Factors Influencing Entrepreneurial Intention among Foreigners in Kazakhstan

Abstract

Entrepreneurship is essential in contributing to economic growth, job creation, technological advances, etc. in all countries, including Kazakhstan. Hence, the purpose of this study was to find out how to further facilitate the development of (foreign) entrepreneurship in Kazakhstan. In doing so, the authors attempted to identify factors influencing entrepreneurial intention (EI), specifically among the foreigners in Kazakhstan, and highlight the role of government support (GS) in general and under COVID-19. The study conceptualized the extended “TPB + Moderator (GS)” model. The hypotheses were tested on a sample of 362 new and established foreign entrepreneurs in Kazakhstan by means of descriptive analysis, Pearson’s correlation analysis, and multiple regression analysis. The study found that the foreigners’ personal attitude (PA) toward entrepreneurship was the strongest factor motivating their EI, followed by their perceived behavioral control (PBC) as the second strongest factor. The role of GS strengthened each effect of PA and PBC on EI. The moderating effect of GS and PA is greater than that of GS and PBC; each of the moderating effects is lower in magnitude than each of the direct effects. Their subject norms (SNs) and the moderating effect of GS and SNs are both insignificant.

1. Introduction

Entrepreneurship contributes significantly to the economy, society, and humankind [1] primarily in that it: (1) stimulates innovation and technological advancement; (2) enhances products, services, and quality management; (3) creates capital, employment, and social mobility; (4) exploits local natural and human resources; (5) adds to productivity and government revenue; (6) distributes wealth, mobilizes savings, and reduces poverty; (7) diversifies economic activities; and (8) speeds up the country’s economic development [2,3,4,5,6,7]. Apparently, the significance of entrepreneurship also holds true in the Republic of Kazakhstan.

1.1. Context of This Study

Statistics show that entrepreneurship as well as small and medium-sized enterprises (SMEs) represent about 90% of businesses and more than 50% of employment worldwide. In addition, the contribution of entrepreneurship and SMEs to the gross domestic product (GDP) accounts for up to 40% in emerging economies [8]. By comparison, the entrepreneurial contribution to GDP in the case of emerging Kazakhstan was only 28.5% (much lower than the average level of 40%) in 2019 according to the latest updates [9]. In recent years, SMEs in Kazakhstan have accounted for only 25% of the value added and 37% of the employment, compared with figures of 57% and 60–70%, respectively, in most Organization for Economic Co-operation and Development (OECD) economies, and most (60%) of them operate in low value-added sectors [10]. The government of Kazakhstan aims to double the contribution of SMEs to the GDP by 2030 (to 36% from the baseline of 17.5% of GDP at the end of 2011) and to 50% of GDP by 2050 [10].

Along with the ambition to enhance general entrepreneurship, “foreign entrepreneurs” shall be emphasized as they also play an important role in stimulating and contributing to the economy of Kazakhstan. “Foreign entrepreneurs” are defined by the U.S. Department of Commerce as “minority entrepreneurs” who are not of the majority population [11]. According to the press service of the Ministry of Labor and Social Protection (MLSP) of the population of Kazakhstan (as of 1 January 2020), 19,145 foreign citizens are employed (based on work permits issued to foreign citizens and the involvement of foreign labor). The number of employers who attracted foreign labor is 2040; they have employed 508,785 Kazakhstani citizens, accounting for 96.3% of the total number of employees [12]. It is evidenced that, compared with the population of Kazakhstan (approximately 19.1 million) [13], there is a considerable need to attract foreign entrepreneurs and increase the share of foreign entrepreneurship in this country.

Nevertheless, Kazakhstan is reputed to have a very good investment climate. Since declaring independence from the Soviet Union in 1991, Kazakhstan has passed a series of reforms to liberalize its economy and attract foreign investment, and the government has been gradually improving the business climate for foreign investors and plans to create a national company, “Qazaq Invest”, which would facilitate the activities of foreign investors in Kazakhstan [14,15].

The indices of political, economic, socio-cultural, technological, environmental, and legal (PESTEL) factors (see

Table 1. Indices of PESTEL factors.

Factor	Index	Description	Result	Assessment
P	Freedom	Based on 10 political rights and 15 civil liberties [16]	22/100 (2019 scores)	Not free
	Political stability index	Based on 195 countries [17]	0 (2018 scores)	Politically stable
E	GDP per capita	Based on International Monetary Fund data [18]	69th/192 (nominal, 2019 rank); 52nd/192 (PPP, 2019 rank)	The largest economy in Central Asia
S	Population growth rate	Estimated population is 19.1 million (68% Kazakhs, 19.3% Russians, and others (Uzbeks, Uighurs, etc.)); the growth rate is only 0.89% [19]	89th/171 (2019 rank)	Low population growth rate
	Net migration rate	0.4 migrants/1000 population [19]	51st/121 (2019 rank)	Low migration rate
	Life expectancy	Median age is 31.6 years; life span is 72 years [19]	85th/138 (2019 rank)	Relatively low life span
	Prosperity index	Strong in education and the investment environment [20]	68th/167 (2019 rank)	Increasingly promising
	Global social mobility index	Key indicators are health, education, technology, work, protection, and institutions. Overall index score: 64.8 [21]	38th/82 (2020 rank)	High capacity for a child to experience a better life than their parents
T	Global competitive index	Drivers of productivity and long-term economic growth reported by the World Economic Forum; a score of 63 points at 103 indicators or 12 pillars [22]	55th/141 (2019 rank)	Increasingly competitive
	Global innovation index	Based on innovation capabilities; 80 indicators [23]	79th/129 (2019 rank)	Increasingly innovative
E	The world’s best countries	Adventure (5.2), citizenship (1.7), cultural influence (0.0), entrepreneurship (0.6), heritage (3.2), movers (26.6), open for business (29.9), power (7.8), and quality of life (4.9) [24]	66th/80 (2019 rank)	Relatively low among the best countries
	Environmental performance index	Scores: 54.56/100 [25]	101st/180 (2018 rank)	Medium environmental performance
	Global tourism ranking	Reported by the World Economic Forum [26]	81st/136 (2018 rank)	Medium tourism ranking
L	Rule of law index	Modernization of the legal framework and adherence to international best practices reported by the World Justice Project [27]	62nd/128 (2020 rank)	Highest in Central Asia
	Doing business ranking	Ease of doing business [28]	22nd/190 (2020 rank)	High ranking
	Global entrepreneurship index	[29]	59th/137 (2019 rank)	Medium entrepreneurial ranking

) demonstrate that Kazakhstan in general is politically stable, economically flourishing, socio-culturally promising, technologically advancing, environmentally appealing, and legally upgrading. It is a country with a favorable business climate that provides opportunities, potentials, and attractiveness for foreign investment and entrepreneurial activities.

To date, Kazakhstan tops Central Asian countries in terms of attracted investments, accounting for more than 70% of all foreign direct investments (FDIs) into the region [30]. The abundance of natural resources has attracted considerable interest among international investors, particularly in the oil and gas sectors as well as the metallurgy industry [13,31]. Kazakhstan’s main challenge remains attracting investment in sectors and activities other than oil extraction and natural resources, which account for more than 70% of the total FDI stock, and to retain investors already in the economy [32].

However, in 2020, Kazakhstan’s economy collapsed due to the COVID-19 outbreak, reporting a negative growth balance of 2.7%. The pandemic halted global activity and depressed global demand and oil prices. In April 2020, the average oil price fell to 21 USD per barrel, the lowest in two decades. Since Kazakhstan is still largely dependent on oil prices (35% of the GDP comes from oil and gas revenues) and the economy relies heavily on hydrocarbon exports (accounting for 75%), this has made the economy even more vulnerable. Moreover, the pandemic has severely hit the retail, hospitality, wholesale, and transport sectors in Kazakhstan, which account for around 30% of employment and are the most concentrated in cities [14].

COVID-19 has significantly affected entrepreneurship in a dual manner. On the one hand, most enterprises have difficulties with operating, such as business cancellations or closures, reduced income, a considerable decline in demand, and a lack of funds. Consequently, many SMEs are expected to go out of business during and after COVID-19 [33]. COVID-19 could negatively impact the risks associated with entrepreneurship, and ultimately hinder business start-ups [34]. On the other hand, COVID-19 could alter perceptions of entrepreneurship for the better and give rise to more or new entrepreneurial activity [34]. The experience of a crisis leads entrepreneurs to become more rational and guided by planned behavior when making a decision [35]. After all, entrepreneurs are fighters and by nature optimistic and resilient; they will overcome this difficult period and bounce back [34].

The government of Kazakhstan promptly responded to the situation with a package of urgent targeted measures to, among other things, stabilize the macroeconomic situation and mitigate COVID-19’s impact on SMEs [36].

Generally speaking, governments, irrespective of the country, are customarily engaged in fostering entrepreneurship and SMEs during the process of their start-up and while they are surviving and thriving [37]. In recent years, governments have become increasingly active in designing and applying policies to encourage and support entrepreneurial efforts [7]. The main reason for this is that entrepreneurship has become one of the most significant drivers of sustainable economic growth and development [38].

Government support can be in any of the forms summarized in

Table 2. Forms of government support (GS).

