Study on the promotion effect of informationization on entrepreneurship: an empirical evidence from China

Family plays an active role in enterprise formation and development in China (Pistrui, Huang, Oksoy, Jing, & Welsch, 2010). Many businesses are family owned. Therefore, we explore the impact of informationization on entrepreneurship from the perspective of family entrepreneurship. The measure of family entrepreneurship and family characteristic variable is based on the “2011 China Household Finance Survey” conducted by Southwestern University of Finance and Economics. The main reason of choosing this data is that it contains a wealth of family entrepreneurial information, such as entrepreneurial choice, type, and income. It would provide a good sample for the research of this paper. In addition, the “2011 China Household Finance Survey” is used in some empirical literature (Chai, Chen, Huang, & Ye, 2018; Zhang, Xu, Zhou, Zhang, & Xie, 2014), indicating the research data is credible. The data covers 25 provinces and 8438 sample households. The survey mainly includes household demographics, assets and liabilities, insurance and security, spending, and income. Specifically, the “2011 China Household Finance Survey” surveys family entrepreneurial information in 2010.

In addition, the measure of provincial-level control variables is from China Statistics Yearbook in 2011. The 2011 China Statistics Yearbook actually reports the economic data in 2010.

To find the relationship between informationization and entrepreneurship, the paper matches informationization at the provincial level with the household entrepreneurial data of each province. In this way, the problem studied in our paper becomes the influence of regional informationization level on family entrepreneurship in this region. In the study of Islam (2015), he adopted a probit model to discuss the relationship between regional government spending and individual entrepreneurial choice. Since Islam (2015) also studied the influence of regional factors on micro-individual entrepreneurship, this paper refers to his research to construct the econometric model shown below:

where entrep_fp represents the entrepreneurial choice of family f in province p and informationization_p represents the informationization level of province p. To attenuate the endogenous effect from missing important explanatory variables, the model incorporates two control variables: the characteristic variable at family level (F_fp) and the characteristic variable at regional level (P_p). β is the core estimation coefficient of interest in this paper and is expected to be significantly positive.

The dependent variable is a dummy variable. If the ordinary least squares method is used to estimate Eq. (1), the result would be biased and deviating. Thus, the probit model is used instead. We can also use logit model to estimate Eq. (1), but there is no significant difference between the probit model and logit model. Therefore, this paper mainly adopts the probit model to study the effect of informationization on entrepreneurship and uses logit model for robustness test.

entrep is a dummy variable measuring whether a family is engaged in entrepreneurial activities. The “China Household Finance Survey” asks whether a family is engaged in industrial and commercial production and operation in 2010. For this question, a positive answer indicates that the family is entrepreneurial and entrep is assigned the value of 1; a negative answer indicates that the family is not entrepreneurial and entrep is 0.

F is a family characteristic variable that affects family entrepreneurship. Referring to existing research (Aldrich & Cliff, 2003; Paulson & Townsend, 2004; Yueh, 2009), we add political network (PN), social network (SN), number of members (NM), risk aversion (RA), and house ownership (HO) to F. Political network is a dummy variable. If a family has at least one member as civil servant, the political network variable is valued 1; if none of the family members are civil servants, the value of this variable is zero. Compared with families that have none of their members as civil servants, those with members as civil servants have a more sophisticated political network. The social network is the natural logarithm of the amount of donation from a family to the Wenchuan earthquake-stricken area in 2008. There are two reasons for choosing the amount of donation as a measure of family’s social network: For one thing, the amount of donation, to a certain extent, reflects the social status of the family, as the more it donates, the higher social status it enjoys and families with high social status usually have more extensive interpersonal networks. For the other, the Wenchuan earthquake is an exogenous shock, and this measure could be regarded as exogenous. The number of members is the number of family members. Risk aversion is divided into five levels. The higher the level, the more inclined for the family to avoid risks. House ownership is a dummy variable. If a family owns a house, this variable is valued 1; otherwise, it is valued zero.

According to existing research (Acs, Desai, & Hessels, 2008; Ghani, Kerr, & O’connell, 2014; Islam, 2015; Wennekers, Wennekers, Thurik, & Reynolds, 2005), we add unemployment rate (UR), population density (PD), educational level (EL), economic development level (EDL), minimum wage standard (MWS), and proportion of fiscal revenue to GDP (FR). The unemployment rate is the urban registered unemployment rate. The population density is the natural logarithm of the ratio of the total population to the land area at the end of the year. The educational level is the natural logarithm of the number of graduates with a college degree or above per 100,000 people. The economic development level is measured by the per capita GDP and is the natural logarithm from empirical research. The minimum wage standard is the natural logarithm of the minimum wage based on the lowest standard. The proportion of fiscal revenue is the proportion of fiscal revenue to regional GDP.

