Bribery, regulation and firm performance: evidence from a threshold model

The negative consequences of corruption on economic growth have been long examined (Mauro 1995; Mo 2001; Lambsdorff 2003; Cieslik and Goczek 2018; Gründler and Potrafke 2019, among many others), yet another strand of literature argues that corruption ‘greases the wheels’ by overcoming some of the ineffective institutions, regulations and bureaucracy (Lui 1985; Acemoglu and Verdier 2000; Meon and Weill 2010). A similar set of findings are observed at the firm level. One stream of literature finds that corruption at the firm level is negatively associated with firm performance (Fisman and Svensson 2007; Faruq et al. 2013; Şeker and Yang, 2014; Hanousek and Kochanova 2016; Yang et al. 2021; Demir et al. 2022), while another finds that bribe payments affect firm performance positively (Vial and Hanoteau 2010; Dreher and Gassebner 2013; Jauregui et al. 2020).

Country-level regulations are usually designed to benefit countries by eliminating market failures caused by monopolies and externalities (e.g., Pigou 1920). For instance, regulations mitigate negative externalities (e.g., air pollution) and enable competition. However, arguably, firm-level regulations increase bureaucracy and could create additional barriers to entrepreneurship activity and firm performance. For instance, Dreher and Gassebner (2013) argue that regulations required to carry out business activities (e.g., the number of procedures required to start a new business, the number of days required to start a new business, the costs of starting a new business, and the minimum capital required to start a new business) deter firms entering the market and corruption could reduce the negative impact of regulations on entrepreneurship activity. Similarly, Jiang and Nie (2014) found that corruption could act as a shortcut to firm-level regulations and bureaucracy and increase profitability. In other words, even though country-level regulations are required to overcome some market failures, the business-related regulations (e.g., number of procedures needed to start a business, minimum capital requirements, time and capital required to carry out regulations related to the taxes, customs, labor regulations, licensing and registration, and so on) may be detrimental for entrepreneurship activity, firm entry and performance.¹ In a recent paper, Amin and Ulku (2019) examined the interrelationship between firm-level corruption and firm performance and found a significant negative relationship between corruption and firm productivity when regulation is high and an insignificant relationship between corruption and firm productivity when regulation is low. However, their paper examines this relationship through an interaction term to capture the nonlinear effect, ignoring the possible heterogeneous impact of other variables (including regulation itself). Other papers in the literature examine the indirect effects of bribery and regulation on different characteristics of the firm. For instance, Wellalage et al. (2020) demonstrate that firms exposed to more bribes tend to face more financial difficulties when applying for credit, while Kitching et al. (2015) provide a theoretical framework that regulation could be a burden or a beneficial factor for firms. Given these interlinkages between regulation, bribery and other firm characteristics, this paper uses a threshold model estimation by using firm-level regulation and corruption as threshold variables to identify the heterogeneous effects of regulation and corruption and other firm characteristics on firm performance by splitting the sample endogenously using regulation and bribery as threshold variables.²

Using an extensive firm-level data set that covers 20,343 firms in 78 developing countries and a threshold regression methodology, we show that firm-level corruption and regulation play a vital role in the factors affecting firm performance. The existing literature argues that bribery could be helpful if firms face rigid business-related regulations that forces them to pay high bribes to ‘get things done’. We find that bribery intensity affects firm performance negatively, but this effect is higher for firms that are exposed to lower bribe levels, while for the same firms with lower bribe exposure regulation could be beneficial as it could promote firm innovation (e.g., Weiss et al., 2019; Tchorzewska et al. 2022). Our findings confirm that access to external finance improves firm performance if firms are exposed to bribes and firm-level regulation is below a given threshold. While bribery harms firm performance, our findings reveal that spending time with regulation could improve firm performance if firms are exposed to low levels of bribery, confirming the so called ‘sand the wheels’ hypothesis.

The remainder of the paper is organized as follows. Section 2 provides a literature review on the role of regulation and bribe for firm performance and sets the hypotheses of the paper. Section 3 provides the details of the data and the variables used in this paper. Section 4 offers the linear and threshold models, and the results are given in Sect. 5. Finally, Sect. 6 concludes.

