Foreign entrepreneurs engage in less misconduct than native entrepreneurs: Evidence from U.K. director disqualifications

Although entrepreneurship is commonly viewed as a positive force for society, entrepreneurs may also engage in harmful activities. We explore whether foreign versus native entrepreneurs have a higher propensity to engage in misconduct, as evidenced by them being formally disqualified from being a director by the government following unfit conduct. Comparing entrepreneurs in the United Kingdom who were solely responsible for their venture and who were disqualified with a matched sample of entrepreneurs who did not engage in such misconduct, we find robust evidence that foreign entrepreneurs are substantially less likely to commit misconduct than native entrepreneurs. We also observe that female entrepreneurs are less likely to engage in misconduct than male entrepreneurs. We discuss the study's contributions to the entrepreneurship and organizational misconduct literatures and examine its practical implications.


Introduction
Although entrepreneurship is commonly associated with growth, innovation, and prosperity, entrepreneurs may also destroy societal value by engaging in misconduct. Despite the societal implications of this dark side of entrepreneurship, strikingly little is known about the characteristics of the entrepreneurs who engage in such behavior. In this paper, we focus on whether foreign or native entrepreneurs have a higher propensity to engage in misconduct. We do so, because foreign entrepreneurs have an increasingly influential role in society, yet we still have little understanding of the behavior of their ventures (Kulchina, 2017). This is also a theoretical conundrum as, on the one hand, foreign entrepreneurs may be hampered by their liability of foreignness such that their lack of institutional knowledge (Drori et al., 2009) may make them (accidentally) commit misconduct. On the other hand, the limited opportunities foreign entrepreneurs have in the labor market (Akresh, 2006;Bates, 2011) will make the penalties of being caught engaging in misconduct more severe, such that their incentive to not commit misconduct is higher than for native entrepreneurs.
We study our research question by focusing on disqualifications from being an entrepreneur following unfit conduct, such as not paying taxes or using company money for personal benefit. Such disqualifications have large effects on the individual-excluding the individual from entrepreneurship for a minimum of 2 years at the risk of jail time. We analyze all 665 disqualifications of entrepreneurs who were the founding and sole director for their ventures in the United Kingdom (U.K.) between November 2017 and November 2019. Taking a matched-sample approach, we find robust evidence that foreign entrepreneurs are substantially less likely to engage in misconduct. Post-hoc analyses offer preliminary evidence with respect to the mechanisms at play, as we do not find evidence consistent with superior contextual knowledge leading to reduced misconduct rates for native entrepreneurs, but instead that the greater consequences of misconduct may have led to reduced rates for foreign entrepreneurs.
Our study has three main contributions. First, we offer novel insights to the (foreign) entrepreneurship literature by analyzing how a micro-level characteristic (i.e., foreignness) relates to the under-studied concept of misconduct by entrepreneurs. Second, we contribute to the organizational misconduct literature (Schnatterly et al., 2018;Zahra et al., 2005) by extending insights on the influence of demographics to entrepreneurship. This is important, as the separation of ownership and control limiting opportunities for misconduct in public firms (Denis et al., 2006;Zahra et al., 2005) is rarely present in ventures, as entrepreneurs commonly own and control the venture. Third, by focusing on disqualifications, we offer one of the first large-scale investigations with respect to actual misconduct by entrepreneurs. This is unique, as many cases of misconduct by entrepreneurs fail to obtain the same widespread attention as those in larger organizations (e.g., Arthaud-Day et al., 2006;Marcel and Cowen, 2013)-leading to scholars to rely on perceptual or self-reported indicators of misconduct (often without being able to attribute misconduct to specific actors) rather than actual cases of misconduct (e.g., Ding and Wu, 2014;Ramdani and Van Witteloostuijn, 2012).

