Abstract
At present, the development of China’s economy is inseparable from the promotion of innovation and entrepreneurship. However, there are huge differences in the level of entrepreneurship in different regions, so it is urgent to analyze the influencing factors of entrepreneurship. This paper uses panel data of 31 provinces in 2008–2017 to test the impact of urban unemployment rate, regional financial development level, market scale, agglomeration effect, entrepreneurial cost, scientific and technological innovation level, government support and other factors on entrepreneurial level. To solve the issue of factors influencing the enterprises, we utilised an effective way of fuzzy logic along with decision estimation strategy. The empirical results show that regional financial development level, agglomeration effect, technological innovation ability, government support and so on have a positive role in promoting the level of entrepreneurship, while urban unemployment rate has no significant effect on the level of entrepreneurship. Based on the empirical conclusion, the paper puts forward some policy suggestions.
Introduction
This document provides instructions for style and layout, information on installing the Word template and how to submit the final version. The instructions are designed for the preparation of a camera-ready and accepted paper in MS Word and should be read carefully.
China has accessed to the “new normal economy” mode. In this context, the Chinese government has launched the “Widespread Entrepreneurship and Innovation” since 2014. President Xi Jinping emphasized: development is the first priority, talent is the first resource, and innovation is the first impetus. If China does not follow the path of innovation drive, the old and new kinetic energy cannot be converted smoothly, and it will not be truly powerful. It can only be big but not strong. Being strong depends on innovation, and innovation depends on talent. Talent policies and innovation mechanisms are the focus of the next reform. This paper, based on the panel data from 31 provinces (exclusive of Hong Kong, Macao and Taiwan) between 2008 and 2017, applies the multiple regression model to carry out an empirical study on some factors that affect entrepreneurship in the practical situation. The empirical findings have revealed several determinants, which not only contributes to creating a model that affects the level of China’s entrepreneurial activity, but also provides a certain theoretical support for China governments at all levels to develop local policies for encouraging business start-ups. The fuzzy logic (FL) is procedure of identifying the human reasoning. The method of FL initiates the path of decision making involving every possibilities amongst the values ranging YES or NO, this FL proceeds on the phases of significant possibilities of input to obtain the output which are defined [5, 16–19].
Literature review
After collating the researches of foreign scholars, we find that the topics involving the regional entrepreneurship have been covered more extensively and profoundly; they have made more studies with different variables aiming at different areas and different countries. It was hereby found that there were regional disparities in the level of entrepreneurship. They also attempted to interpret the reasons why these disparities emerged by a dozen of elements. Staniewski et al. studied the influences of factors such as public infrastructure, unemployment rate, employment policy, industry concentration, innovation level, marginal populations, economic pluralism on entrepreneurship level [13]; Castaño et al. analyzed the correlations of market growth, population agglomeration and urbanization effect, start-up capacity and behavior, unemployment rate and private wealth with local entrepreneurship level [12]; Keebled found in his studies that there were two factors that widened the gaps in the areas of United Kingdom, i.e. they are private wealth and SME survival environment [4]; Aldrich et al. analyzed the relationship of gross capitals and enrichment of local human capitals with start-up of a service business [6]. Carree argued that the local population, per capita disposable income, population density, population structure, etc. are the determinants to affect regional entrepreneurship [10]. Staniewski also believed that the major causes for these disparities in the regional entrepreneurship level rooted in the proportions of peanuts, professional operation and management personnel and high-tech personnel, the financial development degree and industry access barriers [14].
