Is Self-Employment a Career Trap? A Large-Scale Field Experiment in the Labor Market

Abstract

We conduct a large-scale labor-market experiment using more than 8,000 fictitious résumés to uncover the demand-side mechanisms behind the adverse treatment of the self-employed entering wage employment. We find that, compared to wage earners, self-employed individuals face lower callback rates across industries and skill groups. This adverse treatment is concentrated in the lower-skilled, non-managerial market of associate professionals, with a 28% drop in callback. Results suggest that self-employment leads to the development of generalist skills (useful for managerial roles) at the cost of specialist skills. Furthermore, perceived behavioral fit of the self-employed seems to be penalized in associate positions.

Keywords

self-employment field experiment skills mismatch managerial skills specialist skills

Introduction

Governments encourage entrepreneurship, a call to which many respond by taking up or continuing in self-employment. However, whether self-employment results in net benefits is an open question. Pecuniary benefits in entrepreneurship on average are lower than in wage employment due to lower earnings or slower income growth (Hamilton, 2000; Hyytinen et al., 2013; Moskowitz & Vissing-Jorgensen, 2002). Entrepreneurs may enjoy non-pecuniary benefits such as additional job satisfaction (Åstebro et al., 2014; Georgellis et al., 2007; Stephan, 2018) or generalist experience (Lazear, 2004), but these are hard to quantify. Crucially, extant evidence shows that former entrepreneurs face challenges in the wage labor market (Baptista et al., 2012; Bruce & Schuetze, 2004; Kaiser & Malchow-Møller, 2011; Koch et al., 2021). The core question is whether these challenges are due solely to self-selection (supply-side factors) or if they are compounded by adverse treatment from prospective employers (demand-side factors).

To study the demand-side factors in isolation, correspondence experiments have emerged. These methods are used to investigate the job prospects of entrepreneurs by randomly varying the experience of candidates in applications and measuring the callback rate for interviews (Aldén et al., 2021; Botelho & Chang, 2023; Kacperczyk & Younkin, 2021; Koellinger et al., 2015). Koellinger et al. (2015) and Aldén et al. (2021) showed that applicants with self-employment experience receive adverse treatment in the job market. Botelho and Chang (2023) also found that candidates are less likely to receive a callback for an interview if they have founding experience. While there are specific boundary conditions, such as Kacperczyk and Younkin (2021), who rely on the stereotype conflict theory showing that the adverse treatment of founders is less pronounced for female candidates with a history of co-founding a new venture that employs people, a consistent pattern remains evident. That is, some evidence shows adverse treatment of the self-employed compared to wage earners on the demand side, and that founding experience seems to be penalized in the recruitment process.

However, it is unknown to which extent differential treatment of the self-employed by prospective employers is generalized across skill groups and industries, which limits understanding of the underlying demand-side mechanisms. Our study addresses this critical gap. The main contribution of this study is to systematically uncover the demand-side mechanisms underlying adverse treatment for solo self-employed individuals by conducting a large-scale correspondence field experiment that, within the same experimental context, simultaneously tests vertical heterogeneity (across skill groups) and, additionally, examines the generalizability of adverse treatment horizontally across distinct industries.

To assess job prospects for solo self-employed individuals and wage earners at the selection stage of the recruitment process, we sent over 8,600 fictitious résumés to 2,150 real job vacancies advertised in the Russian labor market in a relatively stable economic and entrepreneurship period (from March to August 2017). We randomly assigned employment status (solo self-employed or wage earner) and other applicant characteristics. We chose three industries and, within each industry, targeted positions requiring different skill levels: high-level managerial or lower-skilled associate professional jobs, based on the International Standard Classification of Occupations (ISCO).

We find that the solo self-employed, compared to wage earners, are adversely treated at the application stage of the recruitment process on average across industries. More importantly, we find that this pattern of adverse treatment of the self-employed is concentrated in associate professional positions, with almost equal treatment of the self-employed and wage earners in high-skilled managerial positions. Exploratory analysis reveals the same relationship between managerial skill level and unfavorable treatment of self-employed applicants at a more granular level of occupations. That is, evidence shows that prospective employers appreciate the development of self-employment experience in managerial positions, but are concerned that it may not be a good fit for associate professionals’ roles.

Theoretical Framework

The ISCO (2008) defines a job as “a set of tasks and duties performed, or meant to be performed, by one person, including for an employer or in self-employment.” Research on occupational choice echoes this definition by assuming equal employment opportunities for self-employed workers and wage earners with similar characteristics (Evans & Leighton, 1989; Hamilton, 2000; Praag & Ophem, 1995). Moreover, Manso (2016) showed that short-term self-employed workers, once they return to wage employment, earn a comparable rate as those who were never self-employed. Similarly, Hyytinen and Rouvinen (2008) reported the same effect after controlling for unobserved differences in productivity. Thus, candidates with self-employed or wage-earning experience, but otherwise comparable skills should be treated equally in the labor market.

However, demand-side research increasingly contradicts this assumption (Aldén et al., 2021; Botelho & Chang, 2023; Kacperczyk & Younkin, 2021; Koellinger et al., 2015). A growing body of experimental and quasi-experimental work documents that applicants with self-employment backgrounds receive less favorable responses from employers than otherwise identical wage earners. Because field experiments isolate employer behavior, they provide direct evidence of demand-side responses, making this literature particularly informative (Aldén et al., 2021; Botelho & Chang, 2023; Kacperczyk & Younkin, 2021; Koellinger et al., 2015). One of the earliest field experiments examining self-employment was conducted by Koellinger et al. (2015), who sent around 1,200 applications for human resource manager positions in the Netherlands and found that former self-employed HR professionals were significantly less likely to receive callbacks. The authors argue that employers may interpret self-employment as a sign of difficulty adapting to hierarchical structures or organizational norms. Kacperczyk and Younkin (2021) showed that in the United States, candidates with founding experience are penalized, suggesting that employers view former founders as a poorer fit for traditional organizational structures and less inclined or prepared to work in standard employee roles.¹ Their study also shows that penalties for nonlinear career paths are not fixed but can change depending on who the applicant is, as some demographic characteristics appear to signal commitment or fit in ways that reduce employer concerns about entrepreneurial backgrounds. Botelho and Chang (2023), using a large-scale field experiment among software engineers, found that former founders, especially those with successful ventures, receive fewer callbacks than wage earners, potentially because employers infer lower commitment or a higher likelihood of turnover. Aldén et al. (2021) provided evidence from Sweden showing that former self-employed candidates applying for accountant and software developer positions receive fewer interview invitations than comparable wage earners, even when the self-employed period is short. Taken together, these studies consistently document adverse treatment of the self-employed, yet they focus on specific occupational niches and therefore do not assess whether this treatment generalizes across industries or skill levels.

Insights into the mechanisms underlying differential treatment, which have tested whether labor-market penalties vary vertically across skill hierarchies and horizontally across distinct industries, can be drawn from parallel experimental literature on other characteristics, such as ethnicity, gender, and unemployment duration. Focusing on those specific categories rather than self-employment, this external literature reveals vertical heterogeneity in the differential treatment of applicants. Some studies find larger callback gaps for ethnic minorities in lower-skilled occupations (Bursell et al., 2021; Carlsson & Rooth, 2007), while others observe stronger penalties for ethnic or gender groups in managerial roles or no systematic relationship (Adamovic & Leibbrandt, 2023; Park & Oh, 2025). Simultaneously, significant horizontal heterogeneity has been documented, where the differential treatment of ethnic and gender groups varies across industries regardless of skill level (Galos & Coppock, 2023; McGinnity et al., 2009). Related studies on unemployment duration, which provide a closer comparison to the self-employment context, similarly present inconsistent patterns regarding whether penalties for that specific career interruption are more severe in low- versus high-skilled positions (Bonoli, 2014; Eriksson & Rooth, 2014). Together, these contrasting findings suggest that employer reactions to career interruptions are not uniform but are highly dependent on the specific signaling mechanisms active across different industries and skill levels.