Government Support	Description
Educational support	Entrepreneurial education Entrepreneurial training
Cultural support	Cultural and social norms on entrepreneurship Building confidence in new and growing firms Increasing the knowledge and empathy of government officials toward business
Environmental support	Access to the market Creation of opportunities Sharing of information Provision of resources Improvement of commercial, legal, and physical infrastructures Favorable external ties
Financial support	Investment Financial aid, grants, loans, and subsidies Funds
Government policies	Employment-related measures Health system and medical actions Fiscal and monetary policies Stabilizing and stimulating markets Tax incentives Regulatory measures on customs duties and tariffs National and state emergency measures Simplified regulations for doing business Anti-bureaucracy measures Anti-corruption measures Anti-red-tape measures Agency services E-governance
Government programs	Public procurement Specific initiatives for new and growing firms Science parks and business incubators Innovation activities

Source: compiled by the authors based on Amsden [39]; Choi et al. [40]; Storey and Techer [41]; Kamrul and Patrick [42]; Wilson et al. [43]; Fajnzylber et al. [44]; Doh and Kim [38]; Maxat [45].

. Two of these are the most fundamental: government policies and government programs [29]. Thus, this study aims to shed light on government policies and programs to explain the role of government support.

From the very first days of its independence, the government of Kazakhstan started to pay close attention to entrepreneurship and SME development. The government has done a lot of reform to stimulate Kazakhstan’s economy [13], such as E-governance. It is regarded as one of the most successful examples of ICT-driven reform of public administration in Central and South Asia [46]. Moreover, the government has launched numerous supporting programs over the years. Among them are the “Innovative Industrial Development Strategy for 2003–2015”, the long-term “Kazakhstan Strategy 2030”, which was later expanded to the “Kazakhstan Strategy 2050”, and “30 Corporate Leaders of Kazakhstan” [47]. The outbreak of COVID-19 has once again testified to the importance of government support in reshaping confidence in, stabilizing, and recovering the economy in the country [34,48].

In summary, positively speaking, the government support for entrepreneurship in Kazakhstan is beyond doubt. Entrepreneurship programs are working. The physical, commercial, and legal infrastructures provide support to new ventures [49]. Yet, negatively speaking, entrepreneurial education, both at the school and post-school stages, needs improvement. The research and development (R&D) transfer also requires significant work as it is currently insufficient. Research institutions should share knowledge with new and growing firms. Legislation seeking to prevent anti-competitive behavior from established firms should be improved and firmly enforced. Furthermore, science parks and business incubators should receive more support, their number should be increased, and they should learn how to operate more effectively. The equity financing is currently inadequate, which makes it difficult for high-tech ventures to get started and scale up [49]. In addition, bureaucracy, corruption, and red tape (the long time required for documents to be processed) are three major barriers to effective state–business interactions [50].

1.2. Problem Statement and Purpose of This Study

The above-discussed background reveals that fostering SMEs and foreign entrepreneurship in particular is vital for economic, societal, and human benefits to flow to Kazakhstan. Kazakhstan has a generally favorable business climate as well as strong government support; however, there remain many weaknesses and constraints to be addressed. The state of its general and foreign entrepreneurship requires continual enhancement. Moreover, COVID-19 has brought about severe challenges to entrepreneurial activities in this country. It is of great concern and interest to scholars, entrepreneurs, and politicians to investigate and react to the major trends during and after the crisis [33].

Accordingly, this study identified a threefold issue (foreign entrepreneurship, the impact of COVID-19, and the role of government support (GS)) in the context of Kazakhstan. This issue entails research of great necessity and urgency in the search for appropriate solutions to the encouragement of foreign entrepreneurship in particular.

1.3. Approach to and Objectives of This Study

Contemporary research has discovered that entrepreneurial intention (EI) is the strongest, most fundamental, enduring, and frequently used predictor of entrepreneurial activity and business success [51,52]. This is due to the crucial role EI plays during the decision process of starting a business [53]. EI represents mental perspectives, such as desires, wishes, and hopes, that influence individuals’ choice of entrepreneurship [54]. Hence, we shifted our attention away from promoting entrepreneurship to promoting EI. Taking a further step backward, the factors stimulating EI were initially identified and examined in order to address the promotion of EI and foreign entrepreneurship in the end. Thus, this study focuses on the factors influencing EI among foreigners in Kazakhstan.

There is still a severe scarcity of research on this specific subject based on the extant literature to date. First, the function of government support (GS) in facilitating entrepreneurship in Kazakhstan has not received much attention. Second, the population of foreigners and their entrepreneurial practices have not been surveyed. Third, the complexity incurred by COVID-19 has produced a void in the research area of EI and its motivating factors in this country. The present study attempted to fill these research gaps by investigating the EI of foreigners and the role of GS in the context of Kazakhstan during the COVID-19 pandemic.

The objectives of this study are as follows:

(1)

To explore the miscellaneous factors, theories, and models that explain EI.

(2)

To develop and examine a research model containing the role of GS and other factors influencing EI among foreigners in Kazakhstan.

(3)

To provide insights into improving (foreign) entrepreneurship and suggestions on further research for academics, researchers, practitioners, and policy-makers.

2. Literature Review

Entrepreneurial intention (EI) is related to the concepts of entrepreneur, entrepreneurship, and intention. Entrepreneurs, in the business world, use social and environmental resources to improve economies and people’s lives. They create jobs, develop new solutions to problems, create technology that improves efficiency, and exchange ideas globally. They are the bridge between invention and commercialization [29]. Entrepreneurship, accordingly, emphasizes invention, innovation, and creativity in the process of creating something new or better for society. Entrepreneurship requires the capacity and willingness to develop, organize, and manage a business and handle uncertainties, challenges, difficulties, and risks. Entrepreneurship aims to pursue a profit, which is more related to achievement and success than to capital gain [55,56]. Intention, according to Ajzen [57], refers to “the indication of how hard people are willing to try, of how much of an effort they are planning to exert, in order to perform the behavior”. Intention represents a conscious state of mind that precedes the action [58].

Combining the above separate concepts, EI can be defined as the level of cognitive awareness [51] that aims to set up a new business or to form a new organization in the future [59] and represents the “individual commitment to commence a new business” [60].

2.1. Approaches to EI Research in General

Two types of approach have emerged in the field of EI research: (1) the content-oriented approach, which searches for the specific things within individuals that initiate, direct, sustain, and stop behavior; and (2) the process-oriented approach, which explains how behavior is initiated, directed, sustained, and stopped [61].

The content-oriented approach can be explained by the “pull” theory and the “push” theory proposed by Gilad and Levine [62]. The “pull” theory contends that individuals are attracted by internal forces into entrepreneurial activities. The “push” theory argues that individuals are pushed into entrepreneurship by negative external forces [63,64]. Empirical research [65,66] indicates that individuals become entrepreneurs primarily due to “pull” factors rather than “push” factors.

Table 3. Single factors influencing EI.

Category	Typical Factors
Demographic factors [67,68,69,70,71,72]	Gender Age Educational level Ethnic background	Religion Nationality Geography
Personality or Personal factors [61,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88]	Lifestyle Need for independence Need for achievement Tolerance of stress Tolerance of ambiguity Capacity or skills Intelligence or talents Leadership	Innovativeness or creativity Competitiveness Risk-taking propensity Locus of internal control Self-efficacy Rebelliousness Empathy Optimism
Situational factors [89,90,91,92,93,94,95,96,97,98,99]	Triggering events Time constraints Unemployment	Ideas and opportunities Job satisfaction Low income
Cognitive factors [100,101,102,103,104,105,106,107,108]	Knowledge structures Cognitive styles Decision-making	Arrangement scripts Willingness scripts Ability scripts
Social factors [69,73,83,90,94,98,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130]	Life experiences Prior experiences Role models	Entrepreneurship education Business incubators
Environmental factors [37,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146]	Cultural and social norms Entry regulation Market dynamics Access to capital Availability of information Human resources Supply of materials Social relations or networks Government support	R&D transfer Techniques of production Physical infrastructure Commercial and legal infrastructure Economic and political infrastructure Institutional infrastructure

summarizes and categorizes the typical factors based on an exhaustive review of past research using the content-oriented approach.

The process-oriented approach focuses on people’s cognitions, perceptions, motivations, and intentions that result in activities [61]. Humans can think about possible future outcomes, decide which of these are most desirable, and decide whether it is feasible to pursue these outcomes. It is not reasonable to expect people to pursue outcomes that they perceive to be either undesirable or unfeasible [147]. As a reaction, a variety of influential intention-based models and theories were developed to offer another way of predicting and understanding entrepreneurship [148], as displayed in

Table 4. Established intention-based models.