Through sorting out the domestic literatures, we found that scholars have devoted themselves to those determinants of entrepreneurial intentions (birth of business) [22]. (1) Identification of entrepreneurial opportunities. Si et al. studied the process of identification for entrepreneurial opportunities, and analyzed the origin of entrepreneurial opportunities. They believed that the government should encourage and support entrepreneurship because the supportive policies the government issued play a very important role in simulating entrepreneurship [21]. (2) Relationship between entrepreneurship and unemployment rate. Dong et al. [26] based on 17-year panel data of 21 areas in Guangdong from 1991 to 2007, examined the relationship between entrepreneurship and unemployment rate. The study revealed that the influence of the unemployment rate on the entrepreneurial rate was not evident, but on the contrary, the latter produces a great adverse impact on the former. (3) Influence of unofficial systems on entrepreneurial activity. Tian and Wang [23] probed into the influence of unofficial system on entrepreneurial activities from two perspectives of the national culture and regional subculture.(4) Comprehensive factors. Zhang [9] explored the differentials that affect startups, Gao and Zhang based on spatial panel data, conducted an empirical test on unemployment rate, agglomeration effect, market scale and other factors, Niu and Wen [2] went by ISM to build a hierarchical structure model based on the public venture influencing factors, Shanxi. The studies of Zhang and Li [20] and other scholars can also be available. Ngan et al. [15] proposed unique representation of Intuitionistic fuzzy system based on their application to decision making scenario. Roopa et al. [3] proposed fuzzy network based on rule information for analyzing automated ECG. Prathik et al. [1] mentioned various application of graph theory which can be utilised for finding the factors influencing the entrepreneurship.
In conclusion, we discovered from collated literatures from the domestic and foreign scholars that both domestic scholars and foreign scholars only have a superficial knowledge on the determinants affecting China’s entrepreneurship, especially lack empirical studies based on nationwide data.
Study design
Variable selection
(1) Explained variables: the level of entrepreneurship in the area.
In the academic circles abroad and domestic, two methods are usually used to measure the level of entrepreneurship in an area. One is to adopt the number of self-employers in the local area, and the other is to adopt the number of start-ups in that area. This paper considers the availability of data, imbibes the methodology adopted by Dong Zhiqiang, et al., and based on the number of self-employers to measure the level of entrepreneurship. The specific formula is the number of self-employer ÷ aggregate labor force (or total employment population or the total population). As a caveat, the number of self-employers in the formula is approximated substituted by the number of private and individual employees.
(2) Explanatory variable
(2.1) Agglomeration effect. In this paper, the size of the agglomeration effect is measured by an index, i.e. the ratio of the urban population to the total population (i.e. the so-called urbanization rate).
(2.2) The market scale. The population density (population per square kilometer) and per capita GDP are used to measure the market scale.
(2.3) The level of technological innovation. Given the availability of data, the proportion of population with college degree or above is used as a proxy index.
(2.4) The level of financial development in the area. The index used in this paper is the local resident’s deposit balance at year end in financial institutions.
(2.5) Urban-registered unemployment rate. Given the availability of data. The urban-registered unemployment rate is hereby selected as an index to be incorporated into the panel measurement model.
(2.6) Strength of policy support from local government. An aerial annual expenditure per capita from government fiscal is selected as proxies to measure this support strength.
(2.7) Initial cost. This paper uses the average wage of urban employee in local area as an index to measure the time cost. The higher the average wage of urban employees, the higher the cost of startups.
Table 1 consolidates the explanatory variables and the selected explanatory variables in this paper and lists their corresponding variable codes.
Summary of variables
Summary of variables
There are three data sources in this paper: (1) NBS website; under the data query item at the NBS website, there is a provincial annual data option, where the urban-registered unemployment rate, average wages of urban employees, the annual government expenditure each area, the total population each area, etc., are all from the channel. As the local residents’ deposit balance at year end, the proportion of population with college degree or above of the local financial institutions are missing in the provincial data, this paper presents the sources of the two indexes, namely, they can be available from the Report on Regional Finance Development released by the People’s Bank of China and from the China Statistical Yearbook; (2) Consult other statistical yearbooks, e.g. the index of urbanization rate in any area is from China Statistical Yearbook, and the acreage of local area is sourced from the China Geography Yearbook.
