Like father, like son? Intergenerational transmission of household housing preference: Evidence from China

Abstract

Based on the cross-generational data of the China Family Panel Studies (CFPS) in 2010, 2014, 2018, this paper empirically explores the intergenerational transmission of household housing preference and its underlying mechanism from a perspective of intergenerational transmission. It finds that: (1) There is a distinct intergenerational transmission of housing preference. (2) For offspring under the age of 45, or those who are female, the intergenerational transmission of housing preferences from parents to their children is more pronounced. (3) Fertility intentions exhibit a resource dilution effect on the intergenerational transmission of housing preference. (4) The stronger the belief within a family that a child's future success depends on the family's economic status, the more likely it is to negatively influence the child's housing preferences. (5) Parents not only affect their children's consumption preferences regarding housing but also shape their attitudes towards borrowing for home purchases. This paper offers new insights into understanding household housing preference, and provides a foundation for government initiatives aimed at guiding households in rational asset allocation through social mechanisms.

1. Introduction

With the increasing awareness of household financial management and the expanding array of financial products, residents are now facing more pressing demands for asset allocation and wealth management. A well-structured asset portfolio is not only crucial to the essential interests of households but also influences social progress and development [1]. The efficient allocation of household assets plays a significant role in enhancing household wealth and fostering entrepreneurial activities [2]. Thus, the study of household portfolios has emerged as a prominent research topic.

Over the past three decades, the composition of household wealth in China has been predominantly characterized by investments in housing and fixed bank deposits. Concurrently, notable features of the Chinese economy include a high homeownership rate, an elevated savings rate, and an increasingly concentrated distribution of wealth [3]. The underlying motivations for these trends merit further investigation.

The existing literature indicates that key determinants influencing household portfolio choices encompass several factors: low intergenerational mobility, China's entrenched cultural tradition of “building a nest to attract the phoenix”, national family planning policies, and the social phenomenon characterized by a declining birthrate coupled with an aging population [[4], [5], [6], [7]]. Consequently, intergenerational transmission has garnered significant attention as a critical determinant influencing household behavior and decision-making processes.

This paper empirically investigates the intergenerational transmission of household housing preference and its underlying mechanism based on the cross-generational data of the China Family Panel Studies (CFPS) conducted in 2010, 2014, 2018. Compared to previous studies, the primary contributions of this paper can be summarized in two aspects: On one hand, it empirically examines the intergenerational transmission effect of family housing preferences using cross-generational household micro-data. This approach provides a novel analytical perspective for exploring the pronounced housing preferences observed among Chinese households. On the other hand, it enhances the existing literature on family finance and behavioral finance, providing valuable scientific insights that can inform government housing policy formulation and assist residents in making rational asset allocation decisions.

The structure of the paper is organized as follows: The second section offers a comprehensive review of the literature on intergenerational transmission and associated theoretical frameworks. The third section presents experiential facts regarding asset allocation among Chinese households. The fourth section details on the data selection, sorting, and analysis for the econometric model, thereby establishing a foundation for the subsequent empirical analysis. The fifth section constructs an econometric model to empirically test the intergenerational transmission effect of housing preferences. Finally, the sixth section summarizes and offers relevant policy recommendations.

2. Literature review

The literature on intergenerational transmission examines the persistence of welfare and family wealth across generations, aiming to elucidate the underlying mechanisms involved [8]. For example, research indicates that individuals save not only as a buffer against income shocks but also to bequeath greater wealth to their children due to “bequest motives” [9]. Children's relative positions within the income distribution are significantly influenced by their parents' incomes, revealing a notable correlation between child and parental income levels [10]. As society evolves, the intergenerational transmission of both income and occupation reinforces class dominance, primarily driven by an intensifying transfer of cultural capital across generations [11]. In China, there exists a pronounced intergenerational transmission of higher education; children whose parents have attained higher education are more likely to pursue similar educational paths and subsequently secure better career opportunities [12]. Within the framework of parental influence through financial resources, the distinct role of parental time investments is crucial in enhancing human capital among offspring from disadvantaged backgrounds [13]. Historical events shape contemporary trends in intergenerational mobility; moreover, generational changes partially mitigate the effects of rising skills premiums on income mobility in the United States [14].

Financial literacy and financial decision-making play a crucial role in contributing to the significant concentration of wealth and diminishing liquidity among social classes [15]. The transfer of financial knowledge predominantly occurs within the family, primarily from parents, rather than through external sources [16]. Students hailing from families with greater financial resources tend to exhibit substantially higher levels of financial literacy compared to their less affluent counterparts [17]. Financial literacy significantly influences consumer financial behaviors [18]. Social dynamics exert considerable influence on the daily lives of privileged youth. These individuals enjoy distinct advantages in both housing and labor markets relative to their peers [19]. The intergenerational transfer of wealth and resources from parents to Millennials has afforded those with such privileges an opportunity to prepare for an increasingly competitive housing and labor market through parental support [20]. This intergenerational transfer is instrumental in alleviating the substantial pressures associated with homeownership faced by younger generations [21]. There exists a notable correlation between parental income, economic status, and social standing on one hand, and offspring housing wealth on the other hand [22].