The Table 1 shows summary statistics of all variables based on variable measurement. The mean value of entrep is 0.1333, which indicates that 13.33% of families choose to start a business in China. Prieger et al. (2016) argues that the optimal entrepreneurship rate for middle-income countries is 25.1%. Therefore, China’s entrepreneurship rate still has a lot of room for improvement. Another variable we are concerned with is the provincial informationization level. In our research sample, the highest level of informationization is Shanghai, and the natural logarithm of the informationization development index is 4.5078; the lowest level of informationization is Gansu, and the natural logarithm of the informationization development index is 3.5885. The maximum value is 1.26 times the minimum value. The data shows that there is obvious inter-provincial difference in the level of informationization (Additional file 1).

The dependent variable (entrep) is a dummy variable, and thus, this study uses the probit model to study the influence of informationization on entrepreneurship. In the probit model, the estimated coefficient could only reflect in what direction the explanatory variable influenced the dependent variable and could not reflect the magnitude of such influence. Therefore, the average marginal effect of the explanatory variable is calculated to reflect the influence of the explanatory variable on the dependent variable. In addition, this study adopts cross-section data which is usually vulnerable to heteroscedasticity. For this reason, robust standard error is used in the model estimation to reduce the impact of heteroscedasticity on the estimation result.

Column 1 of Table 2 reports the benchmark regression results of the influence of informationization on entrepreneurship. The estimated coefficient of informationization is positive at the significance level of 1%, indicating that the higher the level of informatization, the higher the tendency of family entrepreneurship in the province. In terms of the magnitude of influence, when the informationization level increases by 1%, the probability of the family choosing to start a business increases by 22.21 percentage points. This indicates that the role of informationization in promoting entrepreneurship is not only statistically significant, but also economically significant. The basic regression results show informationization plays a significant role in promoting entrepreneurship, mainly because it can provide abundant opportunities and convenient conditions for entrepreneurship. According to the Report of China Informationization Development Assessment released by the China Center for Information Industry Development in 2015, 11 provinces have informationization development indexes higher than the national average, while the case is opposite for 20 provinces. In addition, this paper also calculates the entrepreneurship rate of each province according to the “China Household Finance Survey”. The calculation results show that 8 provinces have entrepreneurship rates higher than the national average and the case is opposite for 17 provinces. This indicates that the informationization level and the entrepreneurship rate remain low in most parts of China. Under this context, it is of practical significance to the promotion of “mass entrepreneurship” and economic development if the government further strengthens informationization construction.

Next, we briefly explain the influence of control variables on entrepreneurship based on benchmark regression results. The estimated coefficient of the political network is negative at the significance level of 1%, which indicates that families with political networks are 9.95 percentage points less likely to start a business than families without political networks. Necessity-push entrepreneurship is the main type of family entrepreneurship in China. With well-off conditions, families of civil servants are naturally less motivated to start a business. The social network has a positive estimated coefficient and passes the 1% significance test, indicating that an extensive social network plays a significant role in promoting entrepreneurship. This finding is consistent with the conclusions of Renzulli, Aldrich, and Moody (2000) and Yueh (2009). Entrepreneurship requires a range of resources such as information, capital, technology, and labor. People can get entrepreneurial resources they need through their social networks. Hence, residents with more extensive social networks are more likely to venture into entrepreneurial activities. The estimated coefficient of the number of members is positive at the significance level of 1%. Families with more members are more likely to start a business. This is mainly because more family members mean wider social networks. The estimated coefficient of risk aversion is significantly negative, indicating that the higher the risk aversion, the lower the probability that families would choose to start a business. Empirical studies by Cramer, Hartog, Jonker, and Praag (2002) and Hvide and Panos (2014) also support this conclusion. The estimated coefficient of house ownership is negative, but it does not pass the significance test, which means that house ownership is not significantly related to the probability of starting a business.

