Housing Security, Relative Deprivation, and Subjective Well-Being: Empirical Evidence Derived From CFPS Data

2.1 Literature Review

Happiness, alternatively known as SWB, is considered a “hidden national wealth” and serves as a vital metric for assessing residents' quality of life and societal welfare (Lu et al. 2017). During the 1960s to 1980s, research on SWB emerged as a focal point in psychology, gradually integrating into sociological and economic domains (Ouyang and Zhang 2018). Given its relevance to economic–social welfare, scholars have suggested incorporating SWB as an indicator for evaluating national economic and life quality (Diener 2000). In the economic field, maximizing social welfare is the ultimate goal of public policy, whereas subjective evaluations of individual form a crucial aspect of exploring bounded rationality. Consequently, an increasing number of scholars are probing the determinants of residents' SWB. Existing research recognizes personal income, employment status, health, housing prices (Kang and Park 2023), public housing environment (Latreille et al. 2024), social structures, government quality, natural environment, and urban planning as influential factors.

Housing is residents' essential living need and a primary determinant of their well-being (González et al. 2024), especially in China, where “settledness” embodies traditional familial aspirations. Both the Universal Declaration of Human Rights (1948) and the International Covenant on Economic, Social, and Cultural Rights recognized adequate housing as a component of the right to an adequate standard of living (González et al. 2024). The United Nations General Comment No. 4 (1991) on adequate housing encompasses legal guarantees of housing rights, affordability, appropriateness, and equitable housing opportunities. In other words, the right to housing includes the right to housing security, access to and ownership of housing in the market (Lyu et al. 2020). This signifies that housing transcends mere shelter, symbolizing dignified living and reflecting living standards and well-being.

As a societal welfare system, housing security must transcend focusing solely on socioeconomic performance, considering holistic influences on public welfare (Ferrara et al. 2022). The system chiefly aims through governmental or societal intervention to aid those economically disadvantaged in acquiring suitable housing, thereby diminishing residential cost pressures, reducing anxiety and discontent stemming from housing burdens, and subsequently enhancing residents' SWB (Leng and Zhu 2021; González et al. 2024; Latreille et al. 2024). Therefore, we propose:

Housing security aspires to eliminate economic-based residential disparities, enhancing residents' welfare (Juvenius 2024). However, individual characteristics, familial circumstances, and regional differences create varied housing preferences and conditions, which in turn modulate housing security's effect on SWB across demographics. Some scholars suggest complex similarities and differences between young and old populations, underscoring public housing's crucial role in young individuals' well-being (Latreille et al. 2024), while living environments disproportionately affect happiness across age groups (Reece et al. 2024). Additionally, housing security has such a close relationship with the burden of housing costs that it can be seen as a governmental wealth transfer, enhancing financial capability or reducing housing expenditures (H. Li et al. 2020), subsequently decreasing overworking and enhancing happiness (C. Zhang et al. 2023). Studies also indicate that housing welfare policies, such as public rental housing, alleviate housing cost burdens for disadvantaged and low-income families, improving psychological well-being (Kim et al. 2023; Gu and Zhu 2024). Moreover, given the vast regional disparities in urbanization, population growth, and housing demand–supply structures, rental costs and housing prices exhibit localized segmentation (Ozabor et al. 2024). This disparate pressure on housing affordability presents an inconsistent welfare effect across regions (Sun et al. 2024), leading to differential impacts of housing policy. Consequently, we hypothesize:

Housing security primarily manifests through in-kind allocations and monetary subsidies. The former distributes based on stipulated per capita space usage, assigning rents slightly below market rates (generally 10%–20% discounts), whereas the latter issues monetary subsidies based on “market pricing, tiered subsidy, subsidy-separation” principles. These methods essentially mitigate household housing burdens or enhance financial capacities. Moreover, policy implementation alters residents' previous perceptions of relative deprivation due to suboptimal housing or financial pressure, a state where individuals compare their situation to others and feel discontent upon recognizing possessed deficits (Runciman 1966). Intensified feelings of relative deprivation in consumption can lead to heightened status anxiety, resulting in multifaceted psychological, physiological, and societal repercussions (Hastings 2019; Chen et al. 2024). Thus, we propose:

4.1 Data

The data utilized in this article are sourced from the China Family Panel Studies (CFPS) conducted in 2014 and 2018. These datasets encompass three levels: individual, family, and community, and incorporate various economic indicators such as household income, expenditure, housing, assets, and debts. Additionally, they include personal information of family members, including age, gender, education, employment status, health conditions, subjective attitudes toward relevant issues, and cognitive test results. The indicator data actually employed in this study primarily originate from the household economic questionnaires and individual self-reported questionnaires within the CFPS database. The dependent variable in this article is SWB, whereas the core explanatory variable pertains to whether housing security has been obtained. Due to the significant deficiency of the “SWB” indicator data in the databases for 2016 and 2012 (with a missing rate of 95%) and the absence of questions or options related to affordable housing in the 2010 questionnaire, this study exclusively utilizes the family economic questionnaires and individual self-reported questionnaires from the CFPS database for 2014 and 2018.