Data
We combined different data sources concerning the directors of ventures offered by the Companies House in the U.K. These directors are legally responsible for their ventures and its records, accounts, and performance. They are not self-employed but instead are all involved with limited companies-which implies a number of duties as determined by the Companies Act (2006), including to act within powers and to promote the success of the company, to act in the interest of their employees, to consider the consequences of any decision in the long term, to foster the company's business relationships with suppliers and customers, and to exercise independent judgment. Most directors pay themselves via dividends while taking little to no salaries (Institute of Directors, 2020) and can therefore be seen as entrepreneurs-bearing both the responsibilities and the risks of their ventures.
Our starting point is all disqualifications between November 2017 and November 2019 in the Register of Disqualifications (more information on what disqualifications entail, further below). This register provides basic demographic information on the disqualified director and the name of the venture where the misconduct took place. However, it contained no other company identifier, such that we manually searched the Company Register (often yielding multiple hits for a given company name). We confirmed that the disqualified director was indeed at the identified venture by inspecting the venture's director list.
By focusing on disqualification of entrepreneurs as a consequence of failing to meet their legal duties in a culpable manner, we align ourselves with the definition of misconduct as a labeling act done by a social-control agent that can impose sanctions on the individual and organization (Greve et al., 2010)-commonly defined as "behavior in or by an organization that a social-control agent judges to transgress a line separating right from wrong; where such a line can separate legal, ethical, and socially responsible behavior from their antitheses" (Greve et al., 2010: 56). Consistent with this, we investigate behavior by an entrepreneur that was judged by a government investigator to be unfit conduct.
We focus exclusively on ventures where a single director was the founder of the venture and remained the sole director throughout the lifetime of the venture. We choose to do so, as in multi-director ventures it is often not possible to identify which director actually was the cause of disqualification, given that frequently the whole director team gets disqualified and thus making it difficult if not impossible to investigate the role of individual characteristics. Moreover, focusing only on founder/sole directors rather than also including directors who took over and managed a venture at a later stage helps ensure that these directors are actually entrepreneurs rather than hired managers. This also allows us to identify the director that committed the misconduct, as that person was the only one ever running the venture. This initial sampling approach yielded a set of 665 ventures where the entrepreneur was disqualified and without missing information on our variables of interest-covering a total of 194 unique 5-digit SIC code combinations and thus encompassing a wide range of entrepreneurial activity.
Then, following common practice in the organizational misconduct literature (e.g., Cumming et al., 2015;Schnatterly, 2003), we constructed a matched sample based on the population of U.K. ventures active between 2017 and 2019. Matched-pair design is a recommended approach when studying why otherwise similar subjects have different outcomes (Schnatterly, 2003). Here, one member of the pair has a disqualified entrepreneur where the other did not. We matched the 665 sampled ventures using all available information from the company profiles: jurisdiction (England, North-Ireland, Scotland, and Wales), founding year, exact five-digit SIC combination, and having a sole founder-director.
We were able to identify exact matches for all but 117 of the cases. Our empirical approach necessitates removing these cases from our analyses, but results are unchanged when including them-reported below. For the remaining 548 cases, we identified a total of 367,643 matches (an average 670.88 matches per case). 1 Given this large number, and the associated risk of identifying false positive patterns, we match each misconduct case with only one non-misconduct case, randomly selected without replacement, following common practice in organizational misconduct studies (Cumming et al., 2015;Schnatterly, 2003;Troy et al., 2011). Results are entirely robust to using all matches, shown below.

Outcome variable
Our main variable of interest is whether the venture's director was disqualified or not. In the U.K., directors can be disqualified for a minimum of two years from being a company director if the director did not meet her or his legal responsibilities by engaging in unfit conduct. Specifically, directors can be investigated when the venture is in insolvency proceedings or if a complaint has been levied against the venture or the director. Unfit conduct includes (but is not limited to) allowing a company to continue trading when it cannot pay its debts, not paying taxes owed by the company, and using company money or assets for personal benefit. The Insolvency Service, the Companies House, the Competition and Markets Authority, the courts, or a company insolvency practitioner can apply to have a director disqualified between two to 15 years. Disqualifications therefore have serious personal and professional ramifications, as the director cannot be a director of any company registered in the U.K. or an overseas company that has connections with the U.K., nor be involved in forming, marketing, or running a company. Moreover, breaking the terms of disqualification results in fines or prison time of up to two years.

