Based on the 2016 China Family Panel Studies (CFPS), this paper investigates the impact of China’s family social network on the total consumption level of residents and other three levels: survival consumption, enjoyable consumption and developmental consumption. The research shows: 1) Family social network has a significant effect on the consumption expenditure of residents, especially on the enjoyable consumption expenditure; 2) Social network can promote family consumption by alleviating the restriction of family mobility and demonstration effect; 3) The effects of social networks on consumption are different among different types of families. Compared with the eastern families, the social network of central and western families plays a more prominent role in promoting consumption expenditure. Families with lower income level re p ly heavily social network, and the effect of social network to stimulate consumption is more obvious.
Residents’ consumption has always been one of the hotspots of economic development. In the report of the 19th National Congress of the CPC, it is pointed out that we should focus on cultivating new growth points of middle and high-end consumption, improving the system and mechanism of promoting consumption, and strengthening the basic role of consumption in economic development. With the development of economy, the consumption level and mode of residents have also changed. From the perspective of different consumption levels, it can be summarized as the proportion of survival consumption in the total consumption of residents’ declines and changes to the consumption pattern of enjoyment and development, that is, residents’ consumption develops towards middle and high-end consumption. In order to explore the influencing factors of consumer behavior changes, many scholars have conducted investigations and studies, but most of the studies are carried out from the perspective of mainstream economics, such as credit market, asset allocation, monetary policy and so on. With the rise of socioeconomics and behavioral economics, the perspective of economic analysis has begun to expand to the interaction and behavior between economic subjects. China is a traditional relationship-based society, which pays more attention to interpersonal relationships. Social networks play an important role in economic and social life. As a special kind of family capital, social network of family also arouses attention and research on its influence on residents’ consumption behavior. On the one hand, consumption can not only bring people material enjoyment, but also enhance the exchange and transmission of information. For families with a wider social network, it also carries social significance and status attributes. Therefore, social networks can make consumption decisions for families through the communication and demonstration effects produced by residents in the process of social interaction. It will affect the consumption level and consumption structure of residents. On the other hand, social networks can give people such as job opportunities, financing channels and access to information to help, as an informal system to provide security for families, reduce the sensitivity of families to risk, and risk attitudes are important factors affecting household consumption decisions, so families can share risk through social networks, so as to mitigate the inhibitory effect of risk on Residents’ consumption. This shows that the impact of social networks on consumption worthy of more in-depth study.
The rest of the article is structured as follows. The second part combs the domestic and foreign literature about social network and resident consumption; the third part empirically analyzes the impact of family social network on different levels of consumer level and its impact mechanism; the fourth part is the robustness test; the fifth part is the conclusion.
Social network is a broader concept that covers sociology, economics and psychology. As a social capital, a non-market and informal institutional force, social network plays a special role in family and society, and has attracted more and more scholars’ attention. Scholars at home and abroad have done a lot of research on it. Research has yet to form a unified definition. Mitchell (1969) [
As a kind of social capital owned by family, social network plays an important role in information sharing and risk sharing. Like the material capital and human capital owned by family, social network can affect family decision-making and consumption behavior. Angelucci (2010) [
At present, there is little empirical research on the impact of social network on consumption at different levels and its mechanism. This paper will introduce the consumption level into the research framework from the micro perspective, and use micro-household data to empirically study the impact of family social network on consumption level and consumption structure and its communication. The mechanism of smoothing consumption is through risk sharing. This study will help to understand the impact of family social network on household consumption, provide a new perspective for the study of household consumption behavior, and provide preliminary empirical evidence for promoting household consumption in China.
