The Impact of Commercial Medical Insurance Participation on Household Debt

Abstract

Household debt is an important part of household financial decision-making, and commercial medical insurance has gradually become an important tool for households to use in improving their household balance sheets. Based on 2017 China Household Finance Survey (CHFS) data, this paper studies the impact of commercial medical insurance participation on household debt and analyzes the heterogeneity of household conditions, such as the location of the household, the age of the household head, and the health status of members. The study found that households participating in commercial medical insurance are more likely to be indebted, and their degree of debt is higher than that of households without commercial medical insurance. For urban households, young households, and households with healthy members, the participation of commercial medical insurance has a high effect on the likelihood and the degree of debt. Therefore, while strengthening household insurance awareness, the government should promote the strengthening of the risk-resistance function of commercial medical insurance and encourage financial institutions to design products that combine insurance and credit to release households’ consumption and investment potential.

1. Introduction

The wealth status of households has a significant impact on the healthy development of a country’s economy. Household balance sheets, corporate balance sheets, and government balance sheets together describe the complete national economic fundamentals and the development of a country’s economy [1]. Healthy and reasonable household balance sheets can establish a healthy and stable national financial ecology, which is of great significance in enhancing the economy’s capability of coping with external market shocks and preventing systemic financial risks. Traditional theory suggests that household debt is generated to smooth out the intertemporal consumption patterns of households and to maximize consumption and capital utility to improve household welfare.

Compared with developed countries, China is a late starter in household debt. However, the price of housing, which is a rigid demand of households, has risen substantially, and the housing debt ratio has gradually increased in recent years [2]. Therefore, it is of great practical importance to promote the healthy operation of China’s economy by comprehensively exploring the influencing factors of household debt in the country to guide households in consuming moderately and, at the same time, reasonably controlling the debt scale of households.

As an important household financial behavior, insurance participation is closely related to household debt [3]. However, little literature on commercial medical insurance in household finance focuses directly on the relationship between purchasing commercial medical insurance and household debt [4,5]. More studies are concerned with the impact of insurance on households’ savings, consumption, and financial asset allocation [6,7,8,9]. Most studies on the impact of insurance on households’ saving and consumption behavior have concluded that participation in social medical insurance, pensions, and commercial insurance causes uncertainty about what will occur in the future [10,11,12,13]. Therefore, medical insurance will reduce precautionary saving and promote consumption to a certain extent [6]. Meanwhile, commercial medical insurance offers households access to higher medical coverage at lower costs, satisfies diversified medical demands, and reduces the uncertainty of households’ future medical expenditures [7,14]. It also results in lower precautionary savings and increased consumption by households.

On the other hand, many studies have found that households’ risk tolerance is enhanced when they are covered by medical and life insurance and that the likelihood and proportion of risky financial asset allocation are significantly higher [10]. Overall, the crucial factors affecting household debt and the impact of insurance participation on various household behaviors have been widely examined. However, fewer studies have analyzed the impact of commercial medical insurance participation on household debt. Therefore, this paper examines the impact of commercial medical insurance participation on household debt and analyzes the heterogeneity of household conditions, such as the household’s location, the household head’s age, and the members’ health status.

In China, commercial medical insurance is an important part of the medical insurance system. It is more targeted than public medical insurance and can provide households with more diversified medical insurance options [15]. As the government encourages households to participate in commercial medical insurance and companies actively improve insurance products, the threshold for commercial medical insurance participation in China is not high, especially after 2015.

First, the government issued “Implementing the Pilot Program of Individual Income Tax Policy for Commercial Medical Insurance” in 2015, encouraging households to participate in commercial medical insurance by reducing individual income tax [16]. Second, in order to cover more consumers, insurance companies designed various types of commercial medical insurance based on the consumption levels and needs of households, with prices ranging from about ten to millions RMB. Insurance with a low price also covers some diseases with a very low probability of occurrence but high medical costs, which can be affordable for poor households. Finally, due to the rapid popularization and development of the Internet in China, sales of commercial medical insurance have been available online since 1997. After the China Insurance Regulatory Commission issued the “Interim Measures for the Management of Internet Insurance Business” in 2015, Alipay, WeChat, and many banks quickly launched online sales of commercial medical insurance through apps, which made it easy for everyone to participate in commercial medical insurance [17]. Therefore, in China, almost every household can afford and conveniently participate in commercial medical insurance, although the types of insurance may be different. Since almost all Chinese households have basic medical insurance, which weakens their reliance on commercial medical insurance, the participation rate of commercial medical insurance is not high [18].

Based on the analysis related to commercial medical insurance and household debt, this paper examines the impact of commercial medical insurance participation on household debt through Probit and Tobit regression models. Using CHFS2017 household microdata as the sample, it further analyzes the heterogeneous role of multiple household characteristics. This study recognizes a significant positive effect of commercial medical insurance on household debt and its degree. Urban households, young households, and households whose members are in good health are more likely to incur household debt after participating in commercial medical insurance, and the degree of debt is elevated even more.

This paper is innovative in the following ways. First, previous studies on China’s household debt have mainly concentrated on household credit constraints and farm borrowing. This paper examines China’s household debt from a different perspective of commercial medical insurance participation, which complements and improves the theoretical system regarding household debt and finance. Second, existing research on insurance in China focuses on the effects of insurance participation on household savings, consumption, and investment, largely concentrating on social medical insurance and basic medical insurance, with less emphasis on commercial medical insurance. This study explores the economic effects of commercial medical insurance and provides suggestions for its better services to household finance.

