Unpacking the Link Between Top Manager Gender and Firm Involvement in Bribery

The Relationship Between Top Manager Gender and Firm Involvement in Bribery

In society, there is a perception that men and women tend to behave differently. The SRT provides a plausible explanation for such perception (Eagly & Wood, 2016; Wood & Eagly, 2010). According to SRT, behavioral differences are primarily the result of the distribution of gender roles within society (Eagly et al., 2000). The division of labor is further supported by the behavior of men and women through socialization and the formation of gender roles, which become a consensually shared expectation within the society (Eagly & Wood, 2016). Because of the different roles that men and women have had throughout history, men have generally developed agentic traits, while women have mainly matured communal traits (Eagly & Wood, 2012, 2016; Schneider & Bos, 2019). Consequently, the SRT posits that women tend to show more compassion and care for others, and have a greater awareness of social responsibility, than men do (Carli & Eagly, 2016; Liao et al., 2019).

An important mechanism that explains this is the personal adaptation of gender norms, where the norms become standards for judging one’s own behavior, and the individuals comply with the norms to the extent that they integrate the gender roles into their self-concepts. “Women internalize a self-concept that involves behaving communally; they identify communal behavior as desirable for themselves and experience greater self-esteem when their behavior is in alignment with communal ideals” (Schneider & Bos, 2019, p. 189). While involvement in bribery could be portrayed as something voluntary and a rational choice by decision-makers (Uribe, 2014), it is also affected by the social norms and the perceived roles that decision-makers have about themselves. For instance, Alhassan-Alolo (2007) reported that individuals in sub-Saharan Africa assume and behave in accordance with the expectations set by the community, where the social pressures on women lead them to be less involved in bribery. Similarly, Asomah et al. (2023, p. 57) argue that “women in Ghana are less likely to be corrupt because they are under intense pressure to conform to social expectations.” This gender role categorization developed through social interactions not only affects the involvement of women in bribery but also the distribution of rewards from it (Zaloznaya et al., 2024).

From a social network perspective, and in line with the SRT, women are also not likely to be part of networks that facilitate bribery. As, historically, men have had more opportunities and greater access to social networks through their participation in labor markets (Chowdhury et al., 2018), women tend to have a lower degree of social capital (Neumeyer et al., 2019). As a result, women are often excluded from accessing networks dominated by men, such as politics (Neumeyer et al., 2019), through which corruption may flow (Esarey & Chirillo, 2013). Accordingly, Mocan (2008) has argued that, because men are likely to have more contact with public officials in civil and business life, as well as a higher propensity for illicit activity or tolerance toward it, they are more likely to engage in corruption than women.

In line with above reasoning based on the social network perspective and the SRT, several empirical studies have shown that female managers behave differently compared to their male counterparts, when it comes to embracing bribery. For instance, in the context of developing countries, Breen et al. (2017) found that female managers are associated with a significantly lower level of bribery. In their study covering firms in 21 countries, Tuliao and Chen (2017) have also shown that firms with female CEOs are less likely to engage in bribery. Similarly, using a survey of enterprise owners and managers in Georgia, Swamy et al. (2001) have found that firms managed by women are significantly less likely to be involved in bribe-giving than those managed by men. Likewise, Trentini and Koparanova (2013) studied 5,471 firms across 31 countries in Europe and Asia, and found that female entrepreneurs have a significantly lower propensity to engage in bribery than their male counterparts. Based on the above rationale and the previous empirical evidence, we hypothesize that:

Bribe Demand as the First Stage in the Bribery Process

Despite the abovepresented dominant view in the literature reflecting that female top managers are less likely to engage in bribery than their male counterparts, some studies have shown that top manager gender is not significantly related to firm involvement in bribery (Na et al., 2018; Wellalage et al., 2020). This uncovers the possibility that the relationship between top manager gender and firm involvement in bribery is mediated by certain factors related to bribe payment, being the presence of such factors a stimulator for the bribe payment to occur. A key factor that calls for and explicitly enables bribe payment is the demand from bribe takers for such a payment in exchange of rendering a service or providing preferential treatment in public service provision (Goetz, 2007; Jung & Lee, 2023; Lu et al., 2023).

For bribery to take place, at least two agents—that is, a public official and a client—need to interact (Olimov, 2024; Romero, 2025) and engage in a process involving two stages where one of the agents must initiate it before the payment is made. Given their monopoly as service providers (Jung & Lee, 2023), it is typically the officials who make the first move by demanding a bribe to provide the required service (Hunt & Laszlo, 2005; Olimov, 2024). If the officials are willing to provide the service without requiring any bribe, as rational choice agents (Uribe, 2014), clients would not need to spend resources that could otherwise be invested elsewhere.

Therefore, regardless of whether the top manager is male or female, receiving a bribe demand is likely to accentuate their willingness to engage in bribery, in light of the SRT’s mechanism of “expectation of others” (Eagly & Wood, 2016). In that sense, and independently of the gender, if bribe takers expect a top manager to engage in bribery (e.g., by making a bribe demand), the manager will be more likely to engage in it than without facing such an expectation. Not only the expectation of others but also opportunism plays an important role here. Namely, if a person assesses that by doing something they can obtain a great benefit with little effort, they may be willing to embrace such action or behavior, regardless of the potential negative consequences. Thus, it is plausible to argue that a manager may engage in a bribery relationship, in part, because they receive a bribe demand.

Yet bureaucrats have discretion in selecting targets among firms when making bribe demands (Gauthier et al., 2021). We contend that a manager’s gender is an important factor influencing this discretion. Following SRT, it is plausible to anticipate that female top managers will receive fewer bribe demands than their male counterparts, because women are less likely to be perceived as willing participants in corrupt exchanges. Societal norms tend to associate women with more selfless, ethical, and community-oriented behavior (Anglin et al., 2018; Wood & Eagly, 2012). Thus, if others expect that women would not engage in noncommunal behavior, it can be assumed that these “others” will be less likely to ask them to adopt such a behavior. Subsequently female managers are less likely to be invited to engage in bribery activities than their male counterparts. Accordingly, Mocan (2004) has shown that men are more likely to be asked for a bribe than women.

Second, societal expectations around gender shape the structure and composition of social networks, which in turn affect access to bribe-enabling ties. Bribery, being a risky and tacit practice, depends heavily on trust-based networks (de Jong et al., 2015; Shepherd et al., 2021). Romero (2025) highlights the importance of social proximity and network centrality in sustaining bribery. However, female managers are often excluded from these predominantly male networks (UNODC, 2020). This exclusion limits their exposure to bribe demands. As UNODC (2020, p. 36) notes, “for a network of individuals to coordinate any activity that is illegal or widely disapproved of, there must be strong within-group trust, and this trust may be easier to establish and reinforce among people who have gender in common.” Moreover, repeated interactions foster trust and predictability in bribery networks, but such interactions are less likely to occur when women are excluded from the male-dominated ties that reinforce and sustain these networks (Diviák et al., 2020; Romero, 2025). This is a particularly important mechanism as we conceptualize bribery as a relationship that consists of repeated encounters and interactions between bribe takers (public officials) and bribe payers (firms).