The data used in this empirical study are from the latest China Family Panel Studies (CFPS) published on the official website of the Chinese Social Science Research Center of Peking University in 2016. CFPS is a biennial large-scale micro-household follow-up survey. There are four main types of questionnaires: community questionnaires, family questionnaires, adult questionnaires and children’s questionnaires. The sample covers about 16,000 households in 31 provinces/municipalities/autonomous regions except Hong Kong, Macao and Taiwan. It is a national, large-scale and multidisciplinary social follow-up survey project. The data used in this paper are from CFPS Adult Questionnaire and Family Questionnaire in 2016. Since there is no information about the head of household in the questionnaire, this paper identifies the head of household according to the “financial respondent” and matches the sample of family and adult with “personal ID”. Because the urban household consumption content is more abundant, the level of consumption is more complete and more representative, the rural household data are excluded, and only the urban household data are retained, and the missing and abnormal values of variables are excluded. Finally, 5719 observation values are obtained.
The explanatory variable of this empirical study is household consumption expenditure. Choose the annual per capita consumption expenditure as a dependent variable to analyze the impact of social network on the consumption level of residents, and divide the consumption expenditure into three levels: survival consumption, enjoyable consumption and developmental consumption to analyze the impact of family social network on the consumption structure of residents. The core explanatory variable is social network. According to the previous research literature, the selected indicators are mainly as follows: Ming Lu (2009) [
This paper also controls the variables of individual and family characteristics, including family size, family net income, family net assets, family housing net assets, urban and rural classification, gender, age, educational background, marital status, health status, work status. The definition and descriptive statistics of all variables are shown in
This paper uses the following model to study the impact of family social network on the consumption level of residents, and divides consumption expenditure into survival consumption expenditure, enjoyment consumption expenditure and development consumption expenditure.
ln total i = β 0 + β 1 ln network i + β 2 χ i + μ i (1)
ln survive i = β 0 + β 1 ln network i + β 2 χ i + μ i (2)
ln enjoy i = β 0 + β 1 ln network i + β 2 χ i + μ i (3)
ln develop i = β 0 + β 1 ln network i + β 2 χ i + μ i (4)
In Equation (1), ln total i represents the logarithm of the total consumption expenditure of family I in the whole year. In Equation (2), ln survive i represents the logarithm of the survival consumption expenditure of family I in the whole year. In Equation (3), ln enjoy i represents the logarithm of enjoyment consumption expenditure of family I in the whole year. In Equation (4), ln develop i represents the logarithm of development consumption expenditure of family I in the whole year. ln network i represents the social network of family I and take the household average human relationship and gift expenditure as the proxy variable. χ i is a personal characteristic variable and a family characteristic variable. μ i is a random disturbance term.
The mixed OLS regression method is used to estimate the above models, and the regression results are as follows (
From the above regression results, we can see that the elasticity coefficient of social network to the total consumption level is 0.0462, that is, the total
Variable identifier | Variable description | Mean value | Standard deviation | Minimum value | Maximum value |
---|---|---|---|---|---|
lntotal | The logarithm of total consumption expenditure | 10.821 | 0.805 | 8.380 | 12.913 |
lnsurvive | The logarithm of the sum of food, housing and clothing expenditures | 10.292 | 0.814 | 7.618 | 12.282 |
lnenjoy | The logarithm of the sum of expenditure on household equipment and daily necessities and transportation and communications | 8.831 | 1.212 | 5.480 | 12.427 |
lndevelop | The logarithm of the sum of expenditure on culture, entertainment, education and health care | 8.510 | 1.320 | 4.605 | 11.082 |
lnnetwork | The logarithm of human gift expenditure | 7.139 | 2.666 | 0 | 10.308 |
lnincome | The logarithm of net income of households | 10.525 | 1.892 | 0 | 12.899 |
lnfinanceasset | The logarithm of financial assets of households | 7.282 | 4.933 | 0 | 13.815 |
Size | Number of household population | 3.525 | 1.733 | 1 | 14 |
Gender | Male = 1, female = 0 | 0.463 | 0.498 | 0 | 1 |
Age | Age of head of household | 49.308 | 15.170 | 18 | 95 |
Edu | Illiteracy/semi-illiteracy = 1, primary school = 2, junior high school = 3, senior high school/technical school/vocational high school = 4, junior college = 5, undergraduate = 6, master’s degree and above = 7 | 3.048 | 1.425 | 1 | 7 |
Marriage | Unmarried = 0, married = 1 | 0.941 | 0.234 | 0 | 1 |
Health | Very unhealthy = 1, unhealthy = 2, relatively unhealthy = 3, generally = 4, health = 5 | 2.852 | 1.145 | 1 | 5 |
Work status | Unemployment or unemployment = 0, employment = 1 | 0.652 | 0.476 | 0 | 1 |
Data source: China family panel studies from 2016.