In terms of regional characteristic variables, the estimated coefficient of unemployment rate is significantly positive. In areas with high unemployment rates, starting a business is a more likely choice for families, mainly because unemployment might force people to engage in entrepreneurial activities for their livelihood. The estimated coefficient of population density does not pass the significance test. Ghani et al. (2014) believe that areas with high population densities have more concentrated knowledge, resources, talents, and creativity, which could, in turn, stimulate people’s enthusiasm for entrepreneurship. However, excessive concentration of population may generate congestion costs and thus increase the cost of entrepreneurship. Therefore, on the whole, population density does not have a significant impact on entrepreneurship. The estimated coefficient of educational level is significantly negative, indicating that families in areas with high educational attainment are less likely to start a business. An important reason is that people with higher educational degrees have more external choices and are more likely to get satisfactory jobs, and they are less motivated to engage in high-risk entrepreneurial activities (Van der Sluis, Van Praag, & Vijverberg, 2008). The economic development level of a region does not have a significant impact on the choice of family entrepreneurship. The estimated coefficient of minimum wage standard is negative at the significance level of 10%, indicating that this variable could inhibit entrepreneurship. Areas with high minimum wage standards also entail high opportunity cost and labor cost of entrepreneurial activities, so families are less likely to start a business. The estimated coefficient of fiscal revenue proportion is negative at the significant level of 1%, and families in areas with high fiscal revenue proportion tend not to venture into entrepreneurship. The conclusion is consistent with the finding of Djankov, Ganser, McLiesh, Ramalho, and Shleifer (2010). A high proportion of fiscal revenue suggests a heavy tax burden on enterprises, which increases the cost of entrepreneurship and thus undermines the motivation to start a business. Therefore, it is important to reduce the burden on enterprises when the Chinese government tries to encourage residents into entrepreneurship.

Benchmark regression results have proved the significant effect of informationization on entrepreneurship, yet this conclusion may still be affected by other factors, such as endogeneity, index metrics, econometric model, and research sample. In order to illustrate the robustness of the research conclusion, this paper also conducts a series of robustness tests.

Endogeneity may lead to biased and inconsistent estimates. As far as this study is concerned, the endogeneity mainly comes from missing important explanatory variables and the possible bidirectional causality between informationization and entrepreneurship. To solve the endogenous problem, we have to find an appropriate instrumental variable. With reference to the research of Kolko (2012), this paper uses altitude as an instrumental variable of informationization. The geographical environment of a region can affect the cost of local informationization construction. Altitude is an important geographical factor that affects informationization construction. Areas with higher altitudes need higher costs of informationization construction and have lower informationization levels. In addition, as an exogenous geographical factor, altitude is not affected by family entrepreneurship. Arguably, altitude is an effective instrumental variable of informationization. With regard to the measurement of altitude for each province, this paper considers the average altitude of major cities in each province to approximate the altitude of each province.

Based on the selected instrumental variable for informationization, this paper reestimates Eq. (1) using the IV probit model. The estimation results are shown in column 2 of Table 2. The estimated coefficient of informationization is positive and passes the 1% significance test. In addition, the Wald exogenous test passes the 5% significance test, which indicates that the test rejects the null hypothesis that the informationization variable is exogenous, suggesting that the regression using the IV probit model is effective in this paper. The regression results show that the research conclusion that informationization promotes entrepreneurship after considering the endogenous problem is robust.

The informationization evaluation results vary with evaluation indexes and methods. Therefore, in order to prove that the research conclusion does not vary with informationization measurement methods, this paper uses the informationization development indexes constructed by the National Bureau of Statistics to reevaluate the influence of informationization on entrepreneurship.

Column 3 of Table 2 reports the regression results of the informationization indexes remeasured. The regression results show that the estimated coefficient of informationization is positive at the significance level of 1%, which indicates that the conclusion that informationization greatly promotes entrepreneurship is still well grounded even if the evaluation indexes and methods are changed.

Although there is no essential difference between the estimation results based on the probit model and the logit model, we would also adopt the logit model to estimate the impact of informationization on entrepreneurship in order to show that the conclusion of the paper is not changed due to the change of the measurement model. Column 4 of Table 2 illustrates the regression results based on the logit model. The estimated coefficient of informationization is significantly positive. As we saw in columns 1 and 4, the estimated coefficients of informationization based on the probit model and the logit model are very similar. Therefore, we believe that the research conclusion that regional informationization has a promotion effect on entrepreneurship would not change whether using the probit model or the logit model.