Explanatory variables
To investigate possible correlates of disqualifications, we utilized available information from officer profiles via the Companies House. As our focal variable, we coded whether or not the director held the U.K. nationality based on all the provided nationalities (taking on a one if the director does not hold the U.K. nationality and thus is a non-U.K. entrepreneur). If the director holds a dual nationality, we coded the individual as holding the U.K. nationality (we relax this restriction, further below).
We also coded whether the director was female or not based on the title provided by the director. To classify this, we went through all titles and assessed whether these were unequivocally gendered (e.g., "Sir" or "Mr." for men): 87.72% of all directors in the data provided a gendered title. For the remaining directors, the title was either not gendered (e.g., "Dr.") or not provided, to start. We therefore applied a gender estimator on these remaining directors' first names using the "gender" package in R with the "ssa" option and take a first name to be gendered if it obtains a score of least 90% for a given gender. This results in 94.31% of the sample being assigned a gender, and the remaining entrepreneurs being assigned to the "unknown" category. We proxy for education based on whether or not the individual's title indicated a doctorate or a professorship. We also measured the director's total entrepreneurial experience by counting the number of ventures that the director has been involved with as a director (excluding the focal venture and up to the disqualification). 2

Method
Given our cross-sectional data and dichotomous outcome variable, we apply conditional logistic regression to model the probability that the director engaged in misconduct-conditional on the matched pair. We also show models where we apply unconditional logistic regression, controlling for characteristics on which the cases were matched, following other work applying unconditional logistic regression (Troy et al., 2011;Cumming et al., 2015). Table 1 contains descriptive statistics for the combined sample, the subsample of disqualified directors, and the sample of matched directors-as well as the correlation table for the combined sample. Table 2 contains the results of our logistic regression models, with Model 1 containing our focal analysis. Based on this model, we observe that directors with a U.K. nationality, on average and ceteris paribus, have a probability of 42.20% of being disqualified, whereas directors without a U.K. nationality have a 28.38% probability.

Results
We also observe that male directors have an estimated 40.50% probability of being disqualified. In contrast, female directors have a 28.43% probability, while those with an unknown gender have a 64.16% chance of being disqualified. We do not observe an effect of the age of the director, as both its coefficient and its average marginal effect are not statistically significant (AME = −0.001, p = 0.368). We also did not observe a curvilinear effect of age. There is no statistically significant effect of having a doctorate on the probability of being disqualified, nor of the director's entrepreneurial experience. Thus, we only observe effects of the nationality and of the gender of the entrepreneur.
We engaged in supplemental analyses to confirm the robustness of our identified patterns and to better understand these patterns. First, we re-ran our model while keeping the full set of matched entrepreneurs rather than a single match per entrepreneur, conditioning on the unique matching combination. As shown in Table 2, model RC1, our results show that our one-to-one matching approach did not drive our results. Similarly, we also re-ran our model using unconditional logistic regression model, controlling for the characteristics on which we matched. This approach additionally allows including entrepreneurs without a match. These results, shown in model RC2, are entirely consistent with our prior results.
Then, to better understand our result regarding non-UK entrepreneurs' lower propensity to engage in misconduct and to glean insights as to possible mechanisms at play, we ran exploratory post-hoc analyses. Here, we use our full set of matched entrepreneurs given that our original results did not differ when using this wider set and because these analyses split the group of non-U.K. entrepreneurs into smaller subgroups and thus reduce our statistical power. We split the group of foreign entrepreneurs into those that are There are 548 disqualified directors and 548 matched directors who were not disqualified. Matching was done on venture age, exact five-digit. SIC code combination, jurisdiction, the number of directors (being one), and the director being the founder and sole director for the venture. from the Commonwealth countries (past or present; 31,767 entrepreneurs) and from other countries (57,027 entrepreneurs). Given the limited number of disqualifications for each of these groups (27 and 40, respectively) we opted not to split the categories any further. As shown in model EA1, the coefficient is negative and marginally significant for the Commonwealth group (coefficient equals −0.370, p = 0.065) and negative and highly significant for the EU/EFTA group (−0.805, p < 0.001). Entrepreneurs from the Commonwealth countries have an estimated 29.68% probability to have been disqualified, entrepreneurs from the other non-U.K. countries a chance of 21.64% probability, and U.K. entrepreneurs a probability of 37.68%. Comparing these probabilities, we observe that entrepreneurs from the U.K. differ significantly from those from the Commonwealth (p = 0.048) and from the other countries (p < 0.001). Those from the Commonwealth, in turn, have marginally higher rates of disqualified from those from the other countries (p = 0.085). We also investigated whether entrepreneurs with a dual nationality (with one being from the U.K.) have different rates of disqualification. So far, we have classified these entrepreneurs as being from the U.K., but in model EA2 we isolate them. There are 440 entrepreneurs with a dual nationality in our sample. Strikingly, none of these entrepreneurs have been disqualified. When we enter this variable into our model, we indeed find an extremely strong, negative coefficient (coefficient equals −11.641, p < 0.001).