To ensure the reliability of our empirical findings, we implemented several rigorous preprocessing steps. First, to maintain sample comparability, we retained only those households currently residing in nonowned accommodations. Second, acknowledging potential biases arising from disparities in housing security policy implementation across cities and uneven urban sample distribution, we excluded cities with a sample size below 30. This exclusion encompassed six cities located in Inner Mongolia, as well as Ningxia, Xinjiang, Hainan, Tibet, and Qinghai. Third, to mitigate bias stemming from the age of household heads, we restricted their ages to the range of 18–65 years. Fourth, in consideration of data integrity and continuity, we eliminated samples with missing or invalid values for key indicators. Fifth, to address the potential distortion of empirical outcomes by outliers among sample indicators, we applied a tail reduction of 1% on each continuous economic indicator, such as household income and debt. Lastly, we matched regional-level indicator data, including housing prices and consumer price indices, primarily sourced from the National Bureau of Statistics. Ultimately, this meticulous preprocessing process yielded a total of 3441 observations.

4.2 Empirical Model

$$\mathrm{Well}-{\mathrm{being}}_{i,j,t}$$ represents the SWB of respondent $$i$$ in province $$j$$ of year $$t$$. $${\mathrm{Housing\_security}}_{i,j,t}$$ indicates whether the respondent $$i$$ has received housing security. $${X}_{i,j,t}$$ is a set of control variables, including individual level, family level, and local level factors. $${Y}_{t}$$ is the year fixed effect, $${P}_{j}$$ is the province fixed effect, and $${\varepsilon }_{i,j,t}$$ is the random error term.

4.3 Definition of Key Variables

5.1 Basic Regression Result

Table 3 presents the regression results derived from Model (2) to examine the influence of housing security on family SWB. Specifically, Column (1) not only incorporates the core independent variable “Housing_security” but also comprehensively accounts for individual characteristics of the household head, family attributes, and economic factors pertinent to the province in which the family resides, thereby constructing a more nuanced and detailed regression model. Furthermore, building upon listing (2) in (1), by introducing province fixed effects and year fixed effects, we effectively control for external factors that may vary over time yet remain unobserved, thus enhancing both robustness and reliability of our research conclusions. This extension not only bolsters model robustness but also enriches our understanding of dynamic changes regarding the impact of housing security. Notably, whether considering Column (1) or Column (2), regression results consistently indicate a significantly positive coefficient for the “Housing_security” variable. This finding strongly implies that households with housing security exhibit higher overall levels of SWB compared to those without it. Such evidence substantiates the beneficial role played by housing security systems in enhancing well-being among protected families while revealing significant positive marginal effects through alleviating housing pressures; furthermore, it reaffirms that housing security is an essential component of social safety nets with critical implications for improving livelihoods and promoting residents' welfare.

5.2 Robustness Check

To ensure a high degree of reliability and credibility in the conclusions drawn from the baseline regression analysis, we have employed a series of rigorous and diversified methodologies to conduct robustness tests, aimed at comprehensively verifying the stability and consistency of the model under varying conditions. Specifically, we first implemented a sample replacement strategy whereby 90% of samples were randomly selected from the original full sample set for regression analysis. This step is designed to assess whether the regression results are contingent upon specific sample selections, thereby ensuring the generalizability of our conclusions. Second, we explored alternative explained variables by introducing new metrics based on the baseline regression model (where SWB level scores greater than 5 are coded as 1; otherwise coded as 0) to investigate whether variations in variable definitions affect the stability of regression coefficients. This approach elucidates the model's sensitivity to different metrics, thus enhancing conclusion robustness. Additionally, we conducted a replacement of the regression model itself by substituting the Ordered Logit model utilized in our baseline analysis with an alternative for further examination. This transformation not only assesses dependence on specific models but also reinforces both robustness and adaptability through comparative evaluation of estimated coefficients across different modeling frameworks. The empirical findings are presented in Table 4. Results from all robustness tests consistently indicate that estimated coefficients for our core explanatory variable “Housing_security” remain significantly positive across various testing scenarios. This outcome not only bolsters confidence in our baseline regression results but also clearly demonstrates that housing security exerts a stable and significant positive influence on study variables, thereby thoroughly substantiating their robustness.