Model	Factors
Shapero’s [116] model of entrepreneurial event (SEE)	Perceived desirability Perceived feasibility Propensity to act
Bird’s [51] implementing entrepreneurial ideas model	Social, political, and economic context Personal history, personality, and abilities Rational analytic thinking Intuitive holistic thinking
Robinson’s et al. [64] entrepreneurial attitude orientation model	Achievement in business Innovation in business Perceived personal control of outcomes Perceived self-esteem in business
Ajzen’s [57] theory of planned behavior (TPB) model	Personal attitude Subjective norms Perceived behavioral control
Krueger and Carsrud’s [60] intentional basic model	Exogenous influences on behavior Attitudes toward target behavior
Krueger and Carsrud’s [60] theory of planned behavior entrepreneurial model	Exogenous influences on entrepreneurial activity Perceived attractiveness of entrepreneurial behavior Perceived social norms of entrepreneurial behavior Perceived self-efficacy for entrepreneurial behavior Exogenous facilitating or inhibiting influences
Krueger and Brazeal’s [149] entrepreneurial potential model	Perceived desirability (social norms, attitude) Perceived feasibility (self-efficacy) Credibility Propensity to act Potential Precipitating event
Boyd and Vozikis’ [92] intention model	Social, political, and economic context Personal history, personality, and abilities Stored information (beliefs) Rational, analytic, and cause–effect thinking Intuitive, holistic, and contextual thinking Attitudes and perceptions Self-efficacy
Davidsson’s [110] economic–psychological model	Personal background General attitudes Domain attitudes Conviction Situation
Douglas and Shepherd’s [150] maximization of expected utility model	Income Individuality (independence) Risk-taking (risk-tolerance) Work effort (workload)
Elfving’s et al. [95] contextual intention model	Trigging event Motivation Self-efficacy Entrepreneurial goal Entrepreneurial behavior Superordinate goal Perceived entrepreneurial desirability Perceived entrepreneurial feasibility Opportunity evaluation

.

These models are to a great extent similar as they all integrate attitudes and social learning theory and include relevant factors, such as individual and contextual factors, that influence the decision to start a business [151]. However, only two of these models (TPB and SEE) have remained competitive and dominate the literature due to their predictive ability; in particular, the superiority of TPB has been acknowledged by many studies [152,153].

2.2. Approach to EI Research in This Study

This study followed the process-oriented approach and applied the well-established TPB model [57] for two reasons: (1) many researchers believe that the focal point of entrepreneurial intention (EI) research should shift from content-oriented or the psychological profile of entrepreneurs to the entrepreneurial process or event as it takes place within a multidimensional social context [59,154]; and (2) the intention-based models predict behavior better than single variables [90]; therefore, they yield more predictive power and improve post hoc explanations of entrepreneurial behavior [60,73,155].

2.3. Application of the TPB Model in EI Research

According to the TPB model, intentions are determined by personal attitude (PA), subjective norms (SNs), and perceived behavioral control (PBC). Alternatively speaking, human behavior is guided by three kinds of considerations: (1) beliefs about the likely consequences of behavior (behavioral beliefs or PA); (2) beliefs about the normative expectations of others (normative beliefs or SNs); and (3) beliefs about the presence of factors that may facilitate or impede the performance of the behavior (control beliefs or PBC) [156].

The TPB model is the most-used model in entrepreneurial intention (EI) research for a variety of reasons. First, “unlike other models, TPB offers a coherent and generally applicable theoretical framework, which enables us to understand and predict EI by taking into account not only personal but also social factors” [157]. Second, entrepreneurship is generally viewed as a form of planned behavior, one that does not occur spontaneously [73]. Third, when considering the influence of SNs on entrepreneurship, approaches that consider a society’s specific feelings about entrepreneurship, as opposed to general cultural norms, are shown to be better indicators of entrepreneurial activity [158]. Fourth, the theory has been empirically shown to be effective across dozens of different behaviors [57,73] and has firmly withstood the test of time over a period of more than 30 years and more than 2000 published empirical studies [156,159].

Reviews of existing research show that intention accounts for approximately 30% of the variance in behavior [160], or, more accurately, 39% [161]. Furthermore, past research showed that the individual TPB components together explain between 21% [162] and 55% [163] of the variance in the intention to develop an entrepreneurial career, varying from study to study. Kolvereid [155] and Liñán and Chen [163] claim that TPB explains 30–45% of the variance in EI. In a meta-analytical review by Armitage and Conner [160], the TPB accounted for 27% and 39% of the variance in behavior and intention, respectively. Kautonen et al. [164] found that PA, SNs, and PBC jointly explained 59% of the variation in intention.

Although TPB has already been acknowledged as a robust theory and proven to be an effective tool in EI research, there is still a chance to add some variables that may influence the model [165,166]. As such, Ajzen [167] refined the TPB model by adding new variables, namely demographic, personal, social, and environmental factors that may be indirect antecedents (mediators) of EI and behavior. Furthermore, Fayolle and Liñán [153] contend that new research regarding moderating effects (moderators) added to the TPB model may also be valuable.

3. Hypothesis Development and Research Framework

This study embraced TPB model to explain the entrepreneurial intention (EI) among foreigners in Kazakhstan, and attempted to incorporate the role of government support (GS) as an external moderator. A moderator variable is defined as one that systematically modifies either the form (direction) or the strength (magnitude) of the relationship between an independent (predictor) and a dependent (criterion) variable through its moderating effect [168,169,170].

The modest attempt to add GS as a moderator is assumed to be important both theoretically and practically. Theoretically, examining a widely tested model—TPB—in the context of EI among foreigners in Kazakhstan contributes to strengthening the generalizability and the explanation capability of the TPB theory, and exploring whether the external variable—GS—has significant moderating effects may enhance the sufficiency of the TPB model. Practically, insights into the role of GS may provide implications for the enhancement of foreign entrepreneurial activities and inspire further research in this area.

3.1. Personal Attitude (PA) and Entrepreneurial Intention (EI)

In context of entrepreneurship, personal attitude (PA) toward the behavior refers to the degree to which the individual holds an overall positive or negative personal valuation about being an entrepreneur [57]. People develop attitudes based on the behavioral beliefs that they hold about the consequences or outcomes and other attributes of performing the behavior. Such consequences include both intrinsic and extrinsic rewards, such as financial rewards, independence or autonomy, personal rewards, and family security, all of which do influence favorably the intention to start a business. Negative or costly outcome expectancies, such as perceiving risks associated with entrepreneurial activities, impact unfavorably the intent to start one’s own business [171,172].

Many studies, including a meta-analysis [160], show that PA toward entrepreneurship is the strongest predictor of EI. The positive link between PA and EI has also been verified in diverse contexts and cross-country studies [173,174]. However, in an opposite and conflicting view, PA has also been reported to be an insignificant predictor of EI [175].

Nevertheless, as a general rule, the more favorable the PA is toward a behavior, the stronger the intention to perform that behavior. Therefore, rending entrepreneurship a more favorable career path and an advantageous option to people may suggest that they would come up with business ideas and may start their own ventures [57]. Accordingly, the following hypothesis is proposed:

Hypothesis 1. (H1)

Personal attitude (PA) toward entrepreneurial behavior positively influences entrepreneurial intention (EI) among foreigners in Kazakhstan.

3.2. Subjective Norms (SNs) and Entrepreneurial Intention (EI)

Subjective norms (SNs) refer to the perceived social pressure to perform or to not perform a specific behavior. This is based on beliefs concerning whether important “reference individuals or groups”, such as family, friends, or significant others, approve or disapprove of an individual starting to take steps to create a new business, and to what extent this approval or disapproval matters to the individual [57]. In other words, SNs consist of two components: normative beliefs and the motivation to comply with these beliefs [176].

There appears, in general, to be a strong relation between norms and intention depending on the individual’s propensity to conform and personality characteristics, especially within collectivistic cultures, where SNs play a positive and important role in explaining intention [163,175,177]. Cialdini and Trost [178] suggest that social norms have the greatest impact when conditions are uncertain. Numerous studies have found support for social norms as a strong predictor of EI [157,175,179].

However, some authors [162,180] controversially mention that perceived social norms have little explanatory power with respect to entrepreneurial intentions. For instance, it has been reported as an insignificant or the least significant determinant of EI [73,162,163,181,182]. Some other authors [183,184] claim that, despite the weak role that SNs play in the TPB model regarding the pattern of relationships, in some studies in the area of EI research this alleged weakness of inconsistency is not so clear.

Thus, in an attempt to probe into the relationship between SNs and EI among foreigners in Kazakhstan, the following hypothesis is developed:

Hypothesis 2. (H2)

Subjective norms (SNs) positively influence entrepreneurial intention (EI) among foreigners in Kazakhstan.

3.3. Perceived Behavioral Control (PBC) and Entrepreneurial Intention (EI)

Perceived behavioral control (PBC) refers to the perceived ease or difficulty of performing a given behavior. It is based on control beliefs regarding the presence or absence of requisite resources and opportunities for performing the behavior in question. The greater the PBC over starting to take steps to create a new business, the stronger the individual’s intention to engage in such activities [156]. In other words, PBC is an individual’s belief and confidence in his or her capability to perform as an entrepreneur and realize control and success in an entrepreneurial activity [57].

Ajzen [57] states that PBC in entrepreneurship comprises both the feeling of being capable of starting an entrepreneurial activity as well as the perception of one’s ability to control the activity. PBC can be operationalized via self-efficacy (SE) [185] or perceived feasibility (PF) [116] in the activity. All three concepts—PA, SE, and PF—deal with the “perceived ability to perform a behavior” [57,186]. Nevertheless, the difference between them lies in the fact that PBC includes not only the feeling of being able, but also the perception of the controllability of the behavior [186].

There is overwhelming empirical support for PBC having a positive impact on EI. Actually, in many prior studies (e.g., Almobaireek and Manolova [181]; Aragon-Sanchez et al. [187]; Armitage and Conner [160]; Autio et al. [162]; and Boyd and Vozikis [92]) PBC also appeared to be the most powerful antecedent or best predictor of intention. As a result, the following hypothesis is suggested:

Hypothesis 3. (H3)

Perceived behavioral control (PBC) positively influences entrepreneurial intention (EI) among foreigners in Kazakhstan.

