The Value of a Happy Population for Relative Engagement in Vertical-Scaling and Horizontal-Scaling Entrepreneurship

Abstract

We investigate the impact of a country’s population happiness level and happiness inequality on the level of entrepreneurship engagement pursued by individual entrepreneurs in that country. Entrepreneurship engagement broadly captures ambitions of entrepreneurs. It is measured along two dimensions: vertical-scaling entrepreneurship (newness-focused scale-up entrepreneurship) and horizontal-scaling entrepreneurship (expansion-focused scale-out entrepreneurship). Adopting the lens of supply and demand theory and occupational choice theory, we argue that a country’s happiness and happiness inequality levels are differently related to these two dimensions. We employ a sample of 71,964 early-stage (nascent or new) entrepreneurs from 79 countries, using the Global Entrepreneurship Monitor dataset. We construct new ordinal scales to measure individuals’ engagement levels in vertical-scaling and horizontal-scaling entrepreneurship. Country-level happiness and happiness inequality data are drawn from the Gallup World Poll (GWP) database. We estimate a pooled ordered logit model to explain individual engagement levels in vertical- and horizontal-scaling entrepreneurship. Explanatory variables include the two country-level happiness indicators plus a set of control variables. We find that country-level happiness significantly increases the likelihood of entrepreneurs within that country to pursue high-end entrepreneurship on the vertical-scaling dimension. At the same time, it decreases the likelihood that they will pursue high-end entrepreneurship on the horizontal-scaling dimension. On the contrary, country happiness inequality increases the likelihood of entrepreneurs’ pursuit of high-end entrepreneurship on the horizontal-scaling dimension while decreasing the likelihood of their pursuit of high-end entrepreneurship on the vertical-scaling dimension. In short, population happiness pushes entrepreneurs toward innovativeness but away from expansion, while happiness inequality does the opposite. This study contributes to the literature on psychological entrepreneurship by bringing the contextual influence of happiness into the dialog of entrepreneurship engagement. Our study also contributes to the high-quality entrepreneurship dialog by decoupling the growth perspective into two dimensions of entrepreneurship: vertical scaling and horizontal scaling.

1. Introduction

Within the past several decades, happiness or affective well-being has attracted considerable attention from scholars, practitioners, organizations, and governments across the globe. Happiness is a positive mental state characterized by frequent experience of positive affects and infrequent experience of negative affects [1,2]. It leads to a wide spectrum of benefits for individuals and societies across a variety of cultural and national backgrounds [3]. The entrepreneurship field, in recent years, has become a new and welcoming home for the ongoing interest in happiness studies [4]. Notable efforts have been made to link happiness with numerous entrepreneurial outcomes.

The surge in these studies is due to numerous reasons. To begin with, studying happiness or affective well-being captures the new trend in the psychological entrepreneurship field where emotions are studied for a better understanding of entrepreneurial choices and processes. This is especially relevant in an unpredictable and uncertain environment [5,6]. Moreover, happiness might be another factor motivating start-up entry beyond any external pecuniary concerns [7,8]. Hence, apart from economic and financial rewards, the expected experience of happiness associated with the act of entrepreneurship is likely to be another key factor influencing entrepreneurial behavior [9,10].

The current study aims to investigate the association between a country’s happiness and an individual entrepreneur’s engagement with high-end (i.e., high quality) entrepreneurship relative to that of the lower end, within the country’s pool of entrepreneurs. In this regard, a country’s happiness is conceptualized in terms of population happiness level and happiness inequality. Population happiness may influence individual entrepreneurial decisions and ambitions as it captures the affective well-being of stakeholders around the entrepreneur, in particular, their employees and customers. As an example, because positive affects at work have been found to improve creativity and cognitive functioning [11,12], a greater supply of happy employees in the labor force will make it easier for entrepreneurs to attract creative employees. In turn, more creative employees in a business may stimulate entrepreneurial ambitions related to innovation. Finally, at the macro level, innovation is a key determinant of economic growth, illustrating the potential economic importance of population happiness.

In our study, we conceptualize “entrepreneurial quality” in terms of the entrepreneurs’ ambition/aspiration levels [13]. Specifically, we classify entrepreneurial activity into two dimensions—vertical-scaling (scale-up) entrepreneurship and horizontal-scaling (scale-out) entrepreneurship—that are inspired by terms from data analytics and cloud computing [14,15]. Vertical-scaling or scale-up entrepreneurship is defined and measured by the degree of an entrepreneur’s ambitions with respect to their innovativeness. Horizontal-scaling or scale-out entrepreneurship is defined and measured by the degree of an entrepreneur’s ambitions with respect to the increase in market share for a given product or service delivered by the entrepreneur (expansion). Hence, the vertical-scaling dimension captures newness/innovation-oriented entrepreneurship while the horizontal-scaling dimension captures expansion/export/growth-oriented entrepreneurship. Importantly, the two dimensions are not mutually exclusive, i.e., an entrepreneur operating at the high end of the vertical-scaling dimension may operate at the high end of the horizontal-scaling dimension at the same time.

Our paper makes a number of contributions addressing several research gaps in the current literature regarding “happiness” or “affective well-being” in entrepreneurship. First, the field is lacking studies investigating the value of happiness when aggregated in terms of a higher-level concept, such as population happiness. Our paper addresses this gap by introducing contextual happiness (country happiness level and happiness inequality) into entrepreneurship research, drawing upon the eclectic theory of supply and demand for entrepreneurship [16] and occupational choice theory [17]. In particular, we investigate the important link between contextual happiness and high-quality entrepreneurship, which is a type of entrepreneurship that is associated with disproportionate economic and social impacts [18].

Second, few studies endeavor to differentiate between the quality and type of entrepreneurship [19], such that the field lacks knowledge about the antecedents of high-quality and distinguished types of entrepreneurship engagement. High-quality entrepreneurial activities are argued to be the real source of new technologies, innovation, competition, job creation, and social prosperity [20,21]. Moreover, we know little about the factors affecting entrepreneurs’ relative preferences for pursuing entrepreneurship at the lower or higher quality end of the entrepreneurship spectrum. The present study focuses on the determinants of entrepreneurs’ relative engagement in the higher or the lower end of the quality spectrum. Consistent with Acs [18] and Hermans et al. [13], we impose growth as the central tenet of high-quality entrepreneurship.

Third, existing research has a biased setting regarding the stage of entrepreneurship, overfocusing on entrepreneurial intention rather than further action and progress beyond the stage of intention. Although forming an entrepreneurial intention is essential [22], initial intentions might not be successfully translated into action due to the difficulties inherent in the nature of the entrepreneurship phenomenon [23,24]. As a consequence, studying the antecedents of entrepreneurs’ concrete action beyond mere intention may be more important.

Fourth, we seek to contribute to dialogs regarding global sustainable development and the Sustainable Development Goals (SDGs). Gross national happiness or well-being is acknowledged to be an essential but neglected dimension of sustainable development [25]. Its importance is emphasized in Goal 3 of the SDGs (good health and well-being). Well-being augments traditional welfare indicators such as gross domestic product and national income in solving poverty and hunger, which are the topics of the first two SDGs. Without questioning the concept of happiness as being the ultimate human desire, we seek to study the consequential meaning of happiness by linking it to entrepreneurship outcomes. Our study therefore also provides new knowledge regarding Goals 8 and 9 of the SDGs, which deal with decent work, economic growth, and innovation. In particular, the decoupling of entrepreneurial activity in two dimensions may provide new insights for policy makers to understand and differentiate between the roles of innovation and economic growth in fulfilling SDGs.

Fifth and finally, we define and measure our dependent variable of entrepreneurship engagement in the implementation phase in order to find the factors influencing actual entrepreneurial action instead of mere intention.

The next section describes our theory and hypotheses, followed by our methodology and data, and an outline of the results. The final section discusses the findings in light of theory, policy implications, and suggestions for further research.

2. Theory and Hypotheses

In this section, we provide further explanation of our two dimensions of entrepreneurial activity: vertical-scaling (scale-up) entrepreneurship and horizontal-scaling (scale-out) entrepreneurship. We also explain how these two types of entrepreneurship are embedded in the literature on high-quality entrepreneurship (Section 2.1). The next subsection then focuses on the theoretical relationship between population happiness and relative entrepreneurship engagement (Section 2.2), culminating in hypotheses related to the level of population happiness (Section 2.2.1) and to happiness inequality (Section 2.2.2).

2.1. Vertical-Scaling Entrepreneurship and Horizontal-Scaling Entrepreneurship

The present paper employs a framework of high-quality entrepreneurship. The literature on high-quality entrepreneurship is relatively recent, dating back no further than two decades. It captures a small subset of entrepreneurs with a disproportionate economic and social impact. Some authors argue that policy makers should exclusively focus on these high-quality entrepreneurs who typically run businesses with growth potential [26]. The phenomenon of high-quality entrepreneurship is known under other labels as well, such as high-impact entrepreneurship [18], high-aspiration entrepreneurship [27], ambitious entrepreneurship [13], and Schumpeterian entrepreneurship [28]. The three most commonly described manifestations of high-quality entrepreneurship are engagement in innovation activities, engagement in export activities, and growth aspirations of entrepreneurs [28,29]. Our two types of entrepreneurship are strongly embedded in the high-quality entrepreneurship literature as they correspond to these manifestations: vertical-scaling entrepreneurship relates to innovation ambitions while horizontal-scaling entrepreneurship relates to export and growth ambitions. Thus, high ambitions in these areas are signals of high-quality entrepreneurial activities, while new entrants with lower ambition levels in these areas, such as necessity and lifestyle-driven new ventures, are deemed to be of low quality [30].

