Housing cost burden and life satisfaction

Abstract

The share of income spent on housing varies across individuals and countries but it has been increasing over time in a wide range of countries, particularly among lower income households, rising housing affordability as a prominent challenge in higher income economies. Variations in share of income spent on housing can reflect variations in household preferences but when more than a certain level of income is spent on housing, households face tradeoffs between housing and non-housing consumption that are expected to negatively affect their overall life satisfaction. Using data from the 2018 European Union Statistics on Income and Living Conditions (EU-SILC) for 14 countries we find that, controlling for household sociodemographic characteristics, households spending more than 30 percent of their income and those spending more than 50 percent of their income on housing report lower levels of life satisfaction, with the latter group reporting the lowest level. The negative relationship between housing cost burden and reported life satisfaction is found across countries but varies in magnitude, which points to the need to further investigate the mechanisms behind the association between housing cost burden and life satisfaction and the role of country-specific effects, including differences in welfare systems, in moderating this association.

Introduction

The large number of households spending a disproportionately high share of their income on housing, experiencing high housing cost burdens, is a well-documented source of concern across the advanced economies (Bratt 2002; Deidda 2015; Shamsuddin and Campbell 2021; Yoko 2008). The main concern is that spending above a certain share of income on housing reflects constraints caused by the high minimum cost of securing access to shelter, a universal basic need, as opposed to household preferences for higher housing services consumption. These constraints can formalize in multiple ways, and we would expect the negative tradeoffs associated with housing cost burdens to be reflected in a person’s stated life satisfaction.³ This paper explores differences in life satisfaction among households experiencing varying levels of housing cost burden in 14 European countries.

While high housing cost burdens are expected to negatively affect a household’s life satisfaction if it is a constrained outcome, there is little direct empirical evidence of this relationship. Existing research looking at the relationship between housing and life satisfaction largely focuses on tenure and on housing satisfaction. There has been far less research on housing cost burdens and wellbeing although in the US context, Shamsuddin and Campbell (2021) provide evidence that high housing cost burden is associated with increased likelihood of experiencing material hardship, which includes food insecurity, difficulty paying bills and forgoing medical care although they do not have a direct measure of overall life satisfaction or wellbeing.

This paper analyzes the relationship between housing cost burden and life satisfaction, controlling for households’ socio-demographic characteristics, and discusses this relationship in the context of a broader set of national factors that likely affect the budget tradeoffs households have to make. We use the European Union Statistics on Income and Living Conditions (EU-SILC) for 14 countries, in particular two measures of housing cost burden: 1) the share of income spent on housing and 2) how burdensome housing costs are perceived to be. We use a question on overall life satisfaction asked as part of the 2018 special module on material deprivation, wellbeing, and housing difficulties as our main dependent variable.

We find that households spending more than 30 percent of their income on housing report significantly lower levels of life satisfaction and those spending more than 50 percent report even lower levels. Many households report feeling housing cost-burdened, when they numerically are not. But although the perception of burden extends more broadly than actual burden, only those with actual housing cost burdens report lower life satisfaction as well. These within-country effects vary across countries. In particular, higher cost burdens are more strongly associated with lower life satisfaction in Anglo-Saxon countries than in Northern and Western European countries. These findings suggest that welfare regimes might mediate the impact of housing cost burden on life satisfaction.

Section I reviews the literature on the link between housing and wellbeing, with a focus on the theoretical determinants of life satisfaction and how welfare and housing costs may affect this relationship. Section II describes the EU-SILC data and the operationalization of the housing cost and wellbeing measures and the empirical model. Section III presents the results, finding strong negative relationships between higher levels of cost burden and reported life satisfaction with some systematic variation across countries. Finally, section IV provides a concluding discussion of the implication of the results and the potential for further research to explore the mechanisms that may explain the results found in this paper.

I. Literature Review:

We start by reviewing the literature on life satisfaction and the contribution of housing to it. We then discuss how housing cost burden might affect life satisfaction and how differences in welfare state might mediate the relationship between cost burden and life satisfaction. We summarize these conceptual relationships in a conceptual model that illustrate our hypotheses.

A). Life Satisfaction and Housing

A household’s life satisfaction is a function of its level of satisfaction among many domains, with housing being one of them, alongside health, financial situation, job, leisure, housing, and environment (van Praag et al. 2003; Rojas 2007). Housing is an important source of ontological security for households, defined by Giddens (1991: 92) as “the confidence that most human beings have in the continuity of their self-identity and in the constancy of their social and material environments.” It is therefore a fundamental element of household wellbeing.

Traditionally, research on housing and household wellbeing has focused on objective measures of wellbeing such as mental health, nutrition, and stress levels (Bentley et al. 2011; Bratt 2002; Rowley and Ong 2012) and material deprivation (Shamsuddin and Campbell 2021). Subjective measures of wellbeing are also important to examine. There is some extant research on tenure, housing conditions, and measures of subjective wellbeing (Bentley et al. 2011; Bratt 2002; Elsinga and Hoekstra 2005; Herbers and Mulder 2017; Lu 1999), as well as on housing cost burdens and subjective wellbeing (Cracolici et al. 2011; Sunega and Lux 2016; Rowley and Ong 2012). This paper explores the relationship between housing cost burden and subjective overall wellbeing as estimated by reported household life satisfaction.

Data on subjective overall wellbeing is important for understanding human welfare (Herbers and Mulder 2017; Diener and Suh 1997) because it is more likely to account for the multifaceted nature of wellbeing (Cracolici et al. 2011). Subjective measures of wellbeing reflect the concerns of individuals, which are the foundation of policy formation (Diener and Suh 1997). While there is often a disconnect between objective and subjective measures of wellbeing (McCrea et al. 2006), such discrepancies also have important implications for policies aimed at improving societal welfare. For example, a household may have an objectively high level of income, but state that they have a low level of wellbeing. On the one hand this subjective measure could be deemed unwarranted and discarded, but on the other, evidence shows that wellbeing is often a function of feelings of relative deprivation (Herbers and Mulder 2017). This means that even when an individual is not objectively poor, they may feel poor compared to their peers. Such feelings translate into consumption levels, policy priorities, and, according to Lu (1999), the public services that households demand.

Much of the existing research on the relationship between wellbeing and housing has focused on difference in housing satisfaction across tenure. One study focusing on eight European countries finds that homeowners are more satisfied with their housing than renters, but less so in countries with unitary housing regimes according to Kemeny’s definition (Elsinga and Hoekstra 2005). Other studies have found that owners have higher social wellbeing than renters overall (Zumbro 2014, Rohe and Bassolo 1997). Additional research points to the fact that owners without a mortgage may have higher social wellbeing than those with one (Cairney and Boyle 2004). The nuanced relationship between tenure and wellbeing is apparent in a study finding that in Germany, homeownership is associated with higher life satisfaction, but only if the home is being maintained well (Zumbro, 2014). As a result, rather than focusing only on owners and renters, we included both tenure types in our analysis but control for tenure.