These insights offer a useful foundation for theorizing about how employers evaluate self-employment experience across industries and skill levels. As the existing experimental evidence on labor market outcomes for applicants with self-employment experience study specific occupations without covering managerial positions mainly focusing on specific occupations (Aldén et al., 2021; Botelho & Chang, 2023; Kacperczyk & Younkin, 2021; Koellinger et al., 2015), it remains difficult to identify systematic patterns in how employers evaluate such candidates across different occupational settings or to determine the main mechanisms driving their adverse treatment compared to wage earners.

Theoretical entrepreneurship and labor literature suggest that several distinct mechanisms may drive these penalties, reflecting different dimensions of employer uncertainty and valuation (Blanchflower & Oswald, 1998; Hartog et al., 2010; Koellinger et al., 2015; Landier, 2005; Lazear, 2012). One explanation relies on the generalist versus specialist skill framework. Self-employment frequently involves handling diverse responsibilities, such as financial planning, marketing, supplier coordination, and client management. These generalist skills may translate well into managerial positions, where coordination and oversight are central (Baptista et al., 2012; Lazear, 2012). In contrast, specialist skills required for lower-skilled technical or associate roles may depreciate during self-employment, particularly when entrepreneurial tasks lack standardization (Hartog et al., 2010). Under this framework, employers may view self-employment backgrounds as mismatch with the requirements of lower-skilled specialist roles, implying stronger adverse treatment in these jobs.

Alternatively, demand-side adverse treatment may arise from concerns regarding behavioral and organizational fit. A prospective employer might have concerns that a self-employed candidate does not fit well into a traditional role, assuming they have non-pecuniary preferences that make them less team-oriented or a lone wolf (Blanchflower & Oswald, 1998; Cooper & Saral, 2013; Hurst & Pugsley, 2011). Supporting this stereotype, research indicates that the self-employed often prioritize personal autonomy over financial gain (Reynolds & Curtin, 2008), and may even be willing to bear costs to avoid group decision-making (Masclet et al., 2009). This concern regarding potential resistance to supervision due to excessive independence generates a specific expectation for vertical heterogeneity. While independence is generally valued in leadership, it is often viewed as a liability in subordinate roles, suggesting that adverse treatment may be more pronounced in intermediate and non-managerial positions.

Another theoretical perspective is derived from the uncertainty perspective formulated by Mahieu et al. (2019), who argued that employers may perceive the self-employed as harder to evaluate, especially for roles involving higher responsibility. From this perspective, the lack of clear information associated with entrepreneurial histories could increase concerns about reliability, organizational fit, or role expectations. Their empirical results, based on data from a developed country (Belgium), show that penalties for self-employment are larger among higher skilled and higher paid occupations. However, this evidence is based on observational evidence not allowing to study the hiring practices in isolation and employer expectations that vary across settings.

In a nutshell, with these opposing theoretical predictions and the limited existing evidence demonstrating the generalizability of adverse treatment of the self-employed applicants across different industries and skill groups, the question of how employers evaluate self-employment experience across the labor market remains open. While previous self-employed experience in non-managerial positions is penalized (Aldén et al., 2021; Botelho and Chang, 2023; Kacperczyk and Younkin, 2021; Koellinger et al., 2015) whether penalties for the self-employed returning to wage employment persist vertically across skill levels or whether they generalize horizontally across industries is unknown. The central contribution of our study is to address this gap by examining whether adverse treatment of the self-employed varies across a spectrum of industries and skill groups, including managerial positions, thereby enhancing the theoretical understanding of adverse treatment of self-employed. We address this question by conducting a large-scale correspondence experiment that systematically varies self-employment experience in applications sent to jobs across multiple industries and skill groups.

Experimental Context

Russia in 2017 provides a valuable setting for examining labor-market outcomes of self-employment because, although its entrepreneurial environment developed along a trajectory shaped by the post-Soviet transition and institutional specificities, by the time of our experiment it had reached a relatively stable and more structured phase, broadly comparable to that observed in many middle- and high-income economies (ILO, 2017). Russia’s levels of self-employment and societal perceptions of entrepreneurship were close to international averages, and 2016 marked the peak of entrepreneurial activity in the country (The Global Entrepreneurship Monitor [GEM], 2016/2017; ILO, 2017; Verkhovskaya & Aleksandrova, 2017).

Following the collapse of the Soviet Union in the 1990s, entrepreneurship emerged as a new model of economic activity, attracting mostly young, well-educated individuals (Gerber, 2001). While the shift to a market economy created substantial opportunities for self-employment, it simultaneously introduced significant barriers linked to institutional instability, weak property rights, and inconsistent regulations (Aidis, Korosteleva, & Mickiewicz, 2008; Puffer et al., 2010), with entrepreneurs historically relying on informal networks and alternative institutional mechanisms to mitigate uncertainty (Aidis, Estrin, & Mickiewicz, 2008; Marques et al., 2020). By the mid-2010s, policy reforms sought to ease these constraints: registration procedures were simplified, tax obligations reduced, and entry barriers for solo self-employed individuals lowered (Kovalenko et al., 2019), though several institutional weaknesses including limited legal enforcement and bureaucratic obstacles persisted (Barinova et al., 2020; Marques et al., 2020).

At the time of our experiment in 2017, however, the landscape had largely settled into a more consistent pattern, which is reflected in national indicators: according to the International Labour Organization (ILO) Report, in 2017, 6.6% of the Russian workforce were self-employed, compared with 6.3% in the United States and 10.2% in Germany (World Bank Group, 2017). Additionally, 63.4% of respondents in Russia considered entrepreneurship a good career choice (global average: 62%), ranking Russia 31st out of 61 countries, while 65.6% assigned high status to entrepreneurs (39th out of 61 countries; GEM, 2016/2017). Conducting the experiment before the subsequent health crises and geopolitical turbulence allows us to examine employer behavior in a comparatively stable period, and to assess the likelihood of self-employed applicants entering wage employment.

Understanding how self-employment fits into broader labor-market trajectories is essential for the external validity of this study. In particular, analyzing the prevalence of transitions between self-employment and wage employment in Russia relative to other countries helps interpret the employer responses observed in our experiment. Existing studies covering a broader set of developing and transition economies document more limited mobility between wage employment and self-employment (Pagés & Stampini, 2009). In contrast, comparative evidence from high-income economies shows that transitions between self-employment and wage employment are common and that self-employment often constitutes a temporary or reversible career phase rather than a terminal labor market state (Bruce & Schuetze, 2004; Hamilton, 2000; Hyytinen & Rouvinen, 2008; Kaiser & Malchow-Møller, 2011).

Longitudinal evidence from Russia points in a similar direction. Using data from the Russian Longitudinal Monitoring Survey, researchers show that transitions between self-employment and wage employment are a regular feature of Russian labor-market careers, indicating that self-employment forms part of integrated career trajectories rather than a closed or absorbing state (Gimpelson & Kapeliushnikov, 2014). For the broader labor force, the likelihood of securing a wage job is not severely hurt if the applicant’s prior experience is in self-employment. In a flexible labor market, this background can actively function as a “stepping stone” rather than a dead end, as transition rates back to wage employment remain comparable to those from other origin states (Slonimczyk & Gimpelson, 2015).

Studying the Russian labor market therefore allows us to analyze how self-employment signals operate in an economy with still-developing market institutions but established patterns of fluid mobility, while assessing whether employer responses align with those observed in more mature entrepreneurial contexts.