2. Theoretical Framework and Hypotheses

2.1. Mechanisms of Commercial Medical Insurance

The fundamentals of commercial medical insurance are designed to allow households to receive a large percentage of cash benefits if a household member suffers from a contracted illness or a prescribed medical event, which shifts the financial risk resulting from the illness to the insurance company. Unlike commodities under the general demand theory, commercial insurance, as a non-desired commodity, suffers from severe information asymmetry. Lack of familiarity with the product can lead to weak demand motivation of consumers. Instead, consumers purchase commercial medical insurance mainly to obtain medical assistance to cover potentially high future medical costs and to realize income transfers [19]. The precautionary saving theory argues that risk-averse consumers will reduce their consumption levels to save in consideration of the uncertainty of future income [20], which will affect the rational allocation of household assets and weaken households’ incentives to expand their consumption and to engage in risky investments. Commercial medical insurance can reduce the uncertainty of future income, lower the concern about future risks, reduce precautionary savings, and increase people’s current propensity to consume [8].

Deterministic consumption theory assumes that current income determines current consumption. However, when income increases significantly, the surplus after satisfying consumption follows, and it moves to the next period in the form of savings and other property [21]. Conversely, people can rely on loans for consumption if their current income and possessions cannot meet their consumption desires. Many studies argue that the assumption of deterministic consumption theory is too strict to analyze practical problems, and income uncertainty should be introduced. Such studies have developed a series of non-deterministic consumption theories, including the theory of precautionary saving, the theory of liquidity constraints, and the buffer stock model [20,22,23]. The non-deterministic consumption theory suggests that consumption depends on consumer credit and the mutual transformation of income and property, which results in great uncertainty [24]. In order to resist future uncertainties and to reduce asset losses that may be brought about by risks, households usually take measures to protect themselves against risk, and the simplest form of protection is precautionary savings. However, when households purchase commercial medical insurance, their future uncertainty and the loss of assets caused by it are also reduced. The households do not have a higher urgency to save in the current period, and precautionary savings are reduced [14]. Consequently, households are more inclined to consume, generating or increasing household debt [11].

Meanwhile, participation in commercial medical insurance increases households’ tolerance for future risk, which makes them more willing to invest in risky assets, thus generating or increasing household debt [9]. In addition, households expect fewer future asset losses and liquidity constraints after participating in commercial medical insurance [25]. Consequently, their consumption and investment propensity will increase, and the possibility of smoothing consumption or investing by borrowing funds will be higher. Based on the above analysis, this paper proposes the following hypotheses:

H1a. 

Households participating in commercial medical insurance are more likely to be indebted.

H1b. 

Households participating in commercial medical insurance have a higher degree of debt than those without commercial medical insurance.

2.2. Households’ Heterogeneity Impact

2.2.1. Difference between Urban and Rural Areas

Lewis’s dual economic model argues that in developing countries, due to industrial differences between urban and rural areas, the production and living conditions of urban and rural residents are quite different [26]. As China is still a developing country, its urban–rural dual structure has resulted in a completely different allocation of accessible resources between urban and rural households. Consequently, households’ savings, consumption, and investment are affected differently. Rural areas are dominated by agriculture, while urban areas are dominated by the manufacturing and services industries. The industrial gap between the two areas causes the income and the stability of urban households to be higher than that of rural households. In the face of future uncertainty, rural households need to have more savings than urban households [27]. While urban areas have more consumable items and higher consumption levels, rural households are limited in their consumption perceptions and behavior due to the relative backwardness of their regional infrastructure and education levels [28]. Thus, urban households are more likely to smooth their consumption with borrowing than are rural households.

In terms of investment, urban households have more convenient access to financial investment channels, due to the distribution of a greater number of financial institutions in urban areas. Compared with urban households, rural ones have an insufficient supply of financial services and limited access to information. Meanwhile, agricultural operations’ high input and low returns lead to a lower possibility of risky financial asset allocation for rural households. As a result, urban households will have higher investment possibilities and riskier financial assets. Rural households are subject to higher liquidity constraints than urban households, due to their lower agricultural production incomes and heavy family burdens [29]. In addition, rural households are subject to more eligibility constraints in terms of borrowing channels, and urban households are more advantaged in borrowing [30]. Therefore, after participating in commercial medical insurance, the likelihood and the degree of debt of urban households will be higher than that of rural ones. Based on the above analysis, this paper proposes the following hypothesis:

H2a. 

Participation in commercial medical insurance elevates the likelihood and the degree of debt of urban households more than rural ones.

2.2.2. Age Differences of Household Heads

The household life cycle theory argues that the constraints faced by households at different stages vary greatly, including income levels, consumption, and members’ physical conditions. In addition, differences in constraints can make households’ expected utility usages vary in different periods, affecting their consumption and investment behavior [31]. Therefore, households’ consumption and investment behaviors are characterized by the life cycle stages of their households. The life cycle stage of the household becomes an essential factor in the process of utility maximization and a necessary basis for households in making decisions on saving, consumption, and investment [32]. The household head’s age usually reflects the household’s life cycle stage.

In terms of savings and consumption, younger households have yet to accumulate wealth in the early years but consume more and save less than older households. As younger married households gradually have children, their consumption expenditure will continuously grow. In older households, children usually already have financial means, and household spending is relatively lower and more stable [33]. Meanwhile, in younger households, the overall income grows at an early stage, peaks, and then declines in the members’ middle ages. Accordingly, younger households have higher income expectations than older households [34] and will have a greater incentive to borrow to smooth consumption across time.

In terms of investment, households allocate appropriate assets according to the different stages of the life cycle [35], and younger households are more concerned than older ones about financial management. At the same time, the relationship between the degree of risk aversion and age decreases and then increases [36], indicating that younger households are less risk-averse than older ones and would be more willing to allocate risky financial assets. Therefore, reducing future uncertainty after participating in commercial medical insurance will significantly impact younger households’ debt decisions. Based on the above analysis, this paper proposes the following hypothesis:

H2b. 

