Received 10 March 2016; accepted 10 April 2016; published 13 April 2016
Since the 1990s, China’s capital market has developed maturely, accompanied by financial globalization and the emergence of a wave of financial product innovation, as well as an increase in per capita income; people are in dire need to administrate finance. Putting money in the bank is no longer the only option of a majority of the people to deal with their money. Nowadays, the allocation of residents’ financial asset has increasingly diversified, with the stocks, funds, bonds, insurance and financial derivatives and other financial products gradually accepted by the people. In this context, it is of great significance to study the factors that give guidance for people to choose various financial assets.
Religion, a special ideology and an important part of culture, affects people’s cognitive competence, thinking concepts, values, and thereby influences people’s economic behavior. Foreign researchers have conducted broad researches on the relationship between religion and economic development, but most of them are based on the macroeconomic level. From the micro perspective, Luc Renneboog and Christophe Spanenjers implemented the empirical analysis on the relationship between religious beliefs and residents’ financial assets selection. In general, fewer researches are conducted from micro aspect, especially on the relationship between religious beliefs and economic development in our country. Kanglin Zeng has explored the development of culture, religion and financial undertakings; Yujiao Ji has demonstrated the relationship between religious beliefs and the economic development. Generally speaking, domestic researches in this field are still in its infancy by words explanation; especially there are even no empirical researches from the micro perspective concerning religious beliefs and financial assets selection. It is believed that lacking authoritative data results in the above result, especially data on the residents’ religious beliefs.
CGSS2010 serves as the more authoritative social survey data, a collection not only of religious beliefs of the interviewees, but also of the risk assets of respondents stocks, funds, bonds, etc., and the detailed data such as their real estate holdings, insurance participation, education background, and demographic characteristics. This paper, taking advantages of CGSS2010 data, selects “Do you believe in any religion” as an explanatory variable, “Would you like to buy the stock”, and “Do you hold risky assets” as the dependent variable, explores the impact of religious belief on residents’ financial assets selection through establishing Probit model.
The empirical results of this paper show that, those who believe in religion are more willing to hold risk assets than those who do not believe in religion, including the stocks and funds, bonds, etc.
In this paper, the following chapters will be arranged as follows: in Section 2, the related literatures will be reviewed; in Section 3, the data and the variables are to be explored; in Section 4, the possible endogenous will be modeled; in Section 5, empirical analysis will be conducted; Section 6 concludes the paper.
2. Literature Review
2.1. Relevant Literatures of Residents’ Financial Assets Selection
Branson has systematically discussed the financial assets selections, holding that the assets selection of the economic agent is mainly based on the expected rate of return, the level of the risk and liquidity strength. Subsequently, the researchers found that resident’s assets selection is not only influenced by the microeconomic factors as wages, house property and demographics. Heaton and Lucas found a high correlation between salary and stock holdings, high-income groups hold greater proportion of stock holdings  . Daimin Shi and Yan Song researched the microscopic data, finding that in family assets, the proportion of risk assets will increase with the increase of wealth  . Yao and Zhang (2005) estimated the relationship between US residents’ equity investments and real estate variables through PSID data, and found that in event of certain income and net worth, higher property values decrease the probability of people’s participation in the stock market  . Campbel found that the factors such as age, gender, education, health status can affect the financial asset allocation of the residents in accordance with econometric analysis  . Shum and Faig noted that, in addition to income, gender, age, education level, the marital status and background risk would also impose an impact on the allocation of the residents’ financial assets  .