consumption level increases by 4.62% with the increase of 1% of the family expenditure on maintaining and expanding the social network, which indicates that the family social network can significantly improve the overall consumption level of residents.
This paper will further divide the eight types of consumption expenditure into three levels: survival consumption, enjoyable consumption and development consumption, and further explore the impact of social network on different levels of consumption.
(1) lntotal | (2) lnsurvive | (3) lnenjoy | (4) lndevelop | |
---|---|---|---|---|
lnnetwork | 0.046*** (0.009) | 0.039*** (0.008) | 0.071*** (0.008) | 0.041*** (0.010) |
lnincome | 0.062*** (0.012) | 0.065*** (0.013) | 0.084*** (0.010) | 0.052*** (0.016) |
lnfinanceasset | 0.008*** (0.002) | 0.013*** (0.002) | 0.016*** (0.002) | −0.004 (0.003) |
Size | 0.106*** (0.006) | 0.079*** (0.005) | 0.140*** (0.013) | 0.205*** (0.013) |
Gender | 0.010 (0.032) | 0.026 (0.025) | −0.002 (0.040) | −0.108*** (0.038) |
Age | −0.010* (0.004) | −0.007 (0.006) | −0.004 (0.006) | −0.003 (0.010) |
Edu | 0.143*** (0.011) | 0.142*** (0.013) | 0.171*** (0.010) | 0.138*** (0.014) |
Marriage | 0.223*** (0.054) | 0.148** (0.058) | 0.254*** (0.057) | 0.536*** (0.109) |
Health | −0.009 (0.008) | 0.025** (0.010) | 0.010 (0.007) | −0.143*** (0.019) |
Work status | −0.097*** (0.020) | −0.178*** (0.022) | 0.051 (0.034) | −0.031 (0.043) |
Provincial fixed effect | YES | YES | YES | YES |
Constant term | 9.585 (0.186) | 9.227 (0.195) | 6.825 (0.196) | 6.852 (0.376) |
adj R-squared | 0.303 | 0.309 | 0.328 | 0.155 |
Sample size | 5298 | 5280 | 5302 | 5242 |
Note: The values in parentheses are robust standard errors. *, **, ***indicate significant levels at the 10%, 5%, and 1% significance levels, respectively, and () is the t-statistic value.
The coefficient of influence of social network on living consumption is 0.0399, which is smaller than that of other kinds of consumption. This shows that the influence of social network on living consumption is weaker. The living consumption consists of food consumption, clothing consumption and residential consumption. It is the basic consumption to meet the necessities of physiological needs, so the impact of social networks is relatively small. With the increase of residents’ income and living standards, the proportion of living consumption is gradually decreasing, while the proportion of enjoying and developing consumption is increasing, which indicates that the consumption structure of residents is in the process of adjustment and optimization and upgrading.
The elasticity coefficient of social network to enjoyable consumption is 0.0712, which is significantly higher than that of other types of consumption. This shows that social network has the greatest effect on the promotion of enjoyable consumption. With the development of economy and society and the improvement of people’s living standards, the residents are pursuing higher level of consumption, such as comfort and enjoyment, so they are more inclined to enjoy consumption, which mainly includes household equipment and daily necessities expenditure and transportation and communication expenditure. On the one hand, the larger the scale of social network and the more frequent the interpersonal communication, the more traffic and communication expenditure in enjoyable consumption. On the other hand, the circulation and dissemination of consumption information in social network will greatly affect the consumption preferences of family members, which can be reflected through demonstration effect and comparison effect in the process of interpersonal communication. The consumption behavior of the members of social network can be heard and stimulate consumption, especially for the visible consumption such as household equipment and daily necessities expenditure in enjoyable consumption, which can better show the consumption ability of the members, and then stimulate the consumption desire of the family.