Overall, based on the above reasoning, it is plausible to expect that a top manager, regardless of their gender, may engage in bribery partly because they receive a bribe demand, and that female top managers are less likely to receive a bribe demand than their male counterparts. Accordingly, we hypothesize that:

The Moderating Roles of Firm Location and Investment in Security

As argued above, male and female top managers are not likely to receive the same number and type of bribe demands. While, in light of the SRT, this is primarily due to the perceptions of distribution of roles in society and the expectations associated with each gender (Eagly, 1987; Eagly & Wood, 2016; Wood & Eagly, 2012), firm-level factors can also have an influence. In the corruption literature, there are two key firm-level factors that can influence the relationship between top manager gender and bribe demand: firm location (Dutta et al., 2022; Goel et al., 2021; Seck, 2020) and investment made by firms on security (Moyo, 2011; Zhou et al., 2013).

On the one hand, the factor of firm location specifically relates to the level of urbanization of the area where the firm is located. The phenomenon of urbanization has increased the number of female managers and improved their positions in the labor market. For instance, Fung (2014) and Tsai et al. (2016) have found that, in Chinese urban areas, women are pursuing economic status and, unlike in previous generations, are less expected to participate in family responsibilities. Similarly, Yesilirmak et al. (2023) noted that, rapid urbanization has played an important role in enhancing the representation of women in technical and high-skills professions in Turkey, and thereby helped to improve the status of women in society. From the perspective of SRT, this would mean that the expectation that women impose on themselves in terms of the roles they should perform in their job positions, and how the rest of the society expects them to behave will change. Thus, the interaction of female managers with bureaucrats in urban areas will be less influenced by typical gender stereotypes as compared to similar interactions in less urbanized areas.

Besides changing the division of labor, urbanization has also altered the traditional social structure as evidenced by the shift from an “acquaintance” to a “stranger” society (Xu, 2021). This is manifested in a lower need for strong social ties in urban areas, as compared to rural areas (Bhatnagar & Papatla, 2019). Jancsics (2015) suggested that in smalltown societies personal ties and strong local communities facilitate illicit payments and relationships. Similarly, Tian and Xia (2024) argued that personal ties and social norms are more important in less developed areas and communities. The change in both the division of labor and the social structure, is modifying gender stereotypes and the traditional perception of society toward women (Goktan et al., 2015).

Based on this logic, it can be argued that the relationship between top manager gender and the manager’s likelihood of receiving a bribe demand is contingent on the location of the firm—in urbanized vs. non-urbanized areas. One of the characteristics of urbanized areas is the high intensity of business activities. Since gender stereotypes are different in these areas, with the perception that women and men are more equal (i.e., there is a lower gap between gender role associations), female top managers of firms that are located in the main business cities are likely to receive more bribe demands than female top managers of firms from nonurbanized areas. Accordingly, we hypothesize that:

On the other hand, the factor of investment in security refers to the amount that a firm pays for protection, including the costs of hiring personnel or professional security services. Investment in security may affect firms’ likelihood of receiving a bribe demand in at least two ways. First, it increases firm visibility and perceived legitimacy within the local business environment. For female top managers, this increased visibility can have asymmetric effects. While visibility may enhance legitimacy in formal settings, it can simultaneously increase exposure to informal demands in weak institutional environments, where discretion enables opportunistic behavior by public officials.

In addition, investment in security functions as an observable signal that shapes bribe takers’ strategic calculations. The bribery literature consistently shows that public officials target firms based on characteristics that indicate extractable rents and a high likelihood of compliance, such as firm size, age, foreign ownership, and export orientation (Freund et al., 2016; Gauthier et al., 2021; Kim et al., 2022). Security expenditures similarly signal financial capacity and a strong concern for business continuity. Firms that invest in protection may be perceived as more risk-averse and more willing to comply with informal demands to avoid disruption, making them attractive targets for bribe demands. Furthermore, higher security expenditures may intensify interactions with regulatory and public enforcement authorities, and thus expand firms’ contact with public officials. This expanded contact could partially mitigate the relational disadvantages female top managers face in corrupt networks, ultimately raising their likelihood of receiving bribe demands.

Second, investing in security by hiring professional security service providers may indirectly connect firms to public officials to whom they would otherwise lack access. In many developing-country contexts, prevailing social norms constrain women’s access to political and bureaucratic elites, thereby shaping how female top managers engage with public officials (Audretsch et al., 2022). SRT explains this pattern by emphasizing that women are often excluded from informal power structures characterized by deeply embedded gender-role expectations that associate authority, bargaining, and informal deal-making with men (Eagly, 1987). As a result, female top managers are typically less embedded in the informal networks through which bribe demands are initiated, reducing their exposure to bribe demands. However, bribery is often enabled by intermediaries who facilitate contact, information exchange and trust between otherwise disconnected actors (Jancsics, 2015). In fact, the literature on bribery networks highlights the importance of intermediaries who occupy structural holes and connect firms to public officials without necessarily obtaining direct commissions (Shepherd et al., 2021).

Professional security service providers may serve such an enabling role of connecting public officials with firms. As governments have monopoly over policing, judicial enforcement and licensing, security providers ought to maintain cooperative relationships with public law enforcement authorities to operate effectively (Moyo, 2011; R. Rose & Peiffer, 2013). These institutional ties embed security providers within public-sector networks and, by extension, link other firms using their services with bureaucrats. As Jancsics (2015) notes, intermediaries in bribery networks are often motivated by sustaining access and legitimacy rather than extracting a share of the bribe itself, making them effective conduits for repeated informal interactions. This mechanism might be more consequential for female top managers. Prior research shows that corruption networks are highly gendered, with men dominating both bureaucratic and intermediary roles (Diviák et al., 2020). From the SRT perspective, this exclusion reflects a role incongruity: informal bargaining and illicit exchange are perceived as masculine-coded activities, making women less likely to be directly approached. Through the professional security service providers, female-led firms main gain indirect access to these male-dominated networks. These secondary ties may increase interaction frequency and familiarity with public officials, which can weaken gender stereotypes and reposition female top managers as legitimate participants in informal exchanges rather than peripheral outsiders.

Overall, the above arguments suggest that security investment may reshape both the perceived pay-off structure and the social embeddedness associated with female-led firms. While firms led by female top managers would be negatively associated with bribe demands partly because of gendered exclusion from informal networks, a higher investment in security may counteract this exclusion by increasing visibility, perceived extractable rents and network connectivity. Accordingly, we hypothesize that

Figure 1 portrays our conceptual model capturing the relationship between top manager gender and firm involvement in bribery via bribe demand, as well as the roles that firm location and investment in security play in the relationship between top manager gender and bribe demand.

OPEN IN VIEWER

Figure 1.

Hypothesized Model.