The terms used for our two dimensions of entrepreneurship originate from the field of data analytics and processing. Vertical-scaling refers to adding more resources or powers to a single server, and horizontal-scaling refers to connecting a cluster of machines or entities to work as a single logical unit [14,15]. Applied in the context of entrepreneurship [31], vertical-scaling entrepreneurship focuses on newness, and is defined by the degree of an entrepreneur’s growth ambitions in terms of innovativeness to improve the output per input ratio. Horizontal-scaling entrepreneurship focuses on expansion, and is defined by the degree of an entrepreneur’s growth ambitions aimed at connecting more entities (in this case: employees and customers) to expand the size of the venture. The functions of these two types of entrepreneurship differ in significant ways. Vertical-scaling entrepreneurship emphasizes exploration through the selling of new products to initiate a market, while horizontal-scaling entrepreneurship emphasizes exploitation through the selling of one specific product to more consumers to gain a competitive advantage in an existing product market.

2.2. Population Happiness and Relative Entrepreneurship Engagement

Happiness has various definitions, from long-term overall well-being to short-term positive feelings. In the present study, we define happiness as an overall positive emotional state represented by frequent experiences of positive affects and infrequent experiences of negative affects, based on empirical support from Diener et al. [1] and Lyubomirsky et al. [2]. Compared with previous frequently employed forms of positive affect focusing on intensity, our definition of happiness is more psychologically subtle and more individually private, and considers a longer time horizon. More importantly, in this study, we regard happiness as a collective macro-level construct.

Entrepreneurs do not act in a vacuum. It is crucial to understand the contextual determinants of the entrepreneurial process, i.e., how individuals take into account the environment when building up a new business [32]. Incorporating contextual happiness into this dialog is important for two reasons. The first reason is related to affective spillovers. Emerging as an important driver of entrepreneurial action, individual positive affects may be influenced by contextual happiness, as emotions are contagious and tend to spread from one individual to another [33]. Second, the psychological process of how entrepreneurs relate to the external world is influenced by the positive affects of the stakeholders and the environment, i.e., by contextual happiness.

Specifically, we theorize that population happiness may affect entrepreneurial relative engagement primarily through two mechanisms: the supply and demand for entrepreneurship [16], and an individual’s risk-reward profile as argued by occupational choice theory [17,34]. Contextual factors can influence entrepreneurial choice at the individual level from either the supply or the demand side. The supply side can influence entrepreneurial engagement by affecting entrepreneurs’ external resources and internal attitudes, and the demand side can enhance or attenuate the number of opportunities for entrepreneurs [16]. Both sides could affect entrepreneurs’ relative decisions in engaging in the higher or lower end of vertical-scaling (or horizontal-scaling) entrepreneurship. But how do each of these conditions affect entrepreneurial decisions? Occupational choice theory, generated from economic research, is a well-established theoretical anchor for explaining the detailed process. The theory assumes that individuals are rational human beings who seek to maximize expected utility associated with career choice, i.e., they calculate and compare the rewards and risks of the occupational alternatives and choose one with the highest net rewards (e.g., wage-employment versus self-employment). In the current study, the difference is that we are comparing entrepreneurial alternatives (i.e., different levels of entrepreneurship engagement) as opposed to occupational alternatives, whether the entrepreneur would perceive a higher net utility in entering high-newness (or high-expansion) versus low-newness (or low-expansion) entrepreneurship if surrounded by a happier population.

The next two subsections will derive hypotheses on the relationship between population happiness (level and inequality) and relative entrepreneurship engagement. In doing so, we treat happiness as the national affective well-being climate (prevalence of positive affect/low negative affect), and happiness inequality as dispersion in life evaluations. These macro-level affective and evaluative climates shape the collaboration costs, perceived opportunity space, and competitive pressures that entrepreneurs face.

2.2.1. Population Happiness Level and Relative Entrepreneurship Engagement

We first illustrate the effect of a population’s average happiness level on engagement in vertical-scaling and horizontal-scaling entrepreneurship. The effect of population happiness level on entrepreneurial engagement is not only associated with the happiness levels of the entrepreneurs themselves, but is also related to the happiness levels of stakeholders around the entrepreneurs (in particular employees and customers).

In a happy population, there is a greater supply of happy people in the job market and working environment. Positive affects at work improves creativity and cognitive functioning [11,35]. It also influences interpretations of others’ motives and tends to promote attributions of positive motives [36], thus reducing conflict between individuals who work together [37]. In regard to the entrepreneurship field, a greater supply of happier co-founders and co-workers tends to benefit entrepreneurs in generating innovative ideas, obtaining support, and reducing pressure along the venture creation process. Therefore, this enhanced package of external resources is more likely to render the entrepreneurs’ higher expected utility in pursuing entrepreneurial activity with higher, rather than lower, levels of innovativeness. Hence, a happier population is expected to be positively related to the level of vertical-scaling entrepreneurship an individual entrepreneur pursues.

However, this package of supply factors makes entrepreneurs more likely to engage in lower levels of entrepreneurship in terms of the expansion-focused dimension. Within a happier population, entrepreneurs can achieve a satisfactory level of economic utility more easily, as explained above (i.e., because there are more “happy” stakeholders surrounding the entrepreneur). Therefore, in a happier population, entrepreneurs may be less motivated to pursue greater expansions that tend to bring higher risks, uncertainty, and opportunity costs, for instance, international expansion into a much less familiar environment. Hence, a happier population is expected to be negatively related to the level of horizontal-scaling entrepreneurship an individual entrepreneur pursues.

A happy population can also influence entrepreneurs’ internal values and attitudes toward engaging within different ends of entrepreneurship. Feeling positive and happy can not only strengthen people’s creativity, but also the intrinsic motivations to pursue creative and challenging tasks [35,38]. More importantly, this intrinsic tendency of happiness is contagious [39], meaning that an individual tends to perceive more desirability for pursuing high-newness when embedded in a happier working environment and population. When applied in entrepreneurship, a supply package of happier stakeholders tends to promote entrepreneurs’ internal preference to engage in greater levels of newness-focused entrepreneurship. However, this supply package would not denote much difference in influencing entrepreneurs’ desirability in pursuing high or lower levels of expansion-focused entrepreneurship, since a positive environment, as illustrated, is more of a signal for newness and innovativeness, rather than the expansion of market share in an existing product market. In sum, from the supply side perspective, a population with a higher level of happiness is likely to enhance the likelihood of entrepreneurs’ engagement in higher levels of entrepreneurship on the vertical-scaling dimension and lower levels of entrepreneurship on the horizontal-scaling dimension.

There are more happy consumers in a happy population as well. Positive affects can strengthen an individual’s cognitive functioning [12] and broaden an individual’s thought–action repertoire [40]. Therefore, happier consumers tend to favor a wider variety of innovative products and niche markets. Hence, from the demand-side perspective, there is ample room (or demand) for entrepreneurs to engage in high levels of vertical-scaling entrepreneurship, deriving innovative products and initiating niche markets [41]. In contrast, happier consumers may be less interested in standardized products typically sold in bigger markets occupied by dominating brands and products, making it hard for horizontal-scaling entrepreneurs to obtain considerable economic utility from engaging in higher level of venture expansion. In addition, as happy consumers tend to favor a high variety of products and niche markets, they are likely to cause higher uncertainty in a “happier” environment. With such uncertainty, there is less “controllability” for entrepreneurs, who therefore tend to maintain lower expansion ambitions and favor lower engagement levels in terms of horizontal-scaling entrepreneurship. Our sets of arguments all point in the direction of the following hypotheses:

Hypothesis 1a.

The level of population happiness is positively related to the relative engagement level in vertical-scaling entrepreneurship.

Hypothesis 1b.

The level of population happiness is negatively related to the relative engagement level in horizontal-scaling entrepreneurship.

2.2.2. Population Happiness Inequality and Relative Entrepreneurship Engagement

To comprehensively investigate the value of a happy population, it is essential to consider the amount of happiness variation within that population, apart from the average level of population happiness. The current study thus seeks to examine the effect of happiness inequality within a given population on individual entrepreneurs’ entrepreneurship engagement. Framed as a judgment indicator for relative justice, social inequality could profoundly influence social status and progress and thus affect entrepreneurial engagement [42]. Happiness inequality, conceptualized in this study as the dispersion in life evaluations, is a broader and more comprehensive measure of social inequality than income inequality [43]. Similarly to happiness levels, we will separately state the expected effect of happiness inequality on vertical-scaling and horizontal-scaling entrepreneurship engagement.

Following the supply and demand sides of entrepreneurship theory [16], and the aforementioned occupational choice theory, entrepreneurs could be strongly hindered to engage in higher levels of vertical-scaling (newness-focused) entrepreneurship, when they are surrounded by a high level of happiness inequality. The arguments are as follows. Firstly, pursuing high levels of innovation is a challenging process that needs support from a fair-structured environment. When situated in an environment with a high level of happiness inequality, entrepreneurs may not find the pursuit of innovation rewarding. As such, social inequality is associated with politically unstable institutions, which in many cases is translated into uncertainty of property rights and other risks of social conflict [44]. This instability is usually detrimental to creativity and innovation, hampering entrepreneurship engagement on the vertical-scaling dimension.