Participation in commercial medical insurance elevates the likelihood and the degree of debt of younger households more than older ones.

2.2.3. Health Differences of Household Members

The Grossman model of the demand for health argues that the health status of household members affects their future healthcare expenditures and expected incomes [37]. Therefore, the decision-making behavior of households in participating in commercial medical insurance is tightly linked to their members’ health status. Suppose a household member is in good health. In that case, the household is likely to incur lower future health expenditures and have higher and more stable expected future income than a household with poor health status [38]. Since there is no necessity for high savings to cope with potentially large health expenditures, households with good health status have less urgency to save, less precautionary savings, and a higher propensity to consume in the current period. In contrast, households with poor health status participating in commercial medical insurance will be more conservative in saving and consumption because of subsequent medical expenses and healthcare costs, even though a large percentage of future high medical expenses will be covered [39].

Moreover, due to the potential for significant medical expenses from future health shocks, households with poor health status will reduce their need for risky assets and increase their safe-asset holdings to cope with the risks. The goal of their household asset holdings will be more focused on protecting the health of household members and maintaining basic survival, rather than on leveraging risky investments to achieve household wealth appreciation. Consequently, they will have a greater demand to hold assets with high liquidity and security to cope with major medical expenses [40]. In addition, because households with good health status have higher and more stable expected future incomes and greater repayment capacities, they have fewer liquidity constraints. They are more likely to obtain loans from financial or non-financial institutions. Therefore, upon participating in commercial medical insurance, the reduction in future uncertainty will have a greater impact on the debt decisions of healthy households. Based on the above analysis, this paper proposes the following hypothesis:

H2C. 

Participation in commercial medical insurance elevates the likelihood and the degree of debt of households with good health status more than those with poor health status.

3. Data Source, Variable Selection, and Descriptive Statistics

3.1. Data Source

The data used in this paper are from the China Household Finance Survey (CHFS) conducted by the Survey and Research Center for China Household Finance in 2017. The survey collects relevant information about household financial behavior and provides high-quality micro household financial data for academic research and government decision-making. It covers 29 provinces (autonomous regions and municipalities directly under the central government), excluding Xinjiang, Tibet, Hong Kong, Macao, and Taiwan, 355 counties (districts and county-level cities), and 1428 village (residence) committees, with a sample size of 40,011 households. To ensure that the data accurately reflects the household situation in China, CHFS2017 adopted a three-stage stratified sampling method. In that method, all statistical results are adjusted by the number of people in a city (county), as the weight. The survey questionnaire is comprehensive and informative, including demographic characteristics, assets, debts, insurance participation, expenditures, incomes, and financial literacies of households.

3.2. Variable Selection

3.2.1. Explained Variable

In order to study the impact of commercial medical insurance participation on household debt, this paper examines the willingness of households to be indebted and the degree of debt: (1) debt_or_not indicates that a household is indebted if it has any form of borrowing or loans for house purchase, construction, renovation, or other purposes—if so, it is assigned a value of 1; otherwise, it is not indebted and is assigned a value of 0; (2) debt_degree indicates the household debt ratio to its total assets and represents the degree of household debt.

3.2.2. Explanatory Variable

CMI_hh, commercial_medical_insurance_household indicates that if one member has purchased commercial medical insurance, the household is a commercial medical insurance participating household and it is assigned a value of 1; otherwise, it is a non-commercial medical insurance participating household and is assigned a value of 0.

3.2.3. Control Variables

Many studies have analyzed the influencing factors of household debt [41,42,43,44,45,46,47,48,49], with empirical results indicating that optimistic household income expectations increase the likelihood and the degree of household debt [41]. At the same time, current household income levels also significantly impact household debt [42]. Moreover, the allocation of household assets has a major impact on household debt [43]. Holding more financial assets raises household debt and owning more property positively affects household debt [44]. A higher level of financial literacy facilitates households’ financial decision-making [45]. Financially literate households are more likely to adopt formal sources of finance and are more likely to manage their household debt appropriately [46]. The urban–rural dual structure in China has resulted in differences in the financial literacy of urban and rural households, which have led to differences in debt behavior [47]. Income uncertainty is lower for urban households than it is for rural households [48]. As the underlying theory of household finance research, the life cycle theory establishes the importance of age in research on household debt. Households with younger heads have higher borrowing demands than households with older heads [49]. Similarly, the health levels of household members, a factor strongly associated with age, negatively correlates with household debt [10]. In addition, household characteristics such as household size, marital status, and education level have varying degrees of impact on household debt [43,49].

Based on existing literature, this paper identified the individual characteristic variables of the household head that affect household debt as control variables, including gender, age, marriage status, education level, risk attitude, and financial knowledge. Among these variables, the following values were applied: (1) if the gender of the household head was male, the value was 1, and if the gender was female, the value was 0; (2) marriage experience was assigned a value of 1, and no marriage experience was assigned a value of 0; (3) the risk attitude of the household head was measured by the question “If you had an asset, which investment option would you like to choose?”—the risk attitude of the household head was assigned a value of 1–5, according to the choice of the risk and the return of the investment, with higher values indicating respondents who were more risk averse, and respondents with no investment knowledge were considered the most risk-averse; (4) to understand the financial knowledge levels of household heads, the answers to eleven questions about financial literacy (H301, H3103, H3105-H3107, H3110-H3115) of the CHFS2017 questionnaire were coded (except for two questions, H3101 and H3111), where a correct answer scored 1 point and an incorrect answer scored 0 points; H3101 and H3111 were divided into a gradient of 0.2, according to the degree of concern, with a total score of 1. Equal-weighted summation was applied to construct financial knowledge indicators for household heads, with data spanning a range of 0 to 11.