Methodology

Experimental Design

To assess job prospects for solo self-employed individuals and wage earners at the selection stage of the recruitment process, we sent over 8,600 résumés to 2,150 job vacancies advertised in the Russian labor market between March and August (data collection was completed in December 2017). We randomly assigned (a) if the applicant was self-employed or worked for a company and (b) if their résumé included an objective section. We applied within-subject design to control for vacancy-specific effects: four types of résumés were sent in random order for each vacancy.²

To indicate a difference in employment status, we randomize within vacancy, if the applicant states that (s)he is self-employed, “Individualnyy predprinimatel,” or has wage-earning experience, indicated through randomly selected names of companies. We cross-randomized the objective section to analyze if reducing uncertainty about non-pecuniary preferences of applicants benefits self-employed individuals more than it does with wage earners; specifically, how it influences the likelihood to be invited for a job interview.

Before performing the experiment, we consulted several human resources managers from large and medium-sized companies on their hiring practices. The consultations show that potential employers are less likely to invite a self-employed person for an interview because of concerns regarding the applicant’s ability to fit in or commit to the company (regarding the mismatch between the preferences of self-employed and organizational culture), or why they are searching for a job in the wage market, which is consistent with theoretical literature.³ Therefore, we designed the objective sections to articulate implicitly or explicitly the reason for changing jobs: “I would like to move to a new professional level,”“gain experience in new areas”; or reduce concerns about not being a team player: “working in a team,”“I see myself as a valuable member of a team.”⁴

The résumé construction follows the standard correspondence experiment framework (Bertrand & Mullainathan, 2004; Lahey & Beasley, 2016), in which only predefined treatment dimensions are systematically randomized within vacancy (type of employment and if their résumé included an objective section), while other résumé characteristics are drawn at random from bank of sections (name, surname, patronymic, age, length of experience, job description based on real résumés, etc.) but in line with predefined rules, for example, if applicant is female, her name, surname, patronymics have specific feminine Russian language norm. Adhering to these established best practices (Lahey & Beasley, 2016) allows us to control for potential bias, maintain a realistic résumé, and increase the external validity to the pool tested (see details in section “Experimental Procedure”).

To keep the setting realistic, we chose occupations which commonly included both solo self-employed individuals and wage earners; associate professionals and managers; and which have high demand for applicants. This narrowed our range of choices to the following three industries: Finance, information technology, and public relations. Further we choose two from the first three major skill groups according to the ISCO: Managers (group 1, skill levels 3 and 4) and associate professionals (group 3, skill level 3). Managers are generally considered high-skilled, while associate professionals are considered lower-skilled. We applied for similar occupations in each industry. See Table 1 for occupations under study.

Table 1.

Occupations Under Study.

Occupational group	Industry
Occupational group	Information technology	Finance	Public relations
Managers (ISCO Group 1)	Information and Communications Technology Services Managers (1330)	Finance Managers (1211)	Advertising and Public Relations Managers (1222)
Associate Professionals (ISCO Group 3)	Information and Communications Technology Operations Technicians (3511)	Accounting Associate Professionals (3313)	Conference and Event Planners (3332)

Note. ISCO codes 2008 in parentheses. ISCO: International Standard Classification of Occupations.

This careful sectoral targeting also follows Lahey and Beasley’s (2016) recommendation to situate experiments in policy-relevant labor market segments where treatment effects can be meaningfully interpreted. All applications and résumés corresponded to the jobs advertised in sector, responsibilities, and duties. All fictitious applicants have been employed in a single position for several years after graduating by the time of the job application.⁵

Experimental Procedure

Résumé-Generating Process

To begin creating our fictional résumés, we first analyzed a large set of real résumés posted online. This allowed us to create a database of résumé sections. All our résumés include the following sections: Heading (name, mobile telephone, e-mail, date of birth), Work experience, Education, Professional skills, Personal qualities, Advanced training, and Language skills. We developed a program that generates résumés based on our database of résumé sections. It randomly chooses applicant’s gender and constructs full names. It randomly selects information from every section to create a full résumé according to the target occupation and the type of résumé sent. The layout of the résumés is also random.

Apart from other elements, we consider that the duties of self-employed individuals can differ from those performed by wage earners, and this can influence an employer’s hiring decision (Lazear, 2005). Since this might influence experimental validity, we constructed a database that combined the job duties of both self-employed persons and wage earners. Neither of the sets of résumés contained information which might make an employer favor a particular candidate based on their job responsibilities. For example, the data did not provide information on how many workers are or were supervised by the applicant. To avoid a quality gap between applicants with self-employed and wage-earning experience, all résumés included lists of additional training selected at random but relevant to the industry of application.

Gender was randomly assigned, respectively full names were randomly constructed, and dates of birth were drawn from a predefined range. Each applicant’s educational and career trajectory was then constructed algorithmically, following typical Russian educational paths: the program randomly selected a school starting age of 6 or 7, added 10 years of schooling and 5 or 6 years of higher education also selected randomly, after which professional experience began soon afterward.

Application Process

We identified suitable vacancies through systematic, exhaustive monitoring of online job search websites. We used hh.ru (HeadHunter) as the primary search engine for job offers—the dominant portal in the Russian labor market.⁶ We also used other, much smaller platforms, such as Career.ru, Indeed.ru, employment-service.ru, and online newspapers, to exhaustively search for vacancies. The proportion of job offers on these smaller platforms was low (fewer than 10% of vacancies in our sample). Searches were conducted twice per week, selecting five newly posted vacancies for each of the six target occupations, resulting in 60 vacancies and 240 résumés sent weekly. This regular schedule ensured an exhaustive and representative sample of active employers in the Russian labor market and increased comparability within occupations across skill levels over time (see Holland, 1986 on inference based on the homogeneity assumption).

We sent résumés directly to prospective employers via e-mail. In some cases, the advertisement included contact information; in others, when contact details were not listed in advertisements, we obtained it from the Career or Vacancies pages on the companies’ websites.

According to Russian recruitment process, applications at this stage typically consist of a structured résumé submitted directly through online portals. These résumés follow standardized formats, with pre-defined fields for personal details, education, work experience, and skills. Cover letters are optional, usually brief, more like cover notes, and often generic, while reference letters or certificates are rarely required at the initial stage.

Together, these institutionalized recruitment practices created a setting that is unusually well-suited for correspondence experiments. The reliance on standardized résumé formats, the dominance of a few online platforms in the formal sector, and the absence of requirements for supplementary documents at the initial stage mean that experimental manipulations of résumé content are highly visible to employers while other potential confounders remain minimized.

Mailing

Our computer program sent four types of résumés for each vacancy in random order, with a large time lag between them to minimize detection and avoid submissions immediately before weekends. This ensured that employers did not receive all four applications simultaneously (i.e., on Monday) and created intentional intervals between submissions.

All applications included a cover note in which the candidate expressed their interest in the advertised position. Four templates for brief, standardized cover note consistent with local recruitment norms were created and randomly assigned to preserve realism while avoiding uniformity. Examples (translated from Russian) include: “Hello! Please consider my application for the position of <vac> posted on <site_job>. My résumé is attached. Sincerely, <name>,” and “Good day! My professional skills match your requirements for the <vac> position posted on <site_job>. I kindly ask you to review my attached résumé. <name>.”

To ensure the application is realistic and targeted to a specific vacancy, the program includes an electronic form with the following fields: job vacancy, recipient’s e-mail, web source, and name of contact person (if provided). This information was used to generate the cover note for the subsequent four applications and attached to randomly generated résumés. In several cases, the program failed to send all four types of résumé for some vacancies; consequently, both the vacancies and the résumés were excluded from the main analysis. A total of 8,651 résumés were submitted, of which 8,328 are included in our analysis.⁷ Table 2 shows the balance among the résumé characteristics: we achieved balance in applicant characteristics, industries, and occupations, in line with our experimental design.

Table 2.

Summary Statistics of Résumés.