Secondly, innovation is more likely to thrive in a more liberal environment with higher levels of trust in other people and trust in institutions [45]. According to the political economy view, high structural inequality associated with high happiness inequality might lead to higher redistributive pressure, which will severely harm the economic incentives and cause economic distortions [42] (or might lead the rich to lobby and hinder efficient redistribution policies from being implemented [46]. These lobbying activities are associated with rent seeking and corruption [47], severely reducing the perception of fair opportunity and entrepreneurs’ perceived autonomy.

Thirdly, high-level vertical-scaling entrepreneurs are less likely to experience affective connections with stakeholders in an unequal environment. Pursuing challenging work, such as work requiring high levels of innovation, would receive less support in a socio-politically instable society, where the population is suffering from structural inequality. To conclude, the following hypothesis is proposed:

Hypothesis 2a.

The level of population happiness inequality is negatively related to the relative engagement level in vertical-scaling entrepreneurship.

However, happiness or social inequality might induce an opposite effect, enhancing engagement in higher levels of expansion-oriented entrepreneurial activity. First, inequality is supportive of the feasibility (i.e., external resources) in engaging high level of horizontal-scaling entrepreneurship. The classical perspective argues that market inequality relates to incentives for capital accumulation [48], therefore causing a presumed greater propensity to save among the rich, and a higher level of investment and growth [49]. Second, the desirability (i.e., the internal preference) of entering high levels of horizontal-scaling entrepreneurship is boosted under the social norm of an unequal environment. In particular, one important social norm of a high-inequality environment is formed by the motivation for individuals to fend for themselves and compete for economic success and social status [50]. Therefore, such an environment can strengthen entrepreneurs’ incentives to work hard and take risks [51] in order to expand and grow their ventures economically. To conclude, we propose the following hypothesis:

Hypothesis 2b.

The level of population happiness inequality is positively related to the relative engagement level in horizontal-scaling entrepreneurship.

We realize that our arguments on happiness or social inequality partly overlap with those of economic inequality (e.g., Kuznets [52]), particularly for H2b. Therefore, in our empirical analysis, we will also undertake a robustness test including a measure of economic inequality (i.e., the GINI coefficient), next to our measures of happiness and happiness inequality.

3. Methodology

3.1. Data

In order to analyze the relationship between a country’s population-level happiness and the entrepreneurship engagement level (quality) of individual entrepreneurs within that country, we need an empirical design that is able to link macro- and micro-level determinants to our micro-level dependent variable of relative entrepreneurial engagement. Moreover, harmonized measures are required to be applied consistently across countries to ensure valid comparisons [53]. To satisfy these requirements, we constructed our data primarily through two national-comparative sources: the Global Entrepreneurship Monitor (GEM) database (for our dependent variable of entrepreneurial engagement), and the Gallup World Poll (for our main predictors of happiness and happiness inequality).

Conducted by the Global Entrepreneurship Research Association (GERA), the Adult Population Survey (APS) of GEM contains detailed information about entrepreneurial activities at the individual level. It is completed annually in each participating country with strictly harmonized data collection methods to enhance measurement equivalence [54]. The GEM dataset suits our study in that it is a reliable and internationally comparable dataset with various attitude and aspiration measures capturing entrepreneurial activities along the venture creation phases. Country-level happiness and happiness inequality data are drawn from the Gallup World Poll (GWP), which is the first and largest ongoing representative survey across the globe, tracking important issues worldwide including well-being. A common set of questionnaires and statistics, and a harmonized data collection approach are utilized to enhance measurement equivalence, via face-to-face or telephone interviews in more than 160 countries and over 140 languages. Many well-known organizations and renowned projects are based on GWP data. The GWP contains various questions in their well-being section. We measured country happiness average levels using the Positive and Negative Experience Index, which is consistent with our definition of happiness (see Section 2.2). Additionally, positive experience measures of happiness, in contrast with measures of life evaluation or satisfaction, are less related to subjective attitude or judgment and thus better address measurement equivalence. For our measure of country happiness inequality, we adopt the country’s standard deviation of GWP’s life evaluation index. As happiness inequality focuses on deviations rather than absolute levels, we assess that measurement equivalence is not an issue here as well. We refer to

Table 1. Dependent, independent, and control variables.

Variable	Coding	Level a	Source
Dependent variables
Engagement into vertical-scaling e-ship (ordinal variable)	Respondent who is a nascent entrepreneur or owner-manager of a new business is going to answer Yes or No to the following three statements: 1. You are driven by entrepreneurial opportunity as opposed to necessity; 2. You pursued and used technology and procedures only available within the most recent year; 3. At least some of your customers think your venture product is new. Coded 1 if respondent answered No for statement 1; Coded 2 if respondent answered Yes for statement 1 but responded No for both statement 2 and 3; Coded 3 if respondent answered Yes for statement 1 and Yes for either statement 2 or statement 3; Coded 4 if respondent answered Yes for all the three statements.	1	GEM
Engagement into horizontal-scaling e-ship (ordinal variable)	Respondent who is a nascent entrepreneur or owner-manager of a new business is going to answer Yes or No to the following three statements: 1. You expect employ 20 or more employees within five years; 2. At least 25% of the customers come from other countries. Coded 1 if respondent answered No for both statements; Coded 2 if respondent answered Yes for statement 1 but responded No for statement 2, or vice versa; Coded 3 if respondent answered Yes for both statements.	1	GEM
Independent variables
Country happiness	The individual positive experience score is the mean of all valid affirmative responses (yes-coded 1 or no-coded 0) to the following five items: (1) Did you feel well-rested yesterday? (2) Were you treated with respect all day yesterday? (3) Did you smile or laugh a lot yesterday? (4) Did you learn or do something interesting yesterday? (5) Did you experience the following feelings during a lot of the day yesterday? How about enjoyment? The individual negative experience score is the mean of all valid affirmative responses (yes-coded 1 or no-coded 0) to the following five items: Did you experience the following feelings during a lot of the day yesterday? (1) How about physical pain? (2) How about worry? (3) How about sadness? (4) How about stress? (5) How about anger? Country-level scores is aggregated by individual scores by taking the average, which range from zero to one. The final country happiness score is the difference of country positive experience score and negative experience score.	2	GWP
Country happiness inequality	It is measured by the standard deviation of life evaluation index score of each country divided by the mean of the score. The individual life evaluation score is measured by answering the following question “Please imagine a ladder, with steps numbered from 0 at the bottom to 10 at the top. The top of the ladder represents the best possible life for you and the bottom of the ladder represents the worst possible life for you. On which step of the ladder would you say you personally feel you stand at this time?”	2	GWP
Control variables
Gender	Gender was coded 1 for females and 0 for males	1	GEM
Age	Age of respondent between 15 and 64 (inclusive)	1	GEM
Education	Education level is coded 1 for none or some secondary education, 2 for secondary education, 3 for past secondary education, and 4 for graduate experience. None or some secondary education is used as reference category (baseline)	1	GEM
Work status	Work status is coded 1 for those respondents who working full-time or part-time, 2 for those who working part-time only and 3 for students.	1	GEM
Household income	Head of household’s income, dummy coded categorized into three groups of equal number of respondents for each country (baseline: bottom third).	1	GEM
Industry categories	1 = agriculture, hunt, fish; 2 = mining, construction; 3 = manufacturing; 4 = utilization, transport, storage; 5 = wholesale trade; 6 = retail trade, hotels &restaurants; 7 = information & communication; 8 = financial intermediation, real estate activities; 9 = professional service; 10 = administrative services; 11 = government, health, education, social services; 12 = personal/consumer service activities.	1	GEM
Log GDP per capita	The average Gross Domestic Product (GDP) per capita in purchase power parity (in natural logarithm) during the three years preceding the entry of e-ship.	2	WBI b
GDP per capita growth rate	The average Gross Domestic Product (GDP) per capita in purchase power parity annual growth rate during the three years preceding the entry of e-ship.	2	WBI
Population growth rate	The average population growth rate during the three years preceding the entry of e-ship.	2	WBI
Regulatory quality	Reflects perceptions of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development.	2	WGI c
a Level 1 represents the individual level, and level 2 represent the country level
b WBI refers to World bank Indicators
c WGI refers to Worldwide governance indicators

for the exact operationalization of these variables.

We merged the two main datasets, in their overlapping parts, to complete our main analysis. The matched GEM individual-level APS data runs from 2007 to 2013, and the country-level data (happiness level, happiness inequality, and other macro-level controls) were included with a one-year lag, i.e., the years from 2006 to 2012. We thus expect population happiness to influence entrepreneurial engagement with a one-year lag. Considering all variables included in our regression model, our initial dataset included 792,634 individual observations from 79 countries. We constrained our sample to those respondents currently in the start-up implementation process, resulting in a sample of 71,964 early-stage (nascent or new) entrepreneurs. The reason for this choice is that high-quality or high-aspiration entrepreneurship is especially relevant in the early stages of entrepreneurship because ambitions have not been materialized yet at that stage [13]. At later stages, when the business is more established, the potential role of contextual happiness on entrepreneurial ambitions may be smaller and less relevant, because future ambitions at that point are primarily shaped by the accumulated business experience during the venture’s years of existence.