We also studied the household characteristics variables, including household geographic location (i.e., rural), household income, housing status (i.e., house), household size (i.e., count), and members’ health status (i.e., health_avg). Among these variables, the following values were applied: (1) household geographic location (i.e., rural) was assigned a value of 1 in rural areas and 0 in urban areas; (2) housing status (i.e., house) was measured by the number of houses owned by households; (3) household size (i.e., count) was measured by the number of household members; (4) the health status of household members was counted under the respondents’ self-assessment, based on the question “How is your current health status compared to your peers? 1. Very good; 2. Good; 3. General; 4. Bad; 5. Very bad “. This question assigned a score of 1–5 to the health status of household members, with higher scores representing better health, and generated an average health score (health_avg) for household members.

3.3. Model Design

This paper examined the impact of commercial medical insurance participation on household debt. Whether a household has debt was a 0–1 variable, so a Probit model was used to analyze the effect of commercial medical insurance participation on whether a household was indebted, and a regression model was constructed as follows:

debt_or_not = β₀ + β₁CMI_hh + β₂X + μ

(1)

In Equation (1), debt_or_not is whether the household has debt. A value of 1 indicates indebtedness, while a value of 0 indicates non-indebtedness; CMI_hh is whether the household participates in commercial medical insurance. A value of 1 indicates that at least one member of the household participates in commercial medical insurance, and a value of 0 indicates that no member participates. X is the control variable related to household debt stated above; and μ is the random error term.

The degree of household debt is typically censored data. Some households are not indebted because they are not eligible for loans. There are also some households with small amounts of debt that the interviewers consider unnecessary to report. Therefore, this type of sample data was considered censored data, and this degree of debt was 0 in the research results. In addition, some households’ financial situation deteriorated significantly in the current period, due to specific reasons, and their assets were negative; those households with minimal assets were reported as 0. Some data presented negative values or infinity when calculating the degree of household debt. However, their actual economic meanings were that the household was bankrupt, and such sample data were considered as censored data. Based on this, the Tobit model was applied to examine the effect of commercial medical insurance participation on the degree of household debt, and the regression model was constructed as follows:

debt_degree = β₀ + β₁CMI_hh + β₂X + μ

(2)

In Equation (2), debt_degree refers to the degree of household debt, and the meaning of the remaining variables is consistent with the meanings for Equation (1).

Since there may be a two-way causal relationship between commercial medical insurance participation and household debt, the endogeneity problem should be considered. When a household incurs debt, its financial vulnerability increases, and it will likely purchase commercial medical insurance to transfer risk due to resistance to future risk. Moreover, the model may have neglected relevant variables that affect household debt, such as cultural factors, social customs, and other variables that are difficult to observe directly. All these factors may have affected the empirical results. Therefore, this paper introduced instrumental variables to demonstrate the reliability of the results. We selected the household’s commercial medical insurance participation rate (CMI_rate_hh_community) in the community or village as the first instrumental variable for whether the household participated in commercial medical insurance. Moreover, we added the household’s social subjective trust (i.e., trust) as the second instrumental variable. We adopted a two-step approach to solve the Probit and Tobit models with potential endogeneity problems.

The reasons for selecting these two instrumental variables were as follows: first, population distribution usually has a certain degree of commonality. Households in communities and villages have a high degree of similarity. Therefore, the participation rate of commercial medical insurance in a community or village is closely related to whether the household participates in the insurance. Second, commercial insurance sales rely heavily on the promotion of local sales staff based in communities and business centers in a specific range of locations, which makes the correlation between the two high. Furthermore, although the two are correlated, overall household commercial medical insurance participation in the community or village does not affect a specific household’s generation of indebtedness or the degree of indebtedness. Finally, commercial medical insurance requires the purchaser to trust the product and the salesperson. However, since no directly relevant question was asked in the questionnaire, the subjective social trust of the household head was adopted as a substitute, and this variable was not significantly associated with household debt.

4. Empirical Results

4.1. Descriptive Statistics

The CHFS2017 contains low-quality data for some households and missing values on key variables. After excluding such samples and those households in the top 1% of indebtedness, this paper retained a valid sample size of 11,315.

Table 1. Descriptive statistics of the main variables.

Variables	Meaning	N	Mean	SD	Min	Max
debt_or_not	Household indebtedness or not (1: indebtedness; 0: non-indebtedness)	11,315	0.2977	0.4573	0	1
debt degree	The degree of household debt	11,315	0.1873	1.6193	0	90.1786
CMI_hh	Commercial medical insurance participation (1: participation; 0: non-participation)	11,315	0.1158	0.3200	0	1
rural	Household geographic location (1: rural; 0: urban)	11,315	0.2220	0.4156	0	1
income	Household income (take natural logarithm)	11,315	4.7019	0.1525	0	6.3986
house	Housing status	11,315	1.2014	0.5038	0	9
age	Age of household head	11,315	54.0681	14.8414	16	117
gender	Gender of household head (1: male; 0: female)	11,315	0.7681	0.4221	0	1
marriage	Marriage status of household head (1: married; 0: unmarried)	11,315	0.9319	0.2520	0	1
education	Education level of household head	11,315	3.7689	1.8176	1	9
risk attitude	Risk attitude of household head	11,315	4.0785	1.1850	1	5
financial knowledge	Financial knowledge of household head	11,315	1.8381	1.1884	0	9.8000
count	Household size	11,315	3.0493	1.4669	1	15
health_avg	Members’ health status	11,315	3.6038	0.8279	1	5

demonstrates the descriptive statistics of the variables. About 29.77% of households in the sample were in debt, and the mean value of the degree of debt of households was 18.73%. In comparison, China’s nationwide commercial medical insurance participation rate is 11.58%.