Characteristics	Self-employed	Wage earners	Associate professionals	Managers
Number of résumés (N)	4,164	4,164	4,268	4,060
Objective section (%)	50.0	50.0	50.0	50.0
Managerial position (%)	48.8	48.8	0.0	100.0
Experience in months	52.96 (11.32)	52.75 (11.26)	52.92 (11.29)	52.78 (11.30)
Age in years	28.41 (0.58)	28.40 (0.58)	28.41 (0.58)	28.40 (0.58)
Female (%)	49.3	50.8	50.1	50.0
Univ. min. score	48.88 (8.86)	48.63 (8.67)	48.64 (8.68)	48.88 (8.85)
Industry (%)
Finance	35.4	35.4	35.1	35.6
IT	34.4	34.4	35.0	33.8
Marketing	30.3	30.3	29.9	30.6
Occupation (%)
Accounting prof.	18.0	18.0	35.1	0.0
IT technicians	17.9	17.9	35.0	0.0
Event planners	15.3	15.3	29.9	0.0
Finance managers	17.3	17.3	0.0	35.6
IT managers	16.5	16.5	0.0	33.8
PR managers	14.9	14.9	0.0	30.6

Note. The table reports means or percentage for the résumé characteristics listed on the left. The standard deviation for continuous variables is in parentheses. IT: information technology; PR: public relations.

Response Tracking

All résumés included mobile telephone numbers and e-mail addresses in their contact information. We tracked responses via phone or e-mail. We assigned a unique phone number to every type of résumé (out of 8: self-employed or wage earner × Objective section present or absent × High or low-skilled market) and generated a unique e-mail mailbox for each candidate.

Most potential employers in Russia used phone numbers to contact applicants for interviews. An invitation to an interview or a request for additional information was sent by e-mail and not followed up with a phone call. We define our main outcome variable, “callback” rate, as whether an employer calls or emails the applicant to invite them to an interview. We use this “narrow” definition of callback in the main analysis, reflecting the prevailing standard, with 67% of authors in the comprehensive meta-analysis by Lippens et al. (2023) reporting “narrow” callback. Of all applications, 5.7% resulted in a callback in the narrow sense (an invitation to a phone interview), while an additional 1.2 percentage points, or 17% of additional cases, involved only an e-mail request for further information, not followed up by a phone call (broad callback). We use the narrow definition as our primary outcome. For comparability with previous studies (Koellinger et al., 2015), we, however, also analyze and report a “broad” definition that includes a request for additional information (see Supplemental Appendix A).

To ensure the correct identification of the résumés that resulted in a response and to verify the candidate’s interest, we called the recruiter back. Requests for additional information (such as salary expectations, possession of a driving license, or willingness to work a set schedule) and any invitations received were politely declined to avoid interfering with the actual recruitment process. In a few cases, multiple attempts failed to get a response; in some other cases, the employer phone number was not revealed. Since we could not confirm the interest of these employers in the applicant, we dropped these cases from our analysis.⁸

This experimental design follows the recommended technical standards for correspondence studies outlined in Lahey and Beasley (2016), ensuring both methodological robustness and external validity.

Results

Summary

Table 3 and Figure 1 present callback rates by employment status. The solo self-employed received fewer on average callbacks than applicants with wage employment experience. The callback rate is 28% higher for the applicant with wage-earning experience (1.13 percentage points over base rate of 6.39).

Table 3.

Summary Table of Callback Rates.

Occupational group	Self-employed	Wage earners	Ratio
Overall	5.04	6.44	1.28
Associate professionals	6.42	8.95	1.39
Managers	3.6	3.79	1.05
Number of résumés (N)	4,164	4,164

However, this result is driven by the disparity between callback rates for the self-employed and wage earners in lower-skilled associate professional positions. Wage earner applicants in these positions receive 39% more callbacks than the self-employed (2.53 percentage points over the base rate of 8.95), whereas we observed no advantage for wage earners in managerial positions. Wage earners in managerial positions receive only 5% more callbacks than the self-employed in similar positions (0.19 percentage points over the base rate of 3.6).⁹

Pre-Specified Regression Analysis

To assess the significance of the observed results, we use a linear probability model with fixed effects for vacancies and months. We provide the analysis strictly in line with the registered pre-analysis plan. Specifically, we use the next regressions as explicitly pre-specified in the pre-analysis plan, but we use a conservative linear probability model and fixed effects panel structure¹⁰:

\begin{array}{l} {callback}_{i} = β_{0} + β_{1} \cdot {SELF}_{i} + β_{2} \cdot {ASPROF}_{i} + β_{3} \cdot O_{i} \\ + \sum_{v = 1}^{v - 1} γ_{v} \cdot D_{i}^{v} + \sum_{m = 1}^{m - 1} δ_{m} \cdot M_{i}^{m} + ε_{i} \end{array}

(1)

\begin{array}{l} {callback}_{i} = β_{0} + β_{12} \cdot {SELF}_{i} + β_{22} \cdot {ASPROF}_{i} \\ + β_{32} \cdot O_{i} + β_{4} \cdot ({SELF}_{i} \cdot {ASPROF}_{i}) + \sum_{v = 1}^{v - 1} γ_{v} \cdot D_{i}^{v} + \sum_{m = 1}^{m - 1} δ_{m} \cdot M_{i}^{m} + ε_{i} \end{array}

(2)

\begin{array}{l} {callback}_{i} = β_{0} + β_{31} \cdot {SELF}_{i} + β_{32} \cdot {ASPROF}_{i} \\ + β_{33} \cdot O_{i} + β_{5} \cdot ({SELF}_{i} \cdot O_{i}) + \sum_{v = 1}^{v - 1} γ_{v} \cdot D_{i}^{v} + \sum_{m = 1}^{m - 1} δ_{m} \cdot M_{i}^{m} + ε_{i} \end{array}

(3)

In Equations 1 to 3—callback_i is the callback for applicant i, SELF_i equal to one if applicant has self-employment status and zero otherwise; $ASPRO F_{i}$ is equal to one if the application is sent to the high-skilled market of associate professionals and zero otherwise; O_i is equal to one if the résumé contains an objective section and zero otherwise; $D_{i}^{v}$ denotes dummy for vacancy v (vacancy fixed effect); $M_{i}^{m}$ denotes month m when vacancy sent (month fixed effect); $ε_{i}$ denotes error term. Table 4 reports the results from specifications 1 and 2; those from specification 3 are reported in the text.¹¹ In Table 4, as a robustness check, we add a set of controls in models 2 and 4.

Table 4.

Callback Rate and Self-Employed Status.

Variables	Dependent variable:
	Callback
	(1)	(2)	(3)	(4)
Self-employed	−0.01*** (0.004)	−0.01*** (0.004)	−0.002 (0.01)	−0.002 (0.01)
Self-employed × Associate professionals			−0.02*** (0.01)	−0.02*** (0.01)
Associate professionals	0.00 (0.12)	0.25** (0.12)	0.01 (0.12)	0.26** (0.12)
Constant	0.002 (0.09)	−0.01 (0.09)	−0.004 (0.09)	−0.02 (0.09)
Controls	No	Yes	No	Yes
Industry fixed effects	No	Yes	No	Yes
Vacancy fixed effects	Yes	Yes	Yes	Yes
Month fixed effects	Yes	Yes	Yes	Yes
Number of résumés (N)	8,328	8,328	8,328	8,328
Adjusted R²	.43	.43	.43	.43

Note. Linear probability models. In columns (2) and (4), we control for the applicant’s gender, objective section, if the cover note has a contact name, order of the letter, subject line, type of cover message, length of experience, and industry fixed effects.

p < .1. **p < .05. ***p < .01.

Table 4, column (1), shows that the self-employed are one percentage point significantly less likely (p-value = .001; p_adj-value = .002) to be called for an interview than wage earners (6% callback rate among wage earners). Put differently, self-employed are 22% less likely to receive a callback for an interview than wage earners. The magnitude and statistical significance of these results remain unchanged if we include the set of control variables, for example, gender, presence of objective section (see Table 4, column 2).