Table 1 shows the variables utilized, their coding, their analytical level (individual or country level) and their source.

3.2. Individual-Level Dependent Variables

We employed two individual-level dependent variables: relative engagement in vertical-scaling and horizontal-scaling entrepreneurial activities. We constructed two ordinal scales to measure them, by utilizing information in GEM regarding entrepreneurs’ motivations and ambitions related to starting their new businesses. There are four progressive engagement levels in vertical-scaling entrepreneurship. The levels depend on three conditions: the entrepreneur was driven by entrepreneurial opportunity as opposed to necessity, pursued and utilized new technology only available within the most recent year, and at least some of the customers think that the entrepreneur’s venture product is new. The fourth and highest engagement level was reached when all three conditions are met at the same time. The other three engagement levels are illustrated in

Table 2. (a) Ordinal measurement of dependent variables: Vertical-scaling entrepreneurship. (b) Ordinal measurement of dependent variables: Horizontal-scaling entrepreneurship.

(a)
Engagement into Vertical-Scaling Entrepreneurship
	Level	1	2	3		4
Content
Opportunity motivation		No	Yes	Yes		Yes
New technology			No	No	Yes	Yes
New product			No	Yes	No	Yes
(b)
Engagement into Horizontal-Scaling Entrepreneurship
	Level	1		2		3
Content
20 or more employees within five years		No		Yes	No	Yes
25% or more customers from abroad		No		No	Yes	Yes

a, which shows the ordinal measurement schematically.

Our measurement of vertical-scaling entrepreneurial engagement thus links up with our conceptual variable, as both the use of new technology and the creation of new products are clearly connected to innovativeness. Moreover, we consider opportunity motivation a crucial necessary condition to achieve newness, in line with the literature [55,56,57]. More specifically, innovation often requires opportunity discovery or opportunity creation, in the sense that either the supply side of the opportunity (i.e., the new product) and/or the demand side of the opportunity (i.e., the market demand for the new product) is unknown before the opportunity is seized [58]. Although the GEM questionnaire does not distinguish between these types of opportunities, it is clear that the business should have been started from an “entrepreneurial opportunity” (in GEM terms) in order to qualify for a higher level of entrepreneurship on the vertical-scaling dimension.

Horizontal-scaling entrepreneurship, similarly, has three progressive levels that capture the key to expansion. The third and highest level was identified by meeting aspirations regarding both firm growth and international expansion, in particular, expecting to employ 20 or more employees within five years, and claiming that at least 25% of the venture’s product would be exported to customers in other countries. These specific measures of expected job creation and export intensity are collected and analyzed by GEM for defining high job creation expectations and high export intensity. Over the years, these measures have been used frequently in empirical research on entrepreneurial ambitions and entrepreneurial quality (e.g., Wong et al. [59]; Hessels and Van Stel [60]; Levie and Autio [61]; Cieślik et al. [29]; Giotopoulos et al. [30]). Entrepreneurs who just indicated one of the two aspirations were set to be at the second level, while entrepreneurs who answered no to both questions would be at the first (lowest) level. The measurement of horizontal-scaling entrepreneurship is shown in Table 2b.

Also here, our measurement of horizontal-scaling entrepreneurial engagement links up with our conceptual variable as firm growth and international expansion are both connected to expansion of market share in an existing product market. In this case we did not set opportunity motivation as a necessary condition, because for this type of growth, ambition, opportunity recognition—in the sense that both the supply and demand sides of the opportunity are already known to exist [58]—is already enough to pursue such an opportunity. Thus, although the majority of horizontal-scaling entrepreneurs will have started their business to pursue an “entrepreneurial opportunity” in the GEM (i.e., unspecified) sense, a minority of horizontal-scaling entrepreneurs may have started their business out of necessity, for instance, if they were fired as an employee of the business they previously worked under, through no fault of their own (for instance, in a reorganization). Such circumstances may give individuals, especially talented ones, an unexpected opportunity to start up a business––out of necessity––to see if they can make their business grow [62,63]. For vertical-scaling entrepreneurship, which requires more advanced forms of opportunity recognition, i.e., opportunity discovery or even creation, we consider higher engagement levels for necessity-driven entrepreneurs unlikely.

As we are the first to develop scales of vertical- and horizontal-scaling entrepreneurship, it was not possible to validate our measures based on earlier research. However, we ascertain that our measures cover the most important elements of high-quality entrepreneurship, i.e., expansion orientation and innovation orientation [20,26,64,65,66]. We present the descriptive statistics for the dependent variables and the individual-level controls in

Table 3. Descriptive statistics for individual level variables.

Variables	Categories	Frequencies
Engagement into vertical-scaling entrepreneurship	Level 1	23,483 (32.34%)
	Level 2	23,418 (32.25%)
	Level 3	21,690 (29.87%)
	Level 4	4027 (5.55%)
Engagement into horizontal-scaling entrepreneurship	Level 1	50,460 (70.56%)
	Level 2	17,135 (23.96%)
	Level 3	3920 (5.48%)
Gender	Male	43,855 (60.39%)
	Female	28,763 (39.31%)
Age	Young (18–29 years old)	21,465 (29.56%)
	Mid (30–49 years old)	39,315 (54.14%)
	Old (50–64 years old)	11,836 (16.30%)
Education level	None	7837 (10.89%)
	Some secondary	12,648 (17.58%)
	Secondary	23,768 (33.03%)
	Post secondary	21,335 (29.65%)
	Graduate experience	6376 (8.86%)
Work status	Working full-time or part time	65,517 (90.22%)
	Working part-time only	5535 (7.62%)
	Student	1566 (2.16%)
Household income	Low	15,819 (21.78%)
	Middle	23,617 (32.52%)
	High	33,182 (45.69%)
Industry category	Agriculture	4138 (5.70%)
	Mining & construction	3833 (5.28%)
	Manufacturing	7139 (9.83%)
	Utilization, transport, storage	3267 (4.50%)
	Wholesale trade	5226 (7.20%)
	Retail trade, hotels & restaurants	28,787 (39.64%)
	Information & communication	2685 (3.70%)
	Financial intermediation, real estate activities	3654 (5.03%)
	Professional service	4528 (6.24%)
	Administrative services	2856 (3.93%)
	Government, health, education, social services	5289 (7.28%)
	personal/consumer service activities	1216 (1.67%)

.

3.3. Country-Level Predictor Variables

Our main predictors are country-level happiness and happiness inequality. For country-level happiness, we used data from the Positive Experience Index of the Gallup World Poll, which gathered national overall positive emotional status for each participating country. The positive experience score for an individual respondent is the mean of all valid affirmative responses (yes coded as 1 and no coded as 0) to five items about the respondent’s frequency of experienced well-being on the day before the survey. The items are, respectively, in terms of feeling well-rested, being treated with respect all day, smiling or laughing a lot, learning or engaging in something interesting, and experiencing enjoyment. We measured country-level happiness by averaging the individual scores within that country. The country-level score ranges from zero to one; see Table 1.

Our measure of country-level happiness is based on the positive psychological emotional state focusing on frequency rather than intensity, which responds to our definition of happiness and appeals to the recommendation of putting less focus on short-term and high-activated affective experiences [67]. Hence, our measure considers structural long-term happiness. In this regard, there was initially a concern about the setting of this measure. Since the items asked people about their experiences during the previous day, such aggregate data may simply provide fast-changing perceptions experienced only on the day of the survey. Two further checks, however, suggest that the data are meaningful and valid. The Positive Experience Index is significantly and positively related to the Positive Mental Health index from the European Quality of Life Survey [68]. As these two surveys were conducted with different samples on different days and with different questions, the strong relationship at the country level reassures the data validity of GWP’s experience index. In addition, the strong relationship between country scores over the years demonstrates that the data are not just random sample errors or fickle swings of day-to-day moods at the country level.

Happiness inequality is measured in terms of life satisfaction, i.e., an overall life evaluation, which takes into account both perceived positive experiences and positive attitudes [69]. Hence, life satisfaction inequality possesses high content validity and may thus be considered a good approximation of happiness inequality. The variable was measured by taking the standard deviation of the life satisfaction scores of all individual respondents within a given country and divided by the mean of the life satisfaction scores of all individual respondents within the same country; see Table 1.

Table 4. Descriptive statistics for country-level variables, ranked by development stage.