Another area of research on the relationship between wellbeing and housing has focused on satisfaction with housing quality and location. This research generally finds a positive association between housing quality and overall wellbeing (Gandelman et al. 2012; van Praag et al. 2003; Zumbro 2014). In Uruguay, Gandelman et al. (2012) find that housing with higher levels of access to public good is associated with higher levels of reported happiness. Van Praag et al. (2003) find that higher level of housing satisfaction is associated with higher level of general satisfaction controlling for satisfaction with other life domains and individual characteristics using a sample of German individuals. Zumbro (2014) also using German data find that housing quality issues such as need of repairs and renovations is associated with lower levels of overall life satisfaction.

B). Housing Cost Burden and Life Satisfaction

Research suggests there is a relationship between housing expenditures and reported economic wellbeing, with lower housing expenditures being associated with higher levels of economic wellbeing in Baltic countries (Streimikiene and Kiausiene 2013). Using the Italian EU-SILC data, Cracolici et al. (2011) find that self-reported financial strain—most notably housing cost burden, the inability to buy clothes, and the inability to go on holiday—exerts a powerful influence on economic wellbeing measured by the perceived ability to make ends meet. It is important to note that self-reported financial strain may be different than objective measures of such strain. For example, using EU-SILC data, Sunega and Lux (2016) find that in all European countries they study (with the exception of Norway and Sweden), more households report being housing cost burdened than are actually burdened based on standards definition of cost burden. In general, people perceive themselves as having higher financial burdens in areas with greater inequality and these feelings vary across countries (Brandolini et al. 2013). Using EU-SILC data, Balestra and Sultan (2013) find that an objective measure of housing cost burden is associated with worse levels of satisfaction with housing while subjective measures of cost burden are associated with higher levels of satisfaction. In the Australian context, Rowley and Ong (2012) find that higher levels of cost burden (what they call housing stress) is associated with worse reported housing and location satisfaction, and that this relationship increases in magnitude for households with a higher level of cost burden.

Given that housing is often the largest expenditure for households, it is no surprise that its actual and perceived cost has implications for household welfare. Marcuse and Madden (2016) note that high housing costs can contribute to anxiety and feelings of instability. In turn, they argue that such feelings can compromise the ability for housing to contribute to “ontological security” and overall sense of stability in the world. As noted by Bratt (2002), an increased share of household income spent on housing will also come at the cost of consumption of other goods and services, and this tradeoff can be particularly damaging to the lowest-income households. An article by Shamsuddin and Campbell (2021) in the US examine some of these trade-offs and provide convincing evidence of increased likelihood of experiencing food insecurity, bill-paying hardship and medical care hardship among cost burdened households, particularly for those paying more than 50 percent of their income on housing.

Combined, this evidence indicate that housing cost burdens may reduce a household’s welfare by forcing it to make tradeoffs with other basic needs, thus contributing to a larger sense of instability and precarity that affects the household’s overall wellbeing. We would therefore expect that households most burdened by housing costs will have lower levels of reported overall life satisfaction.

However, for some households, a higher housing cost burden may reflect a choice to allocate a higher share of their budget to housing in order to enjoy better housing quality or higher-amenity locations rather than financial constraints and they might still be able to obtain their preferred level of consumption of non-housing goods and services.

For households who attain higher levels of housing quality through allocating a higher share of their income to housing, they could reach a similar or higher overall life satisfaction, consistent with research that finds a positive association between housing satisfaction and overall wellbeing (Gandelman et al. 2012; Healy 2003; Streimikiene and Kiausiene 2013; van Praag et al. 2003; Zumbro 2014). The relationship between housing cost burden and overall wellbeing may therefore be insignificant or even positive. The direction of the overall relationship depends on whether high housing cost burdens are primarily driven by financial constraints or by individual preferences towards allocating a larger share of their budget towards housing services while still being able to reach their desired level of expenditures on other domains.

C). Housing Cost Burden and Welfare Systems

There are several structural realities associated with housing that can make high cost burdens inescapable rather than a matter of preference. First, there is a lower bound on the quantity of housing services consumed by any housed person, both in terms of space (at least enough to accommodate a body) and quality (either based on human dignity or enforced building standards). In many settings, space and quality are a function of both local and macro government policies such as zoning and are thus outside of the control of any given individual to adjust. Further, the lowest possible cost to deliver that minimum shelter can represent a substantial share of a household income without any possibility to consume a smaller amount of housing services. In order to obtain shelter, households need to pay that minimum cost. Paying that cost will require households not only to adjust their housing consumption to that lower bound (likely meaning worse housing and location satisfaction) but also to make tradeoffs in terms of consumption of non-housing goods and services (for example, smaller expenditure on food, leisure, health or education). This means there is a constraint on one’s ability to reduce housing consumption in order to reduce one’s cost burden, and beyond that constraint, the only counterfactuals are to find informal housing or to consume no housing at all (being homeless). These forced tradeoffs can be expected to lead to lower overall life satisfaction.

Housing is not usually considered a public provision (Herbers and Mulder 2017) and housing situations and contexts can be quite different across European countries (Elsinga and Hoekstra 2005). We expect the relationship between housing cost burdens and life satisfaction will be variegated based on the larger welfare system of a country. Robust welfare structures that account for and modify the myriad of factors affecting individual utility levels could mitigate the implications of housing cost burdens on feelings of wellbeing, and thus the necessity for housing-based interventions. This is an example of how housing systems interact with a country’s social and economic structures (Stephens 2011) in ways that make comparative housing research both challenging and useful. The impact of housing cost burden on life satisfaction is therefore expected to be more negative in countries in which there is no effective right to shelter and housing is not defined as a universal benefit with requirements for governments to provide it (Herbers and Mulder 2017).

As noted by Stephens (2016), two common methods for grouping countries when studying housing are Kemeny’s rental regime approach (1995) and Esping-Andersen’s welfare regime approach (1990). Other researchers have broken down some of these macro structures by focusing on homeowning versus cost-rental societies (Kemeny 1995). All of these housing regimes partially overlap with broader social welfare regimes (Stephens 2016; Stephens 2020). None of these groupings are perfect ways to organize housing systems; instead, they highlight the reality that there is variation in housing contexts that are likely to affect subjective feelings toward housing both within and across countries. In this paper, we adopt a broad categorization that allows us to explore how differences in welfare regimes, particularly as they relate to housing markets and policies, may affect households’ experiences of housing cost burden and the effect it may have on their overall life satisfaction.