There is a large body of literature that has been examining the effect of firm-level corruption on firm performance (Fisman and Svensson 2007; Vial and Hanoteau 2010; Faruq et al. 2013; Dreher and Gassebner 2013; Şeker and Yang, 2014; Hanousek and Kochanova 2016; Williams and Kedir, 2016; Krammer 2019; Jauregui et al. 2020; Qi et al. 2020). The literature provides a mixed set of findings (see Martins et al. (2020) for detailed literature on the effect of firm-level corruption on firm performance). On the one hand, firm-level corruption increases the costs of firms, limits entrepreneurship (see e.g., Dutta and Sobel 2016) and leads to inefficient allocation of resources within firms (Boudreaux et al. 2018), and hence harms the firm performance (e.g., Fisman and Svensson 2007; Faruq et al. 2013; Şeker and Yang, 2014; Hanousek and Kochanova 2016). Some other literature argues that firm-level corruption could overcome excessive business-related regulation and bureaucratic obstacles and improve firm performance (e.g., Dreher and Gassebner 2013; Krammer 2019; Jauregui et al. 2020). For instance, Krammer (2019) demonstrates that bribery is positively associated with the innovative performance of firms (i.e., new product introductions) in emerging markets by overcoming bureaucratic obstacles. Similarly, by using World Bank Enterprise Survey data across 40 African countries, Williams and Kedir (2016) show that corruption is positively associated with firm sales, employment and productivity growth rates, while Jauregui et al. (2020) show that corruption leads to an increased formation of new firms in Mexico. Contrary to this literature, Amin and Ulku (2019) find that firm-level corruption affects firm performance negatively if firms face high business-related regulations. Overall, the findings concerning the effect of corruption in high business regulation environments are mixed.

Some studies support the “sand the wheels” hypothesis, and others support the “grease the wheels” hypothesis. With the former hypothesis, corruption hampers firm performance, while corruption could ‘grease the wheels’ in the latter hypothesis and improves firm performance. Nur-tegin and Jakee (2020) examine different types of corruption and find that certain types of corruption may help “grease” business transactions, and some others “sand” the wheels. Similarly, Martins et al. (2020) explore the relationship between firm-level corruption and firm performance in different geographical clusters and show that corruption “greases the wheels” of business for African firms, but it “sands the wheels” for firms in Latin America, the Caribbean, Eastern Europe, Central Asia, and Southern Asia. Furthermore, Sharma and Mitra (2015) show that bribe acts as a tax on the profits of firms and reduces efficiency, but at the same time, increases firms’ exports and product innovation.

Even though most of the existing literature finds that excessive regulation increases corruption and hampers economic growth (e.g., Djankov et al. 2002), recent analyses show that regulation could also improve firm performance (Kitching et al. 2015). For instance, Wu (2019) demonstrates that regulation (i.e., time spent with government officials) increases firms’ innovation capacity and, subsequently, productivity. Similarly, there are also mixed results regarding the role of regulation on the effect of corruption on firm performance (e.g., Dreher and Gassebner 2013; Krammer 2019; Amin and Ulku 2019; Jauregui et al. 2020). Hence, our paper argues that firm-level regulation and bribery may lead to heterogeneous and differential effects on firm performance.

A growing body of literature examines the role of corruption and regulation on other firm characteristics. For instance, Khwaja and Mian (2005) demonstrate a connection between lending and politically connected firms. Furthermore, in a recent paper, Bircan and Saka (2021) find that credit constraints for firms in opposition areas are relatively higher, which then drops employment and sales of these firms. Using 14 transition economies, Fungáčová et al. (2015) demonstrate that bribery is positively associated with firms’ bank debt ratio suggesting that bribery enables firms to access external finance. On the other hand, Galli et al. (2017) show that small businesses in highly corrupt countries face a greater probability of self-restraint regarding their loan applications compared to small firms located in low-corruption economies. Finally, Wellalage et al. (2020) show that firm-level corruption in India is detrimental to accessing credit and the negative effects vary based on the age and size of the firm. The literature mentioned above highlights that the impact of external finance on firm performance could vary based on firm-level bribery.