Beliefs and values that shape individuals' perceptions, preferences, and behaviors can significantly influence their financial decision-making [23]. The intergenerational transmission of views and preferences has emerged as a prominent area of research. Parents often seek to pass on a reflection of their personal traits to their children, viewing parent-child similarity as an anticipated outcome of this transmission process [24]. A widely accepted perspective posits that parents design their children's socialization practices based on their own values. This concept is metaphorically referred to as the “Transmission Belt” [25,26]. Research indicates that the stronger a particular value is endorsed by parents, or the more broadly it is esteemed within society, the greater the parents' desire to instill it in their children [27]. Children may also draw upon cultural and cognitive norms to inform their behaviors [26]; however, long-term development of behaviors and preferences, such as saving habits and consumption patterns, among young people can be subtly influenced by parental guidance over time [28]. Hoellger et al. examined the alignment of values between parents and children, discovering that children's perceived satisfaction was linked to value congruence within mother-child and father-child dyads [29]. Perales et al. noted significant intergenerational connections regarding gender ideology; both paternal and maternal attitudes exert comparable influences on children's beliefs, but these influences are complementary rather than cumulative effects. While fathers' attitudes equally affect sons' and daughters' perspectives, mothers' attitudes tend to have a more pronounced impact on daughters' compared to sons' [30].

In summary, the majority of existing studies on the intergenerational transmission of assets have primarily focused on factors such as occupation, education, income, and housing assets. While many investigations into the intergenerational transmission of preferences have centered around values, filial piety, and perspectives on gender and marriage, there has been a notable lack of emphasis on portfolio choice or preferences. Consequently, this paper empirically examines the intergenerational transmission of household portfolio choice using cross-generational micro-data from the China Family Panel Studies (CFPS) conducted in 2010, 2014, and 2018. This study holds both theoretical and practical significance. It not only offers a novel perspective for understanding intergenerational mobility but also serves as a scientific reference for government policies aimed at guiding household portfolio decisions.

3. Experiential facts

Fig. 1, Fig. 2, Fig. 3 below illustrates the participation rates, asset scale and proportion of various types of assets among Chinese households across six survey rounds (2010–2020). In this study, the assets participation rate refers to the percentage of households that own a specific type of assets relative to all resident households. Meanwhile, the assets proportion denotes the average share of that particular type of asset within the total assets of all households.

Fig. 1.

Participation rates of households by asset type, 2010–2020.

Fig. 2.

Scales of all types of household assets, 2010–2020.

Fig. 3.

Proportion of household assets by category, 2010–2020.

As shown in Fig. 1, Chinese households exhibit three distinct patterns of asset participation, categorized by high, medium, and low participation rates. Specifically, this indicates high participation rates for housing assets and durable consumer goods, medium rates for financial assets and land assets, and a low rate for productive fixed assets. Notably, the participation rates for housing assets and durable consumer goods (with the 2010 data for this category excluded due to unavailability) have consistently surpassed 80 % over the years, with the housing asset participation rate peaking at an impressive 97.98 % in 2018. The participation rate for durable consumer goods has also remained relatively stable, with even its lowest rate at a substantial 88.26 %, reflecting a pronounced preference among Chinese households for housing consumption and practical spending behaviors. Conversely, the overall participation rate for productive fixed assets remains comparatively low, never surpassing 40 % at its highest point. Participation rates for financial assets and land assets typically fluctuate between 40 % and 80 %. In particularly, while the financial asset participation rate generally exhibits an upward trend over time, it experiences significant fluctuations during certain years, likely influenced by market dynamics. Conversely, the land asset participation rate follows an “inverted U-shaped” pattern that mirrors the overall trend of one increasing (financial assets) while another declines (land assets).

In terms of the scale of various types of assets held by Chinese households, housing assets prominently dominate and exhibit a significant upward trend. Specifically, the average value of housing assets increased from 303,300 yuan in 2010 to 651,200 yuan in 2020. Notably, the average household property scale surpassed 600,000 yuan in 2018. The sixth round of surveys indicates that the overall asset scale has more than doubled since its initial measurement. The scales of other asset types have remained relatively stable; however, financial assets and durable consumer goods have shown a steady increase. In contrast, land assets and productive fixed assets have experienced minimal changes over this period. With the exception of financial assets, other asset types have consistently remained below 100,000 yuan in value, with the largest increase limited to approximately 14,000 yuan (see Fig. 2).

Reflecting the trends in asset scale, the distribution of household assets in China reveals a distinct pattern characterized by the dominance of one category: housing assets. These assets constitute a substantial proportion of total household wealth, while land assets, financial assets, productive fixed assets, and durable consumer goods each account for comparatively smaller shares. As illustrated in Fig. 3, the proportion of housing assets has consistently remained elevated, fluctuating between 70 % and 90 %, significantly surpassing the proportions attributed to the other four asset types. This pronounced preference for housing consumption among households—coupled with the considerable scale of housing investments—tends to displace holdings in alternative asset categories.

In terms of household debt, mortgages represent the predominant component. As depicted in Fig. 4, the proportion of mortgages among Chinese households remains consistently high, with even the lowest ratio exceeding 85 %. Although there has been a decline in the overall proportion of mortgages, the participation rate has exhibited a notable upward trend annually, rising from 5.48 % in 2010 to 22.75 % in 2020.

Fig. 4.

Household mortgage proportion and participation rate, 2010–2020.