4.2. Regression Results

4.2.1. Household Indebtedness or Not-Indebtedness

The results in column (2) of

Table 2. The effect of commercial medical insurance participation on whether a household has debt.

	Probit	Probit	IV Probit
Variables	(1)	(2)	(3)
CMI_hh	0.4000 ***	0.1780 ***	0.9678 ***
	(0.0373)	(0.0404)	(0.1994)
rural	-	0.3209 ***	0.3241 ***
	-	(0.0343)	(0.0349)
income	-	0.1041	−0.0252
	-	(0.0856)	(0.0932)
house	-	0.2221 ***	0.1985 ***
	-	(0.0259)	(0.0270)
age	-	−0.0312 ***	−0.0292 ***
	-	(0.0012)	(0.0013)
gender	-	0.0075	0.0194
	-	(0.0330)	(0.0336)
marriage	-	0.1943 ***	0.1574 ***
	-	(0.0554)	(0.0571)
education	-	−0.0162 *	−0.0288 ***
	-	(0.0090)	(0.0097)
risk attitude	-	−0.0320 ***	−0.0237 *
	-	(0.0122)	(0.0126)
financial knowledge	-	0.0111	−0.0076
	-	(0.0127)	(0.0137)
count	-	0.1411 ***	0.1382 ***
	-	(0.0097)	(0.0099)
health_avg	-	−0.1908 ***	−0.1985 ***
	-	(0.0179)	(0.0183)
constant	−0.5809 ***	0.4579	1.0013 **
	(0.0133)	(0.4079)	(0.4380)
N	11,315	11,315	11,315
Pseudo R2	0.0083	0.1276	-
LR/Wald chi2	113.96	1758.19	1494.03
Endogeneity test	-	-	16.92 ***
p value (Wald)	-	-	0.0000
F-statistics of first stage	-	-	129.62

Note: Standard errors are in parentheses; *, **, and *** denote the significance levels at 10%, 5%, and 1%, respectively.

reveal a significant positive effect of commercial medical insurance participation on whether a household has debt, with an average marginal effect of 0.1780. This indicates that participation in commercial medical insurance increases the likelihood of incurring debt in households. When a household member purchases commercial medical insurance, the possibility of the household’s indebtedness increases by 0.1780 on average. This finding supports hypothesis H_1A: households participating in commercial medical insurance are more likely to be indebted. The p-value of the Wald test results in column (3) is 0.000, indicating that the hypothesis of no endogeneity is rejected at the 1% statistical level and, therefore, there is an endogeneity problem. The estimated first-stage F-value of the instrumental-variables method is 129.62, which is greater than the critical value of 10 under 10% bias, so the instrumental variables are appropriate, and there are no weak instrumental variables. After adding the instrumental variables, the core variables’ direction of influence and significance level remain the same. However, there is a significant difference in the magnitude of the values, indicating that when endogeneity is not considered, the impact of households’ participation in commercial medical insurance on debt is underestimated. Therefore, IV Probit estimation is necessary.

4.2.2. Household Indebtedness Degree

The results in column (2) of

Table 3. The effect of commercial medical insurance participation on the degree of household debt.

	Tobit	Tobit	IV Tobit
Variables	(1)	(2)	(3)
CMI_hh	0.7302 ***	0.3518 ***	1.8281 ***
	(0.1140)	(0.1196)	(0.5881)
rural	-	0.8228 ***	0.8290 ***
	-	(0.1029)	(0.1036)
income	-	−0.0325	−0.2684
	-	(0.2374)	(0.2561)
house	-	0.3628 ***	0.3190 ***
	-	(0.0763)	(0.0787)
age	-	−0.0750 ***	−0.0711 ***
	-	(0.0036)	(0.0039)
gender	-	0.0876	0.1098
	-	(0.0999)	(0.1009)
marriage	-	0.4761 ***	0.4061 **
	-	(0.1671)	(0.1704)
education	-	−0.0606 **	−0.0846 ***
	-	(0.0270)	(0.0287)
risk attitude	-	−0.0374	−0.0218
	-	(0.0367)	(0.0374)
financial knowledge	-	0.0076	−0.0284
	-	(0.0380)	(0.0408)
count	-	0.3050 ***	0.2997 ***
	-	(0.0289)	(0.0292)
health_avg	-	−0.5700 ***	−0.5842 ***
	-	(0.0540)	(0.0546)
constant	−2.4786 ***	1.8725	2.8603 **
	(0.0549)	(1.1409)	(1.2115)
N	11,315	11,315	11,315
Pseudo R2	0.0017	0.0420	-
LR/Wald chi2	40.87	1004.30	808.58
Endogeneity test	-	-	6.67 ***
p value (Wald)	-	-	0.0098
F-statistics of first stage	-	-	129.62

Note: Standard errors are in parentheses; ** and *** denote the significance levels at 5%, and 1%, respectively.

reveal a significant positive effect of commercial medical insurance participation on the degree of household debt, with an average marginal effect of 0.3518. This indicates that participation in commercial medical insurance increases the degree of household debt. When a household member purchases commercial medical insurance, the degree of the household debt increases by 0.3518 on average. This finding supports hypothesis H_1B: households participating in commercial medical insurance have a higher degree of debt than those without commercial medical insurance. The p-value of the Wald test results in column (3) is 0.0098, indicating that the hypothesis of no endogeneity is rejected at the 1% statistical level, so there is an endogeneity problem. The estimated first-stage F-value of the instrumental-variables method is 129.62, which is greater than the critical value of 10 under 10% bias, so the instrumental variables are appropriate, and there are no weak instrumental variables. After adding the instrumental variables, the core variables’ direction of influence and significance level remain the same. However, there is a significant difference in the magnitude of the values, indicating that when endogeneity is not considered, the impact of households’ participation in commercial medical insurance on indebtedness degree is underestimated. Therefore, IV Tobit estimation is necessary.

