Real Assets and Subjective Well-Being: Using a Novel Measure for Relative Effects

Across economics, sociology and psychology, there is a widely held assumption that a higher income increases individuals’ ability to satisfy their needs and aspirations, thereby improving their well-being. Based on this premise, numerous studies have investigated the relationship between income and subjective well-being. Subjective well-being typically describes how people experience the quality of their lives, encompassing both emotional reactions and cognitive judgments (Diener, 1984). One of the most commonly used measures of subjective well-being is life satisfaction.¹ A substantial body of research has demonstrated that life satisfaction tends to increase with income, whether measured at the individual or household level although the gains tend to diminish at higher levels of income (Clark, 2018).

However, income is not the only monetary dimension that likely influences well-being; wealth, including housing and other real assets, must also be considered (D’Ambrosio et al., 2020). While income measures an individual’s capacity to consume within a specific period, wealth fulfils multiple functions: it generates income, provides economic security, and enables individuals to smooth consumption over their lifetime, with important consequences for economic and social behaviours at the individual level (Dietz & Haurin, 2003). Crucially, unlike earnings or income, housing and other real assets are visible and can easily be evaluated by peers, thus serving as direct sources of comparison and aspiration, potentially affecting individuals’ life satisfaction.

Studies have shown a positive association between life satisfaction and household net wealth (Headey & Wooden, 2004). Wealth encompasses multiple dimensions – including real estate, financial assets, and business assets – which may each relate differently to well-being. The literature examining homeownership and life satisfaction generally finds that homeowners report higher life satisfaction compared to non-homeowners (e.g., Cheng et al., 2020; Gignac et al., 2024; Hu, 2013; Stotz, 2019; Zheng et al., 2020; Zumbro, 2014). Additionally, both ownership status and dwelling size appear to affect satisfaction levels. For instance, Bellet (2024) demonstrates a positive link between house size and housing satisfaction. Regarding financial assets, Brown and Gray (2016) and D’Ambrosio et al. (2009) report positive correlations between financial assets and life satisfaction.² D’Ambrosio et al. (2020) further indicate that net real estate wealth and business assets are positively associated with life satisfaction, whereas credit debt has a negative impact. This aligns with findings by Brown et al. (2005), who report that unsecured debt, as opposed to secured debt, reduces psychological well-being. Similarly, Keese and Schmitz (2014) show that household debt negatively affects mental well-being, particularly when individuals explicitly recognise or label it as debt, thereby heightening its psychological impact (Greenberg & Mogilner, 2021).

A substantial body of literature also highlights that subjective well-being is affected by relative concerns and depends on comparisons with peers. These comparisons can have two distinct effects. On the one hand, numerous studies have documented negative comparison effects, whereby individual well-being decreases as outcomes (such as earnings or wealth) of the reference group improve. This phenomenon has been consistently observed across survey data (e.g., Brodeur & Flèche, 2019; Clark & Oswald, 1996; Clark et al., 2009; Ifcher et al., 2018; Luttmer, 2005; Neman, 2020; Shields et al., 2009), and experimental settings (e.g., Card et al., 2012; Ifcher et al., 2020). On the other hand, an information (or ‘tunnel’) effect may occur: observing peers with better outcomes can be interpreted positively as a signal of one’s own potential future improvements, thus increasing current subjective well-being.³

But what about relative effects of wealth specifically? Using Australian data, Brown and Gray (2016) find that relative wealth indeed matters; however, the information effect generally dominates the comparison effect, meaning an individual’s subjective well-being is positively influenced by others’ wealth. This implies that individuals interpret the increased wealth of their comparison group primarily as a sign of promising future opportunities rather than as a source of positional disadvantage. D’Ambrosio et al. (2020) reach similar conclusions with German data. Conversely, Bellet (2024), using data from the United States, finds the opposite: comparison effects dominate, and an individual’s housing satisfaction declines as neighbours’ house sizes increase.