Next, we assess whether the treatment of self-employed candidates varies by job ISCO skill level to test theories about the nature of the unfavorable treatment of the self-employed (see Table 4, column 3). We observe heterogeneous treatment of self-employed applicants across markets (p-value = .002; p_adj-value = .012), with unfavorable treatment concentrated among associate professionals. Self-employed associate professionals are two percentage points less likely to receive a callback than associate professional wage-earners—equivalent to 28% less likely to receive a callback than associate professionals wage earners (9% callback rate).¹² We find that self-employed associate professionals are less likely to receive callbacks, consistent with the idea that self-employment as an entrepreneurial activity requires a generalist skill set (Baptista et al., 2012; Lazear, 2012).

Finally, we test whether the objective section mitigates the differential treatment of the self-employed. We see suggestive evidence that the objective section increases the callback rate both for self-employed and wage-earners (p-value = .054; p_adj-value = .27).¹³ However, we find no evidence that the increase in the number of callbacks for self-employed is higher than for wage earners (p-value = .716; p_adj-value = 1).¹⁴ The results remain unchanged when we use robust-clustered errors rather than fixed effects. Except that we find that the callback rate is higher for associate professionals if we use robust-clustered standard errors on vacancy level (p-value = 1.41 × 10⁻⁵), but not with vacancy fixed effects (p-value = 1; p_adj-value = 1). That is to be expected, given that vacancy fixed effects absorb vacancy-type heterogeneity, further allowing us to focus on the treatment of the self-employed across markets.¹⁵

To summarize, we find strong evidence of heterogeneous treatment of solo self-employed at the application stage of the recruitment process across different ISCO skill groups: self-employed associate professionals are treated less favorably than wage-earning associate professionals, whereas this effect is not observed among high-skill managerial positions. However, an objective section does not improve this treatment. This pattern holds across industries under study.¹⁶ Additionally, we do not find support for the uncertainty-based model of adverse treatment for the self-employed. In this case, the self-employed should be less likely to be invited for the interview than wage earners in high-skilled managerial positions, and the objective section, which provides additional information about the qualities of a candidate, should improve their job prospects.

The observed pattern can be explained by the view that the self-employed could be considered generalists as managers (Baptista et al., 2012; Lazear, 2012). In high-skilled managerial positions, prospective employers can expect a good fit for the self-employed candidate, whereas employers suspect that the non-managerial human capital of self-employed associate professionals depreciates over time. Thus, employers might be less interested in candidates with long self-employment experience (see further analysis in Supplemental Appendix A).¹⁷ We elucidate these explanations in supplementary analyses.

Robustness Check (Post- Hoc): Occupational Managerial Skill Requirements (O*NET) and Self-Employment Status

We have looked at the treatment of the self-employed only in the aggregated (associate professionals and high-skilled managers) labor markets. However, within the labor market, the required skills of candidates vary by occupation, as well. Therefore, to assess the robustness of the results and better understand the reasons for observed regularity, we want to test whether the pattern of adverse treatment of self-employed applicants is persistent and holds at a more granular level.¹⁸ If recruiters treat the self-employed less favorably than wage earners depending on their managerial skill level, this effect should be evident at both the aggregate (associate professionals and managers) market level and at the occupation level.

To test this, we use the Occupational Information Network (O*NET, n.d.) classification of the skills required for each occupation. We construct an index of the required skill level for each occupation based on the managerial skills per occupation according to O*NET: Management of Financial Resource, Management of Material Resources, Management of Personnel Resources, and Time Management. As in Deming (2017), we convert the average score by occupation from the O*NET scale to a 0 to 10 scale in order to obtain a weighted percentile ranking.

We use the U.S. Bureau of Labor Statistics crosswalk with O*NET Standard Occupational Classification and ISCO codes to match the managerial skills weighted percentile rankings, which are based on O*NET values, and occupation-based rankings, which are based on ISCO classifications. Compared to associate professional jobs, managerial jobs have higher required skill values, according to the index based on O*NET values.¹⁹ The index shows some variability, which helps to explain whether adverse treatment of the self-employed depends on the managerial skill level of the occupation.

We use a linear probability model with vacancy and month fixed effects to assess whether treatment of the self-employed compared to wage earners depends on the level of skills required for the occupation. The results are reported in Table 5.

Table 5.

Callback Rate and Level of Managerial Skills by Occupation (O*NET).

Variables	Dependent variable
	Callback
	(1)	(2)	(3)	(4)
Self-employed × Level of managerial skills required	0.01*** (0.003)	0.01*** (0.003)	0.01*** (0.003)	0.01*** (0.003)
Level of managerial skills required	−0.005 (3.67)	−0.08** (0.04)	−0.005 (3.67)	−0.01 (0.11)
Self-employed	−0.06*** (0.02)	−0.06*** (0.02)	−0.06*** (0.02)	−0.06*** (0.02)
Associate professionals			0.00 (0.22)	0.22 (0.28)
Constant	0.03 (22.22)	0.48* (0.25)	0.03 (22.22)	0.07 (0.66)
Controls	No	Yes	No	Yes
Industry fixed effects	No	Yes	No	Yes
Vacancy fixed effects	Yes	Yes	Yes	Yes
Month fixed effects	Yes	Yes	Yes	Yes
Number of résumés (N)	8,328	8,328	8,328	8,328
Adjusted R²	.43	.43	.43	.43

Note. Linear probability models. In columns (2) and (4) we control for the applicant’s gender, objective section, if the cover note has a contact name, order of the letter, subject line, length of experience, and type of cover message.

p < .1. **p < .05. ***p < .01.

We focus on the interaction term between employment status and required skill level (see Table 5, row 1). We see that self-employed candidates are more likely to be called for an interview at higher levels of required managerial skills (p-value = .012): The self-employed are treated better with an increase in managerial skill level. Put differently: The lower the managerial skill level of a self-employed candidate, the more likely they will be treated adversely compared to a candidate with wage-earning experience.

As a robustness check, we also control for the aggregate associate professional and manager level (ISCO classification) to check whether the pattern holds for the skill groups (see Table 5, columns 3 and 4). The relation does not change (p-value = .002), which suggests that the self-employed are treated adversely compared to wage earners depending on the required managerial skill level of the occupation. Moreover, it shows that this association is robust to the difference in the tightness of the labor market (non-managerial market is less tight; Baert et al., 2015; Carlsson et al, 2018) and inclusion of the control of the level of non-managerial skill required.²⁰

Discussion and Conclusions

We conducted a large-scale correspondence field experiment to assess how employers evaluate solo self-employed individuals switching to wage employment, with the aim of uncovering the demand-side mechanisms behind their adverse labor market outcomes. Contributing to the existing research (Aldén et al., 2021, Botelho & Chang, 2023; Kacperczyk & Younkin, 2021; Koellinger et al., 2015), we assessed the adverse treatment of former self-employed applicants across several industries and ISCO skill groups. We found that at the application stage of the recruitment process employers treat the self-employed adversely on average compared to wage earners across different occupational groups, with a 22% decrease in callback chance compared to wage earners. To benchmark this estimate, consider that (commonly studied) ethnic minority candidates have on average approximately 29% fewer positive responses compared to ethnic majorities (see meta-analysis Lippens et al., 2023; Zschirnt & Ruedin, 2016).

However, we show that this is only part of a larger, more interesting story: specifically, the adverse treatment of self-employed individuals is observed in lower-skilled associate professional positions but not in high-skilled managerial positions. Self-employed associate professionals are 28% less likely to receive a callback than their wage-earning counterparts (9% callback rate for the latter group). We confirm this observation at a more granular level by showing the same relation applies to occupation level: the lower the level of managerial skills required for the occupation, the less competitive are the applications with self-employed experience.