The influence coefficient of social network on developmental consumption is 0.0410, which is only next to enjoyable consumption. This indicates that social network plays a significant role in promoting developmental consumption. Human capital is the core factor to construct the family social network. In order to satisfy their own better development and cultivate the family human capital, the family’s consumption demand for health and education will be greater. The development-oriented consumption includes medical and health care consumption and cultural and educational entertainment consumption, so the larger the scale of family social networks, the greater the demand for developmental consumption.
To sum up, social network can improve the overall consumption level of families, especially for the enjoyment consumption and development consumption. Therefore, social network can help release the consumption potential of residents and promote the adjustment and upgrading of residents’ consumption structure.
From the perspective of the impact of family characteristics on consumption, the influence coefficients of household income on total consumption and various types of consumption are 0.0628, 0.0653, 0.0840 and 0.0524, respectively, which are significant at the level of 1% and larger than that of social network, indicating that household income is the main factor affecting current consumption. Family financial assets have a positive effect on residents’ consumption in general, but the influence coefficient is smaller than that of social network and family’s total annual income. This shows that the more family financial assets are, the more wealth they have, the more consumption expenditure they will have. But families will plan the consumption level of each period according to their own wealth level, so the level of family financial assets has a positive effect on current consumption. The impact factor on development consumption is negative but not significant, which indicates that household financial assets may bring crowding-out effect as well as wealth effect. Families allocate more assets to financial assets, thus reducing investment in development consumption. The influence coefficient of household population size on household consumption is significantly positive, that is, the more the household population, the more the household consumption expenditure.
From the perspective of the impact of personal characteristics on household consumption, the gender, age and health status of the head of household have no significant impact on consumption, because the level of household consumption is determined by all family members, and these indicators of the head of household cannot represent the situation of the whole family, so the impact of household consumption is not significant, while the head of household’s educational background and marital status have significant impact on various consumption coefficients. In general, the higher the educational level of the head of the household, the higher the educational level of the family members, the stronger the overall strength of the family and the higher the consumption level.
In this paper, according to the practice of previous literature and considering the availability of data, we choose the human gift expenditure as the proxy variable of social network. On the one hand, there may be consumption expenditure in the form of festival gifts. To a certain extent, social network needs consumption expenditure to maintain and expand. On the other hand, human gift expenditure and consumption expenditure are also affected by many variables, such as family background, consumption habits, personality characteristics and so on. Therefore, in the process of the regression of social network and consumption expenditure, which takes the human relationship gift expenditure as the proxy variable, there are problems of reverse causality and missing variables, which will lead to serious endogenous problems.