To mitigate potential sample bias, we judiciously employed a methodology known as propensity score matching (PSM) to further enhance and validate our analysis. Specifically, we devised an experimental design framework wherein the success of households in securing housing served as a core variable; consequently, the sample was distinctly categorized into two groups: individuals receiving housing security formed the treatment group (designated treated = 1), whereas those not receiving such support were classified as the control group (designated treated = 0). In implementing the PSM strategy, we systematically utilized various matching techniques to ensure balance and representativeness within samples, including one-to-one matching, one-to-two matching, one-to-three matching, and one-to-four matching. The application of these diverse matching ratios aims to capture and address potential heterogeneity by enhancing the flexibility of matches, thereby improving the generalizability of our findings. As illustrated in Table 5, following data processing and statistical analysis, we observed that the T-value significantly surpassed the critical statistical threshold of 2 across all four adopted matching methods. This outcome not only underscores the statistical significance of the treatment effect but also robustly substantiates the reliability and validity of our benchmark empirical results presented herein. In essence, even after accounting for potential confounding factors related to sample selection bias, our conclusions remain resilient—providing substantial evidence for relevant policy formulation and practical interventions.

Considering that individuals affected by housing security policies may also benefit from other poverty alleviation policies that may affect their well-being. However, it is difficult to control the influence of these related policies in the empirical process. In this regard, this paper takes the land transfer data (logarithm) with the purpose of affordable housing at the provincial level1 as the instrumental variable to alleviate the endogeneity problem. The results are shown in Column (4) of Table 4. The regression coefficients of the key explanatory variables are all significantly positive, strengthening the verification of the robustness of the benchmark regression results in this paper.

5.3 Heterogeneity Analysis

6.1 The Mediating Effect of Income Relative to Sense of Deprivation

Runciman (1966) was the first to define the concept of “relative deprivation,” arguing that relative deprivation should meet the following four conditions at the same time: (1) the individual does not possess X; (2) other individuals in the reference group have X; (3) the individual expects to have X; (4) The individual believes that such expectations are justified. Since then, the concept has been gradually applied to areas such as consumption and income (Fehr and Schmidt 1999; Huang et al. 2019), derived from the concepts of “relative deprivation of income” and “relative deprivation of consumption.” Of course, this can also be understood as an upward social comparison triggered by the psychology of comparison, which is more likely to produce psychological gaps between individuals, inducing “illusions of loss,” which will directly affect people's sense of gain and happiness (Y. Zhang et al. 2022).

In Equation (3), $$\mathop{{M}_{Y}}\limits^{̅}$$ indicates the mean per capita disposable income of all the households in sample group Y, and $$\bar{{M}_{{y}_{i}}^{+}}$$ represents sample group Y in which the per capita disposable income of the households exceeds the mean per capita disposable income of $${y}_{i}$$. $${\gamma }_{{y}_{i}}^{+}$$ represents the percentage of the total sample size in which household per capita disposable income exceeds household $${y}_{i}$$ per capita disposable income in sample Y.

The concept of relative deprivation in household income is inherently relative. Before analyzing the impact of housing security on the sense of relative deprivation in household income, this paper first examines the influence of housing security on household savings2 at the conceptual level of absolute figures. The findings are presented in Column (1) of Table 9, demonstrating that households receiving housing security experience a significant increase in savings. Furthermore, the effect of housing security on the sense of relative deprivation in household income is illustrated in Columns (2) and (3) of Table 9. At a significance level of 5%, the housing security policy exhibits a substantial mitigating effect on the sense of relative deprivation in household income. Specifically, following the implementation of this policy, the perceived degree of relative income deprivation among families has significantly diminished. This change, in turn, positively enhances the SWB of families, as evidenced by a notable improvement in their self-reported well-being.

6.2 The Moderating Effect of Consumption Differences

In Formula (4), $$\mathop{{M}_{C}}\limits^{̅}$$ represents the average per capita household consumption of all samples in sample group C, and $${M}_{{c}_{i}}^{+}$$ represents sample group C in which the per capita household consumption exceeds the average per capita consumption of $${c}_{i}$$. $${\gamma }_{{c}_{i}}^{+}$$ represents the percentage of samples in sample group C where household per capita consumption exceeds household $${c}_{i}$$ per capita consumption.

The concept of relative deprivation in household consumption (consumption difference) is also a relative term. Before analyzing the relative deprivation in household consumption (consumption difference) induced by housing security, this study first examines the impact of housing security on household consumption at the conceptual level of absolute figures. As shown in Column (4) of Table 9, the results indicate that household consumption significantly decreases for those receiving housing security, primarily due to reduced housing-related expenditures. Furthermore, the influence of housing security on the relative deprivation of household consumption (consumption difference) is presented in Columns (5) and (6) of Table 9. After incorporating consumption disparity as a variable into the model, the coefficient of housing security demonstrates a significant upward trend compared to the corresponding coefficient in the benchmark regression model. This suggests that after effectively controlling for household consumption differences, the SWB of households has correspondingly improved. In other words, the existence of household consumption disparities weakens the happiness effect generated by housing security. Thus, consumption disparities function as a moderating variable, and its moderating effect has been empirically validated.