3.4. Effects of Government Support (GS)

As previously discussed, government support (GS) in Kazakhstan has played an essential role in enhancing entrepreneurship over the years and stabilizing entrepreneurship since the outbreak of the COVID-19 pandemic.

However, there is still a surprising scarcity of research on GS in the intention-based models, and hardly any research regarding the effects of GS on EI among foreigners in Kazakhstan. Hence, this study argues that TPB components predict EI and the predictive power can be dependent on the role of GS, which is referred to as a moderator in statistics [188]. Hence, the following hypotheses are formulated:

Hypothesis 4. (H4)

With the moderation of government support (GS), the relationship between personal attitude (PA) toward entrepreneurial behavior and entrepreneurial intention (EI) among foreigners in Kazakhstan strengthens.

Hypothesis 5. (H5)

With the moderation of government support (GS), the relationship between subjective norms (SNs) and entrepreneurial intention (EI) among foreigners in Kazakhstan strengthens.

Hypothesis 6. (H6)

With the moderation of government support (GS), the relationship between perceived behavioral control (PBC) and entrepreneurial intention (EI) among foreigners in Kazakhstan strengthens.

The hypotheses are alternatively depicted in the research framework shown in Figure 1.

4. Research Methodology

In order to test the hypotheses, this empirical study involved a complete package of data work composed of data collection, data examination, and data analysis.

4.1. Data Collection

The target population of this study was narrowed down to new and established foreign entrepreneurs currently working in Kazakhstan, who most probably hold a middle–top managerial or specialist position in a sampling unit (a SME). “New entrepreneurs” are those who are running new businesses that have been in operation for between 3 months and 42 months, while “established entrepreneurs” are those who are running a mature business that has been in operation for more than 42 months [10]. The focus on new and established foreign entrepreneurs instead of other types (i.e., potential, intentional, nascent, and discontinued entrepreneurs) ensures the maximum likelihood that the firm entrepreneurial intentions led to the subsequent start-up of the venture.

Despite the lack of statistics on the exact number of and contact information for foreign entrepreneurs, and the failure to obtain further assistance from the relevant state bodies, this study estimated a target population of 2040–6000 people. The authors applied a “networking” approach [189,190] through stratified (probability) and convenience (non-probability) sampling techniques to gather data. To do this, this study used referrals, networks, and gatekeepers to circulate the questionnaires. To be more effective, the authors initially resorted to their “circle of friends”; then, the network expanded randomly to new respondents individually or to referrals/gatekeepers of foreign SMEs for further circulation. The participants are mainly from the major cities in Kazakhstan, including Almaty, Nur-Sultan, Aktau, Aktobe, Kyzylorda, Atyrau, and Shymkent. All efforts were made to guarantee the randomness of the data.

The original sample size was determined to be 360, which is adequate for the above-mentioned range according to the Morgan table [191]. The final sample size after a missing data diagnosis was 362 with the following demographic characteristics.

Most of the surveyed respondents were male (72.7%), primarily aged from 42 to 49 years (30.7%) or from 34 to 41 years (26.0%). The majority of them are either married (44.8%) or single (35.0%). They tend to have a high level of education with a graduate degree (42.0%) or an undergraduate degree (33.1%). They usually hold a top/senior (39.2%) ora middle/specialist (33.1%) position. They are mainly from Asian countries, such as China (9.9%), India (3.3%), or the Commonwealth of Independent States (CIS) countries, such as Russia (8.8%), Kyrgyzstan (8.0%), Belarus (4.7%), Ukraine (4.7%), Uzbekistan (4.7), and Georgia (2.2%). The Euro-Asian country Turkey also has a substantial number of representatives (5.2%). The leading European countries that present foreign entrepreneurs are Romania (4.4%), the Netherlands (3.6%), the United Kingdom (3.6%), Italy (3.0%), and Czechoslovakia (2.1%). The principal North American representative is Canada (3.3%). The middle-east countries mainly contributed respondents from the UAE (2.5%). In summary, foreign entrepreneurs from a total of 60 countries participated in this survey.

The survey questionnaires were administered, instruments for the constructs (EI, PA, SNs, PBC, and GS) (see

Table 5. Instruments and reliability of this study.

Construct	Source of Instruments	Cronbach’s Coefficient Alpha *	CR **
EI	Liñán and Chen [163] Solesvik et al. [192]	0.898	0.901
PA	Liñán and Chen [163] Solesvik et al. [192]	0.909	0.913
SNs	Liñán and Chen [163] Solesvik et al. [192]	0.923	0.927
PBC	Solesvik et al. [192] Liñán, Urbano, and Guerrero [177]	0.878	0.880
GS	GEM [49] for government policies GEM [49] for government programs	0.955	0.956

Note: * An alpha value of 0.6 is considered reliable, and a value closer to 1 indicates that the instrument is more reliable and has a higher internal consistency [193]. Nunnally and Bernstein [194] suggest that, for the early stage of basic research, a reliability coefficient of 0.5–0.6 is sufficient, and Hair et al. [195] propose that a Cronbach’s alpha value of 0.6–0.7 should indicate the lower limit of acceptability. Crano and Brewer [196] suggest that the degree of internal consistency is to be considered acceptable if the alpha coefficient is 0.75 or better. Cronbach [197], as well as Hu and Bentler [198], suggest the minimum alpha cut-off point of 0.7 to ensure internal consistency. ** A CR value of 0.7 or above suggests good reliability [195].

and

Table 6. The standardized factor loading of each item on its represented construct.

Construct	Item	Factor Loading
EI	(1) I am ready to do anything to be an entrepreneur.	0.799
	(2) My professional goal is to become an entrepreneur.	0.751
	(3) I will make every effort to start and run my own firm.	0.838
	(4) I am determined to create a firm in the future.	0.749
	(5) I have very seriously thought of starting a firm.	0.749
	(6) I have the firm intention to start a firm some day.	0.771
PA	(1) Being an entrepreneur implies more advantages than disadvantages to me.	0.819
	(2) It is desirable for me to become an entrepreneur.	0.771
	(3) It is attractive for me to become an entrepreneur.	0.923
	(4) If I had the opportunity and resources, I would love to start a business.	0.810
	(5) Being an entrepreneur would give me great satisfaction.	0.745
	(6) Among various options, I would rather be an entrepreneur.	0.704
SNs	(1) My closest family members think that I should pursue a career as an entrepreneur.	0.784
	(2) I do care about what my closest family members think as I decide on whether or not to pursue self-employment.	0.745
	(3) My closest friends think that I should pursue a career as an entrepreneur.	0.836
	(4) I do care about what my closest friends think as I decide on whether or not to pursue self-employment.	0.787
	(5) People that are important to me think I should pursue a career as an entrepreneur.	0.887
	(6) I do care about what people important to me think as I decide on whether or not to pursue self-employment.	0.901
PBC	(1) If I wanted to, I could easily become an entrepreneur.	0.890
	(2) Starting a business and keeping it viable would be easy for me.	0.766
	(3) I am able to control the process of creation of a new business.	0.758
	(4) If I tried to start a new business, I would have a high chance of being successful.	0.714
	(5) I know most of the practical details needed to start a business.	0.721
GS	(1) Government policies (e.g., public procurement) consistently favor new firms.	0.857
	(2) The support for new and growing firms is a high priority for policy at the national government level.	0.922
	(3) The support for new and growing firms is a high priority for policy at the local government level.	0.757
	(4) New firms can get most of the required permits and licenses in about a week.	0.769
	(5) The amount of tax is NOT a burden for new and growing firms.	0.879
	(6) Taxes and other government regulations are applied to new and growing firms in a predictable and consistent way.	0.753
	(7) Coping with government bureaucracy, regulations, and licensing requirements is easy for new and growing firms.	0.720
	(8) A wide range of government assistance for new and growing firms can be obtained through contact with a single agency.	0.756
	(9) Science parks and business incubators provide effective support for new and growing firms.	0.772
	(10) There are an adequate number of government programs for new and growing businesses.	0.791
	(11) The people working for government agencies are competent and effective at supporting new and growing firms.	0.743
	(12) Almost anyone who needs help from a government program for a new or growing business can find what they need.	0.821
	(13) Government programs aimed at supporting new and growing firms are effective.	0.735

Note: All factor loadings should be statistically significant to achieve convergent validity. Kim and Mueller [199] suggest that 0.3 is the minimum level of significance, and above 0.7 is very significant. Comrey and Lee [200] suggest loadings where 0.71 is considered excellent, 0.63 very good, 0.45 fair, and 0.32 poor. A good rule of thumb is that standardized loading estimates should be 0.5 or higher, and ideally 0.7 or higher [195].

) of the questionnaire were borrowed directly from previous studies (with a few wording changes made for GS only), and a seven-point Likert scale (“1”= “Strongly disagree”, “2” = “Disagree”, “3” = “Somewhat disagree”, “4” = “Neutral”, “5” = “Somewhat agree”, “6” = “Agree”, and “7” = “Strongly agree”) was used, while the demographic questions were developed by the authors based on other similar research. The respondents were requested to answer the questions based on the time when they planned to start up their own business in the past because they are already real entrepreneurs in the present.