It has been suggested that medical insurance can reduce the precautionary saving of households [12]. In further analysis, most studies prefer to explore the impact of public medical insurance on household finance. Public medical insurance reduces large unexpected medical expenditures of households [13], so that it reduces the risk of medical debt [4], thereby reducing household bankruptcy risk [5]. Due to the reduction of medical costs, public medical insurance reduces the risk of rent arrears [3] and increases future access to credit [50]. Based on the above research, we further found that households participating in commercial medical insurance are more likely to be indebted and have a higher degree of debt than those without commercial medical insurance. Our findings support the research progress on the impact of medical insurance on household finance. At the same time, the current research on commercial medical insurance focuses more on the impact of migration status, urbanization, and other factors on insurance participation [15,18], and less on the impact of commercial medical insurance participation on household finance. Our research also supplements the literature on commercial medical insurance.

4.2.3. Difference between Urban and Rural Areas

In order to investigate the urban–rural differences in the relationship between commercial medical insurance participation and household debt, this paper classified the samples into rural and urban households. The sample of rural households in

Table 4. Difference between urban and rural areas.

	Rural Households		Urban Households
	IV Probit	IV Tobit	IV Probit	IV Tobit
Variables	debt_or_not	debt_degree	debt_or_not	debt_degree
CMI_hh	0.0310	−1.2190	1.0966 ***	2.2095 ***
	(0.4089)	(1.3103)	(0.2302)	(0.6505)
income	−0.4655 **	−0.9578 *	0.0630	−0.1687
	(0.2239)	(0.5549)	(0.1093)	(0.2887)
house	0.2760 ***	0.2921	0.1775 ***	0.2849 ***
	(0.0756)	(0.2373)	(0.0292)	(0.0816)
age	−0.0249 ***	−0.0625 ***	−0.0301 ***	−0.0708 ***
	(0.0025)	(0.0082)	(0.0015)	(0.0044)
gender	0.1082	0.1438	0.0063	0.0976
	(0.0939)	(0.3023)	(0.0364)	(0.1048)
marriage	0.1319	0.4692	0.1785 ***	0.3901 **
	(0.1346)	(0.4353)	(0.0639)	(0.1830)
education	−0.0537 *	−0.1092	−0.0334 ***	−0.0975 ***
	(0.0288)	(0.0924)	(0.0105)	(0.0299)
risk attitude	−0.0512 **	−0.1478 *	−0.0077	0.0411
	(0.0256)	(0.0811)	(0.0147)	(0.0421)
fFinancial knowledge	−0.0462	−0.1241	0.0055	−0.0014
	(0.0289)	(0.0942)	(0.0157)	(0.0447)
count	0.1461 ***	0.3085 ***	0.1437 ***	0.3235 ***
	(0.0170)	(0.0536)	(0.0125)	(0.0354)
health_avg	−0.3073 ***	−0.9291 ***	−0.1425***	−0.4049 ***
	(0.0339)	(0.1087)	(0.0220)	(0.0634)
constant	3.6348 ***	8.2476 ***	0.3348	1.5049
	(1.0471)	(2.6612)	(0.5121)	(1.3613)
N	2512	2512	8803	8803
Wald chi2	271.72	158.23	1190.47	604.26

Note: Standard errors are in parentheses; *, **, and *** denote the significance levels at 10%, 5%, and 1%, respectively.

indicates that participation in commercial medical insurance had no significant impact on whether the household was indebted or on the degree of indebtedness. However, there was a significant positive effect among urban households. Thus, there were specific differences in the impact of commercial medical insurance participation on urban and rural household debt. This finding supports hypothesis H_2A: participation in commercial medical insurance elevates the likelihood and the degree of indebtedness of urban households more than rural ones.

The urban–rural dual structure in China has resulted in a completely different allocation of accessible resources between urban and rural households, leading to significant differences in the items that households can consume and invest in. First, compared with urban households, rural ones have lower incomes and consumption levels and greater income instability; in addition, they usually save in higher proportions. Taking housing as an example, most of the self-owned houses of rural households are self-built, while urban households mainly purchase commercial houses. There are major differences between the two in terms of the scale and proportion of household funds that are used. Second, due to regulations on the qualifications of formal borrowing channels, rural households are generally subject to more borrowing restrictions than urban households and face tighter credit constraints. In addition, the distribution of financial institutions is greater in towns and cities, so urban households are more financially literate than rural ones and are more likely to engage in risky financial assets, resulting in a more rational household asset structure. Consequently, participation in commercial medical insurance reduces the uncertainty of future expenditures to some extent for both urban and rural households. However, differences in income and expenditure levels, borrowing channels, financial literacy, and other conditions may cause participating urban households to unleash their consumption and investment potential more significantly, increasing the likelihood and the degree of household debt.

4.2.4. Age Differences of Household Heads

In order to investigate the impact of commercial medical insurance participation on household debt among household heads of different age groups, the paper classified the samples into young households whose heads were under 60 years old and old households whose heads were over 60 years old. The results in

Table 5. Age differences of household heads.