With in-depth research, more and more scholars tried to investigate the behavior of financial assets selections of residents from the perspective of behavioral finance. Stango and Zinman believed that overconfidence of the residents may lead to centralized holding of the risk assets by analyzing the impact of cognitive biases on the investment decisions of the residents  . Hong studied the impact of social interactions and peer effects on people’s involvement in stock investment, and found that people who regularly participated in social activities can access to broader capital market information, with higher stock holdings  . Georgarakos and Pasini integrated the social intercourse and credibility into the residents’ investment decisions, and there is a significantly positive correlation between the credibility and stock holdings that the higher the credibility of investors is, the more loosening the expectations for risk is, and therefore there are a higher proportion of risk assets  . Domestically, by analyzing the data of “Investor Behavior Survey” from Beijing Investment Advisory Centre of Aldo, Cong Wang (2006) concluded the impact of consumer expectations, credibility, social interaction, investor sentiment and other factors on residents’ involvement in the stock market  . Tao Li studied the 2004 survey data on the residents of Guangdong Province and found that social interaction and credibility would promote resident participation in the stock market, but also the social interaction can obviously impose positive effects on those undereducated participants in the stock market  . Dezhu Ye and Liyan Zhou conducted empirical analysis on the data from Chinese family financial investigations, and found that there is a significant negative correlation between the sense of happiness and the risk assets of the residents. The more happy the people are, the more willing the people would rather maintain their current living conditions, the more risk-averse, the higher degree of concern for the future, and the less corresponding short-term behavior is  .
2.2. Literatures Relevant to Religious Economics
Blum and Dudley found the fact by analyzing the relationship between the religious beliefs and the early economic growth in Central Europe, that the cultural background of Catholic is more conducive to a country’s economic growth  . Guiso stated that different religious beliefs can lead to such differences in economic attitudes as savings, credibility, and social responsibility and so on, which in turn lead to different asset selection tendency of individuals  . Based on the data of DNB residents’ survey from 1995 to 2008, Luc Renneboog and Christophe Spaenjers analyzed the influence of religious beliefs on New Zealand residents’ selection of their financial assets. Studies have shown that differences in the credibility, risk attitude, individual responsibility, social capital, the extent of thrifty will be reflected in different faiths of people  . For example, Catholics and Protestants have a more long-term plan for life; they generally tend to save money, with weak willingness to buy risky assets while the evangelical is more inclined to take time to manage finance and purchase various financial assets to manage their wealth.
China started late to explore this area and the exploration was limited to words explanation and thus there are fewer relevant empirical researches. Kanglin Zeng pointed out that the development of the financial industry in the region depends on the liquidity of the asset and financial consciousness. People’s financial awareness not only is influenced by the change of social system, but also stimulated and constrained by such superstructures as cultural traditions, and religious beliefs  . Fan Luo analyzed the impact of religion on the concept of wealth. In his view, different religions and the variance of public understanding and acceptance of doctrine can cause people to form different views on wealth: firstly, money and wealth are considered as something dirty that wealth because they think wealth is the temptation of the devil, which will make people become greedy, narrow- minded and utilitarian; secondly, some people remain neutral attitude towards money and wealth, neither exclude the wealth brought by their labor, nor too much in pursuit of utilitarian; thirdly, the wealth is regarded as the gift of god, as to realize the wealth is to follow the God’s will. Fan Luo further pointed out that more people in today’s society have the third value of fortune  .
From the discussion of the above documents, a hypothesis can be drawn: people of faith are more likely to buy stocks and risky assets.
3. Description of Data and Variables
3.1. Introduction of Samples and Data
The data applied in the empirical research of this paper is from the 2010 China General Social Survey Projects, which is the first national, comprehensive, continuous large-scale social survey projects in China, conducting investigation on the individual of the 125 counties (districts), 500 street ( countryside ,township), 1000 (village) committees, 12,000 families. The results of the questionnaire collected and sorted include not only the religious situation of the interviewees, but also a detailed record of such data as risky assets holdings including respondents stocks, funds, bonds, real estate holding, insurance participation, education background, demographics and so on, as well as the subjective data such as their social participation, credibility, degree of happiness. The survey provides a good sample source for the empirical study. According to the needs of research, 9662 effective samples are screened out with the missing values excluded with the required data for all empirical analysis in the paper from CGSS2010 database.