Data Sources and Sample

The data for this study were obtained from the WBES. A representative sample of firms from each country’s non-agricultural formal private economy was drawn using the stratified random sampling technique, by taking into account firm size, business sector, and geographic region. Firm size was stratified using a composite measure of the number of permanent and temporary employees as follows: small (5–19); medium (20–99); and large (100+). Firms were also stratified by sector, such as manufacturing or service. In larger economies, an additional disaggregated stratification of firm sector was conducted, using the number of employees, value added, and type of establishment. Firms that belong to a public utility, government service, health care, or financial services were excluded. The geographic region strata comprised regions within a country where the largest production or economic activity is carried out. The data covers over 140 countries using uniform methodology, enabling the comparison of indicators across countries and years (from 2006 to 2020). The data include a range of indicators: firm characteristics, firm performance, firm workforce, regulation and taxes, corruption, crime, informality, gender, finance, infrastructure, innovation and technology, as well as the biggest obstacles firms face. The WBES surveys are administered by private contractors and anonymity of the respondents is guaranteed, which limits desirability bias. Moreover, the surveys are administered face to face in the local language to firm owners and top managers, who are allowed to ask human resource specialists and accountants to answer questions about labor and sales, which limit also gender reporting differences.

In addition, we obtained our institutional variables from the Heritage Foundation. The dataset comprises indices for the following areas: property rights, government integrity, judicial effectiveness, tax burden, government spending, fiscal health, business freedom, labor freedom, monetary freedom, investment freedom, trade freedom, financial freedom, and overall score. The indices cover over 180 countries, in the period from 1995 to 2019. To capture the prevalence of corruption practices across countries, we gathered a corruption perception index from Transparency International. We also collected the GDP per capita data from the online World Bank World Development Indicators (WDI).

The WBES has recently extended its coverage to some developed countries in Western Europe. However, since our study focuses on bribery in the context of developing countries (where institutional voids and corruption challenges are more salient) we restricted our sample to developing countries only. We also excluded observations with missing values on the key variables of interest. After merging the WBES data with additional country-level datasets (WDI, Corruption Perceptions Index [CPI], the Heritage Foundation), our final sample for the analysis covered 57,758 firms from 94 countries over the period from 2007 to 2018.

Definition of Variables

First of all, our independent variable is top manager gender, and it was measured in a binary fashion, taking value 1 if the top manager is female and 0 if the top manager is male. Second, our dependent variable, bribe payment, was measured via the following WBES question: “It is said that establishments are sometimes required to make gifts or informal payments to public officials to ‘get things done’ with regard to customs, taxes, licenses, regulations, services, etc. On average, what percentage of total annual sales, or estimated total annual value, do establishments like this one pay in informal payments or gifts to public officials for this purpose?” Hence, our dependent variable is a dummy variable with value 1 if the answer to the above question is greater than 0, and 0 otherwise. Prior studies investigating bribery with the same dataset have used this measurement as well (Birhanu et al., 2016; Webster & Piesse, 2018; Wellalage et al., 2019).

Third, our mediating variable, bribe demand, measured if a firm had received a payment request to “get things done.” The indicator is constructed by six questions asking firms whether an informal gift or payment was demanded or expected when they applied for an/a (a) electrical connection, (b) water connection, (c) import license, (d) operating license, (e) construction related permit, or (f) inspections, or meetings with tax officials. Hence, it is a dummy variable with value 1 if firms had received a bribe demand related to at least one of the factors mentioned above, and 0 otherwise. Fourth, firm location and investment in security are the moderating variables in our study. Firm location is a dummy variable adopting value 1 if the firm is located in the main business city of the country, and 0 otherwise. Investment in security refers to the average cost that a firm has paid for security and protection (e.g., equipment, personnel or professional security services), and it was measured as a percentage of total annual sales.

Finally, following prior literature on bribery and firm characteristics (de Jong et al., 2012; Francisco & Pontara, 2007; Mendoza et al., 2015; Ramdani & van Witteloostuijn, 2012; Wu, 2009), we employed a number of control variables, including firm age, firm size, export propensity, financial constraint, manager’s work experience, ownership structure, and domestic ownership. We also control for country-level variables, including GDP per capita and legal system quality. In addition, to capture potential differences in the institutional context across countries that may affect the relationship between top manager gender and firm involvement in corruption, we included the business freedom index. This index ranges from 1 to 100, where 100 denotes a stronger degree of business freedom. Furthermore, we added the CPI developed by Transparency International, to capture the disparity in the prevalence of corruption across countries. This index ranges from 10 to 100, where 100 represents the lowest level of corruption. In addition to these, we controlled for Control of Corruption from World Governance Indicators. Finally, we controlled for potential year-, industry-, and country-specific heterogeneities with fixed effects. Table 1 presents the definitions and measurements of our control variables.

OPEN IN VIEWER

Table 1.

Definition and Measurement of Control Variables Included in the Model.

Explanatory variables	Definition	Measurement
Business freedom index	It is a weighted score from 10 elements related to procedures of starting, running, and closing a business	0–100
Log of GDP per capita	Logarithm of current GDP per capita in USD	Logarithm of the amount
Legal system quality	Perception of the judiciary, based on the answer to the following question: Does the firm strongly agree or tend to agree that the court system is fair, impartial, and uncorrupt?	1 = yes; 0 = otherwise
Financial constraint	Firms that did not apply for a loan for the following reasons: complex application procedure, unfavorable interest rates, very high collateral requirement, insufficient size of loan and maturity, perception it would not be approved, and other reasons	1 = yes; 0 = otherwise
Export propensity	When at least 10% of the firm’s annual sales is derived from direct export	1 = yes; 0 = otherwise
Top manager experience	Top manager’s years of experience working in the firm’s sector	Logarithm of Years
Firm size	Logarithm of number of employees	Logarithm of the amount
Firm performance	Annual employment growth	Percentage
Corruption perception index	It is a composite index about corruption across countries	0–100
Control of corruption	Perceptions of the extent to which public power is exercised for private gain	Standard normal units
Ownership structure
Open share	Firm legal status is open-shareholding	1 = yes; 0 = otherwise
Closed share	Firm legal status is closed-shareholding	1 = yes; 0 = otherwise
Sole proprietorship	Firm legal status is sole proprietorship	1 = yes; 0 = otherwise
Partnership	Firm legal status is partnership	1 = yes; 0 = otherwise
Limited partnership	Firm legal status is limited partnership	1 = yes; 0 = otherwise
Domestic	Firm is domestically owned	1 = yes; 0 = otherwise
Firm age	Number of years since the firm was established	Logarithm of years
Sector	Three sector classifications of the firm: manufacturing, retail, and other services	1 = yes; 0 = otherwise for each sector

Model Estimation

In order to test our hypotheses, we have estimated three models: two models with bribe payment as the dependent variable, and one with bribe demand as the dependent variable. In model (1), we have estimated bribe payment on top manager gender and other control variables with the following equation:

b r i b e p a y m e n t i = α + β 1 t o p m a n a g e r g e n d e r i + X ′ i γ i + ε i

(1)

where X ′ i is a vector of control variables, and γ i represents the associated coefficients. Since we are interested in assessing the mediating role of bribe demand, we have also estimated the bribe demand dummy as dependent variable over top manager gender and control variables:

b r i b e d e m a n d i = ω + θ 1 t o p m a n a g e r g e n d e r i + X ′ i σ i + φ i

(2)

We then augmented Equation 1 with our mediating variable, bribe demand:

b r i b e p a y m e n t i = μ + π 1 t o p m a n a g e r g e n d e r i + π 2 b r i b e d e m a n d i + X ′ i ρ i + ϵ i

(3)

We estimated all the three equations with logistic regression.