Pretesting was conducted among a small group of colleagues and friends of the authors after the questionnaire was formed in three languages (English, Russian, and Chinese), and the result was satisfactory. Next, a pilot study was performed among 50 participants who were designated as part of the initial “network” to further confirm the reliability and validity of the items in the questionnaire. In order to reduce biases to the greatest extent, the criteria for the study group included: (1) typical new or established foreign entrepreneurs; (2) diversity in terms of nationality, age, gender, industry, and position; (3) management skills or expertise; and (4) proficiency in the English or Russian language. The reliability (for internal consistency) (see Table 5) was tested through Cronbach’s coefficient alpha and composite reliability (CR), while the validity tests went through face validity and construct validity. Face validity was based on the experts’ judgment on face value of the test items. Construct validity (including convergent validity and discriminant validity) was analyzed by means of confirmatory factor analysis (CFA) to evaluate factor loadings (see Table 6), average variance extracted (AVE) (see

Table 7. Results for AVE, CR, and correlation coefficients in the construct validity tests.

Construct	AVE *	Square Root of AVE	CR **	Correlations
				EI	PA	SNs	PBC	GS
EI	0.604	0.777	0.901	1
PA	0.637	0.798	0.913	0.709	1
SNs	0.681	0.825	0.927	0.295	0.535	1
PBC	0.597	0.773	0.880	0.493	0.726	0.532	1
GS	0.628	0.792	0.956	0.401	0.509	0.160	0.376	1

Note: * Discriminant validity is achieved when the square root of the AVE is greater than the correlation estimates in that a latent construct should explain its item measures better than it explains another construct [195,201,202]. ** Convergent validity is achieved when CR is significant according to the threshold stated in Table 5.

), and CR (see Table 5 and Table 7). The results of the pilot study indicated that the “goodness of the measures” was adequate.

This study adopted personally administered and electronic survey questionnaires due to the constraints caused by COVID-19 as well as the considerations of convenience and efficiency. The reasons for relying on a “networking” approach to collect data include: (1) the alternative sampling frames were proved to be infeasible due to the insufficiency of the statistics; (2) entrepreneurs are bound to be connected through a network in the business community, which can snowball and continually reach new potential respondents given adequate patience; and (3) the authors either work as entrepreneurs or have opportunities to interact with foreign entrepreneurs quite often, which enabled the process to be initiated by identifying and forming the first network. In addition, the authors took every chance to collect data from encounters with individuals, new acquaintances, etc. who were identified as new or established foreign entrepreneurs based in Kazakhstan. The entire process of data collection lasted for half a year, from 1 August 2020 to 31 January 2021.

4.2. Data Examination

We went through a data examination process to evaluate the impact of missing data, identify outliers, and test for assumptions in order to ensure that the “goodness of the data” was adequate before applying the multivariate data analysis.

4.2.1. Missing Data Analysis

Missing data were diagnosed in terms of their extent and randomness and remedied with a proper imputation method. Among the initially collected 398 questionnaires, (1) the ratio of questionnaires from uncertain or non-new-or-established foreign entrepreneurs (16 copies) was 4.0%; (2) the ratio of omitted or redundant responses (20 copies) was 5.0%; (3) no variable or case had missing data with a percentage of more than 10%; most of these inappropriate questionnaires had a high proportion of the responses missing just one question, and some missed two to three questions at most; and (4) the missing data were found completely at random (MCAR). Based on the above situation (less than 10% of the responses being missing data cases and MCAR), this study applied the complete case approach, i.e., “imputation using only valid data” [195], to retain only those observations with complete data. Luckily, the final sample (362 cases) is sufficiently large to satisfy the designed sample size of 360.

4.2.2. Detection of Outliers

Firstly, univariate boxplots were examined for single variables (EI, PA, SNs, PBC, and GS). No values were denoted with an asterisk (*) as extremes, which are more than 3 interquartile ranges (IQRs) from the end of a box, or labeled as outliers (o), which are between 1.5 and 3 IQRs from the end of a box [195], in the boxplots. Bivariate scatterplots for pairs of variables revealed no presence of substantial outliers that fall outside of the ordinary range for the 95% confidence interval [195] in each bivariate relationship. The Mahalanobis distance (significance level: 0.001) [203] for the entire set of variables identified no case of multivariate outliers. Finally, detection of outliers over the TPB regression results was performed by means of a standardized residuals plot and partial regression plots. No standardized residuals with absolute values greater than 2.0 (the t value is 1.96 for α = 0.05) [195] were observed, which means that no influential outliers were identified from the regression results. We detected no substantial outliers in the standardized partial regression plots between the dependent variable (DV: EI) and the independent variables (IVs: PA, SNs, and PBC) with one upper line and one lower line framing the 95% confidence interval in each graph.

4.2.3. Tests of Assumptions

Tests of univariate and multivariate assumptions cover the aspects of normality, constant variance (homoscedasticity), linearity, and absence of correlated errors (independence) [195]. The results of graphical analyses (i.e., histograms, scatterplots or a scatterplot matrix, partial regression plots, residual plots, and normal probability plots) and statistical analyses (i.e., skewness, kurtosis, KS test, SW test, Levene’s test, and the Durbin–Watson value) indicate that: (1) there is a very slight non-normality (e.g., see

Table 8. Kolmogorov–Smirnov (KS) and Shapiro–Wilk (SW) tests of normality.

Construct	Kolmogorov–Smirnov * Statistic	df	Sig.	Shapiro–Wilk Statistic	df	Sig.
EI	0.093	362	0.000	0.971	362	0.000
PA	0.131	362	0.000	0.970	362	0.000
SNs	0.110	362	0.000	0.979	362	0.000
PBC	0.112	362	0.000	0.972	362	0.000
GS	0.118	362	0.000	0.969	362	0.000

Note: * Lilliefors significance correction. All variables are statistically non-normal (p < 0.05) according to Ho and Yu [204].

and

Table 9. Skewness and Kurtosis statistics of normality.

Construct	N	Skewness		Kurtosis
	Statistic	Statistic	Std. Error	Statistic	Std. Error
EI	362	−0.166	0.128	−0.886	0.256
PA	362	−0.194	0.128	−0.568	0.256
SNs	362	0.084	0.128	−0.695	0.256
PBC	362	0.268	0.128	−0.658	0.256
GS	362	−0.125	0.128	−0.429	0.256

Note: All the values fall within the threshold of ±2 [205], revealing a very slight non-normality.

), whose detracting effect on the results can be compensated for by the large sample size (N = 362) of this study according to Hair et al. [195]; (2) the homoscedasticity of the variance is met (e.g., see

Table 10. Levene’s test of homogeneity.

Constructs	Levene Statistic	df1	df2	Sig.
EI vs. PA	0.946	21	340	0.531
EI vs. SNs	2.395	24	333	0.052
EI vs. PBC	1.479	21	338	0.082

Note: No unequal variances are present because all the values are nonsignificant at p < 0.05.

); (3) the relationships of linearity among the DV (EI) and IVs (PA, SNs, and PBC) are evidently present (e.g., see Figure 2 where the right three bivariate graphs in the first upper row—except for the left graph of the histogram—all exhibit an obvious shape of a straight thick line with all dots evenly distributed along the middle of this line, typical of linear relationships); and (4) no correlation between the error terms was found as, e.g., the Durbin–Watson value is 1.917 in comparison with the standard value for the independency of the observations (1.5–2.5) [164]. Thus, all the assumptions for each single variable and the entire set of variables are adequately supported.

4.3. Data Analysis

This study applied descriptive statistics (descriptive analysis) and inferential statistics (Pearson’s correlation and multiple regression) with the aim of assessing the “goodness of the model’s fit”. These techniques have frequently been adopted by prior researchers using the TPB model, such as Tkachev and Kolvereid [179], Autio et al. [162], Souitaris et al. [206], Gird and Bagraim [207], and Kibler et al. [208].

Descriptive analysis was used to describe variables, focusing on the central tendency and the dispersion. Pearson’s correlation was employed to determine the association between any two of the variables (including interaction terms) (EI, PA, SNs, PBC, PA × GS, SNs × GS, and PBC × GS) in the extended TPB model. The association identifies whether there is any relationship (in terms of strength and direction) between each pair of variables. In order to further test the hypotheses of the entire extended model (“TPB + Moderator”), we employed the multiple regression technique, which includes simultaneous and hierarchical multiple regression analyses.

5. Results and Findings

This section presents the respective empirical results yielded from the three analytical techniques using SPSS 18.0, which lead to the primary research findings of this study.

5.1. Descriptive Analysis

The descriptive statistics (

Table 11. Descriptive statistics for each construct and the items that comprise it (N = 362; seven-point scales).