Model setup
The panel data model is assigned with a dual nature, i.e. the cross-section data and the time series data. The following part briefly describes the panel data model.
where, y represents the dependent variable of this article: the level of entrepreneurship in the area; X represents the set of influencing factors and μ represents the error terms.
To analyze the specific effect of individual, we usually assume the random error term is set to μ it :
where, the special effect from individual is represented as α i , reflecting the differences between different individuals. If such individual difference is supposed to be systematic and definite, assume α i is a fixed constant, now the model is called a fixed-effect model; if it is supposed to be random, indefinite, assume α i is not unfixed constant, but random, and the corresponding model is called the random-effect model. Whether to set a fixed effect model or a random effect model depends on the empirical analysis results of this paper.
In the panel data model, there are several analytical procedures as below: (1) smoothness test; (2) cointegration test; (3) regression model and its estimation.
Smoothness test
In this paper, we adopt the Common root-Levin, Lin, Chu test, i.e. LLC method, by which we carry out a data smoothness test on nine variables with final results deduced as below:
As can be seen from the above Table 2, the above nine sequences are non-smoothness, so we will have the first order differential in next step, where the sequence that corresponds to variable growth rate can be available. We express them as DY, DCity, DPop, DGdp, DScin, DFin, DGov, DIncome, DUnemp, respectively, and use Eviews software to conduct LLC unit root test with the results conclude as below:
LLC Unit Root Test
LLC Unit Root Test
It can be found from the above Table 3, at the significance level of 5%, the above nine economic variables all refute null hypothesis, namely that the unit root exist. We hereby concluded that these variables are the smoothness sequences, the integrated of order. We learn by the knowledge of econometrics that the next step we carry out is the cointegration test to demonstrate whether there is a long-term steady equilibrium relationship.
First order differential LLC unit root test
In the panel data analysis, the traditional practice is that the sequence used as required must be stable, otherwise it may generate pseudo-regression phenomenon. The cointegration test used in this paper is the third method, i.e. the Johansen test, see Table 4 for the results.
Cointegration test results
Cointegration test results
By analyzing the results listed in the above table, it is found that at the 5%significance level, there is a cointegration relationship among the eight combinations as above, that is, we can draw a conclusion that at the 5%significance level, there are long-term steady equilibrium relationships among the eight explanatory variables and the explained variables.
As previously mentioned, the panel data model is divided into two types, i.e. the fixed effect model and the random effect model. For which model we choose, Eviews software provides a Hausman test method proposed by Hausman in 1978.
The following section describes the results obtained using the Hausman test method.
As can be seen from the above Table 5, P value from Hausman test in the national model we have built is 0.0000. At the 5%significance level, we shall reject the null hypothesis: a random effect model shall be set up. We can draw a conclusion that the individual effect in the random impact model are irrelevant to the explanatory variables.
Hausman test
Hausman test
We then analyze the fixed effect model, the results from it are as follows:
Based on the test results in the Table 6 above, we found that the urbanization rate, population density, per capita GDP, high-tech innovation level, per capita savings deposits, per capita GDP and per capita wage income all pass the significance test at the 5%significance level; whilst the urban unemployment rate does not pass the significance test.