To test Hypothesis 2a, we employed nonlinear mediation analysis (Imai, Keele, & Tingley, 2010; Imai, Keele, & Yamamoto, 2010; Imai et al., 2011). We conducted the analysis with a user written package in STATA, ldecomp (Buis, 2010). This is because marginal effects are not constant when the dependent variable and mediating variables are binary. In other words, the marginal effects are dependent on the value of covariates used in the estimation. Therefore, it was not possible to calculate the mediation effects by taking the product of coefficients from Equations 1 to 3, as it is the case in linear regression.

Hence, the causal mediation effect in the nonlinear regression models was calculated as follows:

δ i ( t ) = Y i ( t , M i ( 1 ) ) − Y i ( t , M i ( 0 ) )

(4)

for each unit of analysis i and treatment status 0 and 1, where δ i , Y i , and M i denote causal mediation effect, outcome variable, and mediating variable, respectively. The causal mediation effect is the difference between the outcome variable that would be observed under the treatment condition as a result of a change in the mediator, M i ( 1 ) , and from the value that would be realized under no treatment or control condition, M i ( 0 ) , while the status of treatment is hold to be constant at t .

Similarly, the direct effect of the treatment was calculated as follows:

ς i ( t ) = Y i ( 1 , M i ( t ) ) − Y i ( 0 , M i ( t ) )

(5)

for each unit of analysis i and Treatment status 0 and 1.

Descriptive Statistics and Regression Results

Table 2 presents the mean, standard deviation, minimum, and maximum of the variables used in the study. In our sample, 8,707 firms (15.07% of the total) have a female top manager, 7,835 firms (13.57%) made a bribe payment, and 7,207 (12.48%) received a bribe demand. Among firms that made a bribe payment, the depth of bribery, which is measured as a percentage of annual sales, ranged from 0.001% to 100%t with a mean of 8.26% of annual sales. However, the mean of the depth of bribery in the total sample (which includes all firms irrespective of their involvement in bribery) was 1.12% of their annual sales. Our institutional proxy variables, business freedom index, corruption perception index and control of corruption had a mean of 57.99, 34.06, and −0.55 respectively.

OPEN IN VIEWER

Table 2.

Summary Statistics of the Variables.

Variable	Obs.	Mean	Std. dev.	Min	Max
Bribe payment	57,758	0.14	0.34	0	1
Bribe demand	57,758	0.12	0.33	0	1
Bribe depth	57,758	1.12	5.61	0	100
Top manager gender	57,758	0.15	0.36	0	1
Firm location	57,758	0.40	0.49	0	1
Investment in security	57,758	1.67	4.26	0	92.50
Legal system quality	57,758	0.49	0.50	0	1
Financial constraint	57,758	0.22	0.41	0	1
Export propensity	57,758	0.14	0.34	0	1
Top manager experience	57,758	2.76	0.65	0	4.11
Firm size	57,758	3.41	1.29	0.69	9.39
Domestic	57,758	0.91	0.28	0	1
Firm age	57,758	2.80	0.66	0	5.24
Firm performance	57,758	3.72	16.59	−96.56	100
Business freedom index	57,758	57.99	13.50	20	90
Log of GDP per capita	57,758	8.03	0.89	5.69	9.92
Corruption perception index	57,758	34.06	10.16	8	74
Control of corruption	57,758	−0.55	0.51	−1.53	1.38

Table 3 shows the correlation matrix of the main variables used in our analysis. Bribe payment is correlated positively with bribe demand, and negatively with top manager gender. Similarly, the correlation between bribe demand and other control variables is in line with the theoretically expected sign. Furthermore, to identify potential multicollinearity problems, we conducted collinearity diagnostic tests following Belsley et al. (2005) using coldiag2, a user-written STATA package. The collinearity diagnostics is based on the interrelationships among the independent variables, and hence appropriate for models other than linear regression. This procedure examines the conditioning of the matrix of independent variables and reports condition indices along with variance-decomposition proportions. The maximum condition index is 19.78, which is below the commonly used threshold of 30 that would indicate potential collinearity problem requiring further investigation. ⁷

OPEN IN VIEWER

Table 3.

Correlation Matrix of the Main Variables Used in the Study.

Variables	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. Bribe payment	1
2. Bribe demand	.39***	1
3. Bribe depth	.50***	.16***	1
4. Top manager gender	−.01**	−.02***	−.01**	1
5. Firm location	−.04***	.01***	.00	−.01**	1
6. Investment in security	.07***	.06***	.10***	.01	.02***	1
7. Top manager experience	−.03***	−.03***	−.03***	−.06***	−.05***	−.02***	1
8. Firm size	.00	.04***	−.01***	−.08***	.03***	.02***	.09***	1
9. Firm age	−.02***	−.01***	−.02***	−.04***	.00	.01	.42***	.24***	1
10. Firm performance	.02***	.02***	.01***	−.01	.00	.01**	−.10***	.06***	−.17***	1
11. Business freedom index	−.03***	−.08***	.01**	.05***	−.21***	.02***	.17***	.00	.03***	−.01	1
12. Log of GDP per capita	−.11***	−.16***	−.05***	.04***	−.25***	−.05***	.19***	.06***	.08***	−.02***	.49***	1
13. Corruption perception index	−.16***	−.17***	−.06***	−.01	−.02***	−.06***	.10***	.06***	.10***	−.01***	.31***	.38***	1
14. Control of corruption	−.16***	−.17***	−.07***	.00	−.07***	−.07***	.10***	.06***	.09***	.00	.34***	.40***	.93***	1

*

p < .10. **p < .05. ***p < .01.

Table 4 shows the results of the logistic regressions for models (1), (2) and (3). In model (1), we estimated Equation 1—bribe payment on top manager gender and control variables. The coefficient for top manager gender is negative and significant ( β = − . 138 , p < . 01 ) indicating that having a female top manager is negatively associated with a firm’s likelihood of engaging in bribery. This result empirically supports Hypothesis 1.

OPEN IN VIEWER

Table 4.

Regression Results.

Variables	Bribe payment	Bribe demand	Bribe payment
	Coef/SE	Coef/SE	Coef/SE
Top manager gender	−.138*** (0.040)	−.090** (0.041)	−.131*** (0.042)
Bribe demand			1.933*** (0.033)
Financial constraint	.355*** (0.033)	.349*** (0.032)	.281*** (0.034)
Export propensity	.203*** (0.042)	.188*** (0.041)	.154*** (0.044)
Top manager experience	.003 (0.025)	.036 (0.024)	−.008 (0.027)
Firm size	.035*** (0.013)	.157*** (0.012)	−.015 (0.014)
Domestic	.142*** (0.050)	.151*** (0.049)	.092* (0.052)
Firm age	.030 (0.025)	−.036 (0.025)	.048* (0.027)
Firm performance	.001 (0.001)	.000 (0.001)	.001 (0.001)
Legal system quality	−.563*** (0.030)	−.551*** (0.030)	−.434*** (0.032)
Ownership structure
Open share	.167 (0.125)	−.193 (0.120)	.251* (0.132)
Closed share	.201* (0.113)	.007 (0.105)	.219* (0.119)
Sole proprietorship	.086 (0.114)	−.130 (0.104)	.144 (0.121)
Partnership	.111 (0.120)	.076 (0.110)	.096 (0.128)
Limited partnership	.288** (0.118)	.037 (0.108)	.318** (0.124)
Control of corruption	.539 (0.338)	−.193 (0.288)	.638* (0.344)
Corruption perception index	−.003 (0.014)	−.020* (0.012)	.003 (0.014)
Business freedom index	−.065*** (0.006)	−.014*** (0.005)	−.068*** (0.007)
Log of GDP per capita	−.078 (0.376)	−1.677*** (0.340)	.213 (0.381)
Constant	5.959** (2.839)	11.958*** (2.502)	3.403 (2.856)
Sector	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes
Country fixed effect	Yes	Yes	Yes
Log-likelihood	−18,077.51	−18,674.09	−16,459.53
Number of observations	57,758	57,758	57,758
Chi2	7,336.112	5,178.163	9,196.803
Pseudo R2	.212	.141	.282

Note. Standard errors in parentheses ***p < .01, **p < .05. *p < .1.