There are two main surveys to investigate the entrepreneurship of Chinese families, namely China Household Finance Survey and China Family Panel Studies. In comparison, China Household Finance Survey includes more entrepreneurial information, such as entrepreneurial industries, entrepreneurial types, and number of employees employed. This information is conductive to study the heterogeneous influences of informationization on entrepreneurship and influence of informationization on entrepreneurial performance. Therefore, this paper eventually chooses the China Household Finance Survey to analyze the effect of informationization on entrepreneurship. However, China Household Finance Survey only supplies cross-section data in 2010, which may lead to inaccurate estimation results due to the missing time factor. Fortunately, China Family Panel Studies provide family entrepreneurship information for 2011, 2013, and 2015. Hence, we use this dataset for robustness test.

Due to the replacement of research sample, we need to redefine some control variables in family level, but the definition of control variables in the provincial level has not changed. The social network is the natural logarithm of the amount of family donations per year. Risk aversion is the dummy variable. If a family purchases a financial product, this variable is valued 0; otherwise, it is valued 1. Column 5 of Table 2 shows the regression results based on China Family Panel Studies. China Family Panel Studies do not provide any information regarding the political network of family. Therefore, column 5 does not include the political network variable. The estimated coefficient of informationization is significantly positive, which indicates that the research conclusion is still true when we expand the study sample and consider the time factor.

Table 3 reports the regression results of the influences of different informationization dimensions on entrepreneurship. Column 1 shows the influence of information infrastructure on entrepreneurship. Its estimated coefficient is positive at the significance level of 1%, indicating that the more developed the information infrastructure in a region, the more likely the families there would choose to start a business. Specifically, every 1% increase in the information infrastructure level in a region would lead to an increase in the probability of family entrepreneurship by 13.8 percentage points. Information infrastructure not only spurs significant entrepreneurial opportunities, but also enhances connectivity and interaction between people, facilitates exchange and sharing of knowledge and ideas, and thus promotes entrepreneurship (Audretsch et al., 2015).

Column 2 shows the regression results of government informationization and family entrepreneurial choice. The estimated coefficient of government informationization is positive and passes the 1% significance test. Every 1% increase in government informationization would lead to an increase in the probability of family entrepreneurship by 3.71 percentage points. Das and Das (2015) also agree that government informationization can promote market entry for enterprises. Government informationization promotes entrepreneurship mainly from three aspects. Firstly, it helps promote inter-departmental cooperation, breaks information islands, and realizes information transfer and resources scheduling between departments, thus helping to optimize administrative approval processes and improve administrative approval efficiency, which can better motivate residents to start a business. Secondly, government information disclosure made possible by government informationization can effectively restrain the government’s control of the market economy, make the exercise of power more transparent, reduce the cost of rent-seeking, and create transparent business environment for entrepreneurship. Thirdly, government informationization can promote innovative supply of public services and integration of service resources, result in an integrated online public service system, provide a full range of entrepreneurial services, and reduce the cost of entrepreneurship.

Column 3 explores the influence of resident informationization on entrepreneurship. The estimated coefficient of resident informationization is positive, but it is less statistically significant than the estimated coefficients of informationization infrastructure and government informationization, and it only passes the 10% significance test. Every 1% increase in resident informationization would result in an increase in the probability of family entrepreneurship by 6.46 percentage points. Resident informationization has a positive influence on entrepreneurship for at least three reasons. Firstly, it helps residents access the global network; in other words, residents can expand the market scope and seek opportunities for entrepreneurship (McQuaid, 2002). Secondly, entrepreneurial activities usually involve collecting and processing a large amount of information. Resident informationization is conducive to this task and thus encourages residents to engage in entrepreneurial activities. Thirdly, the high requirements for capital, system, and information in traditional industries make them prohibitive to entrepreneurs, while the development of e-commerce and the improvement in resident informationization help residents to engage in entrepreneurial activities in the field of e-commerce with less capital and more flexible time.