Construct	Range		Mean	Std. Deviation	Mode	Percentiles
	Min	Max				25th	50th (Median)	75th
EI	4.00	6.83	5.52	0.639	5.67	5.00	5.58	6.00
(1)	3	7	5.22	0.772	5	5	5	6
(2)	3	7	5.38	1.025	6	5	5	6
(3)	3	7	5.41	0.818	5	5	5	6
(4)	3	7	5.39	0.884	6	5	5	6
(5)	4	7	5.86	0.720	6	5	6	6
(6)	4	7	5.84	0.755	6	5	6	6
PA	4.33	6.67	5.49	0.531	5.67	5.00	5.50	5.83
(1)	3	7	5.00	0.829	5	4	5	6
(2)	4	7	5.78	0.682	6	5	6	6
(3)	4	7	5.74	0.708	6	5	6	6
(4)	4	7	5.86	0.730	6	5	6	6
(5)	4	7	5.26	0.639	5	5	5	6
(6)	3	7	5.30	0.701	5	5	5	6
SNs	3.00	6.50	4.74	0.706	4.50	4.17	4.67	5.33
(1)	3	7	4.72	0.829	5	4	5	5
(2)	2	7	4.70	0.820	5	4	5	5
(3)	3	7	4.81	0.872	5	4	5	5
(4)	2	7	4.69	0.849	4	4	5	5
(5)	3	7	4.80	0.907	5	4	5	5
(6)	2	7	4.71	0.885	4	4	5	5
PBC	3.40	6.20	4.77	0.602	4.40	4.40	4.70	5.20
(1)	2	6	4.56	0.814	4	4	5	5
(2)	2	7	4.35	0.861	4	4	4	5
(3)	3	7	4.58	0.752	4	4	4	5
(4)	3	7	5.45	0.713	5	5	5	6
(5)	2	7	4.92	0.838	5	4	5	5
GS	1.46	5.54	3.59	0.929	3.77	2.92	3.77	4.08
(1)	1	7	3.97	1.302	5	3	4	5
(2)	2	7	4.12	1.270	5	3	4	5
(3)	1	7	4.07	1.258	5	3	4	5
(4)	1	7	4.02	1.154	4	3	4	5
(5)	1	6	3.46	1.114	4	3	4	4
(6)	1	7	3.97	1.136	4	3	4	5
(7)	1	7	3.24	1.199	3	2	3	4
(8)	1	6	3.37	1.043	3	3	3	4
(9)	1	7	3.48	1.109	3	3	3	4
(10)	1	6	3.46	1.047	3	3	3	4
(11)	1	6	3.22	0.997	3	3	3	4
(12)	1	5	2.90	0.927	3	2	3	3
(13)	1	6	3.40	1.146	3	3	3	4

Note: The values for each construct are the aggregates computed from its corresponding items.

) report the range, mean, standard deviation, mode, and percentiles for each construct.

As can be seen from Table 11, on average, the surveyed new and established foreign entrepreneurs have a fairly strong entrepreneurial intention (EI) (Mean = 5.52), ranging from a moderate level (Min = 4.00) to an extremely high level (Max = 6.83), to start up a business. No individuals provided less than a moderate score for each item, and most frequently they strongly agreed with becoming an entrepreneur (Mode = 5.67).

The surveyed entrepreneurs tended to have a very positive personal attitude (PA) toward entrepreneurship (Mean = 5.49). Similarly to the extent of their intention, nobody showed less than a moderate attitude (Min = 4.33) and, in most cases, they expressed a strong desire to start a business (Mode = 5.67).

The surveyed entrepreneurs generally scored moderately on subjective norms (SNs) (Mean = 4.74). Some of them cared little about others’ opinion on self-employment or received less recognition (Min = 3.00), while some others did show a lot of attention to the recognition and suggestions from individuals important to them (Max = 6.50). Most frequently, they were substantially influenced by those important to them (Mode = 4.50).

Collectively speaking and most often, the surveyed entrepreneurs were rather modest about their perceived behavioral control (PBC) (Mean = 4.77; Mode = 4.40). Before deciding to create their business, a lot of them did not believe they could easily have success afterwards (Min = 3.40), while some others were very confident about their future entrepreneurship (Max = 6.20).

The level of government support (GS) was mediocre in the overall opinion of most of the surveyed entrepreneurs (Mean = 3.59; Mode = 3.77). Some of them gave very low scores on GS (Min = 1.46), while some others had a positive assessment of GS (Max = 5.54). Individually, the entrepreneurs offered extreme scores at two ends (1 and 7) for GS, indicating they have very different opinions on the current conditions of GS in Kazakhstan.

As noted, in general, the scores of the Modes, Means, and Medians for each construct do not substantially differ, which indicates that the most frequently occurring responses represent both the average level of the perceptions of the entire sample and the middle values in magnitude.

5.2. Pearson’s Correlation Analysis

The results of the Pearson’s correlation analysis are given in the correlation matrix (

Table 12. Pearson’s correlation matrix.

Variables	1	2	3	4	5	6	7
1. EI	1
2. PA	0.750	1
3. SNs	0.594	0.633	1
4. PBC	0.665	0.639	0.680	1
5. PA × GS	0.473	0.541	0.351	0.490	1
6. SNs × GS	0.362	0.396	0.629	0.529	0.877	1
7. PBC × GS	0.350	0.335	0.345	0.702	0.890	0.872	1

Note: Correlations are significant at the 0.001 level (p < 0.001) (two-tailed).

), which shows that EI is positively and significantly correlated with PA, SNs, and PBC from the basic TPB model and with PA × GS, SNs × GS, and PBC × GS in the extended model. The correlation coefficients range from 0.594 to 0.750 in the TPB model, and from 0.335 to 0.890 in the extended model. The three variables of the TPB model (PA, SNs, and PBC) are significantly intercorrelated with each other, with the correlation coefficients ranging from 0.639 to 0.680. The three moderation terms (PA × GS, SNs × GS, and PBC × GS) are also significantly intercorrelated, with coefficients ranging from 0.335 to 0.890.

The bivariate correlations imply the possible existence of predictive effects on EI from the TPB components and the moderation terms, whose significance values were subsequently assessed through a multiple regression analysis.

5.3. Multiple Regression Analysis

This study employed a simultaneous regression analysis for the “TPB” model and a hierarchical regression analysis for the “TPB + Moderator” model.

Table 13. Results of simultaneous regression on the “TPB” model (DV = EI).

Overall Model Fit
Multiple R									0.790
Coefficient of Determination (R2)									0.624
Adjusted R2									0.620
Standard error of the estimate									0.394
Analysis of variance
	Sum of squares			df		Mean square			F	Sig.
Regression	91.890			3		30.630			197.657	0.000
Residual	55.477			358		0.155
Total	147.367			361
Regression coefficients, correlations, and collinearity statistics
Variables	Regression coefficients			Statistical significance		Correlations			Collinearity statistics
	B	Std. error	Beta	t	Sig.	Zero-order	Partial	Part	Tolerance	VIF
(Constant)	0.318	0.075		4.184	0.000
PA	0.632	0.054	0.526	11.672	0.000	0.750	0.525	0.379	0.518	1.930
SNs	0.064	0.043	0.070	1.488	0.138	0.594	0.078	0.048	0.471	2.122
PBC	0.299	0.051	0.281	5.915	0.000	0.665	0.298	0.192	0.465	2.152

Note: N = 362; EI: entrepreneurial intention; PA: personal attitude toward entrepreneurship; SNs: subjective norms; PBC: perceived behavioral control; GS: government support.

and

Table 14. Results of hierarchical regression on the “TPB + Moderator” model (DV = EI).

Overall Model Fit
Multiple R									0.868
Coefficient of Determination (R2)									0.753
Adjusted R2									0.749
Standard error of the estimate									0.320
Analysis of variance
	Sum of squares			df		Mean square			F	Sig.
Regression	110.991			6		18.498			180.528	0.000
Residual	36.376			355		0.102
Total	147.367			361
Regression coefficients, correlations, and collinearity statistics
Variables	Regression coefficients			Statistical significance		Correlations			Collinearity statistics
	B	Std. error	Beta	t	Sig.	Zero-order	Partial	Part	Tolerance	VIF
Block 1
(Constant)	0.224	0.065		3.413	0.001
PA	0.514	0.133	0.506	3.845	0.000	0.750	0.518	0.414	0.593	1.687
SNs	0.023	0.138	0.025	0.169	0.866	0.594	0.009	0.006	0.586	1.705
PBC	0.253	0.087	0.240	2.888	0.004	0.665	0.245	0.209	0.626	1.598
Block 2
PA × GS	0.026	0.006	0.151	4.110	0.000	0.473	0.152	0.127	0.314	3.182
SNs × GS	−0.013	0.041	−0.082	−0.310	0.757	0.362	−0.016	−0.010	0.782	1.278
PBC × GS	0.022	0.008	0.138	2.669	0.008	0.350	0.131	0.109	0.401	2.495

Note: N = 362; EI: entrepreneurial intention; PA: personal attitude toward entrepreneurship; SNs: subjective norms; PBC: perceived behavioral control; GS: government support; PA×GS: moderating effect of PA and GS on EI; SNs×GS: moderating effect of SNs and GS on EI; PBC×GS: moderating effect of PBC and GS on EI.

, respectively, report the regression results in detail.

5.3.1. Assessing the Overall Model Fit

The values of R², adjusted R², and the F ratio are 0.624, 0.620, and 197.657, respectively, in the “TPB” model. The F ratio is substantially large at the p < 0.001 level; hence, the overall TPB model is statistically significant. TPB explains 62.4% of the variance in predicting and estimating entrepreneurial intention (EI).

The respective values of R², adjusted R², and the F ratio are 0.753, 0.749, and 180.528, in the extended “TPB + Moderator” model. The F ratio indicates that the extended model fit is also statistically significant at the p < 0.001 level. The extended model explains 75.3% of the variance in intention with a large improvement of 12.9% in the adjusted R² attributable to the moderating effect, and the standard error of the estimate decreased from 0.394 in the basic model to 0.320 in the new model, which means an improvement in the overall model fit was achieved. It is evidenced that the role of government support (GS) plays an important role in facilitating EI.