We follow the Elsinga and Hoekstra (2005) model in clustering countries by welfare and housing systems as follows: Anglo-Saxon countries with more limited government interventions in housing markets and with a mix of renters and owners but with owning being the tenure of choice (Ireland and the United Kingdom); Southern European countries with limited welfare policies focused on housing and where homeownership is the norm (Italy, Spain, Portugal and Greece); and Northern and Western European countries with extensive affordable housing policies and a substantial rental sector (Austria, Denmark and the Netherlands). We also include Belgium, France, Germany, Sweden, and Switzerland in this latter group.⁴

D). Conceptual Model

On the most basic level, an individual’s reported life satisfaction or wellbeing is a statement about their utility level, which in an economics framework is an individual’s decision-making point. As a result, understanding the relationship between housing and life satisfaction sets the foundation for exploring the tradeoffs that households make between consuming housing and other goods and services in an effort to increase their overall utility. Because housing is a fundamental need, cost burdens associated with it are likely to produce tradeoffs that have direct implications for individual wellbeing and formalize in statements of subjective wellbeing. Such tradeoffs also have macroeconomic implications because they involve decisions about the consumption of housing services and other goods, as well as individual investments in things like education and health. Finally, this relationship has implications for policy intervention because a reduction in wellbeing due to housing cost burdens serves as the foundation for political will and action for policy change.

Figure 1 provides a conceptual model illustrating how higher housing costs burden can lead to lower levels of overall wellbeing by affecting housing satisfaction and non-housing satisfaction. It includes four scenarios for cost burdened households depending on whether or not they are able to attain their preferred level of housing and non-housing satisfaction.

Figure 1:

Conceptual Model of Potential Pathways from Housing Cost-Burden to Overall Wellbeing

Existing research on the relationship between housing cost burdened and subjective wellbeing has focused on housing satisfaction rather than overall wellbeing (Balestra and Sultan 2013; Rowley and Ong 2012). We extend this literature by looking at overall wellbeing as high housing cot burdened might negatively affect overall wellbeing even if households are able to meet their housing consumption preference.

The first scenario in Figure 1 shows that if cost burdened households are able to attain their housing and non-housing consumption they would be expected to have as high or higher levels of life satisfaction than not cost-burdened households. However, in the other three scenario they would have lower levels of overall life satisfaction even in the scenario in which they are satisfied with their housing situation. The reason this would happen is if meeting their housing preference by spending a high share of their income on housing means they are not able to meet their preference for consuming non-housing goods and services. These cost burdened households would have similar housing satisfaction to not cost burdened households but worse satisfaction in other domains than housing (i.e. alimentation, health, education, leisure,…) and therefore lower levels of overall life satisfaciton. But it is also possible that cost burdened households are not able to reach their desired level of housing consumption despite spending a high share of their income on housing and whether they are able to still reach their desired level of non-housing good consumption (Scenario 3) or not (Scenario 4) they would experience lower levels of life satisfaction than cost burdened households. Individual circumstances and welfare regime are likely to affect which of these scenario takes place, but in this paper we are able to explore whether on average Scenario 1 or either of Scenario 2–4 appears to reflect the situation in European countries.

In sum, the extent research establishes that subjective wellbeing is by a multiple of factors including housing and is increasingly used as part of the public policy discourse but there is limited evidence on the direct relationship between housing cost burden and wellbeing. In this paper, we seek to establish the relationship between reported and actual housing cost burden levels and subjective measures of wellbeing, measured through reported life satisfaction, within and across European countries, controlling for observable socio-demographic household characteristics. The focus is therefore on the relationship between housing cost burden and overall life satisfaction but further work is needed to examine the relationship between housing cost burden and the individual domains that contribute to overall life satisfaction.⁵

II. Data and Methods

A). Data

This paper uses the 2018 Statistics on Income and Life Conditions (SILC) microdata, which include information from 32 countries, and which are made available to researchers by Eurostat.⁶ The survey includes nationally representative samples of several thousand individuals for European countries, including some that belong to the European Union and others that do not. The core set of survey questions in these data ask for information about the respondent’s sociodemographic characteristics and housing situation, as well as some limited geographic information. In addition, in each wave, a special topic is covered in more depth. This paper uses the 2018 wave, which included questions about overall life satisfaction as part of a special module on material deprivation, wellbeing and housing difficulties.

This paper focuses on 14 countries representing a range of welfare regimes and housing markets (Austria, Belgium, Denmark, France, Germany, Greece, Ireland, Italy, Netherlands, Portugal, Spain, Sweden, Switzerland and the United Kingdom). We restrict our sample to countries that have the required housing cost and wellbeing information, and we do not include Eastern European countries.⁷ Focusing on householders (or household heads) in these countries alone, we arrive at a total sample of 155,815 households, with a range of 4,000 to 24,000 households surveyed in each country. Table 1 reports the variables used in the analysis along with the original variable names from the EU-SILC.

Table 1:

Dependent and Control Variables Definition

Variable	Definition	Source
Dependent Variable
Overall Life Satisfaction	Individual level life satisfaction. Ordinal scale from 0 (Not at all satisfied) to 10 (Completely satisfied)	PW010T
Independent Variables
Objective Housing Cost Burden	Total housing cost divided by total household gross income. Total housing cost includes: for owners: mortgage payments; for tenants: rental payments and for all: insurance, mandatory services and charges, regular maintenance and repairs, taxes, and the cost of utilities (electricity, water, gas and heating). Total household income includes the sum for all household members of gross personal income plus gross income components at the household levels. Are included cash benefits, pensions, unemployment benefits, income from property, regular inter-household cash transfers	Housing Cost: HH070; Income: HY010
Subjective Housing Cost Burden	Financial burden of the total housing cost to the household. Categorical variable: Heavy burden; Slight burden; Not a burden at all.	HS140
Tenure	Tenure status. Categorical variable: Own with mortgage; Rent market rate; Rent below market; Housed for free.	HH021
Sex	Sex. Categorical variable: Male or Female	PB150
Age	Age. Derived from year of birth and survey year	PB140
Educational Attainment	Highest ISCED level attained. Categorical variable recoded into: Primary; Secondary; Post-secondary	PE040
Employment Status	Self defined current economic status. Categorical variable recoded into: Employed; Not employed	PL031
Marital Status	Marital status and consensual union. Categorical variables recoded into: Never married; Married or in civil union; Divorced, separated or widowed	PB190 and PB200
Household Size	Combination of all current household members with same current household ID	RB040
Health Status	General health. Ordinal scale from 1 (Very good) to 5 (Very bad)	PH010
Urbanization Degree	Degree of urbanization. Categorical variable: Densely-populated area; Intermediate area; Thinly-populated area	DB100

Outcome Variable:

Our outcome variable is overall life satisfaction. This is a standardized subjective wellbeing measure reported on a Likert scale that ranges from 0 (not at all satisfied) to 10 (completely satisfied) and aims to capture the respondents’ assessment of their degree of satisfaction with their lives. This measure is not for a certain time period, but is instead “intended to represent a broad, reflective appraisal the person makes of his or her life” taken as a whole according to the data documentation and is consistent with the guidelines proposed by Diener’s (2006) recommendation to develop measures of wellbeing that can be used to inform policies.