Model (2) estimated our mediating variable, bribe demand, on top manager gender, and control variables. The coefficient for top manager gender is negative and significant ( β = − . 090 , p < . 05 ). It shows that having a female top manager is negatively associated with a firm’s likelihood of receiving bribe demand. The result empirically supports hypothesis 2b.

Model (3) augmented model (1) with our mediating variable, bribe demand, to check if it mediates the relationship between bribe payment and top manager gender. The coefficient for bribe demand is positive and significant ( β = 1 . 933 , p < . 01 ), whereas the coefficient for top manager gender is negative but lower in absolute value terms than what we obtained in model (1) without the mediating variable as a regressor ( β = − . 131 , p < . 01 ). Although these coefficients suggest that bribe demand plays a mediating role, they cannot be interpreted directly, because calculating the mediation effect by taking the product of coefficients (as it is the case in linear regression) does not work in nonlinear regressions (Buis, 2010; Imai, Keele, & Tingley, 2010). Hence, we carried out nonlinear mediation analysis using Equations 4 and 5.

Table 5 shows the average indirect effect (average causal mediation effect [ACME]), the average direct effect, and the total effect of the observed coefficients. The coefficient of the direct effect is negative and significant ( β = − . 121 , p < . 05 ). Similarly, the coefficient of indirect effect is negative and significant ( β = − . 051 , p < . 01 ), but smaller than that of the direct effect. The sum of the direct and indirect effects gives the total effect. The size of the indirect effect is about 30% of the total effect, indicating a significant mediation by bribe demand. Hence, these results provide empirical support to Hypothesis 2a.

OPEN IN VIEWER

Table 5.

Mediation Analysis: Observed Coefficient.

	Observed coefficient	Bootstrap standard error	Normal-based [95% conf. interval]
Average indirect effect	−0.0506***	0.0115	−0.07306	−0.02808
Average direct effect	−0.1208**	0.0394	−0.19811	−0.04354
Total effect	−0.1714***	0.0419	−0.25212	−0.09067

Note. Bootstrap replications: 6,741. ***p < .01, **p < .05. *p < .1.

Table 6 shows the regression results of the moderation analysis for bribe demand. The coefficient for the interaction term between firm location and top manager gender in model (3) is positive and significant ( β = . 157 , p < . 05 ). This indicates that firm location positively moderates the negative relationship between top manager gender and bribe demand, empirically supporting hypothesis 3. Similarly, in model (5), the coefficient for the interaction term between investment in security and top manager gender is positive and significant ( β = . 014 , p < . 05 ). This means that investment in security positively moderates the negative relationship between having a female top manager and a firm’s likelihood of receiving a bribe demand, empirically supporting Hypothesis 4.

OPEN IN VIEWER

Table 6.

Regression Results (Moderation Analysis).

Variables	(1)	(2)	(3)	(4)	(5)
	Coef/SE	Coef/SE	Coef/SE	Coef/SE	Coef/SE
Top manager gender	−.090** (0.041)	−.094** (0.041)	−.162*** (0.054)	−.093** (0.041)	−.125*** (0.044)
Firm location × Top manager gender			.157** (0.080)
Investment. in security × Top manager. gender					.014** (0.007)
Firm location		.175*** (0.037)	.154*** (0.038)
Investment in security				.020*** (0.003)	.018*** (0.003)
Financial constraint	.349*** (0.032)	.348*** (0.032)	.347*** (0.032)	.348*** (0.032)	.348*** (0.032)
Export propensity	.188*** (0.041)	.185*** (0.041)	.184*** (0.041)	.184*** (0.041)	.183*** (0.041)
Top manager experience	.036 (0.024)	.033 (0.024)	.033 (0.024)	.039 (0.024)	.039 (0.024)
Firm size	.157*** (0.012)	.154*** (0.012)	.154*** (0.012)	.153*** (0.012)	.154*** (0.012)
Domestic	.151*** (0.049)	.152*** (0.049)	.152*** (0.049)	.152*** (0.050)	.151*** (0.049)
Firm age	−.036 (0.025)	−.035 (0.025)	−.035 (0.025)	−.040 (0.025)	−.040 (0.025)
Firm performance	.000 (0.001)	.000 (0.001)	.000 (0.001)	.000 (0.001)	.000 (0.001)
Legal system quality	−.551*** (0.030)	−.544*** (0.030)	−.545*** (0.030)	−.551*** (0.030)	−.551*** (0.030)
Ownership structure
Open share	−.193 (0.120)	−.192 (0.120)	−.192 (0.120)	−.201* (0.120)	−.202* (0.120)
Closed share	.007 (0.105)	.004 (0.105)	.002 (0.105)	.001 (0.105)	−.000 (0.105)
Sole proprietorship	−.130 (0.104)	−.127 (0.104)	−.127 (0.104)	−.135 (0.104)	−.135 (0.104)
Partnership	.076 (0.110)	.073 (0.110)	.073 (0.110)	.061 (0.110)	.061 (0.110)
Limited partnership	.037 (0.108)	.032 (0.108)	.030 (0.108)	.029 (0.108)	.029 (0.108)
Control of corruption	−.193 (0.288)	−.158 (0.287)	−.165 (0.287)	−.209 (0.288)	−.205 (0.288)
Corruption perception index	−.020* (0.012)	−.018 (0.011)	−.018 (0.011)	−.018 (0.012)	−.018 (0.012)
Business freedom index	−.014*** (0.005)	−.015*** (0.005)	−.015*** (0.005)	−.014*** (0.005)	−.014*** (0.005)
Log of GDP per capita	−1.677*** (0.340)	−1.666*** (0.340)	−1.667*** (0.340)	−1.611*** (0.340)	−1.610*** (0.340)
Constant	11.958*** (2.502)	11.965*** (2.498)	11.972*** (2.498)	11.498*** (2.505)	11.502*** (2.505)
Sector	Yes	Yes	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes	Yes	Yes
Country fixed effect	Yes	Yes	Yes	Yes	Yes
Number of observations	57,758	57,758	57,758	57,758	57,758
Chi2	5,178.163	5,211.991	5,217.450	5,211.329	5,216.915
Pseudo R2	.141	.141	.142	.142	.142

Note. Standard errors in parentheses ***p < .01, **p < .05, *p < .1.