The analysis of various types of household assets illustrated in Fig. 1, Fig. 2, Fig. 3 reveals the characteristics of Chinese households, which tend to “prioritize housing assets while neglecting others”. Furthermore, Fig. 4 underscores the phenomenon where mortgages represent a significant portion of household liabilities, with the mortgage participation rate demonstrating a consistent annual increase. This particular structure of asset allocation and loan distribution may potentially crowd out consumption and savings; over time, excessive debt could adversely affect China's macroeconomy. Given the current asset allocation patterns among Chinese households, it is essential to investigate whether there exists an intergenerational transmission of this distinctive housing preference in China. This topic warrants comprehensive examination to enable government guidance for households in achieving a more rational allocation of their assets.

4. Data selection and statistical analysis

This paper utilizes data from the China Family Panel Studies (CFPS) conducted by Peking University. To enhance our understanding of the intergenerational transmission of housing preferences, we have compiled the data from the years in which the CFPS questionnaire included relevant questions, such as “The importance of having children to you” (2010, 2018) or “The importance of having at least one son to you (in order to carry on the family line, people should have at least one son)” (2014). Subsequently, we performed a detailed analysis by matching survey data from 2010, 2014, and 2018.

The advantages of the CFPS database are as follows: (1) It encompasses data from a total of 16,000 households across 25 provinces in China. (2) It incorporates information at individual, household, and community levels. (3) It can accurately identify and match household codes with individual codes. (4) Data collection is conducted on a continuous basis; sample data are tracked and updated at least every two years. (5) It contains rich information, including detailed statistics on social, economic, demographic, educational, and health changes.

The rationale for selecting CFPS data as the foundation for data matching lies in the belief that the influence of parental generation on the preferences of subsequent generations constitutes a long-term developmental process that remains relatively stable over extended periods. Consequently, this influence is not expected to change significantly due to events such as parental death.

The CFPS survey comprises four primary types of questionnaires: community questionnaires, family questionnaires, adult questionnaires, and children's questionnaires. These correspond to five distinct databases: community database, family member relationships database, family economy database, adult database, and children's database. This study specifically focuses on three relevant databases: the family member relationships database, the family economy database, and the adult database.

The specific methodology for data matching is outlined below:

Firstly, we merged the family economy database with the adult database. Financial data for families were obtained from the family economy database, where the financial respondent (typically designated as the head of household) was identified by default in the questionnaire. Subsequently, we matched IDs from both databases to extract statistical information regarding offspring (individuals) and their parents.

Secondly, we integrated the adult database with the family member relationships database. By posing questions such as “Is this individual alive?”, “Does this individual economically belong to this household?”, and “Does this individual share meals with other family members?”, we distinguished data concerning surviving children who are economically linked to their families from those whose parents do not maintain an economic bond with them.

Next, utilizing household IDs, we conducted a cross-year data matching process based on previously described methods; indicators from 2018 were sequentially aligned with relevant datasets from 2014 to 2010 to create preliminary panel data.

Finally, in order to maximize the sample size of our study, missing values for relevant indicators in 2018 were supplemented using data from 2014 to 2010. The complete dataset for 2018 was then applied across subsequent years to ensure continuity and completeness of data pertaining to families within the study period. This process ultimately yielded cross-generational panel data spanning the years 2010, 2014, and 2018.

This paper examines intergenerational transmission of household portfolio choice. In profiling this choice, we draw upon Wu and Wang's measurement of risk preference [31], which is based on the proportion of risky assets. We utilize asset shares to approximate portfolio preferences and assess household investment strategies: a higher percentage of a specific type of asset in the portfolio indicates a stronger preference for that asset class. Building on the selection of control variables identified by Yin and Gan [32] as well as Zhang and Wu [33], we introduce three categories of control variables in this study. The first category includes personal characteristics, including offspring gender, age, years of education, marital status, and political affiliation (specifically whether one is a member of the Communist Party of China). The second category pertains to household characteristics, incorporating family size (number of individuals) and net per capita household income. The third category addresses geographical characteristics, which include offspring household registration status (urban or rural) and regional location (eastern, central, or western China). Furthermore, given that income levels exhibit right-skewness, we apply logarithmic transformation to these data so that the residuals from our estimated results align more closely with a normal distribution.

Based on the data screened according to the aforementioned criteria, outliers at the 1 % and 99 % levels for all continuous variables were truncated. Consequently, valid sample data for matching children with their parents was ultimately obtained. The selected core variables, along with their descriptions and descriptive statistics, are presented in Table 1 below.

Table 1.

Variables selected and descriptive statistics.