In this paper, we use microdata from the 2017 Slovak wave of the Household Finance and Consumption Survey (HFCS) to explore the relationship between the value of real assets and life satisfaction, examining specifically the extent to which relative considerations shape this relationship. Although the literature examining subjective well-being in relation to absolute and relative wealth is growing, our study makes a number of contributions. First, comparison effects have traditionally been captured by computing the mean or median within arbitrarily defined reference groups, usually selected based on similarity criteria such as age, educational attainment or region of residence. However, a recent study by Kudrna (2024) points out that the choice of reference group can dramatically influence conclusions regarding relative effects. Our paper addresses this issue by using a source of data often overlooked: the interviewer paradata. Specifically, we use the interviewers’ assessments of respondents’ dwelling quality relative to their immediate neighbours. By employing interviewer paradata, we arguably obtain a more valid and locally relevant evaluation of respondents’ relative dwelling quality compared to conventional approaches. Second, the number of studies focusing on Central and Eastern European countries remains relatively limited. Yet this region is particularly relevant given its distinctive post-socialist legacies. In the region, real assets play a disproportionately significant role in household wealth due to large-scale privatisation, underdeveloped financial markets and limited portfolio diversification. Consequently, the absolute and relative effects of asset ownership on life satisfaction may differ systematically from those observed in Western Europe. Thus, providing additional evidence from Central and Eastern Europe is essential to test the external validity and generalisability of hypotheses regarding absolute and relative wealth effects on life satisfaction.⁴

Our findings are as follows. First, we report a positive association between the value of real assets and life satisfaction. This association remains robust across a battery of robustness checks and is larger than the correlation between life satisfaction and either financial assets or household income. This result is consistent with the fact that real assets in Slovakia largely consist of housing, accounting for more than 90% of total household wealth on average. Around 90% of Slovak households own their dwellings, and the share of renters in Slovakia is significantly lower than in Western European countries such as Germany (Bover et al., 2016; Mathä et al., 2017). The significance of homeownership in Slovakia has deep historical roots: during socialism, the government of former Czechoslovakia actively encouraged homeownership among the population. This (cultural) emphasis on homeownership persisted after the fall of socialism, with Slovaks continuing to invest more heavily in housing and other real assets rather than financial assets such as shares or bonds. Beyond the fact that real assets are positively correlated with life satisfaction in Slovakia, the very nature of these assets may amplify comparison effects, as real assets (such as housing and real estate) are easily observable and accurately assessed by peers. Using interviewer paradata as an externally assessed and locally relevant measure of dwelling quality relative to neighbours, we find that living in a dwelling rated worse than the neighbourhood average significantly predicts lower life satisfaction. However, relative effects appear asymmetric: occupying a dwelling rated better than average does not correlate significantly with increased life satisfaction. Finally, we estimate that relative effects explain approximately 7% of the observed relationship between real assets and life satisfaction.

The remainder of the paper is structured as follows. Section 2 describes the conceptual framework and the empirical challenges. The data, variables and empirical strategy are described in detail in Sect. 3. Section 4 presents the estimated baseline results along with several robustness checks. Finally, Sect. 5 concludes and discusses the implications of the findings, limitations and directions for future research.

Our analysis builds upon the relative pay model of Card et al. (2012), but instead of focusing on wages, we consider individuals’ real assets. Specifically, suppose each individual \(i\) holds a level of assets \({a}_{i}\) and compares themselves to a reference level \(m\). Unlike the typical labour-market context where information on co-workers’ wages may be imperfect, we assume individuals observe \(m\) perfectly. This assumption is defensible in certain contexts where assets (such as residential property) are highly visible, making relative comparisons particularly salient.

Following Card et al. (2012), individual well-being (or satisfaction) \({S}_{i}\) is the sum of the utility from absolute asset holdings and the effect of the individual’s position relative to \(m\). Formally, we writewhere \(u(\cdot )\) captures utility from absolute assets, \(1\{\cdot \}\) denotes an indicator function, and \({e}_{i}\) represents unobserved factors. This specification allows for the possibility that being above the reference level \(m\) confers an effect \((\alpha )\) different from that of being below \((\beta )\), and we remain agnostic about the sign and magnitude of these parameters. We also simplify by assuming that status enters solely through an indicator function rather than through the distance \({a}_{i}-m\). Although one could include the magnitude of the gap in the model, our empirical results (detailed in Sect. 4) indicate that the precise distance to the reference \(m\) does not significantly affect subjective well-being beyond the basic distinction of being above or below \(m\). Consequently, focusing on these indicators does not diminish the validity of our conclusions regarding how relative position influences satisfaction.⁵

However, researchers do not directly observe \(m\) and must rely on an approximation \(\widehat{m}\) in practice. This approximation may be derived from external datasets, broad neighbourhood averages, or survey questions that imperfectly reflect the individual’s true reference group. Instead of using the correct indicator functions \(1\{{a}_{i}>m\}\) and \(1\{{a}_{i}<m\}\), an empirical analysis often substitutes \(1\{{a}_{i}>\widehat{m}\}\) and \(1\{{a}_{i}<\widehat{m}\}\) into a regression model such as

When \(\widehat{m}\) differs from \(m\), any individual truly above \(m\) can be mistakenly classified as below it, and vice versa. This misclassification arises from treating \(\widehat{m}\) as though it were the genuine threshold. If \(\widehat{m}\) is randomly higher or lower than \(m\), the resulting noise in the dummy variables tends to bias the estimated coefficients toward zero, weakening the role of relative asset position. However, if \(\widehat{m}\) is systematically biased, the direction and magnitude of the bias in the estimated effects can become larger or even reversed.