Independent Variables:

Our main independent variables of interest are: 1) an objective measure of cost burden based on the share of household income spent on housing services, categorized as low cost burdened (less than 30 percent of income spent on housing), cost burdened (30–49 percent of income spent on housing) and severely cost burdened (50 percent or more of income spent on housing);⁸ and 2) a subjective cost burden measure based on whether the household reports that total housing costs represent a heavy financial burden, a slight burden, or no burden at all.⁹

One of the challenges with operationalizing analysis of the relationship between housing cost burden and life satisfaction is the lack of a standard definition of housing cost burden (Balestra and Sultan 2013). Different countries and organizations have adopted a range of measure of housing affordability, with cost burden starting at 25 percent of income for setting rent in social housing in Germany and Australia, to 30 percent in the United States, and 40 percent in the OECD. In Australia, some programs are solely income-based and apply to the bottom 40 percent of the income distribution.¹⁰

Despite the wide variation in methods for measuring and operationalizing cost burden, there is broad recognition that housing costs that exceed a certain share of income represent a burden for households, and that this burden may require some form of policy intervention. In this paper, we adopt two commonly used threshold: 30 and 50 percent of household income. We acknowledge that there is nothing special about these thresholds but that they allow to create ordinal categories with households spending more than 50 percent of their income on housing expected to be worse-off overall than those spending between 30 and 49 percent who are expected to be worse off than those spending less than 30 percent of the income on housing.

Other independent variables included in the model are based on established models of life satisfaction and include the following individual and contextual variables: age, gender, marital status, household size, education level, employment status, housing tenure,¹¹ and degree of urbanization. We also include a measure of general health has it has been shown to be an important predictor of life satisfaction.

B). Methods

The analysis is conducted at the household level using an ordered logit estimator in order to take into account the ordinal nature of the dependent variable capturing life satisfaction (Lu 1999; Zumbro 2014).¹² We report the ordered logit as our main estimates because it uses the information from the ordinal nature of the responses without imposing conditions of equivalence of each interval from a linear model. Further, the results of a Brant test are insignificant, indicating that the proportional odds assumption is not violated and supporting the use of the ordered logit. As a robustness, we report OLS estimates, treating the life satisfaction variable as continuous as well as logit estimates, dichotomizing this variable based on overall and country specific median reported life satisfaction. These variations of the model produced qualitatively similar results (Appendix A).

In results table, we reported odds ratios that are interpreted the increased or decreased likelihood of householder i reporting a one-unit-higher level of life satisfaction (j ranging from 0 to 10) with error term ε.

$$\mathit{\text{logit}}\left[\text{P}\left({\text{LifeSatisfaction}}_{\text{i}}>\text{j}\right)\right]={\alpha }_j+{\beta }_1{\mathit{\text{CostBurden}}}_i+{\beta }_2{X}_i+{\gamma }_i+{\epsilon }_{ij}\mathit{\text{where}}j=0,1,\dots ,J-1$$ (1)

We run three models, 1) with the objective measures of cost burden (CostBurden) and country fixed effects (γ); 2) adding controls for a vector of sociodemographic and geographic characteristics that have been shown to affect individual life satisfaction (X), include age, gender, marital status, household size, education level, employment status, housing tenure, and degree of urbanization; and 3) adding reported health condition. These control variables account for observational differences between individuals that have been shown to be associated with life satisfaction.

III. Results

A). Descriptive Statistics

The 14 countries included in the sample differ substantially in terms of size, socioeconomic demographics, welfare systems, cultural context, and housing markets. Table 2 provides a summary of differences across countries for the variables used in the analysis. There is substantial heterogeneity in the average reported level of life satisfaction as shown in Table 2. Austria, Ireland, and Switzerland report the highest levels of life satisfaction (7.8 to 8) followed by Belgium, Denmark, the Netherlands, Sweden, and the UK (7.5 to 7.6) and France and Germany along with Southern European countries reporting the lowest levels (6.3 to 7.3). The average housing cost burden also varies in these countries from 13 percent in Italy to 27 percent in Greece. The overall homeownership rate in the sample ranges from less than 50 percent in Austria, Germany, and Switzerland to more than 70 percent in Ireland and three Southern European countries (Greece, Italy, and Portugal). Figure 2 breaks tenure by country down further into owners with and without a mortgage and renters in market- and below-market-rate units.¹³ Most owners have a mortgage in Switzerland, Denmark, Sweden and the Netherlands and a substantial share do in Belgium, France, Ireland and the UK. In contrast, Southern European countries display high homeownership rates, but few homeowners have a mortgage. Median income ranges from less than 35,000 euros in Southern European countries to over 60,000 in Denmark and Sweden and almost 100,000 in Switzerland. Countries also differ on age, educational attainment, household composition and the level of urbanization. In sum, there are substantial variations across countries and on several dimensions, which justifies the use of the country-level fixed effects in our models, but countries nevertheless tend to cluster based on their levels of economic development and welfare regimes.