Column 4 shows the regression results of the influence of enterprise informationization on entrepreneurship. Its estimated coefficient is negative at the significance level of 1%. Every 1% increase in enterprise informationization would lead to a decrease in the probability of family entrepreneurship by 9.89 percentage points. Arguably, different from the influences of informationization infrastructure, government informationization, and resident informationization on entrepreneurship, enterprise informationization exerts a negative influence. Higher enterprise informationization in a region means that local enterprises are more competitive in the market, which undoubtedly increases the threshold for market access. To compete with incumbent enterprises, interested entrants need to invest more in information. This increases the cost of entrepreneurship for residents and discourages some otherwise entrepreneurial residents from the market. There is a dilemma here: improving the informationization of enterprises can enhance their competitiveness, while it hinders entrepreneurship. In this case, the government should build an informationization service platform for new ventures in order to reduce the informationization investment needed by resident entrepreneurs in the startup stage.

In the previous part of this study, entrepreneurial activities are considered homogeneous. In fact, this is not the case. In general, entrepreneurship can be broadly divided into corporate entrepreneurship and self-employed entrepreneurship. Compared with the latter, corporate entrepreneurship has to overcome higher thresholds of capital, technology, and management and delivers greater economic and social benefits. The influences of the same factor vary with different types of entrepreneurship. In addition, different industries have different informationization levels, which, in turn, have different influences on business ventures in different industries. Based on the analysis above, this section focuses on the heterogeneous influences of informationization on different types of entrepreneurship and different entrepreneurial industries.

In order to explore the influence of informationization on different types of entrepreneurship, the dependent variable (entrep) needs to be redefined. If a family is not engaged in entrepreneurial activities, entrep is assigned with a value of 0. If the business is started in the form of enterprise, entrep is valued 1. If the business is on self-employed basis, entrep is valued 2. The “China Household Finance Survey” divides family entrepreneurship into joint stock company, limited liability company, partnership, sole proprietorship, self-employed individual/privately, or individually owned business. In this paper, joint stock company, limited liability company, partnership, and sole proprietorship are classified as corporate entrepreneurship. According to statistics, of 1125 entrepreneurial families, 13.78% chose corporate entrepreneurship and 86.22% chose self-employed entrepreneurship. Individually owned business is the main type of entrepreneurship for Chinese residents, which is also evidenced for the prevailing necessity-push entrepreneurship among Chinese residents.

Since the dependent variable here is a multiple-choice variable, this paper borrows the multiple-choice probit model to study the influence of informationization on different types of entrepreneurship. In the multiple-choice probit model, the baseline group is families that are not entrepreneurial, and all the estimated coefficients are compared with the results of the baseline group. Columns 1 and 2 of Table 4 reports the estimates. Column 1 shows the regression results of informationization and corporate entrepreneurship. The estimated coefficient of informationization is positive, but it does not pass the significance test. Every 1% increase in the informationization level of the region would result in an increase in the probability of family corporate entrepreneurship by 4.02 percentage points. Column 2 shows the regression results of informationization and self-employed entrepreneurship. The estimated coefficient of informationization is positive and passes the 1% significance test. Every 1% increase in the informationization level would result in an increase in the probability of family self-employed entrepreneurship by 17.43 percentage points. In contrast, informationization plays a bigger role in promoting self-employed entrepreneurship. The main reason is that informationization provides flexible and convenient conditions for entrepreneurship, which is in line with flexible self-employment entrepreneurship. Thus, informationization promotes self-employment entrepreneurship better.

In order to study the industry heterogeneity in the influence of informationization on entrepreneurship, this paper divides industry into secondary industries, domestic services, and productive services,² and the dependent variable (entrep) is redefined. If a family is not engaged in entrepreneurial activities, entrep is assigned with a value of 0. If it is engaged in a secondary industry, entrep is valued 1. If it is engaged in a domestic service, entrep is valued 2. If it is engaged in a productive service, entrep is valued 3. According to statistics, 18.68% of entrepreneurial families start businesses in secondary industries, 71.61% in domestic services, and 9.71% in productive services. Most Chinese residents choose to start a business in domestic services, mainly because of these industries pose low barriers to entrepreneurship and provide flexible entrepreneurial forms.