R.F.J. Haans and K. van den Oever
Finally, even though our venture-level matching procedure helps in reducing the effects of different potential omitted variables, there may still be individual-level factors systematically correlated both with misconduct and with nationality. While our data preclude the introduction of such factors, including for instance dark triad characteristics that have been found to differ along demographic characteristics (Jones and Paulhus, 2014) and to be related to misconduct (Azizli et al., 2016), we attempted to minimize the confounding effects of such factors by further matching on all other individual characteristics available. Specifically, we divided the age characteristic into three (roughly) equal groups (<40; 40-50; 50+ years old) and did the same for experience (zero; 1; 2+ prior ventures). We then matched on the venture-level characteristics as well as these age brackets, experience brackets, education, and gender. Although this greatly reduces the sample due to the lower number of available matches (leaving 53,776 matches for 283 disqualified entrepreneurs and leaving 382 unmatched), models EA3, EA4, and EA5 show effects consistent to those reported above. These analyses therefore suggest that individual-level confounders do not drive our results.

Discussion and conclusion
In this paper, we studied whether foreign or native entrepreneurs were more likely to engage in misconduct. Analyses on a matched sample of entrepreneurs show that foreign entrepreneurs are less likely to engage in misconduct, with entrepreneurs from the Commonwealth countries having lower rates of misconduct compared to native entrepreneurs but higher rates compared to other foreign entrepreneurs. Entrepreneurs who held a dual citizenship together with the U.K. nationality were never disqualified. In all, these results offer novel insights into the conduct of foreign entrepreneurs-important, given that "we still have limited understanding of the behavior of their ventures" (Kulchina, 2017(Kulchina, : 1589. That native entrepreneurs are relatively more likely to engage in misconduct is somewhat unexpected, as one might expect that they have superior knowledge about the local conditions-including the country's regulative, normative, and cultural-cognitive institutions (Drori et al., 2009). Although the law provides an indication of what is right and wrong, it is the social-control agent that determines whether an act is classified as misconduct or not (Greve et al., 2010). This is especially relevant as these social-control agents are guided by their own professional standards and rational analysis, but also by their own biases and the anticipation of their constituents' biases (Wiesenfeld et al., 2008). A better understanding of local institutions and the way social-control agents classify misconduct could therefore have allowed native entrepreneurs to shy away from misconduct. Foreigners, who have less knowledge about this process, in contrast would be at greater odds of transgressing the line of good and wrong-even if unintentional. Moreover, one would expect those from one of the Commonwealth countries to have relatively higher knowledge of these factors compared to those who lack such cultural and historical ties. However, it appears that native entrepreneurs' understanding of the local context did not cause them to engage in fewer transgressions, nor did those from the Commonwealth have lower rates than other foreign entrepreneurs. Hence, the superior knowledge mechanism does not seem to apply.
One possible explanation for our finding is that foreign entrepreneurs face more pressing consequences from disqualification, as they are generally disadvantaged in the labor market (Miller, 2017;Phizacklea and Ram, 1996). This lack of alternatives to their venturing activities (Drori et al., 2009) may make foreign entrepreneurs less likely to engage in misconduct. Indeed, being disqualified constitutes a violation of the Good Character requirement for obtaining British citizenship (Home Office, 2020: 36), making it a highly salient punishment for those without the U.K. nationality, but less so for those from one of the Commonwealth countries. Foreign entrepreneurs may also simply have a lower baseline for committing misconduct, just as they find solutions to operate more costeffectively (Kulchina, 2016). Taken as such, our results add to the literature emphasizing the economic and societal value that foreign entrepreneurs generate for their host countries (Andersson and Wadensjo, 2009;Hunt and Gauthier-Loiselle, 2010).