Referring to the existing literature, we generally consider the social characteristics and data availability to select the appropriate tool variables, such as Lin Jianhao (2018) [
(1) lntotal | (2) lnsurvive | (3) lnenjoy | (4) lndevelop | |
---|---|---|---|---|
lnnetwork | 0.046*** (0.009) | 0.039*** (0.008) | 0.071*** (0.008) | 0.041*** (0.010) |
lnincome | 0.062*** (0.012) | 0.065*** (0.013) | 0.084*** (0.010) | 0.052*** (0.016) |
lnfinanceasset | 0.008*** (0.002) | 0.013*** (0.002) | 0.016*** (0.002) | −0.004 (0.003) |
Size | 0.106*** (0.006) | 0.079*** (0.005) | 0.140*** (0.013) | 0.205*** (0.013) |
Gender | 0.010 (0.032) | 0.026 (0.025) | −0.002 (0.040) | −0.108*** (0.038) |
Age | −0.010* (0.004) | −0.007 (0.006) | −0.004 (0.006) | −0.003 (0.010) |
Edu | 0.143*** (0.011) | 0.142*** (0.013) | 0.171*** (0.010) | 0.138*** (0.014) |
Marriage | 0.223*** (0.054) | 0.148** (0.058) | 0.254*** (0.057) | 0.536*** (0.109) |
Health | −0.009 (0.008) | 0.025** (0.010) | 0.010 (0.007) | −0.143*** (0.019) |
Work status | −0.097*** (0.020) | −0.178*** (0.022) | 0.051 (0.034) | −0.031 (0.043) |
Provincial fixed effect | YES | YES | YES | YES |
Constant term | 9.585 (0.186) | 9.227 (0.195) | 6.825 (0.196) | 6.852 (0.376) |
adj R-squared | 0.303 | 0.309 | 0.328 | 0.155 |
One-stage F value | 311.547 | 350.136 | 321.217 | 344.39 |
Sample size | 5298 | 5280 | 5302 | 5242 |
Note: The values in parentheses are robust standard errors. *, **, ***indicate significant levels at the 10%, 5%, and 1% significance levels, respectively, and () is the t-statistic value.
From the results of two-stage regression, we can see that the influence coefficients of using the per capita human gift expenditure of community households as the tool variable of social network on the total household consumption, survival consumption, enjoyment consumption and development consumption are 0.0974, 0.0904, 0.1254 and 0.0965 respectively, and are significant at the level of 1%. The influence on survival consumption is the smallest, followed by that on development consumption. Enjoyment consumption has the greatest impact. The results of 2SLS regression are generally consistent with those of OLS regression, but the values of 2SLS regression coefficients are higher than those of OLS regression coefficients, which indicates that OLS may underestimate the positive effect of social network on household consumption. Staiger & Stock (1997) [
From the above analysis, we can see that the family social network has a significant positive impact on residents’ consumption, especially on the promotion of enjoyable consumption. This part will discuss and analyze the mechanism of the impact of social network on consumption.
Members of social networks can understand other members’ information through interaction, and it is easy to find reference groups through social networks. Yu Hu (2013) [
This paper will further test the impact mechanism of social network on household consumption through empirical analysis, using intermediary effect model and introducing intermediary variables to test the risk sharing mechanism and demonstration flaunting mechanism of social network respectively. In the risk-sharing mechanism of social networks, social networks play a more direct role through informal financial channels such as lending between relatives and friends and private lending. Therefore, the sum of relatives and friends’ borrowing and private borrowing is chosen as the intermediary variable. In the demonstration mechanism of social network, social status is the main psychosocial factor that social network affects consumption. Therefore, the judgment of residents’ social status from low to high in the questionnaire survey is chosen as the intermediary variable. From the previous empirical part, we can see that social network has the most significant impact on enjoyable consumption, so the intermediary effect of social network on enjoyable consumption is more obvious, which is conducive to the analysis of the impact mechanism of social network. Therefore, this part mainly investigates the impact mechanism of social network on residents’ enjoyable consumption.
This paper uses the hierarchical regression analysis method proposed by Wen Zhonglin and Ye Baojuan (2014) [
Firstly, the regression model of independent variable social network to dependent variable enjoyable consumption is established. The model is as follows:
ln enjoy i = β 0 + β 1 ln network i + β 2 χ i + μ i (5)
Secondly, the regression models of independent variable social network to intermediary variable informal finance and social status are established respectively. The models are as follows:
Variable identifier | Variable description | mean value | standard deviation | minimum value | Maximum value |
---|---|---|---|---|---|
lninfina | The logarithm of total loan amount between relatives and friends and private loans | 1.070 | 3.171 | 0 | 14.508 |
Social status | Very Low = 1, very low = 2, generally = 3, very high = 4, very high = 5 | 2.734 | 1.043 | 1 | 5 |
Data source: China family panel studies from 2016.