Table 2:

Summary of dependent and independent variables by country

	UK	Ireland	Denmark	Sweden	Belgium	Netherlands	Austria	Germany	Switzerland	France	Italy	Spain	Portugal	Greece
Life Satisfaction	7.5	8.0	7.6	7.6	7.5	7.6	7.8	7.2	7.9	7.1	6.9	7.3	6.6	6.3
Objective Housing Cost Burden	20.7%	14.7%	23.0%	19.2%	18.0%	20.3%	15.8%	22.4%	19.6%	15.5%	12.9%	15.0%	13.7%	27.2%
Heavy Subjective Housing Cost Burden	15.0%	24.8%	8.0%	7.6%	27.5%	9.8%	11.8%	13.0%	24.8%	23.6%	45.0%	48.6%	26.4%	49.9%
Homeowners	64.9%	71.5%	52.9%	57.9%	50.9%	57.7%	45.5%	42.8%	37.7%	63.5%	71.3%	65.4%	73.3%	72.3%
Median Income (Euros)	37,100	43,190	64,641	60,668	40,606	49,298	47,757	39,892	99,096	42,093	32,314	27,743	17,671	16,787
Age	53.7	53.5	51.8	50.8	54.6	52.4	54.5	54.1	53.7	55.0	58.4	55.8	57.8	58.0
Sex	49.9%	55.3%	52.1%	50.0%	34.8%	50.4%	43.9%	39.6%	40.1%	52.2%	35.5%	39.5%	35.7%	28.4%
Post-Secondary Education	55.1%	54.7%	78.7%	47.2%	71.5%	46.6%	41.2%	43.6%	86.7%	71.4%	17.3%	33.4%	20.7%	34.4%
Employed	58.4%	55.9%	51.2%	54.2%	53.0%	57.9%	53.7%	55.4%	61.3%	50.4%	53.5%	51.2%	52.8%	44.5%
Married	56.7%	59.0%	45.0%	46.1%	56.6%	56.9%	55.6%	50.7%	55.0%	57.6%	56.9%	61.8%	65.8%	63.8%
Household Size	2.3	2.6	2.0	2.0	2.3	2.2	2.3	2.0	2.2	2.3	2.3	2.5	2.5	2.6
Good or Very Good Health	90.7%	96.0%	90.8%	93.1%	89.0%	94.0%	89.5%	90.3%	94.4%	90.5%	91.2%	91.5%	82.0%	87.8%
Low Degree of Urbanization	12.8%	40.9%	32.3%	19.7%	16.9%	0.0%	34.8%	15.9%	16.6%	32.6%	24.3%	25.7%	26.2%	29.3%
N	17,113	4,382	5,545	5,754	4,384	12,492	5,466	12,892	6,680	10,424	21,173	11,863	13,342	24,305

Figure 2:

Tenure Mix by Country

Note: N=155,815. UK=United Kingdom, IE=Ireland, DK= Denmark; SE=Sweden; BE= Belgium; NL=Netherlands; AT=Austria; DE=Germand; CH=Switzerland; FR=France; IT=Italy; ES=Spain; PT=Portugal; EL=Greece. The rent below marker measure does not capture all subsidized units and is not reported in countries accompanied by a *.

Source: Eurostat-SILC 2018

B). Differences in Housing Cost Burden Across Tenure and Income by Country

Renters are substantially more likely to spend at least 30 percent of their income on housing than owners in all 14 countries as reported in Figure 3. Overall, at least 30 percent of renters spend 30 percent of their income or more on housing except in Portugal (25 percent) and Austria (26 percent). Conversely, over 50 percent of renters in Belgium, Greece, the Netherlands, Spain and the UK spend 30 percent of their income or more on housing. More than 10 percent of renters spend at least 50 percent of their income on housing in the UK, Denmark, and Greece. In contrast, fewer than 10 percent of owners spend more than 30 percent of their income on housing except in the UK (11 percent), Germany (12 percent) and Greece (28 percent) and fewer than 3 percent spend more than 50 percent of their income on housing (except Greece where 5 percent did).

Figure 3:

Housing Cost Burden Category by Tenure by Country

Note: N=155,815. UK=United Kingdom, IE=Ireland, DK= Denmark; SE=Sweden; BE= Belgium; NL=Netherlands; AT=Austria; DE=Germand; CH=Switzerland; FR=France; IT=Italy; ES=Spain; PT=Portugal; EL=Greece

Source: Eurostat-SILC 2018

There have been substantive discussions in the literature about the best ways to measure housing affordability. The simple ratio of housing costs to income used in this study has several documented drawbacks (Balestra and Sultan 2013; Gabriel et al. 2005; Stone et al. 2011). First, it does not account for the absolute amount of income spent on non-housing costs linked to the location of the unit, such as transportation costs. This measure also does not account for preferences, with some households choosing to spend a higher share of their income on housing because they value a higher consumption level of housing services (including location). In addition, the traditional classification of 30 to 50 percent of income as cost burden and 50 percent and over as severe cost burden is arbitrary, but has been adopted by the U.S. Department of Housing and Urban Development to designate preferred cutoff points for housing assistance and is therefore often used by researchers and practitioners in the U.S. and in comparative work (Balestra and Sultan 2013).¹⁴ Another issue with comparing cost burden across countries is the substantial variation in forms of tenure and monthly cost structures, as shown in Figures 2 and 3. The share of each country’s housing stock for which costs are either subsidized or regulated also affects measures of housing costs and the incidence of cost burden.

Despite these documented limits, we rely on the housing-cost-to-income ratio as our main measure because it directly captures the respondent’s budget allocation and reflects the fact that housing services are not totally fractional, such that households must spend a minimum amount of their income for housing services in order to obtain shelter. Figure 4 shows that spending more than 30 percent and even more 50 percent of income on housing is largely concentrated among households in the lowest income quartile (and to some degree second lowest quartile). In all countries, at least 30 percent of the lowest earners spend more than 30 percent of their income on housing services (except Ireland: 22 percent and Portugal: 23 percent) and at least 5 percent spend more than 50 percent. Conversely, fewer than 5 percent of the highest earners are spending more than 30 percent, and fewer than 1 percent of this group are spending more than 50 percent. Unless low-income households have a systemic preference for a higher level of housing consumption relative to their income, these ratios suggest that spending more than 30 or 50 percent of income on housing largely reflects limited cheaper housing options rather than a preference for spending a large share of one’s income on housing.

Figure 4:

Housing Cost Burden Category by Income Quartile by Country

Note: N=155,815. UK=United Kingdom, IE=Ireland, DK= Denmark; SE=Sweden; BE= Belgium; NL=Netherlands; AT=Austria; DE=Germand; CH=Switzerland; FR=France; IT=Italy; ES=Spain; PT=Portugal; EL=Greece

Source: Eurostat-SILC 2018

C). Housing Cost Burden and Life Satisfaction

Table 3 reports the results of the ordered logit regression for the housing cost burden measure.¹⁵ In all three models, households spending 30 percent of their income or more on housing costs are less likely to report higher life satisfaction than those spending less than 30 percent. In the model with full controls, those spending between 30 and 49.9 percent of their income on housing costs are 22 percent less likely to report a one-unit higher level of life satisfaction than those spending less than 30 percent controlling for socio-demographic and health characteristics. Those spending 50 percent of their income or more on housing are 36 percent less likely to report a one-unit higher level of life satisfaction. This indicates that severe housing cost burden has a particularly strong, negative association with reported life satisfaction. By comparison, householders who are divorced or separated are only 4 percent less likely to report a one-unit higher level of life satisfaction than those never married. Health, by contrast, is a stronger factor in life satisfaction even than housing cost burden; those reporting being in good health relative to being in very good health are 53 percent less likely to report a one-unit higher level of life satisfaction and those in very bad health 97 percent less likely. In the rest of the paper, we focus on models including all controls including health status. However, adding health as a control has only a moderate impact on the estimated relationship between housing cost burden and life satisfaction.