Our study design allowed us to focus on vertical heterogeneity across skill groups, with industry variation serving as a horizontal generalizability dimension. These dimensions are commonly studied to understand the reasons for adverse treatment related to ethnicity (Adamovic & Leibbrandt, 2023; Bertrand & Mullainathan, 2004; Bovenkerk et al., 1995; Bursell et al., 2021; Carlsson & Rooth, 2007; Goldberg et al., 1996; McGinnity et al., 2009), gender (Park & Oh, 2025), and employment status (Bonoli, 2014; Eriksson & Rooth, 2014) in non-entrepreneurial contexts, but they are innovative when applied in correspondence studies in entrepreneurship. Notably, we found that the penalty is concentrated in associate professional positions and holds across all three industries studied, suggesting that the pattern is not specific to any single sector. This underscores that the adverse treatment is primarily driven by vertical skill differences rather than horizontal industry variations. We therefore fill a gap in the entrepreneurship literature by providing novel evidence on the demand-side mechanisms underlying the adverse treatment of the self-employed. We randomly varied all applicant résumé characteristics to achieve comparability across employment and other characteristics. We ran the preregistered experiment with high statistical power following the study protocol to assess the existence of differential treatment of the self-employed and if it is homogeneous for different skill groups and industries (pre-determined sample size, treatment arms, and skill level).

The results obtained support the explanation that, relying on group characteristics (Arrow, 1973; Phelps, 1972), prospective employers treat the self-employed differently conditional on if the job requires generalist or specialist skills. That is, employers might be concerned about a skill mismatch for the self-employed, since self-employment leads to the development of generalist skills (useful for managers) at the cost of specialist skills (Baptista et al., 2012; Lazear, 2004, 2005, 2012).

However, the human capital mismatch is likely not the sole driver. Our findings also can be interpreted with help of complementary behavioral and organizational fit mechanism. Employers may perceive former self-employed workers as overly independent, autonomous, or difficult to manage (Blanchflower & Oswald, 1998; Cooper & Saral, 2013; Hurst & Pugsley, 2011; Mahieu et al., 2019). While we do not have direct evidence, this can also explain why the penalty we observed is concentrated in subordinate roles but disappears in high-skilled managerial roles.²¹

The main limitation of our experiment is that, while we kept the proportion of applications and vacancies homogeneous within occupations across skill levels (on daily basis) and pre-specified the heterogeneity analysis by skill, comparison across (non-random) skill-level subgroups is still not necessarily causal. Moreover, this design allows only indirect testing of uncertainty theory, since we proxy for uncertainty by skill level. Thus, we cannot rule out this theory despite the fact that our finding, adverse treatment of lower-skilled associate professionals, is not in line with uncertainty theory and previous findings (Mahieu et al., 2019) based on it.

Moreover, unlike many observational studies (Hamilton, 2000; Mahieu et al., 2019), we do not observe wages, only callback rates. Thus, the income differential between the solo self-employed and wage earners should be attributed to the adverse treatment of the self-employed at the application stage under the assumption that the recruitment process strongly affects hiring prospects. Also, we applied for jobs commonly filled by both the self-employed and wage earners, and which are advertised publicly: high-skilled managers (ISCO group 1, skill levels 3 and 4) or lower-skilled associate professionals (ISCO group 3, skill level 3) according to ISCO-08. Thus, we do not investigate the treatment of self-employed individuals at the lowest level of the skills ladder in jobs that require simple or routine physical or manual work.

Finally, the context of the study (Russia, 2017) allows us to investigate the treatment of the self-employed in a relatively stable period when economic and entrepreneurial conditions were comparable to those of an average country (GEM, 2016/2017). Because the market was stable, the observed penalty is likely due to structural employer perceptions rather than a temporary reaction to a specific economic crisis. However, the Russian institutional context and period of the study do not necessarily allow one-to-one generalization of results to other contexts, either with different institutions or during turbulent times of health crises or conflicts. For instance, the adverse treatment of the self-employed in the low-skills labor market may be due to the fact that associate professionals are penalized for independence in Russia, or self-employed in the high-skills labor market attract more attention from prospective employers in Russia in line with stereotype threat evidence on founders (Kacperczyk & Younkin, 2021). Further experiments in different countries, during institutional reforms (Agan & Starr, 2018), or on employment-oriented social networking services (Evsyukova et al., 2025) can help assess generalizability.

In short, we show that self-employed associate professionals are treated adversely in the labor market. This unfavorable treatment of the self-employed is consistent with the theory of the development of managerial human capital in self-employment at the cost of non-managerial human capital. This cautionary tale has some important positive implications for policy. Specifically, these findings demand a shift in how entrepreneurship is promoted. Governments often rely on self-employment programs as a default active labor market policy to reduce unemployment (OECD/European Commission, 2021). However, our results suggest that for certain groups, particularly those in associate professional roles, this may become a trap. For these workers, self-employment can be perceived as “unemployment in disguise” or involuntary necessity entrepreneurship (Earle & Sakova, 2000; Hyytinen & Rouvinen, 2008; Thurik et al., 2008). By pushing vulnerable individuals into self-employment without considering their exit options, policies may inadvertently scar their future employability.

Moreover, given that the development of Artificial Intelligence reduces job opportunities for young entry-level workers (Brynjolfsson, 2025), the government can be tempted to double down on promoting entrepreneurial careers among youth, for example, through advertising, training, and entrepreneurial role models (Asanov et al., forthcoming). However, our evidence urges caution in promoting this path to less-skilled workers. Instead, a cost-effective policy could be developed to support opportunity-driven entrepreneurship, for example, by focusing on managerial or potentially soft skills (Heckman & Kautz, 2012) that are important for managerial positions (see O*NET classification). Finally, given that young professionals do not necessarily know the consequences of their career choice and available options (e.g.,Asanov et al., forthcoming; Camerer & Lovallo, 1999), nascent self-employed at the associate professional level should be cautious about remaining self-employed or should consider applying for managerial positions if they plan to return to wage employment.

Supplemental Material

sj-pdf-1-etp-10.1177_10422587261440312 – Supplemental material for Is Self-Employment a Career Trap? A Large-Scale Field Experiment in the Labor Market

Supplemental material, sj-pdf-1-etp-10.1177_10422587261440312 for Is Self-Employment a Career Trap? A Large-Scale Field Experiment in the Labor Market by Igor Asanov and Maria Mavlikeeva in Entrepreneurship Theory and Practice

Footnotes

Acknowledgements

We are deeply grateful to the editor and three anonymous reviewers for their constructive and developmental feedback, which significantly strengthened this work. We gratefully acknowledge Guido Büntorf, Stephan Bruns, Johannes König, and Anastasiya-Mariya Noha for their valuable comments and suggestions. We also thank Irina Slastnikova for her outstanding contribution to the experimental fieldwork. This research was supported by the “Rudolf and Ursula Lieberum” Scholarship of the University of Kassel, which made this research possible, and for which we are sincerely thankful.

ORCID iDs

Igor Asanov

Maria Mavlikeeva

Ethical Considerations

The authors provide this experiment following the ethically sensitive approach recommended for this correspondence study by the Central Ethics Committee at the University of Kassel.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the “Rudolf and Ursula Lieberum” Scholarship of the University of Kassel.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data and code sufficient to reproduce the analysis is available on Open Science Framework (OSF) .

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Igor Asanov is a senior researcher at INCHER, University of Kassel, Germany. His research focuses on large-scale field experiments in entrepreneurship, education, and public health. He leads the Showing Life Opportunities program and has published in PNAS, Nature Human Behavior (forthcoming), and World Development.

Maria Mavlikeeva is a researcher in entrepreneurship and labour economics. Conducting large-scale field experiments at INCHER and EPIE Group, University of Kassel, her research examines labour market outcomes: entrepreneurship, gender and ethnic disparities, and employment during economic crises. Her work appears in Industrial Relations Journal and CEPR COVID Economics.