3.2.1. Explained Variables
The explanatory variables selected in the paper are: the stock holdings, holdings of risky assets. According to the answers of the interviewed for the question “Are you currently engaged in the following investment activities (stocks, funds, bonds, financial products, futures, warrants, gold, foreign investment)”, if one is to invest stocks, the stock holdings value will be of 1, otherwise 0; if one is in possession of one or more such risky assets, the level of risk assets held value is for 1, and otherwise 0.
3.2.2. Explanatory Variables
The explanatory variable selected in this paper is religion. According to the answer of the surveyed to the question that “what is your religion”, no matter what their religious beliefs are, as long as they have their religions, the variable of religious values can be assigned to 1, otherwise zero.
3.2.3. Control Variable
Based on the above literature review, the paper integrates the factors affecting asset selection of the residents as a control variable as much as possible into the regression model, specifically:
1) Income. According to the answer of the respondents to the question that “how much is your personal gross income last year”, the income will include their professional income and other income, including gross income data as a reflection of their income.
2) Housing ownership. Real estate, as a large asset may have crowding effect on the investments in stocks and risky assets. According to the answer of the respondents to the question that “Do you have your own house”, if they have their own houses, the variable of the house holdings will be assigned to 1, otherwise 0.
3) Insurance participation. According to the answers of the respondents to the question that “Are you currently involved in the following social security programs (health insurance, pension insurance, etc.)”, if one is involved in at least one of them, the insurance participation will be assigned to 1, otherwise 0.
4) Education background. In the CGSS2010 questionnaire, according to China’s existing education, the education background is divided into 14 levels from the uneducated to the graduate student. What’s more, the value of the level of the educated is corresponding to be assigned to an integer ranging within 1 - 14.
5) Social interaction. Related literature reveals that the social interactions can promote people’s participation in the stock market. In this paper, the frequency of the respondents to participate in social activities in the idle time is regarded as the reflection of the variable of social interaction. Based on the answer to the question that “Over the past year, are you often involved in social activities in your spare time”, the value of variable of social interaction is respectively assigned as 1, 2, 3, 4, 5, and the corresponding results of the answers are: never, rarely, sometimes, often, and always.
6) Credibility. According to the answer of the respondents to the question that “In general, do you agree that in this society, most people are trustworthy”, the level of credibility can be respectively assigned to 1, 2, 3, 4, 5, with the corresponding answers are: totally disagree, relatively disagree, whatever agree or disagree, relatively agree, totally agree.
7) Demographic variables. Marital status, the married is assigned to 1, otherwise 0; age, calculate the respondents’ age according to the year of birth; gender, men is assigned to 1, women is assigned to 0; ethnic, Han is assigned to 1, other ethics are assigned to 0; political affiliation, communists are assigned to 1, others are assigned to 0.
8) Degree of happiness. According to the answers of the respondents to the question that “In general, do you think that you feel happy about your life”, whose degree of happiness is assigned to 1, 2, 3, 4, 5, and the corresponding results of the answers are: very unhappy, unhappier, between happiness and unhappiness, happier, perfectly happy.
9) Health conditions. CGSS2010 divided the health status of the respondents into five grades from very unhealthy, unhealthier, generally healthy, healthier, very healthy, according to which the health conditions will be in turn assigned to 1, 2, 3, 4, 5.
10) Risk preference. According to the answers of the respondents to the question that “if I have the extra money, my first consideration is to keep it up”, the risk preference of which in turn is assigned to 1, 2, 3, 4, 5, with the corresponding results of the answers are: totally disagree, relatively disagree, whatever agree or disagree, relatively agree, totally agree.
As used herein, the variables are as shown in Table 1.
4. Model Setup
As the explained variable in this paper is a binary dummy variable, the regression analysis is conducted on the relationship between religion and residents asset selection by means of Probit model of determination on the nature. Probit model is set as follows:
And, μi~N(0,σ2), Y1 = 1 at i = 1, represents the resident holdings, Y1 = 0 indicates that the resident does not hold any stocks; Y2 = 1 at i = 2 indicates that the resident holds the risky assets, Y1 = 0 indicates that the resident does not hold risky assets. Religion serves as the religion variable, X represents the control variable used in this paper.