Variables	Variable declaration	Obs	Mean	Std	Min	Max
Explained variable
c_preference	The proportion of the offspring's housing assets to their total assets(%)	647	0.647	0.318	0.000	1.000
Explanatory variable
f_preference	The proportion of the parent's housing assets to their total assets(%)	647	0.666	0.295	0.000	1.000
Control variable
gender	Gender of the household head (1 for male, 0 for female)"	647	0.547	0.498	0.000	1.000
age	Age of Offspring (years)	647	35.35	10.82	23.00	78.00
urban	Offspring residence (1 indicates urban, 0 indicates rural)	647	0.631	0.483	0.000	1.000
marriage	Marital status of offspring (1 indicates married or cohabiting, while 0 denotes unmarried, divorced, or widowed)	647	0.810	0.393	0.000	1.000
education_year	Education levels of offspring are categorized as follows: 0 for never attended school, 6 for primary school education, 9 for junior high school education, 12 for senior high school education, 13 for technical secondary school degree, 15 for college degree, 16 for bachelor's degree, 19 for master's degree, and 22 for doctoral degree.	647	10.42	4.433	0.000	16.00
ln_income_per	Natural logarithm of per capita net household income in offspring (RMB yuan)	647	10.03	0.920	5.991	11.65
assetliability_ratio	Offspring household debt-to-asset ratio	647	0.135	0.378	0.000	3.448
family_size	Family size of offspring (number of individuals)	647	2.991	1.560	1.000	8.000

The above sample primarily illustrates the allocation of housing assets between two generations, along with the relevant characteristics of the offspring. From an asset allocation perspective, the selected samples predominantly consist of housing assets, with 64.7 % of children's family housing assets and 66.6 % of parents' housing assets attributed to this category. When examining the percentages of these assets, it is evident that both generations exhibit similar proportions in terms of housing asset allocation. This similarity may be attributable to the influence exerted by parental preferences regarding housing on their offspring's choices.

From the sample data regarding the personal characteristics of the offspring, it is evident that there are significantly more males than females, with a male-to-female ratio of approximately 1:0.82. The average age of respondents is 35.35, and the mean years of education stands at 10.42, suggesting that most individuals have attained at least a middle school level of education. Furthermore, during the survey year, 81.0 % of participants reported being married or cohabiting. In terms of household characteristics derived from the sample data, the average household size is 2.991 individuals, with sizes ranging from 1 to 8 members. The average net household income per capita amounts to 22,808.99 yuan, with values spanning from as low as 400.00 yuan to as high as 115,000.30 yuan. Additionally, the distribution range for household asset-liability ratios varies between 0 % and 344.8 %, highlighting a pronounced polarization in this indicator.

5. Modeling and empirical test

5.1. Main model and robustness test

In this study, the dependent variable is defined as the proportion of offspring housing assets relative to their total assets. The independent variables consist of the proportions of relevant parental assets and various characteristics of the offspring. Among these, control variables encompass personal, familial, and locational attributes of the offspring. This paper employs a two-way fixed effects estimation approach for analysis, with a detailed presentation of the constructed benchmark regression model provided below:

$${\mathrm{Y}\left(\text{Preference}\right)}_{\text{ijt},\mathrm{c}}={\beta }_0+{\beta }_1\cdot {\mathrm{X}\left(\text{Preference}\right)}_{\text{ijt},\mathrm{f}}+{\beta }_2\cdot {\text{Controls}}_{\text{ijt},\mathrm{c}}+{\delta }_j+{\eta }_t+{\epsilon }_{ijt}$$ (1)

Where, ${\mathrm{Y}\left(\text{Preference}\right)}_{\text{ijt},\mathrm{c}}$ represents the proportion of the offspring's housing assets to their total assets of household i in region j at time t. Similarly, ${\mathrm{X}\left(\mathrm{P}\text{reference}\right)}_{\text{ijt},\mathrm{f}}$ denotes the proportion of the parents' housing assets. The term ${\text{Controls}}_{\text{ijt},\mathrm{c}}$ refers to control variables related to the characteristics of the offspring. Additionally, ${\delta }_j$ and ${\eta }_t$ represent fixed effects for region and time, respectively, while ${\epsilon }_{ijt}$ is the error term. The main regression results from this model are presented in Table 2 below.

Table 2.

Empirical tests of the impact of generational pass-through on housing assets allocation.

VARIABLES	Baseline regression		Robustness test
	(1)	(2)	(3)	(4)	(5)	(6)
	OLS	OLS	IV-2SLS	IV-GMM2S	Tobit	OLS
	c_preference	c_preference	c_preference	c_preference	c_preference	c_preference
f_preference	0.1014∗∗∗	0.0981∗∗	0.1585∗∗∗	0.1157∗∗	0.1344∗∗∗	0.1029∗∗∗
	(0.037)	(0.041)	(0.039)	(0.054)	(0.049)	(0.040)
gender		−0.0600∗∗	−0.0906	−0.0604∗∗∗	−0.0797∗∗∗	−0.0608∗∗∗
		(0.023)	(0.057)	(0.023)	(0.027)	(0.023)
age		0.0036∗∗∗	0.0032	0.0035∗∗∗	0.0036∗∗	0.0033∗∗∗
		(0.001)	(0.003)	(0.001)	(0.001)	(0.001)
urban		0.0645∗∗	0.0499	0.0634∗∗	0.0535	0.0698∗∗
		(0.028)	(0.035)	(0.026)	(0.034)	(0.027)
marriage		0.0130	−0.0416	0.0130	−0.0015	0.0242
		(0.035)	(0.048)	(0.033)	(0.044)	(0.035)
education_year		0.0042	0.0076∗∗∗	0.0041	0.0029	0.0018
		(0.003)	(0.003)	(0.003)	(0.004)	(0.003)
ln_income_per		−0.0252	0.0133	−0.0243	−0.0201	−0.0266
		(0.017)	(0.018)	(0.016)	(0.021)	(0.017)
assetliability_ratio		−0.0279	−0.0560∗∗	−0.0284	−0.0181	−0.0073
		(0.033)	(0.028)	(0.029)	(0.049)	(0.033)
family_size		−0.0192∗∗	−0.0001	−0.0191∗∗	−0.0097	−0.0254∗∗∗
		(0.010)	(0.008)	(0.009)	(0.011)	(0.009)
Time effects	yes	yes	yes	yes	yes	yes
Regional effects	yes	yes	yes	yes	yes	yes
Time ∗ Regional effects	no	no	no	no	no	yes
Constant	0.5670∗∗∗	0.7067∗∗∗	0.2727	0.4617∗∗	0.4596∗	0.8490∗∗∗
	(0.027)	(0.191)	(0.439)	(0.212)	(0.258)	(0.221)
Observations	879	647	653	648	566	648
R-squared/Pseudo R-squared	0.2565	0.2696	–	–	0.4427	0.3381