References

Adamovic

Leibbrandt

(2023). Is there a glass ceiling for ethnic minorities to enter leadership positions? Evidence from a field experiment with over 12,000 job applications. The Leadership Quarterly, 34(2), 101655.

Agan

Starr

(2018). Ban the box, criminal records, and racial discrimination: A field experiment. Quarterly Journal of Economics, 133, 191–235.

Aidis

Estrin

Mickiewicz

(2008). Institutions and entrepreneurship development in Russia: A comparative perspective. Journal of Business Venturing, 23(6), 656–672.

Aidis

Korosteleva

Mickiewicz

(2008). Entrepreneurship in Russia (Economics Working Paper No. 88). University College London.

Aldén

Bastani

Hammarstedt

(2021). Ethnic background and the value of self-employment experience: Evidence from a randomized field experiment. Oxford Bulletin of Economics and Statistics, 83(6), 1287–1310.

Angrist

J. D.

Pischke

J.-S.

(2009). Mostly harmless econometrics: An empiricist’s companion. Princeton University Press.

Asanov

Åstebro

Buenstorf

Crépon

Flores

McKenzie

Mensmann

Schulte

(Forthcoming). Remote delivery of STEM and entrepreneurship role models at scale changes college major choice. Nature Human Behaviour.

Arrow

(1973). The theory of discrimination. In Ashenfelter

Rees

(Eds.), Discrimination in labor markets (Vol. 3, No. 10, pp. 3–33). Princeton University Press.

Åstebro

Chen

Thompson

(2011). Stars and misfits: Self-employment and labor market frictions. Management Science, 57(11), 1999–2017.

10.

Åstebro

Herz

Nanda

Weber

R. A.

(2014). Seeking the roots of entrepreneurship: Insights from behavioral economics. Journal of Economic Perspectives, 28(3), 49–70.

11.

Baert

Cockx

Gheyle

Vandamme

(2015). Is there less discrimination in occupations where recruitment is difficult? ILR Review, 68(3), 467–500.

12.

Banerjee

Duflo

Finkelstein

Katz

L. F.

Olken

B. A.

Sautmann

(2020). In praise of moderation: Suggestions for the scope and use of pre-analysis plans for RCTS in economics (No. w26993). National Bureau of Economic Research.

13.

Baptista

Lima

Preto

M. T.

(2012). How former business owners fare in the labor market? Job assignment and earnings. European Economic Review, 56(2), 263–276.

14.

Barinova

Zemtsov

Smallbone

(2020). SMEs and entrepreneurship policy in Russia. In A research agenda for entrepreneurship policy (pp. 103–120). Edward Elgar Publishing.

15.

Bertrand

Mullainathan

(2004). Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. American Economic Review, 94(4), 991–1013.

16.

Blanchflower

D. G.

Oswald

A. J.

(1998). What makes an entrepreneur? Journal of Labor Economics, 16(1), 26–60.

17.

Bonoli

(2014). Employers’ attitudes towards long-term unemployed people and the role of activation in Switzerland. International Journal of Social Welfare, 23(4), 363–376.

18.

Bovenkerk

Gras

M. J. I.

Ramsoedh

(1995). Discrimination against migrant workers and ethnic minorities in access to employment in the Netherlands (International Migration Papers No. 4). Employment Department, International Labour Office.

19.

Botelho

T. L.

Chang

(2023). The evaluation of founder failure and success by hiring firms: A field experiment. Organization Science, 34(1), 484–508.

20.

Bruce

Schuetze

H. J.

(2004). The labor market consequences of experience in self-employment. Labour Economics, 11(5), 575–598.

21.

Brynjolfsson

Chandar

Chen

(2025). Canaries in the coal mine?: Six facts about the recent employment effects of artificial intelligence (Stanford Digital Economy Lab Working Paper). https://digitaleconomy.stanford.edu/publications/canaries-in-the-coal-mine/

22.

Bursell

Bygren

Gähler

(2021). Does employer discrimination contribute to the subordinate labour market inclusion of individuals of a foreign background? Social Science Research, 98, Article 102582.

23.

Camerer

Lovallo

(1999). Overconfidence and excess entry: An experimental approach. American Economic Review, 89(1), 306–318.

24.

Carlsson

Fumarco

Rooth

D. O.

(2018). Ethnic discrimination in hiring, labour market tightness and the business cycle: Evidence from field experiments. Applied Economics, 50(24), 2652–2663.

25.

Carlsson

Rooth

D.-O.

(2007). Evidence of ethnic discrimination in the Swedish labor market using experimental data. Labour Economics, 14(4), 716–729.

26.

Cooper

D. J.

Saral

K. J.

(2013). Entrepreneurship and team participation: An experimental study. European Economic Review, 59, 126–140.

27.

Czibor

Rosendahl Huber

Koudstaal

(2017). Are entrepreneurs lone wolves? An experimental analysis of sorting into teams (Working Paper). https://munich-innovation-seminar.weebly.com/uploads/1/0/2/8/102821044/laura.pdf

28.

Deming

D. J.

(2017). The growing importance of social skills in the workplace. Quarterly Journal of Economics, 132(October), 1593–1640.

29.

Earle

J. S.

Sakova

(2000). Business start-ups or disguised unemployment? Evidence on the character of self-employment from transition economies. Labour Economics, 7(5), 575–601.

30.

Eriksson

Rooth

D.-O.

(2014). Do employers use unemployment as a sorting criterion when hiring? Evidence from a field experiment. American Economic Review, 104(3), 1014–1039.

31.

Evans

D. S.

Leighton

L. S.

(1989). Some empirical aspects of entrepreneurship. The American Economic Review, 79(3), 519–535.

32.

Evsyukova

Rusche

Mill

(2025). LinkedOut? A field experiment on discrimination in job network formation. The Quarterly Journal of Economics, 140(1), 283–334.

33.

Failla

Melillo

Reichstein

(2017). Entrepreneurship and employment stability: Job matching, labour market value, and personal commitment. Journal of Business Venturing, 32(2), 162–177.

34.

Galos

D.-R.

Coppock

(2023). Gender composition predicts gender bias: A meta-reanalysis of hiring discrimination audit experiments. Science Advances, 9(18), eade7979.

35.

Georgellis

Sessions

Tsitsianis

(2007). Pecuniary and non-pecuniary aspects of self-employment survival. Quarterly Review of Economics and Finance, 47(1), 94–112.

36.

Gerber

T. P.

(2001). The development of self-employment in Russia (PONARS Policy Memo No. 186). Program on New Approaches to Russian Security.

37.

Gimpelson

Kapeliushnikov

(2014). Between light and shadow: Informality in the Russian Labour Market (IZA Discussion Paper No. 8279). Institute of Labor Economics (IZA).

38.

Goldberg

Mourinho

Kulke

(1996). Labour market discrimination against foreign workers in Germany (International Migration Papers, No. 7). International Labour Office.

39.

Gornall

Strebulaev

I. A.

(2024). Gender, race, and entrepreneurship: A randomized field experiment on venture capitalists and angels. Management Science, 71(6), 5308–5327.

40.

Hamilton

B. H.

(2000). Does entrepreneurship pay? An empirical analysis of the returns to self-employment. Journal of Political Economy, 108(3), 604–631.

41.

Hartog

Van Praag

Van Der Sluis

(2010). If you are so smart, why aren’t you an entrepreneur? Returns to cognitive and social ability: Entrepreneurs versus employees. Journal of Economics & Management Strategy, 19(4), 947–989.

42.

Heckman

J. J.

Kautz

(2012). Hard evidence on soft skills. Labour Economics, 19(4), 451–464.

43.

Holland

P. W.

(1986). Statistics and causal inference. Journal of the American statistical Association, 81(396), 945–960.

44.

Hurst

Pugsley

B. W.

(2011). What do small businesses do?Brookings Papers on Economic Activity, 2011(2), 73–118.

45.