It is also worth noting that foreign entrepreneurs have been found to be more aggressive in making bold and innovative moves (Chaganti et al., 2008), and that they are also more overconfident in their exporting strategies than native entrepreneurs (Morgan et al., 2018). Our study adds nuance to this literature by showing that foreign entrepreneurs have a lower propensity to engage in fullblown misconduct, even though they may be more aggressive and overconfident in their strategies. Therefore, this increased aggressiveness and overconfidence seems to not translate into outright misconduct, even if these factors have been found to redirect attention toward possible rewards from unethical behavior and away from potential negative consequences (Gino and Margolis, 2011;Thaler and Johnson, 1990).
We also contribute to the organizational misconduct literature by extending their findings to the entrepreneurial context. Specifically, not only do we corroborate a number of core findings relating to demographic characteristics-such as that boards with more female directors engage in less misconduct (Cumming et al., 2015)-but we do so in a setting where we both precisely identify the individual engaging in the misconduct and in circumstances where a number of important assumptions and mechanisms (such as the separation of ownership and control) no longer apply. Indeed, in the organizational misconduct literature it is commonly assumed that 'the organization' commits the misconduct due to difficulties in identifying how and which individuals actually engaged in misconduct.
Our work also offers practical insights by showing that specific entrepreneurs may be more prone to engage in misconduct. Understanding the different baselines of individuals to engage in misconduct may for instance inform entrepreneurial team formation. For policy makers, future policy may be designed around minimizing the probability that entrepreneurs engage in unfit conduct such as through training and awareness programs targeted at those entrepreneurs most susceptible to such conduct. The results also show that entrepreneurship by foreigners is less likely to be a destructive form of entrepreneurship (Baumol, 1996), such that welcoming foreign entrepreneurs has positive societal implications.
In addition to these contributions, we touch upon a few future research opportunities. First, although we identify systematic patterns based on demographic differences, we can only speculate about the causal mechanisms behind these effects. As such, further work engaging in for instance qualitative analyses of misconduct cases would be a prime avenue for coming to a more complete understanding of misconduct by entrepreneurs. Second, we were limited in our ability to disaggregate the set of countries not in the Commonwealth. Naturally, this entails a highly diverse set of countries, and it would be valuable for future research to further unpack this group to help better identify the mechanisms at play.
Third, we only observe cases where the social-control agent has classified misconduct by the entrepreneur, but our matched cases could also include entrepreneurs who engaged in misconduct but were not caught: a common problem in the misconduct field (Greve et al., 2010). We believe this bias to be less salient for our result, as native entrepreneurs would likely be better able to conceal their misconduct given their superior knowledge on the regulatory context, and foreign entrepreneurs would perhaps also receive more scrutiny from the social-control agent (as social-control agents tend to render judgements that are consonant with their understanding of their constituents' thoughts, feelings, and preferences; Wiesenfeld et al., 2008). Hence, if anything, native entrepreneurs would be more likely to have committed non-detected misconduct, which should result in lower rates of disqualifications if these acts go unpunished. A similar logic would apply for Commonwealth entrepreneurs, where their relatively superior knowledge regarding the local culture should help them avoid punishment, compared to the other foreign entrepreneurs-something we find opposite evidence for. Nevertheless, we invite future research to explicitly study whether foreign or native entrepreneurs are more likely to engage in unnoticed or unpunished misconduct.
Finally, understanding the boundary conditions under which the demographics of entrepreneurs impact their propensity to commit misconduct offers strong potential. Although our focus on ventures with a single director in a single country offered a number of empirical benefits, much entrepreneurial activity occurs in teams (Harper, 2008). As such, investigating the effects of team-level characteristics on the likelihood that an individual engages in misconduct would be highly relevant. In sum, our results offer initial evidence into demographic correlates of misconduct. We hope that it may motivate further study to help better understand what drives this oft-overlooked dark side of entrepreneurship.