Development Stage		Country	Frequencies (Early-Stage Entrepreneurs)	National Start-Up Rate (Total Early-Stage Entrepreneurship)	National Happiness	National Happiness Inequality
Factor-driven economies		Bangladesh	177	11.60	0.34	0.40
		Bolivia	770	18.53	0.41	0.33
		Botswana	545	6.33	0.59	0.48
		Ghana	1033	36.94	0.57	0.45
		Malawi	1061	10.80	0.53	0.65
		Nigeria	1343	11.75	0.60	0.35
		Pakistan	408	4.51	0.28	0.43
		Palestine	368	5.58	0.14	0.51
		Philippines	375	6.61	0.51	0.49
		Uganda	1946	25.95	0.42	0.39
		Vietnam	287	16.35	0.39	0.26
		Zambia	1538	5.69	0.57	0.38
Factor-driven economies (in transition)		Algeria	241	3.17	0.32	0.32
		Angola	643	8.84	0.28	0.44
		Egypt	475	5.55	0.26	0.44
		Guatemala	719	4.47	0.58	0.38
		Iran	774	8.83	0.21	0.47
		Jamaica	335	7.81	0.59	0.37
		SaudiArabia	140	4.06	0.47	0.25
		Venezuela	360	4.34	0.67	0.32
Efficiency-driven economies		Argentina	1056	11.09	0.57	0.33
		Bosnia and Herzegovina	412	6.06	0.21	0.45
		Brazil	4160	13.39	0.50	0.33
		China	3051	12.91	0.65	0.40
		Colombia	5594	11.07	0.55	0.39
		CostaRica	447	4.07	0.63	0.28
		D.R	248	7.79	0.46	0.60
		Ecuador	1476	15.43	0.56	0.39
		ElSalvado	184	9.39	0.54	0.54
		Estonia	300	7.24	0.45	0.45
		India	533	7.72	0.40	0.38
		Indonesia	1096	21.20	0.67	0.30
		Lithuania	449	6.97	0.27	0.36
		Macedonia	284	8.45	0.25	0.47
		Malaysia	490	5.74	0.67	0.28
		Mexico	876	3.11	0.58	0.31
		Montenegro	214	7.81	0.20	0.45
		Panama	828	4.08	0.69	0.30
		Peru	2314	8.75	0.41	0.41
		Poland	420	5.25	0.51	0.33
		Romania	484	3.00	0.28	0.44
		Russia	319	2.29	0.42	0.38
		SouthAfrica	583	2.02	0.56	0.38
		Thailand	1553	27.29	0.74	0.28
		Turkey	2717	8.52	0.30	0.38
		U.A.E	153	1.39	0.50	0.27
Efficiency-driven economies (in transition)		Chile	4437	6.92	0.49	0.35
		Croatia	332	4.02	0.33	0.36
		Hungary	567	5.54	0.40	0.42
		Latvia	618	6.04	0.37	0.43
		T.T	330	7.79	0.70	0.30
		Uruguay	969	6.50	0.53	0.35
Innovation-driven economies		Australia	242	8.47	0.61	0.23
		Austria	286	7.12	0.65	0.25
		Belgium	316	3.24	0.58	0.22
		Canada	242	8.44	0.63	0.24
		CZE	462	5.24	0.37	0.31
		Denmark	391	4.97	0.61	0.19
		Finland	379	8.54	0.59	0.21
		France	316	3.24	0.58	0.22
		Germany	969	5.01	0.59	0.28
		Greece	595	13.17	0.36	0.39
		Ireland	478	8.45	0.65	0.26
		Israel	469	3.36	0.37	0.26
		Italy	185	5.04	0.40	0.31
		Japan	278	7.28	0.59	0.31
		Luxembourg	113	2.39	0.59	0.26
		Netherland	877	7.88	0.64	0.17
		Norway	187	6.55	0.63	0.22
		Portugal	345	5.96	0.40	0.41
		Singapore	188	3.26	0.41	0.24
		Slovakia	513	8.55	0.36	0.32
		Slovenia	325	5.10	0.36	0.36
		SouthKorea	615	11.52	0.47	0.34
		Spain	6081	7.43	0.45	0.29
		Sweden	418	6.07	0.66	0.22
		Switzerland	253	8.42	0.63	0.21
		UnitedKingdom	2625	5.74	0.61	0.25
		UnitedStates	1840	7.37	0.60	0.27

lists the 79 countries and the descriptive statistics for the main country-level variables of this study (i.e., happiness level and happiness inequality) for each country.

3.4. Individual- and Country-Level Control Variables

We controlled for a number of influences in the analyses at both the individual and the country level. At the individual level, previous research showed that men, middle-aged and better-educated individuals, as well as employed people, are more likely to start a business [16,70]. We therefore included a dummy variable for gender, the individual’s age, education (primary, secondary, post-secondary, and graduate experience) and employment status (working full-time, working part-time, or students) as control variables to address these influences. Past research also indicated the importance of access to capital for potential entrepreneurs’ engagement in the start-up process [16]. Therefore, we also controlled for the respondent’s relative household income, measured as being in the lower, middle, or upper part of the household income distribution in the respondent’s country of residence. The industry domain pursued by the entrepreneurs might influence their relative entrepreneurial engagement level. We thus controlled for this factor by including the dummies for the twelve-category business industry variable from GEM.

At the country level, we also included several important control variables, as both individual entrepreneurship and population happiness could be influenced by an omitted third contextual factor (or confounding variable). As a country’s level of economic development and expansion may significantly affect its entrepreneurship rate [71], we incorporated both the level and the growth rate of GDP per capita as our controls. A country’s expansion in terms of population growth may also significantly affect new business activity as it implies increased demand for goods and services. We therefore controlled for population growth. Finally, a country’s institutional environment can exert tremendous influence on new venture establishment [61]. We included a representative control of the institutional environment in our models: the Regulatory Quality index from the World Bank’s Worldwide Governance Indicators (WGI) project [72]. Together, this set of control variables captures the most important macro-level determinants of entrepreneurship [16], making omitted variable bias unlikely. Nevertheless, we cannot fully rule out that unobserved cultural or institutional factors jointly drive both national happiness and entrepreneurial ambition.

As a final observation on our data sources and model variables, we note that administrative and survey harmonization across countries is an important strength of both GEM and Gallup, but that measurement of “access to very recent technology,” “perceived newness to customers,” and “expected 20+ employees/≥25% foreign customers” (see Table 1) still relies on self-report and expectation.

3.5. Statistical Analysis

Our data consist of a repeated cross-section of entrepreneurs, i.e., for each year from 2007 to 2013 we have data for (different) samples of entrepreneurs, distributed over multiple countries. Since our dependent variables are of ordinal nature, we adopted ordered logit models to test our hypotheses. Moreover, since we estimated a model at the individual level, but also including country-level predictors, we clustered standard errors by country. A multi-level model is not feasible here for our multiple year cross-sectional data [73]. Instead, we employed a pooled (ordered logit) model to capture both the within-variation (variation over time) and between-variation (variation between countries) in population happiness and happiness inequality in influencing entrepreneurship engagement. Nevertheless, we did a robustness test estimating a multi-level ordered logit model for all our main analyses, through selecting the most recent year data of 2013. Finally, we included time fixed effects in our model.

One thing to note is that our country-level happiness and country-level happiness inequality variables are not of the same measurement scale, which might be misleading to compare in a single regression. Therefore, in all our analyses, we separately run our models for country-level happiness and country-level happiness inequality. Our data violated the proportional odds assumption, which is the basic assumption underlying ordered logit or probit models, that the relationship between each pair of the targeted groups is the same. Therefore, we chose to use the generalized ordered logit model, which relaxes the parallel lines constraint for those variables that violate it while maintaining it for others. In short, our estimation method fully exploits the ordinal nature of our dependent variables while avoiding the restrictiveness of theoretically more sophisticated estimators. Variance Inflation Factors (VIFs) were computed for all our models. The highest VIF was 2.20 for the variables included in the model, suggesting that multicollinearity was not a concern for our analyses [74].

As a final technical note, our research design makes that the model estimates should be viewed as directional associations rather than strict causal effects.

4. Results

Table 3 presents the descriptive statistics for the individual-level variables, while Table 4 reports the descriptive statistics for the country-level variables, distributed according to the World Economic Forum’s classification of development stages in countries [75]. Table 3 reveals that entrepreneurship frequencies decrease as the engagement level progresses, both for vertical-scaling and horizontal-scaling entrepreneurship. Only 5.55% of entrepreneurs entered the highest level of vertical-scaling entrepreneurship and only 5.48% of entrepreneurs entered the highest level of horizontal-scaling entrepreneurship. Table 4 shows the means of the major country-level variables for each development stage. Going from factor-driven economies to innovation-driven economies, a significant decrease in national start-up rates is observed, confirming the results from Acs et al. [76], while a small increase in national-level happiness and happiness equality is also observed.

Table 5. Means, standard deviations, and correlations.

Variable	Mean	S.D.	1	2	3	4	5	6	7	8	9	10	11	12	13
1. Vertical-scaling e-ship engagement	2.04	0.91
2. Horizontal-scaling e-ship engagement	1.35	0.58	0.16
3. Gender	0.39	0.49	−0.05	−0.13
4. Age	36.96	11.23	−0.04	−0.02	0.01
5. Education	3.09	1.12	0.18	0.17	−0.05	−0.01
6. Work status	1.13	0.41	0.04	0.02	0.05	−0.16	0.02
7. Household income	2.25	0.79	0.13	0.14	−0.09	0.01	0.25	−0.04
8. Industry category	6.02	2.70	0.07	−0.02	0.11	−0.04	0.17	0.04	0.06
9. Population growth rate	1.25	1.31	0.002	0.01	0.00	−0.10	−0.17	0.02	−0.05	−0.07
10. Regulatory quality	0.42	0.87	0.12	0.09	−0.01	0.18	0.28	−0.04	0.04	0.12	−0.31
11. Log GDP per capita	9.61	0.85	0.09	0.09	−0.06	0.18	0.38	−0.02	0.05	0.14	−0.35	0.69
12. GDP per capita growth rate	2.60	6.07	0.01	−0.01	0.02	−0.04	−0.04	−0.003	−0.06	−0.05	−0.08	−0.22	−0.18
13. Country happiness	0.50	0.13	0.03	−0.08	0.10	0.07	−0.02	−0.01	−0.02	0.05	−0.02	0.20	0.12	0.07
14. Country happiness inequality	0.35	0.08	0.004	0.03	0.01	−0.07	−0.10	0.03	0.04	−0.04	0.03	−0.31	−0.28	0.06	−0.25

Note: n = 71,964. Absolute correlations above 0.01 are significant at p < 0.05, above 0.02 are significant at p < 0.01.

reports the correlations for the covariates of vertical-scaling and horizontal-scaling entrepreneurship engagement. It reveals that among the individual-level variables, high-quality entrepreneurship engagement (either vertical- or horizontal-scaling) is positively correlated with male entrepreneurs, education level, and household income. Among the country-level variables, high-quality entrepreneurship engagement is positively correlated with regulatory quality and national GDP per capita.