ln infina i = α 0 + α 1 ln network i + α 2 χ i + μ i (6)
status i = α 0 + α 1 ln network i + α 2 χ i + μ i (7)
Finally, the regression model of independent variable social network, intermediary variable informal finance and social status on dependent variable enjoyable consumption is established. The model is as follows:
ln enjoy i = γ 0 + γ 1 ln network i + γ 2 ln infina i + γ 3 χ i + μ i (8)
ln enjoy i = γ 0 + γ 1 ln network i + γ 2 status i + γ 3 χ i + μ i (9)
Among them, ln enjoy i represents the logarithm of per capita enjoyable consumption expenditure of household I in the whole year, which is the dependent variable of the test of intermediary effect. ln network i denotes the social network of family i, and uses the logarithm of per capita personal gift expenditure as the proxy variable, which is the independent variable of the mediation effect test. ln infina i denotes the logarithm of the total amount of family I’s relatives, friends and private loans, and A denotes the social status of family I’s head of household, both of which are mediating variables. status i is a control variable including personal and family characteristics, and χ i is a stochastic perturbation term.
The coefficient β 1 in Equation (5) is the total effect of the independent variable on the dependent variable. The coefficient α 1 in Equations (6) and (7) is the effect of the independent variable on the intermediate variable; The coefficient γ 1 in Equations (8) and (9) is the direct effect of the independent variable on the dependent variable after controlling the influence of the intermediate variable. The coefficient γ 2 is the effect of the intermediate variable on the dependent variable after controlling the influence of the independent variable.
The first step in the test of the mediating effect is to test whether the coefficient A in Equation (5) is significant, and it is significant that the second step is continued. If not, the test is stopped. In the second step, if the coefficient α 1 in Equation (6) and Equation (7) is significant, the next step is tested. The third step is that if the coefficients γ 1 and γ 2 of Equations (8) and (9) are significantly positive, and the coefficient γ 1 is smaller than the coefficient β 1 of Equation (5), there is a partial mediating effect. If the coefficient γ 1 is not significant and the coefficient γ 2 is significant, It indicates that there is a complete mediating effect. The schematic diagram of the mediation effect is shown in
In this paper, Stata 14 software is used to regression the above models. The results of mediating effect regression are shown in
(1) lnenjoy | (2) lninfina | (3) lnenjoy | (4) lnenjoy | (5) lnenjoy | |
---|---|---|---|---|---|
lnnetwork | 0.071*** (0.008) | 0.033* (0.0198) | 0.019** (0.008) | 0.070*** (0.008) | 0.071*** (0.008) |
lninfina | - | - | - | 0.010** (0.004) | - |
Social status | - | - | - | - | 0.043*** (0.013) |
lnincome | 0.084*** (0.010) | −0.022 (0.027) | 0.018** (0.009) | 0.084*** (0.010) | 0.083*** (0.010) |
lnfinanceasset | 0.017*** (0.002) | 0.095*** (0.013) | 0.004 (0.004) | 0.017*** (0.002) | 0.016*** (0.002) |
Size | 0.141*** (0.013) | 0.108*** (0.029) | 0.022** (0.008) | 0.139*** (0.013) | 0.140*** (0.013) |
Gender | −0.007 (0.041) | 0.165** (0.084) | −0.109*** (0.029) | −0.003 (0.040) | −0.001 (0.040) |
Age | −0.005 (0.006) | −0.006 (0.020) | −0.006 (0.007) | −0.004 (0.006) | −0.004 (0.006) |
Edu | 0.172*** (0.009) | 0.106*** (0.035) | 0.033** (0.014) | 0.172*** (0.009) | 0.171*** (0.009) |
Marriage | 0.254*** (0.056) | 0.154 (0.205) | 0.028 (0.066) | 0.252*** (0.056) | 0.254*** (0.057) |
Health | 0.015** (0.007) | 0.132*** (0.032) | 0.102*** (0.010) | 0.012 (0.007) | 0.010 (0.007) |
Work status | 0.060* (0.035) | 0.107 (0.127) | 0.135*** (0.035) | 0.050 (0.034) | 0.051 (0.034) |
Provincial fixed effect | YES | YES | YES | YES | YES |
Constant term | 6.891*** (0.185) | 2.899*** (0.544) | 1.409*** (0.142) | 6.795*** (0.200) | 6.825*** (0.196) |
adj R-squared | 0.328 | 0.057 | 0.079 | 0.329 | 0.328 |
Sample size | 5314 | 5374 | 5374 | 5302 | 5302 |
Note: The values in parentheses are robust standard errors. *, **, ***indicate significant levels at the 10%, 5%, and 1% significance levels, respectively, and () is the t-statistic value.