The significance level was set at 0.05 for the multivariate analysis in this study, and the t value was used to determine the significance of regression coefficients. As such, the regression results of the “TPB” model indicate that the regression coefficient (termed the ‘b coefficient’) of the constant, as well as the b and beta (β) coefficients of PA and PBC, are all statistically significant at the p < 0.001 level. The “TPB + Moderator” model also has a significant regression coefficient (b) for the constant (p < 0.01), as well as significant b and β coefficients for PA (p < 0.001) and PBC (p < 0.01). Additionally, in the extended model, the interaction terms PA × GS and PBC × GS are significant at the p < 0.001 and p < 0.01 levels, respectively. The coefficients of SNs are, however, insignificant (p > 0.05) in both models, and the other interaction term (SNs × GS) is insignificant (p > 0.05) in the extended model.

5.3.2. Assessing the Variables’ Importance

Relatively speaking, the regression coefficients (b) indicate that: (1) in both models, PA exerts the greatest predictive power over EI, followed by PBC as the second greatest predictor; and (2) in the extended model, the importance of PBC is further followed by the interaction effect of GS with PA and the moderating effect of GS on PBC, respectively. Comparatively, across the two models, the standardized β coefficients reveal the same finding that the relative strength of the variables is ranked (in descending order) as follows: PA, PBC, PA × GS, and PBC × GS.

5.3.3. Identifying Multicollinearity

Highly multicollinear variables can distort the results substantially or make the results quite unstable and thus not generalizable [195]. Hence, this study examined the possible existence of multicollinearity among the independent variables (IVs).

An examination of the correlation matrix among the IVs is the simplest and most obvious means of identifying multicollinearity. As shown in the Pearson’s correlation matrix (Table 12), among the set of IVs—TPB components and moderation terms, i.e., PA, SNs, PBC, PA × GS, SNs × GS, and PBC × GS—the correlations fall between 0.335 and 0.890; thus, no presence of high correlations with values of 0.9 or greater is found [195]. Hence, there is no indication of substantial multicollinearity among the IVs in this study.

In addition, the tolerance and VIF values were also used to test the impact of multicollinearity among the IVs (PA, SNs, PBC, PA × GS, SNs × GS, and PBC × GS). In the extended model, the tolerance values (see Table 13 and Table 14) range from 0.314 to 0.626, while the VIF values range from 1.598 to 3.182. All tolerance values are greater than 0.2 [195], and all VIF values are less than 4.0 [195], indicating that the presence of multicollinearity is not problematic. Similarly, no multicollinearity is present in the basic TPB model in comparison with the criteria [195].

5.3.4. Examining Correlations

According to the correlations displayed in Table 13 and Table 14, in both models, PA has the closest bivariate relationship with EI, PBC has the second-closest bivariate relationship with EI, followed by SNs in association with EI. In the extended model, the moderating effects of GS through PA, SNs, and PBC (in descending order) on EI are the lowest among all the bivariate correlations. All independent variables, except for SNs and SNs×GS, have substantial incremental predictive power in the model(s). In the “TPB” model, the unique variance is significantly derived from PA (0.379² = 14.36%) and PBC (0.192² = 3.69%); thus, 44.35% (62.4%–14.36%–3.69%) of the remaining variance is explained by both PA and PBC together. In the “TPB + Moderator” model, PA and PBC contribute a significant unique variance of 17.14% and 4.39%, respectively, while PA × GS and SNs × GS represent 1.61% and 1.19% of the unique variance, respectively; 50.97% of the common variance is shared among the four IVs.

5.3.5. Validating the Results

The primary concern of this process is to ensure that the results are generalizable to the target population and are not specific to the sample used in the estimation. The most direct approach to validation is to obtain another sample from the population and assess the correspondence of the results from the two samples [195]. However, due to the time and cost constraints, the limited access to respondents, the inconvenience caused by COVID-19, etc., a new sample was unable to be drawn. Hence, the authors alternatively relied on: (1) the assessment of the adjusted R²; and (2) the comparison of split samples.

Examining the adjusted R² values reveals little loss of predictive power when compared with the R² values (0.624 versus 0.620 in Table 13; 0.753 versus 0.749 in Table 14), which indicates a lack of overfitting that would be shown by a more marked difference between the two values.

The study created in a random manner two subsamples of equal size (N = 181 for each) from the original sample. Comparison of the overall model fit in the subsamples for each model demonstrates a high level of similarity of the results in terms of R², adjusted R², and the standard error of the estimate. Yet, the overall predictive power decreases in both models due to the reduction of the sample size (from 362 to 181). After comparing the individual coefficients between the two subsamples, as well as with the previous regression results for each model, no substantial differences appeared; the significance and relative weight reflected in the coefficients remained in the same pattern.

5.3.6. Validation of the Hypotheses

Based on the above analyses, the hypotheses proposed were well examined and the findings are stated as follows:

(1)

Personal attitude (PA) toward entrepreneurial behavior positively influences entrepreneurial intention (EI) among foreigners in Kazakhstan.

(2)

Subjective norms (SNs) insignificantly influence entrepreneurial intention (EI) among foreigners in Kazakhstan.

(3)

Perceived behavioral control (PBC) positively influences entrepreneurial intention (EI) among foreigners in Kazakhstan.

(4)

With the moderation of government support (GS), the relationship between personal attitude (PA) toward entrepreneurial behavior and entrepreneurial intention (EI) among foreigners in Kazakhstan strengthens.

(5)

With the moderation of government support (GS), the relationship between subjective norms (SNs) and entrepreneurial intention (EI) among foreigners in Kazakhstan is insignificant.

(6)

With the moderation of government support (GS), the relationship between perceived behavioral control (PBC) and entrepreneurial intention (EI) among foreigners in Kazakhstan strengthens.

6. Discussions

This section discusses the research findings in comparison with prior studies in similar fields.

Prior studies indicate that the generic TPB model explains 21–55% of the variance in intention, varying from study to study [155,160,162,163,164]. Distinctively, Kautonen et al. [164] found that TPB components jointly explain 59% of the variation in intention. Hence, the high variances (62.4% and 75.3%) obtained in this study are still consistent with past research. This study posits that the markedly improved explanation provided by the extended model is due to the characteristics of the sample. The new and established foreign entrepreneurs have already translated their intention into actual entrepreneurial behavior with 100% probability. This is against the generic 30–39% of the variance in behavior accounted for by the intention [160,161] in previous studies. Therefore, the target population is deemed to have a more affirmative and stronger intention compared with the general foreigners if otherwise surveyed.

The results of this study show that PA is a more powerful determinant of EI than PBC. The relative importance of PA and PBC is consistent with some of the past research that found that PA is the most important antecedent of self-employment intentions in the TPB model [85]. The results also reveal that the higher the level of GS that exists and/or is perceived, the stronger the predictive power of PA and PBC with respect to EI will be. The discovery of the moderating effects of GS is the novelty of this study.

The result about the insignificance of SNs is still consistent with past research in that, while numerous studies have found support for SNs as strong predictors of EI [157,175,179], some authors have controversially discovered that SNs have little explanatory power (e.g., Ajzen [57]; Armitage and Conner [160]) or are even insignificant with respect to EI (e.g., Krueger et al. [73]; Liñán and Chen [163]; Tarek [182]). This study argues that there may be several reasons for the non-significance of SNs and SNs × GS in the models:

(1)

The target population or the sample itself is too broad, with the cultural or contextual variety of 60 countries. Hence, either normative beliefs that are too divergent or heterogeneous compliance with beliefs formed among the foreigners. In other words, a uniform pattern of SNs was unable to be formulated to influence their EI.

(2)

The impact of SNs on the foreigners weakly facilitates overseas entrepreneurship in comparison with their PA and PBC.

(3)

The role of GS may not impact upon the opinions of important others or the entrepreneur’s reaction to these opinions. GS functions mainly to motivate the entrepreneur’s PA and PBC toward his/her business start-up.

(4)

The perceived level of GS in Kazakhstan is relatively low, which limited its interaction effect with SNs.

(5)

The high uncertainty and risks incurred by the outbreak of COVID-19 might change the general opinions about self-employment, to say nothing of going abroad to create ventures, especially for the new entrepreneurs in this survey.

7. Implications for Academics, Entrepreneurs, and Politicians

Based on the findings of this study, any adopted mechanism that can influence PA and PBC could also foster EI among foreigners. Apparently, entrepreneurial education and training programs provided by universities or institutions in Kazakhstan are deemed to be indispensable to promoting the PA and PBC of potential entrepreneurs (students and trainees), including foreign students or generic foreigners who attend the courses. Therefore, faculties and academics play a vital role in identifying, motivating, and nurturing potential entrepreneurs, including the foreign ones in this country. We suggest that training platforms regularly invite successful entrepreneurs, trade union leaders, or officials from government agencies to deliver business talks in their classes. These talks could provide potential entrepreneurs with confidence in opportunities and successes, as well as motivate their PA to “dream creatively” and “think big” toward actions to do business in Kazakhstan. Developing the ambitions and talents of potential entrepreneurs through entrepreneurship education enables the potential to visualize and evaluate opportunities [209,210].