Note: Standard errors of heteroscedasticity are presented in parentheses. Significance levels are indicated as follows: ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1 are significant levels. As a structural model, IV-2SLS and IV-GMM2S estimation do not account for the residuals of the predicted values of endogenous variables. Furthermore, the R-squared value associated with the instrumental variable method is not statistically significant; therefore, it is omitted from the table.

From the benchmark regression results presented in Table 2, it is evident that at the 1 % significance level, the ratio of parental housing assets positively influences the proportion of offspring's housing assets to their total assets. This effect remains significant even after controlling various variables, indicating an intergenerational transmission of housing preferences; specifically, a greater parental preference for housing correlates with a stronger inclination among offspring toward similar preferences. Notably, within family units spanning multiple generations, the strategic allocation of resources toward housing assets plays a crucial role.

Furthermore, to assess the robustness of the model, this paper adopts the processing approach proposed by Yang and Zhang [34], which utilizes the mean value of a specific region as an instrumental variable for the relevant variable. Specifically, it calculates the average value of the instrumental variable concerning housing asset proportions based on time (year fixed effects) and district (regional fixed effects) within CFPS data.

Additionally, this study draws upon Luo and Li [35] for GMM2S (Two-Stage Generalized Method of Moments) estimation employing instrumental variables. To further validate the robustness of our baseline regression model, we incorporated the Tobit model and analyzed the region-time interaction effect within the control model, capturing the interplay between regional and temporal dummy variables. As illustrated in columns (3)–(6) of Table 2, testing was conducted using 2SLS (Two-Stage Least Squares Method), GMM2S, Tobit models, and a model that controls for the region-time interaction effect respectively. The research findings remain consistent across these methodologies, thereby affirming both the robustness and reliability of the model.

5.2. Heterogeneity analysis

5.2.1. Age heterogeneity

The age structure of the population may significantly contribute to individual income disparities [36]. Elevated housing prices can influence residents' savings rates and exacerbate wealth inequality [37]. Furthermore, the age effect curve related to housing consumption, after a prolonged period of monotonic growth, tends to stabilize in later years. Given that housing consumption represents a substantial expenditure throughout an individual's life cycle, its variation across different age groups merits thorough investigation. Drawing upon life cycle theory [38] and the age classification methodology established by the World Health Organization [39], this paper categorizes sample age groups into three distinct phases: (1) the growth phase for individuals under 45 years old (youth), (2) the maturity phase for those aged between 45 and 60 years old (middle-aged), and (3) the decline phase for individuals over 60 years old (the elderly). This classification seeks to explore how housing consumption varies among households belonging to different age cohorts.

As indicated in Table 3, the influence of parents on their children's housing asset allocation is more pronounced among individuals under the age of 45. Further analysis using the Chow Test confirmed significant differences between these groups (p-value <0.05). Currently, the age at which Chinese adults first marry continues to be delayed; thus, residents under 45 are predominantly active in the marriage market and are influenced by phenomena such as the “mother-in-law economy” and the notion that “home is where the house is”, resulting in a relatively high demand for housing assets. The intergenerational transmission of housing preferences from fathers to sons is evident. As young people enter into matrimony, families' inclination towards consuming housing assets tends to diminish gradually. Subsequently, as residents' transition into retirement and pension stages, their demand for housing assets decreases and stabilizes.

Table 3.

Results of age heterogeneity analysis.

VARIABLES	<45	45–59	≥60
	(1)	(2)	(3)
	c_preference	c_preference	c_preference
f_preference	0.0983∗∗	−0.0091	−0.0592
	(0.045)	(0.148)	(0.152)
Control variables	yes	yes	yes
Timeeffects	yes	yes	yes
Regional effects	yes	yes	yes
Constant	0.8403∗∗∗	1.0248∗∗	0.9004∗
	(0.230)	(0.462)	(0.520)
Observations	530	84	68
R-squared	0.3239	0.4108	0.5886
Chow test (p-value)	0.0357

Note: Standard errors of heteroscedasticity in parentheses, ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1 are significant levels.

5.2.2. Marriage heterogeneity

The analysis of age effects indicates that marital status plays a significant role in moderating the inhibitory impact of cultural diversity on household asset allocation. This phenomenon may be attributed to the fact that, in contrast to single households, married households typically incorporate the incomes and wealth of both partners into their budgets. Consequently, this results in less stringent budget constraints and more flexible options for asset allocation. Furthermore, the influence of cultural exchange and dissemination on asset allocation decisions is likely to be more pronounced within married households. Therefore, this paper examines “marriage heterogeneity” by categorizing responses based on the “marital status” question from the CFPS data.