As with the study on the heterogeneity of entrepreneurship types, the multiple-choice probit model is used to study the influence of informationization on entrepreneurship in different industries. Column 3–5 of Table 4 reports the estimates. Column 3 explores the influence of informationization on family entrepreneurship in secondary industries. The estimated coefficient of informationization is positive and passes the 1% significance test. Every 1% increase in the informationization level of a region would result in an increase in the probability of families choosing to start businesses in secondary industries by 11.66 percentage points. Column 4 reports the regression results of informationization and entrepreneurship in domestic service industries. The estimated coefficient of informationization is positive, but it does not pass the significance test. Every 1% increase in the informationization level would result in the probability of families choosing to start businesses in domestic service industries by 7.76 percentage points. Column 5 lists the regression results of the influence of informationization on family entrepreneurship in productive service industries. The estimated coefficient of informationization is positive, but it does not pass the significance test. Every 1% increase in the informationization level of a region would result in an increase in the probability of families choosing to start businesses in productive service industries by 2.35 percentage points. The estimation results in column 3–5 shows that informationization plays a greater and more significant role in promoting family entrepreneurship in secondary industries. Secondary industries are more informationalized, and the improved informationization level of a region can significantly increase the informationization level of secondary industries and thus promote the level of entrepreneurship in secondary industries. However, productive service industries are similarly informationalized. Why does informationization not significantly promote the entrepreneurial level in these industries? An important reason can be attributed to the strict regulations in productive service industries that disqualify a large number of entrepreneurial residents from venturing into these industries. Even if informationization creates opportunities and conditions for entrepreneurship, most residents are still not able to start their businesses in these industries. Therefore, informationization does not play a significant role in improving the entrepreneurial level in productive service industries. The government should lift unnecessary regulations in these industries to fully exploit the role of informationization in promoting entrepreneurship and fuel the development of productive service industries.

From the perspective of the life cycle of entrepreneurship, entrepreneurship includes not only the establishment of a new organization, i.e., entrepreneurial choice, but also the development of the new organization, i.e., entrepreneurial performance. The previous part of research focuses on the influence of informationization on entrepreneurial choice and finds that informationization can significantly increase the probability of residents to start a business. What is the influence of informationization on entrepreneurial performance? This is the main research issue in this section. Entrepreneurship not only addresses the income of entrepreneurs, but also solves the social employment. Therefore, entrepreneurship has two basic effects: income effect and employment effect. Income and employment are important indicators to measure entrepreneurial performance. The foothold of this section is to study whether informationization can help realize the income effect and the employment effect of entrepreneurship.

This section focuses on the income and employment effects of entrepreneurial families. However, if this study covers only entrepreneurial families and excludes families not engaged in entrepreneurial ventures, it would produce self-selected samples rather than random samples (Heckman, 1979). If the self-selected samples are estimated by ordinary least squares, the results might be biased and inconsistent (Heckman, 1979). To overcome the sample selection bias, the Heckman two-stage model is used to study the influence of informationization on entrepreneurial performance. In the study of Magri (2011), he also used the Heckman two-stage model to analyze the effect of initial net wealth of household on the size of entrepreneurship. Therefore, it is suitable for studying the relationship between informationization and entrepreneurial performance using the Heckman two-stage model. The Heckman two-stage model is described briefly as follows.

The Heckman model for the first stage is the entrepreneurial choice model shown in Eq. (1). The Heckman two-stage model requires that the first-stage model should include at least one exclusive explanatory variable. This explanatory variable influences the entrepreneurial choice of a family but not the entrepreneurial performance in the second-stage model. Thus, in this section, the average age of family members and its square are added to the family characteristic variable (F) in Eq. (1). The average age of family members has a direct impact on family entrepreneurial choice, and there is a non-linear relationship between age and entrepreneurial choice. In addition, when regressing entrepreneurial performance based on the average age of family members and its square in Eq. (2), we find that the average age of family members and its square have no significant effect on entrepreneurial performance. It could be concluded that the average age of family members and its square are suitable exclusive explanatory variables.

The Heckman model for the second stage is a linear equation that studies the influence of informationization on entrepreneurial performance, as shown in Eq. (2):where performance_fp represents the entrepreneurial performance of family f in province p. We measure entrepreneurial performance by entrepreneurial income and the number of employees in the new business and take their natural logarithms in empirical research. Since both of the measures are zero for some families, no natural logarithms could be taken. In this case, the entrepreneurial incomes and the number of employees plus 1 is the solution, and then, natural logarithms are taken, for all families. F_fp, P_p, and μ_fp represent family characteristic variable, regional characteristic variable, and random error term, respectively. Once again, F in Eq. (2) excludes the average age of family members and its square, which are instead included in Eq. (1). λ_fp represents the inverse Mills ratio to overcome the selection bias of the studied samples. We first get the inverse Mills ratio from regression in the first stage, and then add this inverse Mills ratio to the entrepreneurial performance model in the second stage. If the inverse Mills ratio is significantly non-zero, a sample selection bias is strongly suggested and the use of the Heckman two-stage model turns out to be valid; if it is significantly zero, it means that the Heckman two-stage model is unnecessary. The inverse Mills ratio is calculated as shown in Eq. (3) (Heckman, 1979):

where f(·) and F(·) represents the probability density function and the probability distribution function of standard normal distribution, respectively.