Hyytinen

Ilmakunnas

Toivanen

(2013). The return-to-entrepreneurship puzzle. Labour Economics, 20, 57–67

46.

Hyytinen

Rouvinen

(2008). The labour market consequences of self-employment spells: European evidence. Labour Economics, 15(2), 246–271.

47.

Kacperczyk

Younkin

(2021). A founding penalty: Evidence from an audit study on gender, entrepreneurship, and future employment. Organization Science, 33(2), 716–745.

48.

Kaiser

Malchow-Møller

(2011). Is self-employment really a bad experience? The effects of previous self-employment on subsequent wage-employment wages. Journal of Business Venturing, 26(5), 572–588.

49.

Klapper

L. F.

Parker

S. C.

(2011). Gender and the business environment for new firm creation. The World Bank Research Observer, 26(2), 237–257.

50.

Koch

Park

Zahra

S. A.

(2021). Career patterns in self-employment and career success. Journal of Business Venturing, 36(1), Article 105998

51.

Koellinger

P. D.

Mell

J. N.

Pohl

Roessler

Treffers

(2015). Self-employed but looking: A labour market experiment. Economica, 82(325), 137–161.

52.

Kovalenko

K. E.

Kovalenko

E. V.

Kovalenko

N. E.

(2019). Tax regulation of the self-employed in Russian law. Estação Científica (UNIFAP), 9(2), 45–53.

53.

Kroft

Lange

Notowidigdo

M. J.

(2013). Duration dependence and labor market conditions: Evidence from a field experiment. Quarterly Journal of Economics, 128(3), 1–61.

54.

Lahey

Beasley

(2016). Technical aspects of correspondence studies (NBER Working Paper No. 22818). National Bureau of Economic Research.

55.

Landier

(2005). Entrepreneurship and the stigma of failure. Stern School of Business, New York University. https://doi.org/10.2139/ssrn.850446

56.

Lazear

E. P.

(2004). Balanced skills and entrepreneurship. American Economic Review, 94(2), 208–211.

57.

Lazear

E. P.

(2005). Entrepreneurship. Journal of Labor Economics, 4(23), 649–680.

58.

Lazear

E. P.

(2012). Leadership: A personnel economics approach. Labour Economics, 19(1), 92–101.

59.

Lippens

Vermeiren

Baert

(2023). The state of hiring discrimination: A meta-analysis of (almost) all recent correspondence experiments. European Economic Review, 151, Article 104315.

60.

Mahieu

Melillo

Reichstein

Thompson

(2019). Shooting stars? Uncertainty in hiring entrepreneurs. Strategic Entrepreneurship Journal, 29(9), 1–42.

61.

Manso

(2016). Experimentation and the returns to entrepreneurship. Review of Financial Studies 29(9), 2319–2340.

62.

Marques

L. A.

II Aidis

Estrin

Mickiewicz

(2020). Entrepreneurship and institutions: How institutional context shapes entrepreneurial behavior. Journal of Institutional Economics, 16(2), 193–211.

63.

Masclet

Colombier

Denant-Boemont

Lohéac

(2009). Group and individual risk-taking: A comparison of decision-making processes and outcomes. Southern Economic Journal, 75(3), 637–667.

64.

McGinnity

Nelson

Lunn

Quinn

(2009). Discrimination in recruitment: Evidence from a field experiment. The Equality Authority & Economic and Social Research Institute.

65.

Merida

A. L.

Rocha

(2021). It’s about time: The timing of entrepreneurial experience and the career dynamics of university graduates. Research Policy 50(1), 104135.

66.

Moskowitz

T. J.

Vissing-Jorgensen

(2002). The returns to entrepreneurial investment: A private equity premium puzzle. The American Economic Review, 92(4), 745–778.

67.

Nüß

(2018). Duration dependence as an unemployment stigma: Evidence from a field experiment in Germany (Economics Working Paper, No. 2018-06). Kiel University, Department of Economics.

68.

OECD/European Commission. (2021). The missing entrepreneurs 2021: Policies for inclusive entrepreneurship and self-employment. OECD Publishing.

69.

Pagés

Stampini

(2009). No education, no good jobs? No education, no good jobs? Evidence on the relationship between education and labor market segmentation. Journal of Comparative Economics, 37(3), 387–401.

70.

Park

S. Y.

(2025). Getting a foot in the door: A meta-analysis of U.S. audit studies of gender bias in hiring. Sociological Science, 12, 26–50.

71.

Phelps

E. S.

(1972). The statistical theory of racism and sexism. American Economic Review, 62(4), 659–661.

72.

Praag

C. M. V.

Ophem

H. V.

(1995). Determinants of willingness and opportunity to start as an entrepreneur. Kyklos, 48(4), 513–540.

73.

Puffer

S. M.

McCarthy

D. J.

Boisot

(2010). Entrepreneurship in Russia and China: The impact of formal institutional voids. Entrepreneurship Theory and Practice, 34(3), 441–467.

74.

Reynolds

P. D.

Curtin

R. T.

(2008). Business creation in the United States: Panel study of entrepreneurial dynamics II initial assessment. Foundations and Trends® in Entrepreneurship, 4(3), 155–307.

75.

Roshchin

Solntsev

Vasilyev

(2017). Recruiting and job search technologies in the age of Internet. Foresight and STI Governance, 11 (4), 33–43.

76.

Simmons

S. A.

Wiklund

Levie

(2014). Stigma and business failure: Implications for entrepreneurs’ career choices. Small Business Economics, 42(3), 485–505.

77.

Slonimczyk

Gimpelson

(2015). Informality and mobility: Evidence from the Russian labor market. Economics of Transition, 23(2), 299–336.

78.

Stephan

(2018). Entrepreneurs’ mental health and well-being: A review and research agenda. Academy of Management Perspectives, 32(3), 290–322.

79.

The Global Entrepreneurship Monitor (GEM). (2016/2017). Global report 2016/17. https://www.gemconsortium.org/report

80.

The International Labour Organization (ILO). (2017). World employment and social outlook: Trends 2017. International Labour Organization. https://www.ilo.org/publications/world-employment-and-social-outlook-trends-2017

81.

The International Standard Classification of Occupations (ISCO). (2008). International standard classification of occupations 2008 (ISCO-08): Structure, group definitions and correspondence tables (Vol. I). International Labour Organization. https://www.ilo.org/publications/international-standard-classification-occupations-2008-isco-08-structure

82.

The Occupational Information Network Classification (O*NET). (n.d.). O*NET OnLine classification. U.S. Department of Labor, Employment and Training Administration. https://www.onetonline.org/find/quick?s=classification

83.

The Organization for Economic Co-operation and Development. (OECD) (2016). Gender differences in self-employment rates, in Entrepreneurship at a Glance 2016. OECD Publishing. https://www.oecd-ilibrary.org/industry-and-services/entrepreneurship-at-a-glance-2016_entrepreneur_aag-2016-en

84.

The World Bank. (2017). Self-employed, total (% of total employment) (modeled ILO estimate). World Bank Open Data. https://data.worldbank.org/indicator/SL.EMP.SELF.ZS

85.

The World Bank. (2022). Self-employed, female (% of female employment) (modeled ILO estimate). World Bank Open Data. https://data.worldbank.org/indicator/SL.EMP.SELF.FE.ZS

86.

Thurik

A. R.

Carree

M. A.

van Stel

Audretsch

D. B.

(2008). Does self-employment reduce unemployment? Journal of Business Venturing, 23(6), 673–686.

87.

Verkhovskaya

Aleksandrova

(2017). Entrepreneurial activity in Russia: Global entrepreneurship monitoring project 2006–2016 results. Russian Management Journal, 15(1), 3–26.

88.

Zschirnt

Ruedin

(2016). Ethnic discrimination in hiring decisions: A meta-analysis of correspondence tests 1990–2015. Journal of Ethnic and Migration Studies, 42(7), 1115–1134.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.44 MB

0.00 MB