Robustness Tests

We checked for robustness to ensure that our main results hold for sensitivity, different specifications, and subsamples. First, we conducted sensitivity analysis to make sure that the sequential ignorability (SI) assumption is not violated. The SI assumption—the treatment and mediator are exogenous and that no omitted variables are confounding the mediation effect—must be met for identification in the mediation analysis in nonlinear regressions models (Imai, Keele, & Tingley, 2010; Imai, Keele, & Yamamoto, 2010; Tingley et al., 2014). By estimating a sensitivity parameter ρ—a correlation between the error terms of the mediator and the outcome models (Imai, Keele, & Tingley, 2010; Imai, Keele, & Yamamoto, 2010; Tingley et al., 2014)—it is possible to see to the degree to which the key identifying assumption must be violated to reverse the conclusions obtained in the mediation analysis. A high value of ρ would indicate the existence of a strong confounding between the mediator and outcome variable, and hence a violation of the SI assumption. We conducted the sensitivity analysis by using linear probability modeling for binary outcome and mediating variables (Tingley et al. 2014; Vaznyte et al., 2020). Figure 1 depicts the estimated ACME of the bribe demand mediator against values of the sensitivity parameter ρ . The ACME equals zero when ρ is .4, indicating that the average mediation effect obtained in our analysis is robust and can be causally interpreted (Figure 2).

OPEN IN VIEWER

Figure 2.

Average Causal Mediation Effect as a Function of Degree of Violation of Sequential Ignorability Assumption.

Second, we conducted matched sample analysis to ensure that firms that have a female top manager are comparable with firms that do not have it. We used coarsened exact matching approach in STATA. To match the sample, we used the following firm level variables: Firm size, firm age, ownership structure, export propensity of the firm, financial constraint, top manager experience in the sector, firm’s perception about impartiality of the court, the proportion of full-time female workers, sector and country. The matched sample includes 8,894 firms, of which half of them (4,447) firms have a female top manager and other half do not have a female top manager. Table 7, depicts the regression results of the matched sample, which is consistent with our main findings.

OPEN IN VIEWER

Table 7.

Regression Results With Matched Sample.

Variables	Bribe payment	Bribe demand	Bribe payment
	Coef/SE	Coef/SE	Coef/SE
Top manager gender	−.149** (0.075)	−.154** (0.075)	−.104 (0.079)
Bribe demand			2.073*** (0.099)
Financial constraint	.442*** (0.101)	.563*** (0.099)	.296*** (0.106)
Export propensity	.300* (0.157)	.127 (0.153)	.242 (0.171)
Top manager experience	.044 (0.076)	.036 (0.070)	.043 (0.080)
Firm size	.043 (0.044)	.247*** (0.040)	−.035 (0.049)
Domestic	−.027 (0.244)	.634** (0.307)	−.246 (0.252)
Firm age	.058 (0.082)	.118 (0.077)	.002 (0.086)
Firm performance	−.004* (0.002)	.000 (0.002)	−.005** (0.002)
Legal system quality	−.288*** (0.087)	−.468*** (0.087)	−.146 (0.094)
Ownership structure
Open share	1.112 (1.077)	.359 (0.715)	1.153 (1.029)
Closed share	1.625 (1.007)	.696 (0.634)	1.528 (0.955)
Sole proprietorship	1.585 (1.012)	.490 (0.639)	1.561 (0.960)
Partnership	1.754* (1.025)	.840 (0.654)	1.617* (0.976)
Limited partnership	1.604 (1.014)	.412 (0.640)	1.592* (0.966)
Control of corruption	−.769 (0.944)	−2.334*** (0.878)	−.326 (1.015)
Corruption perception index	.083*** (0.027)	.021 (0.031)	.084*** (0.029)
Business freedom index	−.082*** (0.021)	−.027 (0.018)	−.094*** (0.024)
Log of GDP per capita	1.687 (1.385)	−2.187* (1.205)	1.802 (1.467)
Constant	−11.981 (9.929)	8.329 (8.493)	−10.984 (10.572)
Sector	Yes	Yes	Yes
Year fixed effect	Yes	Yes	Yes
Country fixed effect	Yes	Yes	Yes
Log-likelihood	−2,479.15	−2,492.22	−2,255.24
Number of observations	8,894	8,894	8,894
Chi2	1,064.172	799.984	1,329.046
Pseudo R2	.234	.162	.303

Note. Standard errors in parentheses ***p < .01, **p < .05, *p < .1.

Third, there may be a potential concern of endogeneity in our results because prior research suggests that female top managers may not be assigned randomly to firms (De Cabo et al., 2012). In order to address this concern, we adopted a two-stage least square instrumental variable regression. We used fitted probabilities of top manager gender as an instrument for a potential endogenous binary variable, as Wooldridge (2010, pp. 903–953) suggested. In addition to control variables, we added the Women Business and Law Index (WBL index) from the World Bank and the proportion of permanent full-time workers that are female to fit top manager gender using probit model. The WBL index quantifies discriminatory laws by gender as a measure of legal gender disparity across countries. Angrist and Pischke (2008) suggested the use of two-stage least squares (2SLS) for limited dependent variables with endogenous regressors as the IV method captures the average treatment effect regardless of whether the dependent variable is binary, nonnegative or continuous. Hence, we applied 2SLS with fitted probabilities of top manager gender as an instrument. Table 8 depicts the results of the IV regressions. The results from the 2SLS IV approach are consistent with our main findings. Since our model has one endogenous regressor and one instrument, tests of overidentifying restrictions cannot be performed. Nonetheless, we tested the validity of our instrument (i.e., it must be sufficiently correlated with the included endogenous regressor but uncorrelated with the error term). When there is only one endogenous regressor, the F statistics should exceed 10 for the inference based on the 2SLS estimator to be reliable. As reported in Table 8, the F statistics is sufficiently high to confirm the validity of our instrument. We also reported the eigenvalue statistics as a further test of weak instruments. As can be seen in Table 8, it is also sufficiently high to rule out that our instrument is weak. The 2SLS size of nominal 5% Wald test is 16.38, whereas the minimum eigenvalue statistic is 4,782.53. Since our test statistics exceed the critical value, we can conclude that our instrument is not weak.

OPEN IN VIEWER

Table 8.

Two-Stage Least Square Instrumental Variable Regression (Second Stage Results).

Variables	Bribe payment	Bribe demand
	Coef/SE	Coef/SE
Top manager gender	−.055*** (0.013)	−.060*** (0.013)
Financial constraint	.040*** (0.004)	.038*** (0.004)
Export propensity	.018*** (0.004)	.019*** (0.004)
Top manager experience	.001 (0.002)	.002 (0.002)
Firm size	.003** (0.001)	.015*** (0.001)
Domestic	.012** (0.005)	.013** (0.005)
Firm age	.003 (0.002)	−.003 (0.002)
Firm performance	.000 (0.000)	.000 (0.000)
Legal system quality	−.055*** (0.003)	−.054*** (0.003)
Ownership structure
Open share	.022* (0.012)	−.015 (0.012)
Closed share	.020** (0.010)	.003 (0.011)
Sole proprietorship	.017* (0.010)	−.008 (0.011)
Partnership	.020* (0.011)	.013 (0.012)
Limited partnership	.034*** (0.011)	.010 (0.011)
Control of corruption	−.081*** (0.023)	−.082*** (0.021)
Corruption perception index	.000 (0.001)	.000 (0.001)
Business freedom index	−.008*** (0.001)	−.002*** (0.001)
Log of GDP per capita	.113*** (0.033)	−.121*** (0.037)
Constant	.367 (0.223)	1.425*** (0.251)
Sector	Yes	Yes
Year fixed effect	Yes	Yes
Country fixed effect	Yes	Yes
Number of observations	54,152	54,152
Chi2	11,671.087	5,249.859
Adjusted R2	.188	.114
F-test	3,023.55	3,023.55
Minimum eigenvalue statistic	4,782.53	4,782.53

Note. Robust standard errors in parentheses ***p < .01, **p < .05, *p < .1.