As illustrated in Table 4, at the 1 % significance level, the influence coefficient of marriage on housing preferences across the two generations is 0.1084. In contrast, the coefficient for the unmarried group is not statistically significant. This suggests that marriage may exert a positive effect on the intergenerational transmission of housing preferences. Further analysis using the Chow test indicated that the grouping coefficient did not meet the criteria for significance. This suggests that the differences between groups based on marital status are not particularly pronounced.

Table 4.

Results of marriage heterogeneity analysis.

VARIABLES	Married	Unmarried
	(1)	(2)
	c_preference	c_preference
f_preference	0.1084∗∗∗	−0.0315
	(0.038)	(0.067)
Control variables	yes	yes
Timeeffects	yes	yes
Regional effects	yes	yes
Constant	0.6629∗∗∗	0.9122
	(0.200)	(0.803)
Observations	524	119
R-squared	0.2505	0.5790
Chow test (p-value)	0.9188

Note: Standard errors of heteroscedasticity in parentheses, ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1 are significant levels.

5.2.3. Gender heterogeneity

Currently, the gender imbalance has intensified competition in the marriage market, prompting families to adopt diverse strategies for asset allocation. This shift subsequently influences the asset allocation methods employed by families with offspring of different genders [40]. Therefore, this study investigates whether housing preferences are transmitted differently based on gender through a process of gender grouping.

As illustrated in Table 5, the coefficient for male offspring is 0.1178, which is lower than the coefficient for female offspring at 0.1202. The results from the Chow test further suggest a significant difference in this coefficient across groups. This finding implies that parental housing preferences exert a more pronounced influence on the asset allocation of female offspring. On one hand, this may be attributed to the enduring impact of traditional “family culture” in China, where houses carry greater symbolic meaning. In contemporary society, residential properties are increasingly perceived as investment assets [41]. Consequently, financially capable parents may choose to acquire housing for their daughters in order to ensure their future well-being. On the other hand, the gradual reduction in gender bias favoring sons over daughters also plays a crucial role in this outcome [42]. Since the implementation of China's one-child policy, there has been a decline in family size alongside an increase in daughters' roles, responsibilities, and rights within familial lineage structures, significantly enhancing their status concerning intergenerational inheritance. Positive transformations in intergenerational relationships are reflected through parent-child equality. Liu et al. [5] found that persistently high housing prices also influence fertility rates. Specifically, rising property costs have a notably negative effect on male births while families residing in areas with elevated housing prices exhibit increased willingness to have daughters.

Table 5.

Results of gender heterogeneity analysis.

VARIABLES	Male	Female
	(1)	(2)
	c_preference	c_preference
f_preference	0.1178∗	0.1202∗∗
	(0.063)	(0.057)
Control variables	yes	yes
Timeeffects	yes	yes
Regional effects	yes	yes
Constant	1.0068∗∗∗	0.1475
	(0.234)	(0.272)
Observations	354	292
R-squared	0.3662	0.2023
Chow test (p-value)	0.0289

Note: Standard errors of heteroscedasticity in parentheses, ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1 are significant levels.

5.2.4. Income heterogeneity

The economic foundation determines the superstructure. Keynes's absolute income hypothesis posits that consumption is contingent upon current disposable income [43]. To further investigate the influence of income on housing preferences, the sample was categorized into high and low-income groups (or alternatively, high, middle, and low-income groups) based on the median (or quartile) of the logarithm of per capita household net income. An analysis of “income heterogeneity” was subsequently conducted.

The results presented in Table 6 indicate that the housing consumption of children from high-income families (as well as those from middle-income families) may be significantly influenced by their parents' consumption preferences. However, a thorough comparison between the two groups reveals that the Chow test results are not statistically significant, suggesting no substantial difference exists. It is important to note that, given the considerable size of the middle-income group, this study cannot overlook the impact of social class. Additionally, this study attempts to employ the EGP (Erikson-Goldthorpe-Portocarero) occupational classification derived from the CFPS questionnaire. The samples are categorized into high, middle, and low classes (or managers and skilled workers versus unskilled workers and farmers), following references [44,45], with an aim to conduct a heterogeneity analysis among middle-income families. Nevertheless, due to insufficient data matched with relevant variables, further analysis is not feasible.

Table 6.

Results of income heterogeneity analysis.

VARIABLES	Two-group test		Three-group test
	High income	Low income	High income	Middle income	Low income
	(1)	(2)	(3)	(4)	(5)
	c_preference	c_preference	c_preference	c_preference	c_preference
f_preference	0.0802∗	0.0338	−0.3045	0.1674∗	0.0786
	(0.045)	(0.091)	(0.228)	(0.093)	(0.050)
Control variables	yes	yes	yes	yes	yes
Timeeffects	yes	yes	yes	yes	yes
Regional effects	yes	yes	yes	yes	yes
Constant	0.4421∗∗∗	0.3414∗∗	0.0181	0.5033∗∗∗	0.5096∗∗∗
	(0.087)	(0.149)	(0.350)	(0.176)	(0.094)
Observations	551	177	47	141	450
R-squared	0.3166	0.2984	0.5145	0.2774	0.3322
Chow test (p-value)	0.4054		0.2368

Note: Standard errors of heteroscedasticity in parentheses, ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1 are significant levels.