	Young Households		Old Households
	IV Probit	IV Tobit	IV Probit	IV Tobit
Variables	debt_or_not	debt_degree	debt_or_not	debt_degree
CMI_hh	1.0993 ***	1.9257 ***	0.7227	2.7602
	(0.1903)	(0.5276)	(0.9119)	(3.4406)
Rural	0.2773 ***	0.6589 ***	0.3587 ***	1.1193 ***
	(0.0423)	(0.1176)	(0.0631)	(0.2381)
income	0.0059	−0.1851	−0.8310 *	−2.9800 *
	(0.0935)	(0.2457)	(0.4553)	(1.6964)
house	0.1529 ***	0.1733 **	0.2465 ***	0.6207 **
	(0.0297)	(0.0817)	(0.0651)	(0.2421)
gender	0.0146	0.0894	0.0665	0.2747
	(0.0395)	(0.1116)	(0.0653)	(0.2472)
marriage	−0.0277	−0.0104	−0.3642 **	−0.6915
	(0.0583)	(0.1657)	(0.1637)	(0.6273)
education	0.0072	0.0034	−0.0455 **	−0.1683 **
	(0.0113)	(0.0316)	(0.0200)	(0.0759)
risk attitude	−0.0464 ***	−0.0501	−0.0236	−0.0708
	(0.0140)	(0.0391)	(0.0302)	(0.1135)
financial knowledge	−0.0016	0.0092	0.0067	−0.0949
	(0.0158)	(0.0442)	(0.0281)	(0.1058)
count	0.1174 ***	0.2226 ***	0.2528 ***	0.7203 ***
	(0.0122)	(0.0338)	(0.0183)	(0.0690)
health_avg	−0.1358 ***	−0.4116 ***	−0.2252 ***	−0.8577 ***
	(0.0220)	(0.0614)	(0.0330)	(0.1253)
vonstant	−0.4675	−0.6634	2.9459	10.3928
	(0.4503)	(1.1919)	(2.0759)	(7.7360)
N	7284	7284	4031	4031
Wald chi2	327.04	170.69	380.57	230.55

Note: Standard errors are in parentheses; *, **, and *** denote the significance levels at 10%, 5%, and 1%, respectively.

suggest that young households’ participation in commercial medical insurance had a significant positive impact on whether the household was indebted and on the degree of debt. The participation of old households in commercial medical insurance had no significant impact on whether the household was indebted or on the degree of debt. The results of group regressions revealed that the participation of commercial medical insurance had a distinct difference in the impact of household debt and the degree of debt between different ages. This finding supports hypothesis H_2B: participation in commercial medical insurance elevates the likelihood and the degree of debt of younger households more than the older ones.

The external constraints households face at different life-cycle stages varies significantly, which can lead to different expected utility usages of a household at different stages, affecting their consumption and investment behavior. Most young households do not accumulate wealth early in life and have a greater likelihood of debiting funds to smooth consumption across time. After they enter different stages of the household, such as marriage, childbirth, and children’s education, they may generate more consumption and have a stronger demand to utilize risky financial assets for wealth accumulation. At the same time, since households’ incomes are usually growing when they are young, the income expectations are higher than in their old age. In contrast, many members of old households may have retired or be resting at home due to poor health, and their expected income has declined. After their household enters the final stage, they have accumulated a certain amount of wealth, their children have the financial ability to support their own families and start businesses, and their expenditures will be relatively stable. Therefore, as young households have higher loan demand and fewer liquidity constraints, participation in commercial medical insurance may reduce the uncertainty of future expenditures, which will have a more pronounced effect on young households in increasing the likelihood and the degree of household indebtedness.

4.2.5. Health Differences of Household Members

In order to analyze the differences in the impact of commercial medical insurance participation on household debt among households with different health statuses, this paper classified the samples into households with poor health status (average score not exceeding 3) and households with good health status (average score exceeding 3), based on the average health score of household members. The regression results are summarized in

Table 6. Health differences of household members.

	Households with Poor Health Status		Households with Good Health Status
	IV Probit	IV Tobit	IV Probit	IV Tobit
Variables	debt_or_not	debt_degree	debt_or_not	debt_degree
CMI_hh	0.2500	−0.3956	1.0458 ***	1.8266 ***
	(0.5938)	(2.2131)	(0.2103)	(0.5451)
rural	0.4972 ***	1.5246 ***	0.2133 ***	0.4953 ***
	(0.0563)	(0.2095)	(0.0450)	(0.1182)
income	−0.6469 **	−2.4417 **	0.0293	−0.1613
	(0.3077)	(1.1953)	(0.0979)	(0.2358)
house	0.1431 **	0.1919	0.2086 ***	0.2922 ***
	(0.0658)	(0.2565)	(0.0300)	(0.0760)
age	−0.0303 ***	−0.0932 ***	−0.0275 ***	−0.0577 ***
	(0.0024)	(0.0090)	(0.0015)	(0.0041)
gender	−0.0559	−0.1623	0.0471	0.1809 *
	(0.0602)	(0.2271)	(0.0406)	(0.1071)
marriage	0.1067	0.4464	0.1618 **	0.3031
	(0.0944)	(0.3547)	(0.0719)	(0.1886)
education	−0.0548 ***	−0.1682 **	−0.0211 *	−0.0606 **
	(0.0197)	(0.0740)	(0.0112)	(0.0293)
risk attitude	−0.0339	−0.0938	−0.0166	0.0110
	(0.0240)	(0.0881)	(0.0149)	(0.0390)
financial knowledge	−0.0083	−0.1130	−0.0033	0.0093
	(0.0268)	(0.1010)	(0.0160)	(0.0419)
count	0.1731 ***	0.4504 ***	0.1407 ***	0.2749 ***
	(0.0190)	(0.0683)	(0.0121)	(0.0314)
constant	3.6665 ***	12.5976 **	−0.2626	−0.4628
	(1.3976)	(5.4374)	(0.4601)	(1.1174)
N	3623	3623	7692	7692
Wald chi2	580.97	334.93	946.73	475.75

Note: Standard errors are in parentheses; *, **, and *** denote the significance levels at 10%, 5%, and 1%, respectively.