4.1. Country-Level Happiness and Individual Engagement in Entrepreneurship (H1a, H1b)

Table 6. Generalized ordered logit models explaining relative engagement levels in vertical-scaling and horizontal-scaling entrepreneurship ^a.

	Vertical-Scaling e-Ship Relative Engagement			Horizontal-Scaling e-Ship Relative Engagement
VARIABLES	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Main predictors
Country happiness		1.458 *			−2.422 *
		(0.660)			(1.001)
Country happiness			−1.678 **			3.346 ***
inequality			(0.628)			(0.853)
Individual-level controls
Gender (baseline =	−0.140 ***	−0.165 ***	−0.169 ***	−0.402 ***	−0.382 ***	−0.402 ***
male)	(0.0284)	(0.0260)	(0.0323)	(0.0447)	(0.0379)	(0.0440)
Education (baseline = no edu)
Some secondary	0.194 *	0.214 *	0.206 *	0.281 *	0.272 *	0.316 **
	(0.0925)	(0.0974)	(0.0857)	(0.121)	(0.122)	(0.120)
Secondary	0.456 ***	0.465 ***	0.438 ***	0.542 ***	0.554 ***	0.589 ***
	(0.0895)	(0.0903)	(0.0812)	(0.125)	(0.130)	(0.128)
Post-secondary	0.656 ***	0.680 ***	0.679 ***	0.812 ***	0.792 ***	0.825 ***
	(0.101)	(0.106)	(0.0780)	(0.146)	(0.152)	(0.146)
Graduate	0.702 ***	0.735 ***	0.735 ***	1.003 ***	0.972 ***	1.033 ***
	(0.109)	(0.115)	(0.0903)	(0.159)	(0.168)	(0.155)
Work status (baseline = full or part-time)
Only part-time	0.0694	0.0573	0.108 *	0.0494	0.0804	0.0514
	(0.0501)	(0.0549)	(0.0552)	(0.0562)	(0.0526)	(0.0509)
Student	0.243 *	0.246 **	0.272 **	0.258 **	0.267 **	0.235 **
	(0.0982)	(0.0929)	(0.0981)	(0.0853)	(0.0860)	(0.0843)
Household income (baseline = low)
Mid	0.170 **	0.174 **	0.241 ***	0.0328	0.0312	0.0289
	(0.0577)	(0.0583)	(0.0553)	(0.0485)	(0.0457)	(0.0472)
High	0.410 ***	0.427 ***	0.523 ***	0.361 ***	0.370 ***	0.356 ***
	(0.0691)	(0.0684)	(0.0738)	(0.0513)	(0.0510)	(0.0507)
Age	−0.00798 ***	−0.00842 ***	−0.0118 ***	−0.00303 *	−0.00241	−0.00286
	(0.00160)	(0.00165)	(0.00211)	(0.00150)	(0.00156)	(0.00151)
Country-level controls
Population growth	0.168 ***	0.162 ***	0.133 ***	0.0449	0.0733	0.0730
	(0.0388)	(0.0431)	(0.0383)	(0.0628)	(0.0454)	(0.0461)
Regulatory quality	0.347 *	0.283 *	0.0958	0.274 **	0.326 ***	0.346 **
	(0.140)	(0.133)	(0.0785)	(0.0884)	(0.0935)	(0.111)
LNGDP	−0.151	−0.145	−0.0126	−0.102	−0.0483	0.143
	(0.150)	(0.140)	(0.103)	(0.115)	(0.119)	(0.127)
GDP growth	0.0157 ***	0.0142 **	0.0143 ***	0.00517	0.00763	0.00754
	(0.00461)	(0.00479)	(0.00424)	(0.00445)	(0.00393)	(0.00438)
Observations	75,595	71,964	71,964	74,477	71,964	71,964
Likelihood ratio test (d.f.)		255.4 *** (1)	147.39 *** (1)		432.84 *** (1)	355.64 *** (1)

^a Industry dummies and fixed year dummies have been included in all models, but are not reported. For the likelihood ratio tests, Models 2 and 3 must be compared to Model 1; and Models 5 and 6 to Model 4. Robust standard errors in parentheses. *** p < 0.001, ** p < 0.01, * p < 0.05.

reports the results of our generalized ordered logit regressions with clustered country standard errors. The coefficients reported in Table 6 reflect statistical associations between our dependent and independent variables. We interpret these associations as directional patterns consistent with our theoretical arguments, rather than as definitive causal effects, given the observational design.

A first thing to note in the results is that adding country-level happiness significantly improves the overall model fit for both vertical-scaling entrepreneurship engagement (Model 2: $\chi \left(1\right)$ = 255.4, p < 0.001), and horizontal-scaling entrepreneurship engagement (Model 5: $\chi \left(1\right)$ = 432.84, p < 0.001). As can be seen in Model 2 in Table 6, a country’s average happiness level is directly and positively associated with relative engagement of newness-focused vertical-scaling entrepreneurship (B = 1.458, p < 0.05), supporting Hypothesis 1a. This result shows that within a happier population, entrepreneurs are more likely to engage in higher levels of newness. For horizontal-scaling start-up engagement, we observe a direct and negative coefficient of country happiness in Model 5 (B = −2.422, p < 0.05), which means it decreases the likelihood of entrepreneurs’ engagement in higher levels of expansion-focused entrepreneurial activity. This finding supports H1b.

In order to gain insight into the magnitude or size of the effects,

Table 7. Marginal effects of country happiness level and happiness inequality on relative engagement of vertical-scaling and horizontal-scaling entrepreneurship.

	Vertical-Scaling Entrepreneurship Engagement				Horizontal-Scaling Entrepreneurship Engagement
VARIABLES	Y = 1	Y = 2	Y = 3	Y = 4	Y = 1	Y = 2	Y = 3
Country Happiness	−0.249 * [−0.47 −0.03] a	0.006 [−0.03 0.04]	0.205 * [0.01 0.39]	0.039 * [0.001 0.08]	0.365 ** [0.09 0.64]	−0.307 ** [−0.54 −0.08]	−0.057 ** [−0.10 −0.02]
	(0.112) b	(0.018)	(0.097)	(0.019)	(0.138)	(0.117)	(0.021)
Country happiness	0.312 * [0.02 0.60]	−0.142 [−0.34 0.06]	−0.173 * [−0.33 −0.02]	−0.139 * [−0.28 −0.001]	−0.504 *** [−0.77 −0.24]	0.426 *** [0.21 0.64]	0.079 *** [0.03 0.13]
inequality	(0.147)	(0.103)	(0.079)	(0.071)	(0.133)	(0.110)	(0.024)

^a Confidence interval of the marginal effect. ^b Robust standard errors. *** p < 0.001, ** p < 0.01, * p < 0.05.

displays the average marginal effects of country-level happiness and happiness inequality, which provides a valuable extension to the coefficients in Table 6. To elaborate, these marginal effects indicate the effects of happiness and happiness inequality on entrepreneurs’ probability of engaging in each level of vertical-scaling or horizontal-scaling entrepreneurship, rather than indicating an average effect across all engagement levels as the coefficients do. We here explain the results for country-level happiness and entrepreneurship engagement. For level 1 of vertical-scaling start-up engagement (the lowest level), the marginal effect is negative and indicates that a unit-rise in a country’s average happiness would lower the probability of an entrepreneur engaging in this level by 0.249 (24.9%). For all other engagement levels of vertical-scaling entrepreneurship, the marginal effect is positive, indicating that a unit-increase in country happiness would raise the probability of entrepreneurs engaging in level 2 by 0.6%, level 3 by 20.5%, and level 4 by 3.9%. For horizontal-scaling entrepreneurship engagement, the effect of country-level happiness is opposite for vertical-scaling entrepreneurship engagement. For level 1 of horizontal-scaling start-up engagement, the marginal effect is positive at 0.365, indicating that a unit-rise in country average happiness would increase the probability of entrepreneurs engaging in this lowest level by 36.5%. For the other two engagement levels, the marginal effect is negative, indicating that a unit-increase in country happiness would lower the probability of entrepreneurs engaging in level 2 by 30.7%, and level 3 by 5.7%.

Regarding control variables, our individual level controls demonstrate clear demographic patterns related to both types of entrepreneurship relative engagement. Specifically, male, younger, and better-educated entrepreneurs are more likely to enter high-end relative to low-end entrepreneurship, within both the vertical-scaling and horizontal-scaling dimensions. At the country level, regulatory quality demonstrates a positive association with both dimensions of entrepreneurship engagement, as expected. Population growth and GDP per capita growth show significantly positive associations with vertical-scaling entrepreneurship engagement, but not with horizontal-scaling entrepreneurship engagement.