Table 5 shows the influence of informationization on entrepreneurial performance. Columns 1 and 2 report the influence of informationization on entrepreneurial income based on the Heckman two-stage model. Column 1 is the entrepreneurial choice model for the first stage, and column 2 is the entrepreneurial income model for the second stage. In column 1, the estimated coefficient of informationization is significantly positive, which indicates that informationization has a significant effect on entrepreneurial promotion. In column 2, the estimated coefficient of informationization is positive at the significance level of 5%. Every 1% increase in the informationization level of a region would result in an increase in the entrepreneurial income of a family by 3.86%. The inverse Mills ratio does not pass the significance test, which means that it is unnecessary to apply the Heckman two-stage model in studying the influence of informationization on entrepreneurial income, which is thus remeasured by using the ordinary least squares method. The regression results are shown in column 3. In column 3, the estimated coefficient of informationization is positive at the significance level of 5%. Every 1% increase in the informationization level of a region would result in an increase in the entrepreneurial income of a family by 4.03%. Clearly, both the results from the Heckman two-stage model and the ordinary least squares method prove that informationization could help realize the income effect of entrepreneurship.

Columns 4 and 5 of Table 5 report the regression results of informationization and entrepreneurial employment effect based on the Heckman two-stage model. Column 4 is the entrepreneurial choice model for the first stage, and column 5 is the entrepreneurial employment effect model for the second stage. As shown in column 5, the estimated coefficient of informationization is positive at the 10% significance level, indicating that entrepreneurial families employ more labor forces in regions with higher informatization levels. In addition, the estimated coefficient of inverse Mills ratio passes the 5% significance test, indicating that it is significantly not zero and thus it is valid to apply the Heckman two-stage model in studying the informationization and entrepreneurial employment effect. In order to compare with the estimation results from the Heckman two-stage model, column (6) reports the estimation results of informationization and entrepreneurial employment effect based on the ordinary least squares method. The estimation results show that informationization could significantly increase the number of people employed by entrepreneurial families, and the conclusion is consistent with the finding obtained from the Heckman two-stage model.

There are some shortcomings in the research of this paper. Since China Household Finance Survey can provide abundant information about family entrepreneurship, the paper adopts this dataset for our study. However, China Household Finance Survey offers cross-section data that may lead to inaccurate estimation results due to missing time factor. Although we have used the China Family Panel Studies in 2011, 2013, and 2015 for robustness test, we are unable to adopt the database for research to consider the time factor due to the lack of information in other parts of the study. Therefore, we need to use panel data to reassess the influence of informationization on entrepreneurship in future research.

Some policy implications can be drawn from the research conclusion herein.

Firstly, the government should strengthen its own informationization construction, reform and upgrade various information management systems in accordance with unified norms and standards, establish an information sharing exchange platform to promote wide sharing and effective application of information among government departments, and reduce the degree of control and improve service levels to encourage mass entrepreneurship.

Secondly, since informationization has a significant role in promoting entrepreneurship in secondary industries, government departments should strengthen the planning and arrangement of industrial Internet infrastructure construction, and build an industrial Internet with low latency, high reliability, and wide coverage. They should also speed up the deployment and construction of fiber optic networks, mobile communication networks, and wireless local area networks in areas with manufacturing clusters to achieve information network broadband upgrade.

Thirdly, the reform of administrative management system needs to coordinate with informationization to fully exploit the role of informationization in promoting entrepreneurship in productive service industries. Government departments should relax or cancel unnecessary market access administrative approval regulation over producer service industries, encourage and guide private capital to enter industries and fields that are not prohibited explicitly by laws and regulations, and try to create a market environment of fair competition and equal access.

Fourthly, enterprise informationization construction improves the market competitiveness of incumbent enterprises on the one hand and raises the threshold of entrepreneurship on the other hand. To resolve this dilemma, the government should establish and improve informationization service platforms for enterprises to help new ventures improve their informationization level.