In addition, the regression analysis between the religious beliefs and asset selection may be plagued by endogenous problems. Based on the characteristics of the applied data and the variables, the endogenous problems arising from the applied model are analyzed from the perspectives of omitted variables, measurement error and reverse interaction.
4.1. Omitted Variable
If there is no variable related to both the residents’ asset selection and religious beliefs in the Probit model, then the errors of the estimation results of the model may occur. From the above literature review, the variants related
Table 1. Model specific description of each variable.
to the residents’ asset selection mainly include income, age, health, sex, education background, in the paper, these factors as the control variants are incorporated to regression model as much as possible to reduce the impact of omitted variables.
4.2. Measurement Error
Measurement error will cause bias errors to the system. However, fortunately, the variables used herein are basically dummy variables, which can better prevent data abnormalities.
4.3. Adverse Interaction
If the residents’ asset selections in turn affect people’s selections on religion, then the endogenous problems will occur. Reference to the existing literature on the solution to the similar problems, in the paper, the instrumental variable method is applied to get rid of the potential impact of asset selection on religious beliefs. According to the CGSS2010 questionnaire, the answers to the three problems that “What does your father’s religion?”, “What does your mother’s religion?”, “What is the religious environment in which you are brought up?” are served as the instrumental variables. The reasons are as follows: Intuitively, parents’ religious beliefs and the religious environment have imposed larger impact on one’s future religion, which is not significantly related to his or her asset selection. Therefore, the three instrumental variables are theoretically appropriate. In the following empirical analysis, the selected instrumental variables are tested to verify if there is any endogenous problem in the model in this regard.
5. Empirical Analysis
5.1. Descriptive Statistics of Variable
As can be seen from Table 2, the people who hold stocks and risky assets respectively account for 5.8% and 8.1%, indicating that the residents in China is inactive to participate in the capital market, and thus people’s financial awareness has yet to be improved. 12.3 percent of respondents said they had their own religious beliefs.
Table 2. Descriptive statistics of variable.
The high percentage of people of religious beliefs may deviate from intuitive feelings, but this is precisely mapping out a larger pressure of people’s lives in today’s society, more and more people are turning to religion to get inner peace. In addition, the mean value of risk preference is 1.907, indicating that most people are risk averse, which also reflects low participation in the stock market and the capital market.
5.2. Multicollinearity Test
Many variables applied in the paper may contribute to the multicollinearity problems, leading to bias errors to the regression results. Therefore, it is necessary to examine the variance inflation factor of the independent variables.
The criteria of the statistical analysis to judge multicollinearity are: largest VIF value is greater than 10, and the average of VIF value is greater than 1. As can be seen from Table 3, the maximum value of VIF corresponding to each variable is 1.66, with the mean value at 1.19, which cannot meet this standard, thus multicollinearity between independent variables is weak.
5.3. Basic Regression Results
As noted above, if the asset selection of residents in turn affects people’s selections of faith, then the endogenous problems will occur in the model. Therefore, in this paper, the parents’ religious beliefs and the religion environment are selected to serve as the instrumental variances to conduct the IVProbit regression and the Wald exogeneity test method is applied to detect the effectiveness of selected instrumental variables. If the results of the Wald test support the endogenous assumptions, the results of IVProbit regression can then be accepted; otherwise the results of ordinary Probit regression should be accepted. In the following analysis, these three instrumental variables are sequentially named as f_religion, m_religion and e_religion: if the respondent’s father believes in religion, f_religion can be assigned at 1, otherwise 0; if the respondent mother believes in religion, m_religion can be assigned at 1, otherwise 0; if the respondents were raised in a religious environment, e_religion should be assigned at 1, otherwise 0 otherwise. Regression results of the model are as shown in Table 4.