5.2.5. Fertility heterogeneity

Against the backdrop of continuously evolving fertility policies, family fertility intentions can significantly influence consumption and savings behaviors. This, in turn, may lead to adjustments in traditional asset allocation concepts and alter the “parent-child similarity” observed in housing asset allocation across generations. To further explore whether the housing preferences of Chinese residents are affected by their fertility intentions (intergenerational concepts), this study utilizes data from the CFPS database. The analysis is based on responses to questions regarding “the importance of having a successor” (2010, 2018) or “the importance of having at least one son” (2014), employing median scores for grouped analysis.

As illustrated in Table 7, at the 5 % significance level, there exists a positive correlation between housing preferences of parents and offspring within the low fertility intention group. Conversely, the high fertility intention group does not exhibit a significant relationship, and the differences between these groups were confirmed by the Chow Test. This finding indirectly suggests that when a family's fertility intentions are lower (i.e., the importance of having a successor is less pronounced), paternal influence on offspring's' perspectives becomes more evident.

Table 7.

Results of fertility heterogeneity analysis.

VARIABLES	High fertility intention	Low fertility intention
	(1)	(2)
	c_preference	c_preference
f_preference	0.0330	0.3061∗∗
	(0.052)	(0.121)
Control variables	yes	yes
Timeeffects	yes	yes
Regional effects	yes	yes
Constant	0.8793∗∗∗	0.7260
	(0.257)	(0.461)
Observations	351	117
R-squared	0.2829	0.3906
Chow test (p-value)	0.0310

Note: Standard errors of heteroscedasticity in parentheses, ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1 are significant levels.

This phenomenon may arise because families with lower fertility intentions typically have fewer children, facilitating more comprehensive communication between generations. In contrast, families with higher fertility intentions often have multiple offspring, which can dilute housing preference discussions during intergenerational communication and consequently diminish the impact of intergenerational transmission to some extent.

5.2.6. Mechanism analysis

This paper employs the mediation effect analysis approach introduced by Baron and Kenny [46] to elucidate the influence mechanism that connects a family's economic status and children's housing preferences. Specifically, it investigates how a family's belief in the notion that a robust economic foundation can enhance their children's future success impacts the intergenerational transmission of housing preferences.

$${\mathrm{Y}\left(\text{Preference}\right)}_{\text{ijt},\mathrm{c}}={\beta }_0+{\beta }_1\cdot {\mathrm{X}\left(\text{Preference}\right)}_{\text{ijt},\mathrm{f}}+{\beta }_2\cdot {\text{Controls}}_{\text{ijt},\mathrm{c}}+{\delta }_j+{\eta }_t+{\epsilon }_{ijt}$$ (1)

$${\mathrm{Y}\left(\text{Wealth}\right)}_{\text{ijt},\mathrm{c}}={\lambda }_0+{\lambda }_1\cdot {\mathrm{X}\left(\text{Preference}\right)}_{\text{ijt},\mathrm{f}}+{\lambda }_2\cdot {\text{Controls}}_{\text{ijt},\mathrm{c}}+{\gamma }_j+{\rho }_t+{\xi }_{ijt}$$ (2)

$${\mathrm{Y}\left(\text{Preference}\right)}_{\text{ijt},\mathrm{c}}={\theta }_0+{\theta }_1\cdot {\mathrm{X}\left(\text{Preference}\right)}_{\text{ijt},\mathrm{f}}+{\theta }_2\cdot {\mathrm{X}\left(\text{Wealth}\right)}_{\text{ijt},\mathrm{c}}+{\theta }_3\cdot {\text{Controls}}_{\text{ijt},\mathrm{c}}+{\tau }_j+{\phi }_t+{\omega }_{ijt}$$ (3)

Among them, ${\left(\text{Wealth}\right)}_{\text{ijt},\mathrm{c}}$ is derived from the 2010 CFPS questionnaire and measures the degree of agreement with statements regarding the influence of family wealth on children's future achievements (specifically, children from affluent families tend to achieve more, while those from disadvantaged backgrounds tend to achieve less). The remaining variables are consistent with those utilized in previous studies.

Step 1

Model 1 is regressed to ascertain the significance of ${\beta }_1$;

Step 2

Assuming ${\beta }_1$ is significant, Model 2 and Model 3 are performed. If the coefficients ${\lambda }_1$ and ${\theta }_2$ are significantly different from zero, this indicates the presence of a mediating effect;

Step 3

If at least one of the two coefficients in Step 2 is not significant, a Sobel test is conducted.

The research results are shown in Table 8 below.

Table 8.

The result of mechanism analysis.

VARIABLES	Model 1	Model 2	Model 3
	(1)	(2)	(3)
	property_ratio	wealth	property_ratio
f_property_ratio	0.0981∗∗	1.8124∗∗∗	−0.4604∗∗∗
	(0.041)	(0.429)	(0.110)
wealth			0.1449∗∗∗
			(0.047)
Control variables	yes	yes	yes
Timeeffects	yes	yes	yes
Regional effects	yes	yes	yes
Constant	0.7067∗∗∗	−3.6831∗∗	1.2711∗
	(0.191)	(1.505)	(0.710)
Observations	647	36	36
R-squared	0.2696	0.7481	0.7141

Note: Standard errors of heteroscedasticity in parentheses, ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1.