From the regression results in the
Similarly, after introducing social status variables, social status has a positive impact on enjoyable consumption at the 1% significant level. At this time, the impact of social network on enjoyable consumption is still significant at the 1% statistical level, and the coefficient is 0.0711, which is smaller than model 1. Therefore, social status plays a part of intermediary effect in the process of social network influencing residents’ consumption. In summary, the social network can promote the consumption-type consumption through the non-formal financial channels of relatives and friends, and the social and psychological factors for the pursuit of social status. The theoretical mechanism of the impact of social networks on household consumption summarized in the previous section has been verified.
There are great differences in the degree of economic and financial development and household consumption concepts in the eastern, central and western regions of China. According to the provinces where households are located, this paper divides the samples into three regions: eastern, central and western regions, and studies the heterogeneity of the influence of social networks in different regions on consumption. Families with different income levels will have different quality and utility of social network, which will also have different impact on consumption. Therefore, this paper also divides the sample into high and low income families based on the average of the family’s total annual income. Through empirical analysis of the impact of family social network on consumption in different regions and different levels of income, the robustness test is carried out.
From the regression results, we can see that in the eastern, central and western families, the impact of social network on consumption is significantly positive, and the impact on enjoyable consumption is more obvious, which is consistent with the previous empirical part of this paper. The impact of social network on consumption in the eastern region is weaker than that in the central and western regions. This is because the eastern region has a relatively high degree of marketization, mature financial system and less dependence on social relations than the families in the central and western regions, which makes the impact of the eastern family social network on consumption less significant than that of the central and western families (
Total consumption | Survival consumption | Enjoyable consumption | Developmental consumption | ||
---|---|---|---|---|---|
Eastern Region | Social network | 0.037*** (0.004) | 0.031*** (0.004) | 0.055*** (0.006) | 0.047*** (0.008) |
Control variables | YES | YES | YES | YES | |
Constant term | 9.217*** (0.158) | 8.877*** (0.157) | 6.548*** (0.233) | 6.485*** (0.287) | |
adj R-squared | 0.258 | 0.235 | 0.310 | 0.159 | |
Sample size | 2741 | 2737 | 2737 | 2714 | |
Central region | Social network | 0.076*** (0.008) | 0.063*** (0.008) | 0.122*** (0.013) | 0.060*** (0.014) |
Control variables | YES | YES | YES | YES | |
Constant term | 9.101*** (0.240) | 8.577*** (0.249) | 6.411*** (0.363) | 6.562*** (0.417) | |
adj R-squared | 0.259 | 0.261 | 0.285 | 0.128 | |
Sample size | 1492 | 1486 | 1498 | 1481 | |
Western Region | Social network | 0.054*** (0.010) | 0.048*** (0.010) | 0.088*** (0.014) | 0.003*** (0.017) |
Control variables | YES | YES | YES | YES | |
Constant term | 8.799*** (0.275) | 8.172*** (0.280) | 6.076*** (0.372) | 6.177*** (0.463) | |
adj R-squared | 0.227 | 0.222 | 0.286 | 0.112 | |
Sample size | 1061 | 1054 | 1065 | 1046 |
Note: The values in parentheses are robust standard errors. *, **, ***indicate significant levels at the 10%, 5%, and 1% significance levels, respectively, and () is the t-statistic value. Family and personal characteristic variables were controlled in the regression, and were omitted here due to space limitations.