In summary, we find that population happiness significantly increases relative engagement in vertical-scaling entrepreneurship, especially the probability of reaching level 3 where entrepreneurs have an opportunity-based motivation and engage in one form of newness, related to either the final product or the technology used (see Table 2a). At the same time, population happiness significantly decreases relative engagement in horizontal-scaling entrepreneurship. Specifically, population happiness strongly increases the probability of engaging in the lowest level of horizontal-scaling entrepreneurship, associated with a lack of ambitions, both in terms of expected job creation and in terms of export activity (see Table 2b).

4.2. Country-Level Happiness Inequality and Individual Engagement in Entrepreneurship (H2a, H2b)

Next, we discuss the results about population happiness inequality and relative entrepreneurship engagement. Again, adding country happiness inequality significantly improves the overall model fit for both vertical-scaling entrepreneurship engagement (Model 3: $\chi \left(1\right)$ = 147.39, p < 0.001), and horizontal-scaling entrepreneurship engagement (Model 6: $\chi \left(1\right)$ = 355.64, p < 0.001). Model 3 in Table 6 shows that country happiness inequality is directly and negatively associated with relative engagement of vertical-scaling start-up (B = −1.678, p < 0.01), weakening the likelihood of entrepreneurs’ engagement in high relative to low levels of newness-focused entrepreneurial activity. Hypothesis 2a is supported. We also found a direct and positive association of country happiness inequality with horizontal-scaling start-up engagement (B = 3.346, p < 0.001, Model 6), supporting H2b that country happiness inequality increases the likelihood of entrepreneurs’ engagement in high relative to low expansion-focused entrepreneurial activity. Our findings, hence, fully support all our hypotheses.

Similarly, we interpret the marginal effects of country happiness inequality on entrepreneurship engagement as shown in Table 7. For level 1 of vertical-scaling start-up engagement (the lowest level), the marginal effect is positive, indicating that a unit-rise in country happiness inequality would raise the probability of entrepreneurs engaging in this level by 31.2% on average. For all other engagement levels of vertical-scaling entrepreneurship, the marginal effect is negative, indicating that a unit-increase in country happiness inequality would lower the probability, on average, of entrepreneurs engaging in level 2 by 14.2%, level 3 by 17.3%, and level 4 by 13.9%. For level 1 of horizontal-scaling start-up engagement, the marginal effect is negative, indicating that a unit-rise in country happiness inequality would decrease the probability of entrepreneurs engaging in this level by 50.4% on average. For other two engagement levels, the marginal effect is positive, indicating that a unit-increase in country happiness inequality would, on average, raise the probability of entrepreneurs engaging in level 2 by 42.6%, and level 3 by 7.9%.

In summary, we find that happiness inequality significantly decreases relative engagement in vertical-scaling entrepreneurship. Specifically, happiness inequality strongly increases the probability of engaging in the lowest level of vertical-scaling entrepreneurship, associated with a necessity motive for starting up the business (see Table 2a). At the same time, happiness inequality significantly increases relative engagement in horizontal-scaling entrepreneurship, especially the probability of reaching level 2 where entrepreneurs are ambitious on one dimension, either job creation or export activity (see Table 2b).

When combining our results from Section 4.1 and Section 4.2, two broad contextual profiles stand out. Profile 1 combines high national happiness with relatively lower happiness inequality. Countries with this profile are more conducive to vertical-scaling entrepreneurship (innovation/newness orientation), but place comparatively less emphasis on rapid expansion/export. Conversely, countries with lower national happiness and relatively high happiness inequality (Profile 2) are more conducive to horizontal-scaling entrepreneurship (expansion/export/growth orientation) but place less emphasis on innovation. In practical terms, in happier national contexts, entrepreneurs are more likely to engage in innovative, novelty-based projects, whereas in contexts with higher happiness inequality, they are more likely to pursue aggressive expansion and job creation.

4.3. Robustness Tests

We conducted three robustness checks. The full findings are available on request from the authors. Endogeneity could always be a concern for non-experimental studies. For “happy” or “well-being” entrepreneurship research, two common concerns for endogeneity would be the result of omitted variables and simultaneous causality, specifically an uncontrolled confounder causing both happiness and entrepreneurship engagement, and a loop of causality between the two. We conducted two robustness tests to reduce the concerns for these two issues, though we have both theoretical and empirical reasons to believe spurious effect and reverse causality would be less likely to exist for the current study.

First, a two-stage regression procedure was utilized to further reduce the spurious effect or third factor concern. In individual level studies, self-efficacy and optimism are two specific factors that have been found to be associated with both entrepreneurial engagement (e.g., Baron, 1998 [77] and Krueger, 1993 [78]) and different measures of well-being (e.g., Bradley and Roberts [79], Cooper and Artz [80], Lange [81], and Lucas et al. [82]). At the country level, we identified three variables that can represent country-level self-efficacy and optimism to a certain degree: the business confidence index from OECD statistics (only for OECD countries), healthy life expectancy from the 2017 World Happiness Report, and social support from the Gallup World Poll. Hence in the first stage, we regressed our country happiness variable on these three variables. In the second stage, we took the estimated residuals from the first regression as our measure of country happiness (which is now independent of the regressors from the first stage) and did our main analysis again, estimating the effect of country happiness on individual entrepreneurship engagement, with the same individual and country-level controls. Our main results remained intact.

Second, as mentioned in our statistical analysis Section 3.5, we adopted multi-level random effects modeling as a robustness check for our main results and as evidence against the reverse causality concern. The multi-level analyses also account for the hierarchical structure in our dataset in which individuals represent level one and countries represent level two. As explained, the two-level multi-level model does not allow for cross-sectional data with time variation, i.e., with different samples of individuals every year rather than a pure panel with repeated observations for individual respondents [73]. Therefore, we excluded the time dimension in this robustness check. We selected the entrepreneurship data from 2013 and the data of country-level happiness and other controls as the average value from 2010 to 2012. The regression results further confirm our hypotheses that there is, on relative vertical-scaling start-up engagement, a significantly positive effect of country happiness (B = 4.41, p < 0.01) and a significantly negative effect of country happiness inequality (B = −8.96, p < 0.01), whereas horizontal-scaling entrepreneurship engagement is negatively affected by country happiness (B = −6.12, p < 0.01) and is positively affected by country happiness inequality (B = 10.35, p < 0.01).

Moreover, we applied a series of adjustments to our model and measurement as our third robustness test. First, we added the country economic inequality indicator, i.e., the GINI coefficient. We found consistent result patterns. Specifically, with GINI as a control variable, the negative effect of happiness inequality on vertical-scaling entrepreneurship engagement increases (B = −1.74, p < 0.01), and the positive effect of it on horizontal-scaling entrepreneurship decreases (B = 2.91, p < 0.05). Importantly, direction and significance of our findings for happiness inequality remained intact. We also found that the GINI coefficient is not related to vertical-scaling entrepreneurship engagement, and is positively related to horizontal-scaling entrepreneurship engagement with marginal significance (B = 2.37, p < 0.10). Second, we log transformed our two main independent variables of country-level happiness and country happiness inequality, and added their squared term to our model. Furthermore, we treated our dependent variable as nominal and used a pooled multinomial model with level one as the base outcome to test whether the results are consistent with our main findings from the ordinal model. The main linear effects were all confirmed for the above analysis and there were no significant curvilinear effects for country happiness and happiness inequality on entrepreneurship engagement.

5. Discussion and Conclusions

Drawing on contextual demand and supply theory, occupational choice theory, entrepreneurial entry theory, and affect theory, the present paper studied country-level happiness and happiness inequality as drivers of entrepreneurs’ relative engagement levels in vertical-scaling and horizontal-scaling entrepreneurship, where for both dimensions, higher engagement levels are associated with higher quality entrepreneurship. After controlling for other influences, we found that country-level happiness significantly increases the likelihood of entrepreneurs within that country to pursue high-end entrepreneurship on the vertical-scaling dimension. At the same time, it significantly decreases the likelihood to pursue high-end entrepreneurship on the horizontal-scaling dimension. On the contrary, country happiness inequality increases the likelihood of entrepreneurs to pursue horizontal-scaling entrepreneurship while it decreases the likelihood of pursuing vertical-scaling entrepreneurship. In short, population happiness pushes entrepreneurs toward innovativeness but away from expansion, while happiness inequality does the opposite.

The contributions and implications of our study are two-fold: theoretical and practical. First, we have advanced theory on affect, institutions and entrepreneurship by investigating to what extent contextual country happiness and happiness inequality influence entrepreneurs’ engagement into two forms of high-quality entrepreneurship: vertical-scaling and horizontal-scaling entrepreneurship. Second, our findings also possess practical implications for policy makers. We will illustrate each of these in the following paragraphs.

5.1. Theoretical Contributions

This paper contributes to the literature on psychological entrepreneurship by demonstrating the role of an emotional context, rather than the individual emotion, for entrepreneurship. Our study also contributes to the high-quality entrepreneurship dialog [13,18,28,29] by decoupling the growth perspective into two dimensions of entrepreneurship: vertical-scaling entrepreneurship and horizontal-scaling entrepreneurship, along which entrepreneurial quality is well-differentiated. In the meantime, our findings complement theories of firm emergence (e.g., Zander [83]) and firm growth (e.g., Tong et al. [84]) providing further evidence of entrepreneurship being affected by a range of forces at different levels [85].