As can be seen in the regression results of IVProbit in Table 4, it is estimated with the faith of farther and the mother as well as the environment as the instrumental variables, P value in the Wald exogeneity test is much larger than 0.1, indicating that statistically the religious belief should be the assumption for the explanatory variables. Therefore, this paper may accept the regression results of ordinary Probit, showing that religious beliefs and stock holdings are significantly positively correlated with holdings of risky assets. The results are consistent with the assumption proposed in the paper, that is, the people who believe in religion are more likely to buy stocks and risky assets.
In addition, as can be seen from the regression results from ordinary Probit, income has a significantly positive correlation with risk assets and stocks; insurance participation has a significantly positive correlation with stocks and risky assets holdings, suggesting that the mass who access to social security have relatively low expectation towards future risk, while they are more likely to participate in the capital markets; the level of education and social interaction also impose a significantly positive correlation with equities and risk assets holdings, which indicates that the people with higher degree and the more frequently participation in social activities have strong ability to acquire the information concerning their capital markets, while they are more inclined to purchase stocks, funds, bonds and the like. These results agree with the conventional literatures on asset selections from the basic level.
Religion as a significant component of culture will influence people’s value, view of wealth, economic behavior
Table 3. Variance inflation factor test on independent variables.
Table 4. Basic regression results of the model.
Note: 1. The last line of the table represents the P value of the exogenous Wald test; 2. the data in which the variables lie is the estimated coefficients, in parentheses are the Z statistic, as the space is limited, there are no marginal effects concerning variables; 3. ***, **, * respectively denote that the significant levels of statistics are 1%, 5% and 10%.
and also financial decision-making. Western developed countries have made in-depth discussion of this topic, but such kind of discussion in China is still lacking. On basis of CGSS2010 data, this paper attempts to explore this topic, and aims to understand the financial assets selection behavior of Chinese residents more completely.
On the basis of CGSS2010 data, this paper attempts to discuss the relationship between religious belief and residents’ financial assets selection by building Probit regression model. By controlling related variables as much as possible and fully considering the model’s endogenous influence, we discovered that religious belief and residents’ stock purchase and risky assets holding are remarkably and positively related. That is to say, people with religious belief are more inclined to stock purchase and risky assets holding.
 Heaton, J. and Lucas, D. (2000) Portfolio Choice in the Presence of Background Risk. The Economic Journal, 110, 1- 26. http://dx.doi.org/10.1111/1468-0297.00488
 Yao, R. and Zhang, H.H. (2005) Optimal Consumption and Portfolio Choices with Risky Housing and Borrowing Constraints. Review of Financial Studies, 18, 197-239. http://dx.doi.org/10.1093/rfs/hhh007
 Campbell, J.Y. (2006) Household Finance. Journal of Finance, 61, 1553-1604.
 Stango, V. and Zinman, J. (2008) Exponential Growth Bias and Household Finance. Journal of Finance, 6, 2807-2849. http://dx.doi.org/10.2139/ssrn.1081633
 Faig, M. and Shum, P.M. (2006) What Explains Household Stock Holdings? Journal of Banking & Finance, 30, 2579- 2597. http://dx.doi.org/10.1016/j.jbankfin.2005.11.006
 Hong, H., Kubik, J.D. and Stein, J.C. (2004) Social Interaction and Stock Market Participation. Journal of Finance, 59, 137-163. http://dx.doi.org/10.1111/j.1540-6261.2004.00629.x
 Georgarakos, D. and Pasini, G. (2009) Trust, Sociability and Stock Market Participation. http://ssrn.com/abstract=1397236
 Blum, U. and Dudley, L. (2001) Religion and Economic Growth: Was Weber Right? Journal of Evolutionary Economics, 11, 207-230. http://dx.doi.org/10.1007/PL00003862
 Guiso, L., Sapienza, P. and Zingales, L. (2003) People’s Opium? Religion and Economic Attitudes. Journal of Monetary Economics, 50, 225-282. http://dx.doi.org/10.1016/S0304-3932(02)00202-7