Table 3:

Reported Overall Life Satisfaction and Housing Cost Burden, Ordered Logistic Regression, Odd Ratios

	(1)	(2)	(3)
Housing cost burden (ref. = <30%) 30–49.9%	0.565***	0.754***	0.782***
	(−44.72)	(−18.65)	(−16.21)
50% or more	0.404***	0.602***	0.639***
	(−35.01)	(−18.69)	(−16.43)
N	155,815	155,815	155,815
Country Fixed Effect	Yes	Yes	Yes
Sociodemographic controls	No	Yes	Yes
Health control	No	No	Yes

Note: Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001

In Table 4, we take advantage of a question asking respondents to what extent their housing costs represent a financial burden. The objective measure of housing cost burden (housing costs divided by income) and the subjective measure (whether households think their housing expenses represents a burden) are related and correlated but the overlap is far from perfect. In particular, a substantial share of respondents spending less than 30 percent of their income on housing report that it represents a heavy burden, particularly among higher income households. When combining the two measures, we find that across the subjective categories, households who spend a smaller share of their income on housing are likelier to report a higher level of life satisfaction. Using respondents feeling heavily burdened and spending more than 50 percent of their income on housing as the reference category, all other categories are more likely to report higher levels of life satisfaction. Respondents who report feeling heavily burdened but spend less than 30 percent of their income on housing are 79 percent more likely to be in a higher category of life satisfaction. Those not feeling burdened and spending less than 30 percent of their income are more than three times more likely to be in a higher category. These results indicate that both the perception of housing cost burden and actual cost burden are negatively associated with life satisfaction, but the actual share of income spent on housing matters beyond the mere feeling of cost burden. This indicates that our measure not only captures differences in the individual propensity to be satisfied with one’s life but also the direct impact on life satisfaction of housing expenditure as a share of household income.

Table 4:

Reported Overall Life Satisfaction and Subjective and Objective Housing Cost Burden, Ordered Logistic Regression, Odd Ratios

	(1)
Rent Burden: Obj-Subj Category (ref=Feel Heavy Burden, >=50% Income)
Feel Heavy Burden, 30–49.9% Income	1.148* (2.18)
Feel Heavy Burden, <30% Income	1.789*** (7.56)
Feel Moderate Burden, >=50% Income	1.739*** (5.38)
Feel Moderate Burden, 30–49.9% Income	1.759*** (7.56)
Feel Moderate Burden, <30% Income	1.914*** (9.35)
Feel No Burden, >=50% Income	2.632*** (9.12)
Feel No Burden, 30–49.9% Income	2.905*** (12.94)
Feel No Burden, <30% Income	3.020*** (15.44)
N	155,815
Country Fixed Effect	Yes
Sociodemographic controls	Yes
Health control	Yes

Spending a higher share of income on housing implies tradeoffs on other expenditures or saving. Depending on whether the tradeoffs involve spending less on other basic necessities such as food and health services or on other goods, the impact they have on wellbeing may vary. In welfare state systems with a wider range of housing subsidies and universal access to health and education, higher housing cost burden might be more likely to reflect choices rather than constraints and be less associated with tradeoffs that have a negative effect on wellbeing.

Table 5 reports the results of the interactions between the cost burden measure and the three housing regimes described previously. As stated, we use the Elsinga and Hoekstra (2005) model in clustering countries by welfare and housing systems as follows. In this model, the Anglo-Saxon countries with more limited government interventions in housing markets, and with a mix of renters and owners but where owning is the tenure of choice, serve as our reference group. Households that are both moderately and heavily housing cost burdened in Northern/Western European countries have a higher reported life satisfaction than those in Anglo-Saxon countries. Of particular note is that the coefficient size increases for those who are heavily cost burdened, suggesting that more robust welfare structures particularly improve the welfare of those facing the highest cost burdens. Interestingly, those households that are moderately cost burdened in Southern European countries tend to have a lower life satisfaction than those in Anglo-Saxon countries, but this relationship reversed for those who are heavily burdened. One possible explanation is that the means-tested welfare programs that target the lowest-income individuals in Ireland and the United Kingdom are more effective in raising wellbeing levels among households with the highest cost burdens than are the more limited welfare policies of Italy, Spain and Greece. In aggregate, our results suggest that stronger welfare and housing regimes mediate the impact of housing cost burden on life satisfaction.

Table 5:

Reported Overall Life Satisfaction and Objective Housing Cost Burden, Differences Across Housing Regimes, Ordered Logistic Regression, Odd Ratios

	(1)
Housing cost burden (ref. = <30%)
30–49.9%	0.713*** (−8.61)
50% or more	0.580*** (−7.93)
Regime (ref. = Anglo Saxon)
Northern/Western Europe	0.853*** (−8.40)
Southern Europe	0.449*** (−41.53)
Interaction Cost Burden Category * Regime
Ref. = Moderate Burden, Anglo-Saxon
Moderate Burden * Northern/Western Europe	1.196*** (3.98)
Moderate Burden * Southern Europe	0.815*** (−4.72)
Ref. = Heavy Burden, Anglo-Saxon
Heavy Burden * Northern/Western Europe	1.573*** (4.88)
Heavy Burden * Southern Europe	1.221* (2.28)
N	155,815
Country Fixed Effect	Yes
Sociodemographic controls	Yes
Health control	Yes

D). Robustness

Most of the literature on the relationship between housing cost burden has focused on renters, particularly market-rate renters (Rowley and Ong 2012). We include all forms of tenure in our analysis. Restricting the sample to either owners or renters results in similar point estimates (Appendix B). Owners report overall higher levels of life satisfaction than renters (Appendix D) and are less likely to be cost burdened (Figure 2) but when they cost burdened the relationship with life satisfaction is similar as the one estimated for renters. This is consistent with cost burden itself, rather than differences in tenure type, driving life satisfaction outcomes. Whether an individual owns or rents, paying more than a certain share of their income requires similar tradeoffs. As Rowley and Ong (2012) note, households who own without a mortgage or rent in below-market-rate accommodations are unlikely to be rent burdened. However, we do include owners without a mortgage and renters paying below market rent in our study since some of them might still be experiencing high levels of housing costs. In addition, being cost burdened under these forms of tenure would still be expected to result in tradeoffs that may have a negative impact on life satisfaction.