Total consumption | Survival consumption | Enjoyable consumption | Developmental consumption | ||
---|---|---|---|---|---|
low-income family | Social network | 0.046*** (0.005) | 0.039*** (0.006) | 0.069*** (0.008) | 0.037*** (0.010) |
Control variables | YES | YES | YES | YES | |
Constant term | 10.422*** (0.199) | 10.125*** (0.209) | 8.242*** (0.271) | 7.016*** (0.361) | |
adj R-squared | 0.179 | 0.159 | 0.247 | 0.118 | |
Sample size | 1796 | 1790 | 1807 | 1786 | |
High-income family | Social network | 0.018*** (0.004) | 0.009*** (0.004) | 0.043*** (0.006) | 0.035*** (0.008) |
Control variables | YES | YES | YES | YES | |
Constant term | 4.272*** (0.248) | 4.369*** (0.252) | -0.704* (0.403) | 3.235*** (0.518) | |
adj R-squared | 0.258 | 0.215 | 0.284 | 0.129 | |
Sample size | 3502 | 3490 | 3495 | 3456 |
Note: The values in parentheses are robust standard errors. *, **, ***indicate significant levels at the 10%, 5%, and 1% significance levels, respectively, and () is the t-statistic value. Family and personal characteristic variables were controlled in the regression, and were omitted here due to space limitations.
It can be seen from the regression results that whether it is a high-income family or a low-income family, its social network has a significant role in promoting various types of consumption and has the greatest impact on enjoyment consumption. By comparing the influence coefficients of low-income and high-income families’ social networks on consumption, we can see that the influence coefficients of low-income families are generally larger than those of high-income families at each level of consumption, which indicates that social networks are more conducive to stimulating the consumption of low-income groups and play a greater role in low-income groups. On the one hand, low-income families face more liquidity constraints and less ability to resist risks, so they rely more on social networks to obtain support and help. On the other hand, social networks are more likely to play a demonstration role in low-income families to stimulate the consumption of low-income families, while high-income families are more likely to obtain information and show their social status through social networks, so the role of social networks in promoting consumption can be more reflected in low-income families.
Supported by micro-household data, this paper examines the impact of family social network on residents’ consumption level and consumption structure, solves the endogenous problem of social network by using instrumental variables, verifies the risk-sharing mechanism of social network, and further explores the impact of family social network in different regions and different levels of education on residents’ consumption level and consumption structure. Through the empirical analysis, we get the following conclusions: 1) Family social network has a significant positive impact on residents’ consumption level, among which the promotion of enjoyable consumption level is the most obvious. 2) On the one hand, social networks can provide financial support and informal financial channels for families to reduce uncertainty and risk sensitivity of family income, ease liquidity constraints and smooth consumption. On the other hand, it can also drive the consumption of residents through “demonstration effect” and “comparative effect”, and affect the consumption level and consumption structure of residents. 3) Different types of families have different effects on consumption. Families in the central and western regions have more promotive effects on consumption than those in the eastern regions. Social networks are more conducive to stimulating the consumption of low-income groups, and play a greater role in low-income groups.
As a kind of social capital, social network can play an important role in providing social support to families and promoting consumption. Therefore, we should pay attention to cultivating and utilizing the social network of families, effectively exert the risk-sharing mechanism of family social network, so that social network can be transformed into effective social capital and forms an informal system to make up for the lack of formal system development.
The author declares no conflicts of interest regarding the publication of this paper.
Kang, X. (2019) The Impact of Family Social Network on Household Consumption. Modern Economy, 10, 679-697. https://doi.org/10.4236/me.2019.103046