Results from fixed effect model
The urbanization has an apparently positive effect on the level of entrepreneurship. This shows that the population concentration dramatically drives the development of areal start-ups, that is, the agglomeration effect plays a positive role in development of entrepreneurship, which coincides with the findings of scholars at home and abroad. The population density has an adverse effect on the level of entrepreneurship, and the per capita GDP just goes the other way. In the model as proposed in this paper, we use population density, per capita GDP as a proxies of market scale. In the analytical framework, we think that the greater the population density, the broader the market scale of the relevant areas is, and in turn the greater the positive effect on the level of entrepreneurship. The results from empirical analysis contradict our expectations. We agree through analysis that due to over high population density, the thresholds of resources and environment that an area can bear has been surpassed, resulting in that local resources and environment are too finite to meet the requirements of entrepreneurs, and that the cost for entrepreneurship are too high; The per capita GDP produces a significant positive effect which is consistent with expectations on the level of entrepreneurship, while the level of financial development in the areas plays a motivational role on it. The higher the level of financial development, the higher the level of entrepreneurship in the area. The level of technological innovation in the area positively pushes forward the level of entrepreneurship. The higher the level of technological innovation, the higher the level of entrepreneurship will be. This coincides with what we expected. The urban-registered unemployment has a vague effect on the level of entrepreneurship, and is the only one unexamined index among selected variables. This result is in concert with those inferred by Dong et al. [25], Gao and Zhang [24]. Per capita fiscal expenditure dramatically heaves the level of entrepreneurship, i.e. the higher the per capita fiscal expenditure in the area, the higher the level of entrepreneurship will be. It is suggested that the government’s fiscal support policies can stimulate entrepreneurship and play a positive role in driving the development of start-ups.
In certain life scenario, various decision includes fuzziness since aims, constraints and actions which is possible are called exactly. Situation of decision making in fuzzy scenario, the outcome of making decision is significantly reflected by judgments which are subjective that are imprecise and vague. Imprecision sources consists of information pertaining to unquantifiable, information which are incomplete, non-attainable, ignorance which are partial.
To resolve this type of issues related to imprecision, theory based on fuzzy set was initially introduced as way of mathematical representations and monitor decision making vagueness. Scenario such a fuzzy logic every number amongst 0 and 1 depicts a truth that is partial, wherein crisp value illustrates the logic binary ranging from 0 or 1. Therefore, expression of fuzzy logic and vague handling or judgment of imprecise is done mathematically.
To handle the crisp vagueness of thoughts found in humans and decision taking expression, theory of fuzzy set is very significant. In case of to handle the involved ambiguities in the procedures of estimating the linguistic value, converting terms which are linguistic into numbers which are fuzzy are obtained. The variables which contains the sentences or phrases are known as linguistic variables. For finding certain assessments these linguistic variables are found to be significant.
In practical scenarios, ranges of linguistic can be defined by numbers of fuzzy and triangular fuzzy function which is quite normal. In the following situation specific review are carried for providing essential information regarding the fuzzy logic.
Fuzzy set
This Fig. 1 depicts the outcome obtained from the fixed effect model for various sequence such as Dcity, Dpop, D gpd etc.
Outcome of fixed effect model.
This paper takes data from 2006 to 2015 in 31 provinces in China as a base, applies the panel data model to study the determinants of China’s entrepreneurship. The following conclusions is thereby drawn: some factors such as urbanization rate, per capita GDP, per capita residents’ savings deposit, per capita fiscal expenditure, the proportion of college graduates in the population have a significant effect on the level of entrepreneurship, while population density takes on an adverse impact; the urban unemployment produces a vague effect. This paper gives several pieces of advice as follows:
(1) Adhere to carry out the process of urbanization, and continuously improve the urbanization rate, especially in the central and western areas, facilitate the level of entrepreneurship there. (2) Maintain the strategy of rejuvenating the country through science and education, and constantly enhance the scientific and technological innovation, increase inputs in education and scientific research funds, and actively encourage entities to increase R & D investment. (3) Rationally design industrial layout, exert industrial agglomeration effect. The government should formulate relevant measures according to local conditions to actively guide the establishment of industrial clusters. (4) Improve the level of financial development, tackle with the financing puzzle. The government should provide a good financing environment for start-ups so as to avoid their failure because of tough financing. (5) Enhance business support efforts. It is suggested that the government shall expand supports for start-ups by establishing the multilevel business incubator and the industry intermediary platforms, and support Mass Entrepreneurship and Innovation through granting the start-ups with tax preference, interest-free loans, Angel investment and other preferential policies.