Moreover, the study contributes to the happiness research literature by approaching happiness as a driver of socio-economic phenomena (in this case entrepreneurship), as opposed to viewing happiness as an end goal. Our approach helps connect global challenges related to the United Nation’s Sustainable Development Goals. It is not necessary to prove that most people, perhaps all people, are eager to pursue happiness, which is consistently assumed as the ultimate human desire. Instead, we are more interested in what happiness accomplishes, as arguably, the world is made and directed by happiness [86]. From a macro perspective, accomplishing well-being is not only an SDG, but may also be a driver of sustainable development, particularly in affecting job creation and innovation as captured by other SDGs.

Furthermore, we believe our findings contribute to the wider literature on inequality, social trust, and international strategy. Economic inequality in a given country has also been linked to less social trust in that country and thus influences the corporations’ international strategies [87]. The notion of happiness inequality in this study is more adaptable for measuring the degree of a society’s sustainability, as this concept encompasses both layers of economic and social value applications at multiple levels. Similar to population happiness, reducing inequality is not only emphasized as a target in the SDGs, but may be the catalyst for other SDGs, such as inducing decent work and economic growth.

Finally, compared with Naudé et al. [88], a noteworthy attempt in investigating the relationship between country happiness and country entrepreneurship, the current study provides additional merits from several angles, separating two entrepreneurship dimensions and focusing on high-quality entrepreneurship and entrepreneurs’ relative engagement, and taking another indicator of population happiness—population happiness inequality—into consideration.

In summary, the three main contributions of our paper are as follows. First, we contribute to entrepreneurship research by showing that national affective context (happiness and its inequality) is systematically associated with two kinds of ambitious engagement pursued by entrepreneurs. Second, we contribute to well-being research by demonstrating that population-level affective climate is not only a social or public health outcome but also part of the entrepreneurial opportunity structure. Third, as also discussed below in Section 5.2, we contribute to policy debates by showing that “high-quality entrepreneurship” is not a single target; different well-being climates are associated with different ambition profiles (innovation-oriented vs. expansion-oriented), which implies different levers and different risks for policy.

5.2. Policy Implications

An increasing number of countries have adopted happiness as an indicator to assess the progress of nations. It should at least be a critical component of nations’ socio-economic development, apart from indicators such as gross domestic product, exports, unemployment, etc. Using a large dataset, our study provides novel findings on the socio-economic implications of country happiness and happiness inequality. We theoretically and empirically link well-being, inequality, growth and innovation together, and demonstrate that SDGs are interconnected and should not be targeted separately. Specifically, our work may offer two concrete implications for governments or policy makers, from our findings that average country-level happiness (happiness inequality) has a positive (negative) effect on innovation-focused entrepreneurship but a negative (positive) effect on expansion-focused entrepreneurship.

First, contextual well-being, including level and distribution, may be framed and positioned as an innovation or growth enabler apart from merely being a social desire. Second, by monitoring contextual happiness, policy makers may use it as a signal for prioritizing and balancing between different types of entrepreneurship. If governments are aware of the happiness level within their country or sub-region, they can use this information to direct scarce resources to stimulate one entrepreneurship type or another. For instance, for high-happiness—low happiness inequality countries (Profile 1 mentioned in Section 4.2), there is a relatively strong positive impact on vertical-scaling entrepreneurship, such that stimulation by government of this type of entrepreneurs may be less urgent. In contrast, in such countries there is a relatively strong negative impact on horizontal-scaling entrepreneurship, so that government support to this type of entrepreneurs may be more opportune. In this way, happiness policies can complement entrepreneurship policy in supporting the entrepreneurial ecosystem.

In this regard, our results imply a number of policy challenges for the two contextual profiles identified in Section 4.2 (high-happiness versus high happiness-inequality countries). In high-happiness environments (Profile 1) that tend to nurture vertical-scaling (innovation/newness), the relevant policy actors are those supporting R&D, frontier technology, and early-stage opportunity recognition (innovation agencies, research-oriented incubators, IP/commercialization support agencies, etc.). The policy risk for Profile 1 countries is that such environments may generate strong novelty but not necessarily broaden job creation or export expansion unless those channels are also supported.

In high happiness-inequality environments (Profile 2) that tend to nurture horizontal-scaling (expansion/export/employment), the relevant policy audience is labor/employment ministries, export promotion agencies, SME support, and regional development agencies. The policy risk here is that strong growth/expansion dynamics can come with highly competitive pressures and may not include much radical innovation; there is a temptation to celebrate job creation without asking about quality, inclusivity, or resilience.

5.3. Limitations

Our study has limitations related to the databases that we used. First, an obvious limitation is that we cannot control for happiness at the individual level as this variable is not available in the GEM dataset. However, we do control for education and household income, two important determinants of individual level happiness (e.g., Easterlin, [89], Frey and Stutzer [90], Pouwels et al. [91], Chen, [92], and Noddings, [93]), such that we do not expect the coefficients of our country-level happiness measures to be affected by this omission.

Second, our indicators of vertical-scaling and horizontal-scaling entrepreneurship engagement are based on self-reported expectations (perceived newness, expected job creation, etc.). They are well-established indicators in the GEM, but are still subjective. On one hand, we need more objective behavioral evidence to confirm that the self-reports of entrepreneurs are consistent with their real actions. On the other hand, we need more detailed survey questions to reflect the scaling in scale-up (or scale-out) entrepreneurship in order to measure it as a continuous variable rather than a discrete ordinal one.

Third, although omitted variable bias is rather unlikely due to the inclusion of several macro-level control variables in our model, we cannot completely rule out that certain unobserved contextual variables related to national culture may have influenced the results. Future research should investigate possible moderating effects of culture on the relationship between contextual happiness and high-quality entrepreneurship.

Fourth, the use of aggregated national happiness scores may obscure within-country (i.e., regional) differences that could influence entrepreneurship. Once regional-level happiness scores become available, future research should conduct the analysis at the regional level, in order to obtain more accurate insights.

Fifth, the results describe directional associations, not definitive causal effects. Sixth and finally, the analysis focuses on early-stage entrepreneurs only. The ambition structure we capture may not generalize to later-stage firms whose trajectories are already path-dependent.

5.4. Suggestions for Future Research

First, investigating the role of other forms of happiness in entrepreneurship is crucial both for research and for policymaking. According to the field of positive psychology, happiness is not only a concept of emotion, but also a state of engagement [94]. It could also be an indicator of morality, represented by how much you serve causes that are bigger than yourself [95]. If policy makers seek to take happiness into consideration, they need to consider positive emotion and also other forms of happiness such as positive evaluation, positive engagement, and positive virtues.

Second, future work may decompose happiness inequality into structural inequality of opportunity and inequality of outcomes to see the effect of each type of inequality on entrepreneurship engagement. In order to achieve that, concrete measures are needed to obtain proportions of social and economic inequality in the overall happiness inequality scores.

Third, future research may focus on a possible moderation effect of countries’ development stages (see Table 4), i.e., whether the relationship between contextual happiness and relative engagement in vertical-scaling and horizontal-scaling entrepreneurship differs between factor-driven, efficiency-driven and innovation-driven economies. In addition to the economic development level, other potential contextual moderators that could be investigated in future research include culture and institutional conditions, as well as variations in social capital, welfare systems, and entrepreneurial infrastructure.

Fourth, while our empirical model assumes a linear relationship between the key independent variables (happiness level and happiness inequality) and entrepreneurial engagement outcomes, this assumption may be theoretically restrictive. For example, a non-monotonic or inverted-U-shaped relationship could plausibly exist, where moderate levels of happiness are associated with the highest entrepreneurial aspirations. In contrast, low levels of happiness may drive necessity-based entrepreneurship, while very high levels may lead to risk aversion and preference for stability. Future research could explore non-linear functional forms—such as spline regression, categorical transformations of happiness levels, or polynomial specifications—to capture such threshold or saturation effects.

Fifth, future research could conduct deeper investigations on the construct validity of the new measures employed in our paper. We used participation in vertical- and horizontal-scaling entrepreneurship as proxies for high-quality entrepreneurial activity. However, it is not certain that these proxies accurately reflect long-term entrepreneurial outcomes such as profitability, survival rates, or job creation. If these proxies are only weakly correlated with actual performance, then the conceptual linkage between “quality” and scaling behavior may be problematic. Additional validation using longitudinal or outcome-based indicators would help strengthen the construct validity of entrepreneurial quality in future work.

Sixth, happiness might also interact with other factors in affecting entrepreneurship. Country happiness might not be linked to entrepreneurship directly but under certain conditions related to the institutions in place. Institutions often interact in measurable ways and if theory is to guide empirical work, it must develop more sophisticated portrayals of institutional effects.

In conclusion, with this paper we are happy to make a contribution at the crossroads of happiness and entrepreneurship research by introducing the role of contextual happiness into entrepreneurship research. We have made the first attempt to link country-level happiness to two distinct types of individual-level entrepreneurship related to different types of ambitions: innovation-oriented versus expansion-oriented entrepreneurship. Future research is strongly needed to extend this dialog.