Vice versa, practicing entrepreneurs could also resort to the education and training programs to improve their entrepreneurial knowledge and skills and to government agencies for the required support. Through the expansion of their insights into entrepreneurial activity and obtaining adequate government support, entrepreneurs could further enhance their performance in business.

The positive role of GS was empirically evidenced to enhance foreign EI in Kazakhstan. Therefore, we suggest that policy-makers further improve the level of GS, which may facilitate sustainable entrepreneurial development in Kazakhstan. In doing so, the authors propose that the measures shall contribute to overcoming the major constraints of current GS conditions as previously discussed in this paper. The constraints include corruption, bureaucracy, red tape, insufficient entrepreneurial equity finance, entrepreneurial education and training, R&D transfer, the physical infrastructure, market dynamics, and cultural and social norms. The measures may include, among other things, strengthening the social values on entrepreneurship, supporting entrepreneurial innovation and competitiveness, and increasing economic diversification through entrepreneurship. We especially recommend that the general entrepreneurial ecosystem (EE) be improved. An EE can be defined in various ways, yet one of the most important elements is the interaction among the stakeholders, such as entrepreneurs, government, organizations, and institutions [211]. Therefore, building a solid network of interaction should be prioritized in the agenda. More in-depth discussion on the EE is beyond the scope of this paper.

COVID-19 will reshape entrepreneurship in the long term [34]. Academics, entrepreneurs, and politicians should cooperate interactively and complete each other in the new circumstances. Academics should continue to explore how the pandemic affects the attitudes toward and perceptions of entrepreneurship and provide new insights into the factors that influence EI. Entrepreneurs should refer to the research findings and strive to succeed in the adverse environment. In turn, their practice provides down-to-earth evidence for academics to further conduct case studies. Politicians should integrate the achievements of academics and entrepreneurs in effective policy designs and implementations that promote entrepreneurial intention and behaviors. The government should also encourage further academic research in this area.

8. Conclusions

This study embraced the “TPB + Moderator” model and primarily concluded that foreign entrepreneurs in Kazakhstan started up their business with a very strong entrepreneurial intention (EI), a positive attitude (PA) toward entrepreneurship, a moderate level of compliance with subjective norms (SNs), and modest perceived behavioral control (PBC). They generally perceive a medium level of government support (GS) in Kazakhstan, while the individual perceptions are very unevenly distributed. Their EI is closely and positively correlated with PA, SNs, PBC, and the moderating effects of PA × GS, SNs × GS, and PBC×GS. Their PA toward entrepreneurship and PBC positively influence their EI. With the moderation of GS, the relationship between PA, PBC, and EI strengthens. Unexpectedly, the effects of SNs and SNs × GS did not show a significant influence on EI. The predictive powers are ranked in descending order as PA, PBC, GS × PA, and GS × PBC.

8.1. Major Contributions of This Study

This study contributes to the body of knowledge related to EI research mainly in theoretical, methodological, and practical areas summarized below.

From the theoretical viewpoint, this study filled the research void in terms of the target population (foreigners in Kazakhstan), the research framework (the extended TPB model), the research context (Kazakhstan during the COVID-19 pandemic), and the highlighted role of government support (GS). Specifically speaking, there was hardly any previous research on entrepreneurial intention (EI) among foreigners in Kazakhstan. Moreover, this study integrated the theory of planned behavior (TPB) with the moderating effect of GS in developing the research framework. This conceptual novelty was based on the previous calls to understand the direct effects and moderating effects on EI [173,212,213]. Further, GS is an original research subject in the setting under study. The extended TPB model expanded our knowledge of factors that influence EI in general and among foreigners in Kazakhstan in particular. In addition, COVID-19 has added complexity to EI research. Finally, the critical review and classification of individual factors (see Table 3), as well as the eleven established intention-based models (see Table 4), brought together into one piece of work could offer future researchers an accessible study in this domain and serve as another modest contribution to a holistic understanding of the factors influencing EI.

As for the methodology, this study incorporated and synthesized the methods employed by other researchers in past similar studies. The aim is to systematically apply a complete process of data collection, data examination, and data analysis appropriate for the specific research model of “TPB + Moderator (GS)”. The research route formed in this study may contribute to similar research, especially EI research in the context of Kazakhstan in the future.

Practically speaking, this study makes three primary contributions: (1) validation of the direct factors (PA and PBC) influencing EI among the foreigners in Kazakhstan; (2) the vital role of GS in enhancing EI through moderating effects (PA × GS and PBC × GS); and (3) insights into the direction and magnitude of the direct and moderating effects on EI. Any mechanism that aims to improve PA, PBC, and GS needs to encourage EI in Kazakhstan; hence, universities (academics), theorists (researchers), practicing entrepreneurs (practitioners), and politicians (policy-makers) should accordingly contribute to society by means of their daily commitment to nurturing newcomers, providing new insights, sharing and increasing experiences, and improving the entrepreneurial ecosystem (EE).

8.2. Limitations of This Study

The insufficiencies and deficiencies found during the process of carrying out this study are mainly reflected in various aspects of the data collection.

First, the study aimed to identify the factors influencing EI among generic foreigners in Kazakhstan; however, in order to ensure the confirmative presence of EI, new and established foreign entrepreneurs were designated as the target population. Non-entrepreneurs and other types of entrepreneurs (potential, intentional, nascent, and discontinued entrepreneurs) were excluded.

Second, the size of the target population was unable to be accurately determined due to the severe scarcity of relevant statistics; therefore, only an estimation could be obtained from the limited information.

Third, the sample dataset was gathered from a total of 60 countries, which raises the question of whether it is truly representative of the target population. In fact, the nature of the target population (new and established foreign entrepreneurs) might inherently determine the extent of generalizability due to its substantial diversity.

Fourth, in a stricter sense, the data collected through “networking” might lack adequate randomness because the initial “circle of friends” was, to some degree, confined to a relatively restricted range.

Fifth, the accuracy of the responses was suspect. As this study used questionnaire items from previous research in countries other than Kazakhstan, the authors could not eliminate the chance that the participants in this study could misunderstand some of the questions, although the questionnaire went through pretesting and a pilot study.

Lastly, the study planned to collect cross-sectional data. However, due to COVID-19, and the infeasibility of alternative sampling frames, the data were collected during a long period of six months under unpredictable pandemic conditions. This added a longitudinal nature to the data. As COVID-19 continued to spread, the dynamics of perceptions of entrepreneurial activities, especially among new entrepreneurs, could have affected the consistency of their responses.

All the above-mentioned limitations might cause bias in the results, which in turn would yield ungeneralizable findings.

8.3. Recommendations for Further Research

In order to attract more attention to foreign entrepreneurship in the context of Kazakhstan, the authors propose the following recommendations.

First, the integrated conceptual model of “TPB + Moderator” proffered in this study provides a guideline for future researchers in the subject area of entrepreneurial intention (EI) and factors that influence this intention among foreigners in the context of Kazakhstan. Future research can adopt this proposed model to further test its plausibility in general and in Kazakhstan in particular.

Second, the role of government support (GS)—the moderator in the conceptual model—was defined in terms of government policies and government programs. Future research can add more connotations to the definition of GS and develop more comprehensive measures in its operationalization.

Third, this study did not differentiate between foreigners in terms of their citizenship or country of origin; instead, the study took them as a general population for the purposes of simplicity. Hence, future research may consider the differences among foreigners from different countries or regions and collect data separately according to their nationality or country of origin. This separation could enable the researcher to obtain specific research findings for each different group of foreigners. In addition, foreigners may also be further classified according to the nature of their entrepreneurial experience, such as entrepreneurs versus non-entrepreneurs and new and established entrepreneurs (as in this study) versus potential, intentional, nascent, and discontinued entrepreneurs.

Fourth, as all alternative forms of access failed, this study relied upon the “networking” approach to collect data. This approach was not the prioritized one but was the only feasible one under the circumstances during the study period. Future research should strive to collect data through direct contacts with foreign SMEs or individuals based on adequate statistics. They could be obtained through more productive communications with the state bodies of Kazakhstan, perhaps together with the official assistance of universities, embassies, consulates, or business associations.

Fifth, this study only carried out quantitative analyses, which are unable to avoid the probability of the collection of inaccurate responses caused by misinterpretations during the questionnaire survey’s completion. Moreover, the quantitative analyses could not derive sufficient information from the attitudinal answers of the participants. Hence, future research may add qualitative analyses (e.g., focus group interviews) to achieve more in-depth insights into the influencing factors and their underlying rationales.

Sixth, this study employed descriptive analysis, Pearson’s correlation analysis, and multiple regression analysis as the main analysis techniques. Other types of analysis, such as structural equation modeling (SEM), meta-analysis, analysis of variance (ANOVA), exploratory factor analysis (EFA), and confirmative factor analysis (CFA), are also recommended for use in further research, which is expected to provide new insights into the subject area of EI in the context of Kazakhstan.

Lastly, the impacts of COVID-19 on economic and business activities, including entrepreneurship and SMEs both worldwide and in Kazakhstan, are ongoing. The uncertainty and risks caused by COVID-19 have brought new features to the research in various areas. This requires future researchers, including those focusing on EI research, to reassess their assumptions and uncover new findings beneficial to the relevant individuals and societies. Kazakhstan is definitely calling for more empirical studies (including empirical EI research) as well.