Fourth, given that the bribery process involves a sequential dynamic in our conceptualization, typically beginning with a bribe demand from public officials and followed by a bribe payment decision by the firm, we re-estimate our models using a two-stage modeling approach, reflecting the order in our theoretical argument. In stage 1, bribe demand is estimated as a function of top manager gender and other relevant predictors. Then, in stage 2, the predicted values from stage 1 are used as explanatory variables in a model of bribe payment. The signs and statistical significance of the top manager gender and the predicted values of bribe demand are consistent with those obtained from our main mediation analysis. As expected, the coefficient on predicted bribe demand is slightly smaller than in the main analysis, since it is based on fitted values rather than observed measures. Nonetheless, the results remain consistent with our main findings. ⁸

Fifth, although we include a range of control variables that we identified from the literature in our regression, there is a potential concern that firms with a female top manager may differ on unobservable variables. We followed the approach employed by recent studies (Altonji et al., 2005; Assenova & Sorenson, 2017) to ensure that our results are not driven by unobservable heterogeneity. The ratio test is calculated as ρ = | β ^ F | / | ( β ^ R − β ^ F ) | , where β ^ F denotes the estimated coefficient of top manager gender of the unrestricted (full) model, whereas β ^ R denotes the estimated coefficient of top manager gender in a restricted model (with one fewer regressor) in a regression, being the dependent variable the same. The rations obtained in the regressions are interpreted as how much larger would selection on unobserved factors be relative to selection on observed factors to account for the result. The ratio test should be greater than 1 ( ρ > 1 ) for the model to be robust to unobserved heterogeneity, as suggested by Altonji et al. (2005). The value of ρ in our analysis starts from 23.62 indicating that our main result is not driven by unobserved heterogeneity in firms.

Sixth, as there might be a geographic regional dimension of the gender and bribery relationship, we conducted subsample analyses to check that our main findings are not driven by observations from some specific regions. We undertook five subsample analyses by dropping all firms from one region at a time. The regional classification is based on the World Bank regional classification (sub-Saharan Africa, East Asia and the Pacific, Latin America, Middle East and North Africa, and Southern Asia). ⁹ Our main results stand firm for the subsample analyses.

Seventh, we used alternative measures of bribe payment and bribe demand and re-estimate our models with alternative estimation technique. Bribe payment is measured as percentage of total annual sales, or estimated total annual value paid as a bribe to get things done. For bribe demand, we used depth of bribery which is measured as the percentage of instances in which a firm was either expected or demanded to provide a gift or informal payment during solicitations for public services, licenses or permits. This is the closest continuous variable we got from WBES. We used Tobit regression as both variables are left censored at 0. Our main results hold. ¹⁰

Eighth, we controlled the World Bank’s “Women, Business, and the Law Index” (WBL Index), which considers the laws and regulations that can influence women’s economic opportunities, to check if the relationship between top manager gender and firm likelihood of involvement in bribery is affected by the law and regulations across countries. The WBL index quantifies discriminatory laws by gender as a measure of legal gender disparity across countries. Our main results remain robust when we take into account gender equality across countries. ¹¹

Finally, although we have included several indicators capturing institutional variations across the countries included in our analysis, we also tested if institutional variations moderate the relationship between top manager gender and firm involvement in bribery. We utilized data from the World Bank Governance Indicators. The results show that the indicators do not have a statistically significant moderation effect on the relationship between top manager gender and bribe demand.

Theoretical Contributions

While previous research has investigated several managers’ characteristics as antecedents of firm involvement in bribery, there remains a limited number of studies investigating the relationship between a key top manager characteristic (i.e., gender) and firm involvement in bribery (Dollar et al., 2001; Wood & Eagly, 2012). The results of these few studies are mixed—while some have found that female top managers engage less in bribery than their male counterparts (Breen et al., 2017; Hanousek et al. 2019; Tuliao & Chen, 2017; Xia et al., 2018), other have shown no gender-related differences when it comes to embracing bribery at the firm level (Na et al., 2018; Wellalage et al., 2020). These mixed findings uncover the possibility that the relationship between top manager gender and firm involvement in bribery is mediated by a variable that “calls for action,” namely bribe demand (Goetz, 2007), which is likely to prompt firms to embrace bribery.

Following the literature that conceptualizes bribery as a relationship (Jung and Lee, 2023), and proposing bribery as a two-stage process consisting of bribe demand and bribe payment, our findings contribute to the business ethics literature in general, and the bribery literature in particular, by showing that bribe demand partially mediates the association between top manager gender and firm involvement in bribery. This partial mediation effect has proven to be fairly strong, accounting for some 30% of the total effect. Relatedly, and moving into gender-specific idiosyncrasies, our findings contribute to the literature by showing that firms led by female top managers are negatively associated with involvement in bribery. This could be partly because such firms are also associated with fewer bribe demands from public officials than their counterparts led by male top managers. From the SRT point of view, two key factors may explain this.

First, from the opportunities perspective (Agerberg, 2014), firms led by female top managers are associated with lower number of opportunities to engage in bribery, which derives from the commonly held perception of gender role in society. From this perspective, gender difference in bribery could be understood as a function of the difference in opportunities for corrupt behavior. Second, from the expectations perspective, women are expected to act in a more selfless and collectively oriented manner than men (Anglin et al., 2018; Wood & Eagly, 2012). As a result, bribe takers do not expect them to engage in bribery, and hence are less likely to demand them to do so. These arguments are particularly relevant because we approach bribery as a relationship consisting of repeated interactions. Given the illicit nature of bribery and that public officials demanding bribe largely depend on trust-based networks (de Jong et al., 2015; Shepherd et al., 2021), firms with top managers who do have access to such networks are positively associated with bribe payment and bribe demand. Perceived social roles and expectations, however, may limit access to such networks for female top managers. Consequently, firms led by female top managers would be less associated with bribery relationships. That said, our findings also show that firms led by female top managers still do engage in bribery, especially when demanded, albeit doing so to a lower extent than their counterparts led by male top managers. A potential explanation for this relates to the globally increasing perception of equality between gender roles, particularly in developed parts of the world.

In line with this, our findings further contribute to the literature by showing that firm location moderates the effect of top manager gender on bribe demand, such that firms in urbanized areas led by female top managers are more likely to receive bribe demands than firms led by female top managers in nonurbanized areas. This could be because, as the SRT suggests, urbanization has led women to better integrate in the labor market, and hence prompted a change in perception of their role in society, enabling female managers to get more connected and to integrate into social networks that facilitate bribery. However, not only firm location influences the relationship between top manager gender and bribe demand. Investment in security, namely the money that the firm spends on personnel or professional security service and equipment, also does so. In this regard, our research finds that investment in security positively moderates the relationship between top manager gender and bribe demand, such that firms led by female top managers that invest in security are more likely to receive a bribe demand. A potential reason for this could be that the link that they establish with professional security service providers result in indirect social ties and connections with public service providers, including enforcement agencies. This socialization may, lead them to extend demands for bribery.

