaining the probability of a household being classified as vulnerable according to any of the three ratios.

Table 1 Logistic regression for vulnerable according to the debt-to-assets ratio threshold, 2002–2020
Full size table
Table 2 Logistic regression for vulnerable according to the debt-payments-to-income ratio threshold, 2002–2020
Full size table
Table 3 Logistic regression for vulnerable according to the debt-to-income ratio threshold, 2002–2020
Full size table

Overall, the findings suggest that the probability of having overleverage problems decreases, as expected, with household income. Moreover, the level of gross household wealth reduces the likelihood of households experiencing short-term debt servicing problems, but increases the likelihood of over-indebtedness problems as measured by the debt-to-income ratio.

This somewhat disconcerting result can be explained by the fact that, on the one hand, a higher household wealth makes it easier to absorb the negative effects of an unfavourable macroeconomic scenario, reducing the possibility of falling into the group of vulnerable households; but, on the other hand, a relatively large wealth may allow the household to take on more debt to smooth consumption, which may lead to over-indebtedness and financial distress.

Furthermore, as the level of outstanding debt tends, predictably, to increase the probability that households will be unable to repay their debt in the short-term, as measured by the debt-to-income ratio.

In contrast, the coefficients associated with the age of the head of household suggest that financial vulnerability is greater for younger households and decreases as the main respondent gets older, even considering the economic factors mentioned above. This result, in line with previous literature, points to young households as one of the most financially vulnerable groups due, among other things, to greater job insecurity and real estate prices growing at a much faster rate than wages, which forces them to take on increasingly larger amounts of debt in order to access home ownership.

Moreover, the relationship between financial vulnerability and household size are complicated to disentangle fully: while the number of people in the household appears to increase the likelihood of vulnerability in the short run in more recent waves of the survey, according to the debt-to-income ratio, regressions on the other two ratios suggest, with intermittent results, that the effect works in the opposite direction.

In turn, the educational level of the breadwinner appears to be hugely important in explaining the financial problems of Spanish families. Having a university education reduces the probability of exceeding the thresholds in all three ratios over the whole period analysed and, in more recent waves, so does having completed secondary education.

Attainment of tertiary education—and, to a lesser extent, completion of compulsory schooling—appears to be very important in explaining the level of financial vulnerability of a household. Our results suggest that dummy education variables, which are highly correlated with economic ones, are able to capture effects on household financial vulnerability that are not reflected in gross income and wealth. As discussed in previous literature, a higher level of education implies greater financial knowledge and skills to manage household resources, which may lead to avoid over-indebtedness, reckless consumption and irresponsible risk-taking, steering them away from more fragile financial situations.

Finally, households with an inactive or unemployed household head are more likely to be in financial distress according to two of our three measures. Conversely, if the household head is retired, the probability of experiencing problems also decreases according to two of our three measures.

The case of households where the household head is self-employed is worth mentioning separately: our results suggest that these households are more likely to experience financial problems in the short term, as measured by the debt service ratio, although the opposite is true for over-indebtedness, as measured by the debt-to-assets ratio. This could be explained by the fact that this group of households typically has much more unstable income flows, alternating between periods of low and high activity. Nonetheless, this instability may also keep them away from over-indebtedness, either by choice or because they have greater difficulty in accessing credit.

6 Conclusions

Our results point to a steady increase of the debt burden of Spanish households since the early 2000s, both because of the growing debt participation of all household groups, and because of the ever-expanding volume of outstanding debt—an upward trend that seems to have come to a halt in the second half of the 2010s.

While most debt remains concentrated in the upper half of the income and wealth distribution, and by relatively young families, the expansion of borrowing has affected traditionally less indebted segments of the population, such as low-income and low-wealth households, or the families with and economically inactive person of reference.

Meanwhile, the surge in vulnerability captured by our three indicators seems to have slowed down in the second half of the 2010s. Nevertheless, the state of Spanish households to seems to be relatively similar to the one they had just before the Great Recession.

Moreover, our findings, in line with previous literature, suggest that the determinants of financial vulnerability are not founded exclusively on economic variables, but also on socio-demographic variables such as age, household size, and educational level. As an example, for the same income and wealth level, households with a younger or less educated head of household are more likely to belong to a financially fragile group.

These results raise two major concerns. First, in the immediate term, the income loss caused by the ongoing COVID-19 pandemic and subsequent inflation surge had a sizeable impact on the financial situation of many Spanish households, which surely has pushed more households into more strained positions, increasing the general levels of financial instability. Second, every increase in ECB interest rates to curb prices would be immediately passed on to adjustable rate mortgages and other liabilities, which could put households at the bottom and, particularly, middle of the distribution back into the situation of high financial distress experienced during the last recessionary phase.

In short, this scenario shows the intense dependence of a large proportion of Spanish households on the expansionary economic policies—fiscal and monetary—carried out in recent years. At present, given the very high public debt-to-GDP ratio of the Spanish economy and the very low interest rates applied by the ECB, our country is therefore facing a rather complex situation.

Given the underperformance of wage growth, the value of the main residence seems to have a special relevance in explaining the financial situation of Spanish households. In a market that has been defined in recent decades by price increases well above workers' compensation, it is recommended to undertake public policies that allow access to housing at reasonable prices for the younger disadvantaged families, so that the house purchase does not imply an excessive borrowing that may later become unpayable.