One solution is to gather higher-quality data on the reference group itself, ensuring that any aggregate statistics (such as a median or mean) accurately capture individuals’ actual comparisons. Another approach entails using instrumental variables that help isolate errors in \(m\) and improve the reliability of the threshold estimate, although these solutions are often exceedingly difficult to implement in practice. In this paper, we adopt a more direct and new solution by leveraging a unique data source – the interviewer paradata – which is usually not accessible by researchers during the data acquisition process. Specifically, in the dataset used here, interviewers assess each respondent’s dwelling quality relative to that of the surrounding residences, generating an external observation of whether \({a}_{i}\) is effectively above or below \(m\). Such paradata reduce the risk of self-reporting biases, provide an externally assessed and locally relevant view of the housing standard, and thus help pinpoint which individuals lie above \(({a}_{i}>m)\) or below \(({a}_{i}<m)\) the relevant threshold.

This paper makes several contributions to the literature on subjective well-being and wealth. First, it shifts attention from the well-studied role of income to the relatively understudied role of real assets in shaping life satisfaction. Second, it introduces a novel, data-driven approach to capturing relative wealth by using interviewer-based paradata rather than relying on arbitrary reference groups defined by demographic characteristics. This allows us to assess relative effects in a more localised and independently measured way. Finally, we provide new evidence from a Central and Eastern European context – a region still underrepresented in the literature – highlighting how historical and cultural legacies of housing and homeownership continue to shape the economic and psychological lives of individuals.

Our findings provide insights for housing and social policy. First, policies aimed at improving housing quality may yield well-being gains that go beyond material living standards, especially for those living in disadvantaged neighbourhoods. For example, Glaeser (2019) recently questions the consequences of ‘poor doors’ in the context of segregation and opportunities. Second, the evidence that downward housing comparisons are particularly detrimental suggests that spatial inequality in housing conditions can carry psychological costs. Third, our findings underline the importance of considering both absolute and relative dimensions of wealth when designing policies aimed at improving subjective well-being. Measures that address housing disparities – such as targeted renovation programmes or affordable housing initiatives – may not only improve material conditions (Lindenthal, 2020) but also reduce the negative well-being effects associated with positional disadvantage in housing.

At the same time, our analysis has some limitations particularly those relating to policy implications. While our novel approach of using interviewers’ paradata could identify relative effects within neighbourhoods, given that our context of housing is characterised by observability of asset characteristics, we cannot rule out other possible explanations and mechanisms in action. For example, it is possible that the paradata measure captures certain aspects of respondents' material circumstances more effectively than other available indicators. Furthermore, processes related to perceptions of fairness or disadvantage – concepts that are similar, but not identical, to social comparisons – may also be relevant. Furthermore, as our analysis relies on variables assessing subjective perceptions, namely the interviewer's subjective evaluation of housing and the interviewee's subjective reports of life satisfaction, our results may be subject to common method variance (CMV) bias. As emphasised by Lindell and Whitney (2001), CMV can lead to artefactual relationships that do not necessarily reflect the true underlying associations between constructs, thereby threatening the validity of the conclusions. However, assuming independence of interviewer and interviewee reported information, it is unlikely that CMV bias will be present. Nevertheless, we recognise the potential CMV bias as a limitation because our study does not explicitly deal with CMV.

The considerations above point to avenues for future research. Firstly, richer data infrastructures would enable more precise identification of mechanisms. Combining interviewer-level assessments with administrative data at the postcode or municipal level could help to distinguish social comparison processes from other contextual effects. Secondly, direct survey questions regarding comparison groups, perceptions of fairness or reference standards could complement the paradata-based approach, clarifying whether the observed relative effects are driven by social comparisons or alternative psychological mechanisms. Finally, longitudinal data would enable the examination of how changes in housing conditions and neighbourhood environments affect life satisfaction over time. These would allow researchers to address CMV bias concerns and provide stronger grounds for causal inference.