Managerial and Policy Implications

The findings of our study have important managerial and policy implications. At the managerial level, while it seems that firms may benefit from promoting female leadership to reduce exposure to bribery, as female top managers are less likely to elicit bribe demands and participate in bribery transactions, it should be recognized that the effectiveness of such measure is shaped by contextual factors such as firm location and investment in security, which influence firms’ interactions with bribe-seeking officials. Thus, firms’ anti-bribery policies and initiatives ought to consider the broader relational dynamics of bribery, within which gender may play a role. For instance, in urbanized areas, variation in the perception of gender roles in society seems to be attenuated, providing female top managers with greater access to social networks that facilitate bribery relationships. Relatedly, firms that invest in security ought to be aware that such investment may open the door for female top managers to receive bribe demands. This is probably because, when interacting with security service providers, the likelihood that top managers create connections and social bonds with officials who provide other public services may increase, and such socialization process may result in the reduction of gender-related stereotypes and the subsequent reception of bribe demands. Consequently, firms should adopt a more holistic approach that combines leadership diversity with context-specific antibribery strategies.

At the policy level, our findings suggest that policy measures aimed at increasing the representation of women in corporate executive positions and boards may have a positive economic impact by reducing the likelihood that firms engage in bribery relationships. This further builds on the social impact that such policy measures may have on fostering gender equality at the firm level, particularly in management positions. However, since bribe demand seems to be a key mechanism driving firms’ involvement in bribery, policymakers should also explore strategies to reduce opportunities for officials to solicit bribes by simplifying administrative processes, limiting discretionary power, and minimizing direct interactions between firms and public officials. In addition, the importance of firm location as a moderating factor suggests that antibribery efforts should be context-sensitive and ought to recognize gender-based roles in the society where the policy instrument is to be introduced. Furthermore, improving overall security conditions as well as the business environment may help lessen the likelihood that firms with a female top manager receive bribe demands. Overall, it is important to recognize that bribery is a systemic issue and may not be effectively addressed solely by focusing on gender composition within firms. Thus, gender-related policy measures aimed at reducing bribery should be part of a broader, multifaceted approach, given the complexity of this illicit practice.

Limitations and Future Research

Despite its theoretical contributions and managerial and policy implications, our study has certain limitations that open up future research avenues. First, although the WBES dataset enabled us to cover a large number of observations from several countries, which enhance the generalizability of our findings, our research may suffer from the cross-sectional nature of the data. For instance, due to the cross-sectional nature of our data, we were not able to implement a firm level fixed effect in our models, which would have helped us to address firm level heterogeneity. To alleviate this and other potential concerns, the directionality of our hypothesized effects is supported by theoretical plausibility. In addition, we applied different estimation techniques in our robustness analysis. For instance, we utilized propensity score matching to correct the unobserved heterogeneity. Furthermore, we deployed the two-stage instrumental variable regression to mitigate endogeneity concerns in our findings. The results from these robustness checks are similar to our main results, which enhances the confidence in our findings. However, future research should ideally use longitudinal data. In this regard, it would be relevant to compare bribery levels before and after the appointment of a new top manager between firms appointing a female vs. a male top manager. Moreover, as gender norms and social expectations are not static, a longitudinal design would allow future research to examine how evolving societal views about gender roles may influence patterns of bribery over time. For example, an increase in societal acceptance of women in leadership positions might lessen the current gap in bribe demands received by male and female top managers. Tracking such changes could provide deeper insights into how contextual dynamics shape the observed relationships. In addition, longitudinal data would also help to enhance the precision and temporal elements in the bribery process, such as how the demand-payment process unfolds.

Second, we acknowledge that there might be a response bias in the dataset. For example, one could argue that female top managers may have underreported their misconduct, which would still be in line with the SRT. In fact, regardless of the gender of the respondent, response bias is one of the key challenges that studies on bribery and corruption face. However, in the WBES the questions posed to measure bribe demand and bribe payment were framed in an indirect fashion, precisely in order to address any self-censoring biases respondents may have. Although they might be less precise, indirect questions are commonly used in research addressing sensitive topics like bribery, because they are more likely to generate genuine responses (Jung & Lee, 2023). Furthermore, anonymity of both the firm and the respondent were assured in the survey, which may help to reduce response bias. The measurements utilized in this study have also been utilized in prior studies addressing the issue of bribery (Birhanu et al., 2016; Jung & Lee, 2023; Webster & Piesse, 2018; Wellalage et al., 2019), and hence are fairly established. Nevertheless, we acknowledge that our results need to be interpreted cautiously by taking into account the possible response bias that could be caused by the gender of the respondent.

Third, based on prior literature and evidence (Hunt & Laszlo, 2005; Jung & Lee, 2023; Olimov, 2024), we assume that bribery is initiated from the demand side (i.e., bribe takers), and the survey questions measuring our dependent variable were designed in light of such assumption. Accordingly, our study does not capture bribery initiated by bribe payers. Furthermore, given that in most of the countries in this study, men still dominate the political landscape, their perception toward women and gender roles is crucial to our theorizing based on the SRT. However, as the political landscape changes, and the proportion of women in politics increases, it would be interesting to explore if the gender difference in bribery changes along with perceptions toward gender roles. Thus, future research may address this through a longitudinal study linking changes in the political landscape, gender roles in society, and levels of bribery at the firm level. This type of analysis could help determine whether observed gender differences in bribery exposure are persistent and/or whether social and political changes lower or amplify these differences.

Fourth, we only measured bribery in monetary terms. However, women may face different forms of bribe demand from bribe takers (Goetz, 2007), including “acts of a sexual nature as corruption currency” (UNODC, 2020, p. 4). Although the question used in this study to measure bribe depth is a financial measurement (i.e., percentage of sales), the question related to bribe demand may well include other forms of payment, because it was phrased as “whether informal gift or payment” is demanded. Thus, we cannot argue that the question is explicit enough and that respondents have not considered other types of bribe demand as part of their answer. Future research may address this issue by looking at different forms of bribery and whether difference in gender-related behavior still holds.

Fifth, while our study offers important insights into the relationship between top manager gender and firm involvement in bribery, it is important to acknowledge that bribery is a multifaceted phenomenon that takes various forms, such as low-level facilitation payments and large-scale bribes. Our analysis does not distinguish between these different types, and it is possible that gender differences in exposure to or involvement in bribery may vary across these categories. Future research could explore whether and how gender-related dynamics differ depending on the nature and scale of corrupt practices, which would further refine understanding and inform more targeted antibribery strategies.

Finally, future studies could benefit from exploring whether the observed gender dynamics apply similarly across different types of bribery. Qualitative methods, in particular, could offer a deeper understanding of how gender, power dynamics, and organizational context influence experiences with petty versus grand bribery. Mixed-method or multilevel research designs could further illuminate how such distinctions operate in practice.