A concern with our specification is that we are excluding variables associated with housing cost burden that might also affect life satisfaction. In particular, we are not directly controlling for the amount of housing services consumed or for wealth, which might mediate the effect of income cost burden on life satisfaction. In Appendix E, we report results from a model that included a measure of rent and imputed rent for owners.¹⁶ By controlling for the value of housing services consumed, we effectively proxy for a household’s wealth and at the same time for the choice to spend a higher absolute amount on housing. Including that variable only has a minimum effect on our overall estimates. The estimates in the model with imputed rent are similar to those in Table 3 and show even a slightly larger negative effect of being cost burdened.

Another potential concern is that our empirical setup does not account for the potential for omitted variables that might be correlated with housing cost burden and also affect household wellbeing. For instance, housing cost burden might serve as a proxy for income, which is an important factor in overall life satisfaction. As a robustness check, we break our sample by national income quartile and Appendix F shows that a stronger negative relationship between cost-burden and life satisfaction is found among lower income groups, consistent with the interpretation that for individuals with similarly low income housing cost burden itself affects life satisfaction.¹⁷ Being housing cost burdened could also serve as a proxy for overall financial stress; in this case, we would be attributing to housing costs what is actually the effect of financial precariousness on wellbeing. Exploring control for other financial stressors would be valuable. Further work is also needed to control for both time invariant and variant individual characteristics that may affect the measured relationship between housing cost burden and wellbeing.

IV. Discussion

Housing is an essential good and spending a higher share of income on housing results in tradeoffs with the consumption of other goods and a higher sense of ontological insecurity. This is particularly likely if that high share of income spent on housing is the result of an inability to reduce housing costs rather than a preference for spending a higher amount on housing services. In this paper, we find that households spending more than 30 percent of their income on housing report lower levels of overall life satisfaction and those spending more than 50 percent even lower levels across 14 European countries with diverse housing markets and institutions. There are differences in the magnitude of these relationships across countries though, which may be associated with differences in welfare regimes. In countries with a more extensive safety net, the trade-offs for heavily housing cost-burdened households might not require decreasing consumption of other necessities, like health care, that have been shown to have a large impact on wellbeing, although this hypothesis require more direct evidence looking at satisfaction with specific life domains that have been shown to be negatively affected for housing cost burdened households forced to make trade-offs (material deprivation, health,…).

The relationship between cost burden and life satisfaction is similar across tenure. Owners are more likely to report higher levels of life satisfaction but an increase in housing costs is associated with a similar decrease in the likelihood of reporting higher life satisfaction regardless of the form of tenure.

Low- and moderate-income households are substantially more likely to experience housing cost burden and they are the ones for which high levels of burden has the most negative impact on life satisfaction. Given these findings on the relationship between cost burden and life satisfaction, policies that reduce the share of income that low-income households spend on housing can have a substantial impact on their life satisfaction and overall welfare.

The findings also indicate that subjective feeling of burden have a negative impact on life satisfaction, but that households who report feeling burdened but are not considered burdened based on their housing cost to income ratio still report higher life satisfaction than actually cost-burdened households. In other words, actual burdens are more likely than stated feelings of burden to affect household welfare. This finding has particular importance from a policy perspective. When polling public sentiment, many households may report feeling burdened or view housing affordability as a problem but the coalition of households who want to advance affordable housing policy to improve their utility may be smaller than the broader narrative about the need to do so suggests if the issue is most salient for households who are objectively burdened and are also more likely to be lower income.

Further work is needed to examine how the relationship between housing cost burdens and welfare is affected by changes in cost burden. Additional research is required to unpack the mechanisms that explain how higher levels of housing cost burden result in lower life satisfaction, such as the specific tradeoffs households make to afford housing, which then reduce their welfare. Ultimately, the findings in this paper are consistent with the reality that housing is an essential good, and that challenges in affording that good affect an individual’s wellbeing. Stronger welfare systems that support access to a broader range of basic goods and services for the lowest-income households are more likely to reduce the negative implications of housing costs burdens. Households may also be less likely to trade off some base level of consumption of other goods, such as healthcare, when faced with higher housing cost burdens in countries with stronger welfare states.

Appendix A: Reported Overall Life Satisfaction and Housing Cost Burden, OLS and Logit Estimates

Panel A: OLS Results	(1)	(2)	(3)
Housing cost burden (ref. = <30%)
30–49.9%	−0.642*** (−48.46)	−0.326*** (−21.33)	−0.277*** (−19.40)
50% or more	−1.012*** (−38.03)	−0.575*** (−21.12)	−0.483*** (−18.95)
N	155,815	155,815	155,815
Country Fixed Effect	Yes	Yes	Yes
Sociodemographic Controls	No	Yes	Yes
Health Control	No	No	Yes

Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001.

Note: Life Satisfaction measure treated as continuous.

Panel B: Logit Results (ref. = Below Median Life Satisfaction	(1)	(2)	(3)
Housing cost burden (ref. = <30%)
30–49.9%	0.548*** (−42.00)	0.778*** (−14.36)	0.795*** (−12.59)
50% or more	0.424*** (−29.32)	0.673*** (−12.37)	0.695*** (−10.91)
N	155,815	155,815	155,815
Country Fixed Effect	Yes	Yes	Yes
Sociodemographic Controls	No	Yes	Yes
Health Control	No	No	Yes

Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001.

Note: Life Satisfaction measure dichotomized based on median value for a given country (median value included in above median category).

Appendix B: Reported Overall Life Satisfaction and Housing Cost Burden, Difference Between Owners and Renters, Ordered Logistic Regression, Odd Ratios

Panel A: Renters Only	(1)	(2)	(3)
Housing cost burden (ref. = <30%)
30–49.9%	0.616*** (−23.26)	0.736*** (−13.57)	0.755*** (−12.37)
50% or more	0.445*** (−24.22)	0.562*** (−16.46)	0.603*** (−14.33)
N	43,216	43,216	43,216
Country Fixed Effect	Yes	Yes	Yes
Sociodemographic Controls	No	Yes	Yes
Health Control	No	No	Yes

Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001

Panel B: Owners Only	(1)	(2)	(3)
Housing cost burden (ref. = <30%)
30–49.9%	0.542*** (−28.11)	0.711*** (−15.23)	0.749*** (−12.80)
50% or more	0.457*** (−16.70)	0.603*** (−10.65)	0.638*** (−9.38)
N	110,411	110,411	110,411
Country Fixed Effect	Yes	Yes	Yes
Sociodemographic Controls	No	Yes	Yes
Health Control	No	No	Yes

Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001

Appendix C: Alternative Affordability Specifications

Panel A: Definition of Rent Burden at 40% (OECD definition)	(1)	(2)	(3)
Housing cost burden (ref. = <40%)	0.459*** (−42.71)	0.687*** (−19.31)	0.719*** (−16.91)
N	155,815	155,815	155,815
Country Fixed Effect	Yes	Yes	Yes
Sociodemographic Controls	No	Yes	Yes
Health Control	No	No	Yes

Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001

Panel B: Definition of Rent Burden as Continuous Variable	(1)	(2)	(3)
Housing cost burden	0.974*** (−69.33)	0.984*** (−32.86)	0.986*** (−28.37)
N	155,815	155,815	155,815
Country Fixed Effect	Yes	Yes	Yes
Sociodemographic Controls	No	Yes	Yes
Health Control	No	No	Yes

Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001

Appendix D: Ordered Logit, Full Results

	(1)	(2)	(3)
Housing cost burden (ref. = <30%)
30−49.9%	0.565*** (−44.72)	0.754*** (−18.65)	0.782*** (−16.21)
50% or more	0.404*** (−35.01)	0.602*** (−18.69)	0.639*** (−16.43)
Tenure (ref.= own without mortgage)
Own with mortgage		1.004 (0.29)	1.032* (2.17)
Rent market rate		0.774*** (−16.50)	0.834*** (−11.66)
Rent below market		0.667*** (−14.78)	0.789*** (−8.67)
Housed for free		0.811*** (−8.71)	0.852*** (−6.59)
Female (ref.=male)		1.037*** (3.42)	1.080*** (7.24)
Age		0.941*** (−26.63)	0.950*** (−22.43)
Age square		1.001*** (29.19)	1.001*** (29.79)
Marital status (ref. = never married)
Married or in civil union		1.704*** (31.61)	1.677*** (30.51)
Divorced, separated, widowed		0.942*** (−3.52)	0.960* (−2.42)
Education (ref. = Primary)
Secondary		1.732*** (34.72)	1.386*** (20.34)
Post-secondary		2.282*** (48.21)	1.650*** (28.80)
Household Size		0.956*** (−8.33)	0.952*** (−9.00)
Urbanization degree (ref. = high density)
Intermediate density		1.031* (2.52)	1.036** (2.93)
Low density		1.059*** (4.29)	1.087*** (6.24)
Employment (ref=not employed)		1.557*** (32.16)	1.252*** (16.18)
Health (ref.=very good)
Good			0.470*** (−55.92)
Fair			0.207*** (−96.17)
Bad			0.0804*** (−110.48)
Very Bad			0.0310*** (−84.95)
Country (ref.= Austria)
Belgium	0.622*** (−14.58)	0.537*** (−17.06)	0.550*** (−16.41)
Switzerland	1.133*** (3.79)	0.984 (−0.47)	0.895** (−3.21)
Germany	0.573*** (−19.51)	0.566*** (−18.74)	0.646*** (−14.44)
Denmark	1.130*** (3.50)	1.007 (0.20)	1.053 (1.43)
Greece	0.225*** (−55.98)	0.242*** (−49.53)	0.200*** (−55.92)
Spain	0.487*** (−25.44)	0.520*** (−21.64)	0.512*** (−22.19)
France	0.401*** (−30.93)	0.392*** (−30.33)	0.426*** (−27.63)
Ireland	1.103** (2.63)	1.191*** (4.56)	0.941 (−1.58)
Italy	0.323*** (−41.79)	0.368*** (−35.03)	0.361*** (−35.68)
Netherlands	0.691*** (−13.02)	0.669*** (−13.12)	0.652*** (−14.02)
Portugal	0.235*** (−48.54)	0.307*** (−37.47)	0.397*** (−29.31)
Sweden	0.938 (−1.88)	0.824*** (−5.47)	0.784*** (−6.92)
United Kingdom	0.785*** (−8.34)	0.694*** (−12.01)	0.710*** (−11.31)
cut1	0.00381*** (−158.68)	0.00259*** (−82.08)	0.000940*** (−94.93)
cut2	0.00581*** (−161.77)	0.00395*** (−77.96)	0.00145*** (−91.00)
cut3	0.0104*** (−158.80)	0.00707*** (−71.09)	0.00266*** (−84.19)
cut4	0.0185*** (−148.40)	0.0127*** (−63.40)	0.00494*** (−76.32)
cut5	0.0316*** (−133.76)	0.0220*** (−55.77)	0.00890*** (−68.37)
cut6	0.0829*** (−100.55)	0.0596*** (−41.48)	0.0262*** (−53.14)
cut7	0.159*** (−75.52)	0.118*** (−31.55)	0.0551*** (−42.44)
cut8	0.393*** (−38.85)	0.303*** (−17.64)	0.154*** (−27.45)
cut9	1.548*** (18.28)	1.246** (3.25)	0.707*** (−5.11)
cut10	4.088*** (57.51)	3.328*** (17.73)	1.985*** (10.07)
N	155,815	155,815	155,815
Country Fixed Effect	Yes	Yes	Yes

Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001

Note: this is the specification used for the results reported in Table 3.

Appendix E: Reported Overall Life Satisfaction and Housing Cost Burden, Control for Rent and Imputed Rent as Proxy for Unmeasured Wealth and Quantity of Housing Services, Ordered Logistic Regression, Odd Ratios

	(1)
Housing cost burden (ref. = <30%)
30–49.9%	0.765*** (−16.21)
50% or more	0.625*** (−16.53)
Annual Rent or Imputed Rent (000)	1.029*** (10.41)
N	101,818
Country Fixed Effect	Yes
Sociodemographic controls	Yes
Health control	Yes

Appendix F: Reported Overall Life Satisfaction and Housing Cost Burden by Income Quartile, Ordered Logistic Regression, Odd Ratios

	Overall	Lowest	2nd	3rd	Highest
Housing cost burden (ref. = <30%)
30–49.9%	0.782*** (−16.21)	(−16.21) (−5.83)	0.917* (−2.27)	0.985 (−0.66)	0.932 (−1.24)
50% or more	0.639*** (−16.43)	0.564*** (−18.57)	0.780*** (−8.45)	0.782*** (−6.06)	−0.824 (−0.52)
N	155,815	49,137	41,779	38,696	37,685
Country Fixed Effect	Yes	Yes	Yes	Yes	Yes
Sociodemographic controls	Yes	Yes	Yes	Yes	Yes
Health control	Yes	Yes	Yes	Yes	Yes

Exponentiated coefficients; t statistics in parentheses;

^*p<0.05,

^**p<0.01,

^***p<0.001