As demonstrated by the findings presented in Table 8, an intermediary effect is identified wherein “the relationship between economy and achievement”, specifically, the impact of a family's economic foundation on children's future achievements, acts as the mediating variable. In other words, the greater the consensus among families that children's future accomplishments are dependent on their economic background, the more likely this belief is to negatively influence their children's housing preferences.

This phenomenon may occur because families who believe that “a solid economic foundation will positively impact children's future achievements” are more likely to encourage independence in their children. As a result, these children develop diverse values and interpretations of success, prioritizing personal interests and quality of life. This shift in priorities may reduce their inclination towards real estate investment.

5.2.7. Further analysis

In addition to heterogeneity and mechanism analyses, this paper examines the intergenerational transmission of debt acquisition concepts (i.e., household mortgage proportion to total debt) between two generations. Due to limited panel data for 2010, 2014, and 2018, this study uses cross-sectional data from 2018, the most recent year available, for its analysis.

Table 9 presents the regression analysis results of two generations regarding loans. The findings reveal a significant correlation between parents' mortgage-to-debt ratio and their offspring's housing preferences, indicating that parental influence affects children's debt behavior. In recent years, China's gender imbalance has intensified competition in the marriage market, making housing—a low-risk and highly visible asset—an indicator for eligible bachelors. This reflects the deep-rooted Chinese culture of “build a nest to woo the phoenix” [47]. A house has become largely essential for marriage in China; this phenomenon is often termed the “mother-in-law effect” [48]. Many parents insist on including their daughter's name on property deeds even if they do not contribute financially. However, whether couples should share mortgage payments remains “a matter open to discussion”. This study does not explore intergenerational borrowing concepts due to limited focus and data but plans more detailed research when sufficient data becomes available. Additionally, the belief that “home is where the house is” is deeply ingrained in Chinese society, underscoring housing's vital significance for its people.

Table 9.

Results of the offspring's mortgage to liabilities.

VARIABLES	OLS	OLS
	(1)	(2)
	mortgage_rate	mortgage_rate
f_ mortgage_rate	0.2684∗∗	0.2531∗
	(0.105)	(0.138)
Control variables	yes	yes
Regional effects	yes	yes
Constant	0.3719∗∗∗	−1.8376∗∗
	(0.065)	(0.800)
Observations	86	63
R-squared	0.0690	0.3498

Note: Standard errors of heteroscedasticity in parentheses, ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1.

6. Conclusion and implications

Based on micro-data from the China Family Panel Studies (CFPS) conducted in 2010, 2014, and 2018, this paper presents an empirical analysis focused on the intergenerational transmission of household housing preferences. The study reveals significant evidence of such transmission, with key findings as follows:

• (1)There exists a clear intergenerational transmission of housing preferences; specifically, stronger parental inclinations towards allocating housing assets correlate with a heightened propensity for property ownership among their offspring.
• (2)Among individuals under the age of 45, there is a notable distinction in the intergenerational transmission of housing preferences between fathers and daughters compared to other categories of offspring. In contrast, this transmission appears relatively weaker in other familial relationships.
• (3)Both high fertility intention and low fertility intention groups positively influence housing preferences across father-son generations. However, it is noteworthy that the impact coefficient associated with the low fertility intention group is larger, potentially attributable to its close relationship with resource dilution effects.
• (4)Furthermore, when families strongly believe that their children's future achievements are contingent upon their economic foundation, this belief tends to exert a negative influence on those children's housing preferences.
• (5)Additionally, parental influence extends beyond merely shaping their offspring's consumption preferences regarding property; it also significantly affects their attitudes toward borrowing for property acquisition.

As parents serve as the primary agents of socialization for their children, children's portfolio choices often reflect those of their parents to a significant degree. Consequently, it is crucial for parents to instill morally sound values and perspectives on life during their children's formative years. This guidance can aid in cultivating positive financial habits and establishing a well-structured investment portfolio. The younger generation, as vital contributors to society, should adopt a balanced view of property within the context of marriage and place greater emphasis on non-materialistic aspects of life. Furthermore, China should promote its rich traditional culture alongside socialist core values. It is essential to establish and enhance relevant laws and regulations that mitigate the “mother-in-law effect” while providing concrete and rational guidance regarding household financial asset portfolios.

CRediT authorship contribution statement

Lixuan Chen: Writing – review & editing, Writing – original draft, Methodology, Data curation, Conceptualization. Sijia Guo: Writing – review & editing, Writing – original draft, Funding acquisition, Data curation. Wen Zhang: Writing – original draft, Methodology. Xiuting Li: Supervision, Funding acquisition, Conceptualization. Jichang Dong: Supervision, Funding acquisition, Conceptualization.

Data availability statement

The data are from China Family Panel Studies (CFPS), funded by Peking University and the National Natural Science Foundation of China. The CFPS is maintained by the Institute of Social Science Survey of Peking University. Anyone who uses the CFPS data must go through an application process from: http://www.isss.pku.edu.cn/cfps/.

Funding statement

This work was supported by the National Natural Science Foundation of China (Grants No. 71974180 and 72204244).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.