. For households with poor health status, the participation in commercial medical insurance had no significant impact on whether the household was indebted or on the degree of debt. However, there was a significant positive effect on households with good health status. As a result, there were some variations in the impact of commercial medical insurance participation on the likelihood and the degree of debt among households with different health statuses. This finding supports hypothesis H_2C: participation in commercial medical insurance elevates the likelihood and the degree of household debt with members in good health than those with members in poor health.

The health status of members is among the crucial constraints for households. Households with good health status are less likely to incur future medical expenditures, while the expected future incomes are higher and more stable than those of households with poor health status. Consequently, with commercial medical insurance coverage, households with good health status may have lower future instability, higher income expectations, fewer credit constraints, and better conditions to meet consumption needs and risky investments through borrowing. Compared with households with poor health status, participation in insurance by households with good health status will have a more prominent effect on the likelihood and the degree of household debt.

4.3. Robustness Test

In order to test the robustness of the above empirical results, this paper adopted the per-capita household medical insurance premium expenditure (CMI_exp_avg) variable to measure household commercial medical insurance participation and assessed regression once again.

Table 7. The effect of per-capita medical insurance premium expenditure on household debt.

	Probit	Tobit	IV Probit	IV Tobit
Variables	debt_or_not	debt_degree	debt_or_not	debt_degree
lnCMI_exp_avg	0.0291 ***	0.0567 ***	0.1568 ***	0.2966 ***
	(0.0060)	(0.0178)	(0.0325)	(0.0957)
rural	0.3209 ***	0.8232 ***	0.3242 ***	0.8297 ***
	(0.0343)	(0.1029)	(0.0350)	(0.1037)
income	0.0978	−0.0475	−0.0594	−0.3367
	(0.0857)	(0.2376)	(0.0963)	(0.2650)
house	0.2208 ***	0.3598 ***	0.1920 ***	0.3062 ***
	(0.0259)	(0.0764)	(0.0274)	(0.0798)
age	−0.0312 ***	−0.0750 ***	−0.0292 ***	−0.0712 ***
	(0.0012)	(0.0036)	(0.0013)	(0.0039)
gender	0.0074	0.0880	0.0188	0.1094
	(0.0330)	(0.0999)	(0.0337)	(0.1009)
marriage	0.1930 ***	0.4743 ***	0.1528 ***	0.3976 **
	(0.0554)	(0.1671)	(0.0574)	(0.1710)
education	−0.0160 *	−0.0601 **	−0.0274 ***	−0.0819 ***
	(0.0090)	(0.0270)	(0.0096)	(0.0285)
risk attitude	−0.0313 **	−0.0364	−0.0205	−0.0161
	(0.0122)	(0.0367)	(0.0128)	(0.0378)
financial knowledge	0.0107	0.0066	−0.0093	−0.0320
	(0.0127)	(0.0380)	(0.0139)	(0.0412)
count	0.1418 ***	0.3062 ***	0.1415 ***	0.3058 ***
	(0.0097)	(0.0289)	(0.0099)	(0.0292)
health_avg	−0.1915 ***	−0.5712 ***	−0.2018 ***	−0.5905 ***
	(0.0179)	(0.0540)	(0.0184)	(0.0549)
constant	0.4891	1.9495 *	1.1679 ***	3.1939 **
	(0.4085)	(1.1421)	(0.4518)	(1.2504)
N	11,315	11,315	11,315	11,315
Wald chi2	1762.07	1005.85	1488.89	807.71

Note: Standard errors are in parentheses; *, **, and *** denote the significance levels at 10%, 5%, and 1%, respectively.

indicates that the household per-capita medical insurance premium expenditure has a significant positive impact on whether households are indebted and on the degree of debt. When the per-capita household medical insurance premium expenditure increases, households are more likely to incur debt, and the degree of debt increases. This test indicates that the empirical results of this paper have good robustness.

5. Conclusions

This paper constructed the IV Probit model and IV Tobit model by using data from the 2017 China Household Finance Survey (CHFS2017) of the China Household Finance Survey and Research Center to examine the impact of commercial medical insurance participation on household debt. The heterogeneous effects of household characteristics, such as household geographic location (urban or rural), household-head age, and household members’ health status were explored. The results indicated that (1) households participating in commercial medical insurance are more likely to be indebted and have a higher degree of debt than those without commercial medical insurance, and (2) urban households, young households, and households with good health status have a higher likelihood and the degree of debt when participating in commercial medical insurance.

Based on the above findings and considering the current situation of awareness of commercial medical insurance among Chinese households, this paper put forward the following recommendations from the perspective of promoting commercial medical insurance to better serve the lives of households and the national economy.

On the one hand, the government should vigorously promote the popularization of investment and financial knowledge, including insurance; provide more relevant learning platforms for households; strengthen insurance awareness among households; and increase attention to commercial medical insurance. The relevant departments and financial institutions should endeavor to promote commercial medical insurance so that a broader range of households can realize that commercial medical insurance has greater coverage than social medical insurance and is more resilient to future uncertainties. In addition, households should be aware that commercial medical insurance can help them reduce their concern and manage reasonable debt to smooth out household consumption.

On the other hand, the government should regulate the insurance market, clarify the supervision of commercial medical insurance, and strengthen the risk-resistance function of insurance to guide households in consuming and investing better. Moreover, insurance companies should lower the threshold for households to participate in commercial medical insurance and provide more appropriate products based on different demands of households, including rural households, older households, and households with health risks. Commercial banks should design products that combine insurance and credit, allowing the insurance businesses to stimulate credit and investment loans and guide households in releasing their consumption and investment potential.