It has been found that human capital (manager experience) is also an essential factor for firm performance (see e.g., Peni 2014; Staniewski 2016; Wu 2019). However, the effect of human capital on performance is less effective in more corrupt environments (see e.g., Ehrlich and Lui 1999; Boikos 2016; Feldmann 2017; Hoa 2020). Ehrlich and Lui (1999) argue that individuals invest more in political capital rather than human capital to improve their negotiation and bureaucratic power in corrupt environments. Boikos (2016) shows that the marginal effect of public expenditure on education on human capital accumulation decreases with the increase in corruption. Hoa (2020) also shows that corruption harms labor quality. Finally, Feldmann (2017) demonstrates that economic freedom increases the return on human capital investment. Even though there is no firm-level study about the potential effect of corruption on human capital effect, the impact of human capital on firm performance may vary based on firm-level corruption and regulation.

Finally, some papers also show that different firm characteristics are related to firm-level corruption. For instance, Sharma and Mitra (2015) show that firm-level corruption increases exports. Similarly, Ha et al. (2021) show that larger firms benefit more from bribes through their improved export performances. De Jong et al. (2012) argue that small- and medium-sized firms are more likely to pay a bribe than their larger counterparts as larger firms have more resources to influence government officials or use these resources to take legal action. Examining the firm performance in Indian firms, Jain (2020) also demonstrate that bribery is more harmful to smaller and older firms than larger and younger ones. Overall, regulation and bribery intensity have varying effects on firms with different characteristics.

This paper aims to investigate whether the firm characteristics have heterogeneous effects on firm performance at different regulation and bribery levels. While the threshold model could capture such effects, an interaction term between bribery and regulation would only capture the effect of bribery on firm performance at different regulation levels. Therefore, we employ a threshold regression model to explore the potential heterogeneous effects of firm characteristics on firm performance at different bribery (regulation) levels. The next section provides the data set used in this paper and details of the variables employed in the empirical analysis, and Sect. 4 provides the proposed methodology.

We employ firm-level data from the World Bank Enterprise Surveys database to estimate the heterogeneous effects of firm characteristics on firm performance at different bribery and regulation levels.³ This data set covers various firm-level characteristics, allowing us to control for additional factors likely to affect firm performance. Furthermore, given the detailed information regarding the firm’s sectoral and geography, we can also construct an instrumental variable (IV) for bribery intensity.

Similar to the previous literature (see e.g., Fisman and Svensson 2007; Şeker and Yang, 2014), we use the real annual sales growth for firm performance.⁴ For the level of bribery that firms face, we use the bribery variable (B). The bribery variable measures the percentage of instances in which a firm was either expected or requested to provide a gift or informal payment when conducting six specific business transactions (i.e., gift or informal payment requested during the applications made for i) electricity, ii) water connection, iii) construction-related permit, iv) import license, v) operating license, and vi) during the meetings with tax officials). For regulation, we use the natural logarithm of the percentage of senior management time spent in dealing with requirements of government regulation (REG). The regulation variable measures the time required by managers to deal with the government regulations such as taxes, customs, labor regulations, licensing and registration, including dealings with officials and completing forms (see Appendix Table A1 for detailed measurement of each variable), and therefore, measures the firm-level regulation.

Beyond the main variables of interest, we control for the standard set of variables: the age of the firm (Age), the percentage of exports (Exports), size of the firm (i.e., medium and large dummy variables that are equal to 1 if the firm has 20–99 and has 100 or more full-time workers, respectively), foreign ownership (Foreign), a dummy variable measuring whether a firm is solely owned (Sole ownership), external finance (i.e., percentage of the working capital that was financed by banks, suppliers, or other sources), top manager’s experience in the industry (Manager experience), and the percentage of total annual losses are due to crime (CRIME). The detailed construction and description of the variables are presented in Appendix Table A1.

Corruption and bribery are very hard concepts to measure and there are important discussions about their measurement and reliability (see e.g., Reinikka and Svensson 2006; Seligson, 2006; Banerjee et al. 2013; Kraay and Murrell 2016; Boikos et al. 2023). Since bribery is self-reported, it is usually underreported by firms that benefit from it (Banerjee et al. 2013). Furthermore, firm-specific unobserved factors such as a favorable demand forecast could affect a firm’s performance and bribery positively, and therefore, the estimate of the bribery may be biased toward zero (Fisman and Svesson, 2007). As discussed by Pounov (2016), Amin and Soh (2022) and Boikos et al. (2023), using average bribery intensity in the country, location, and sectoral cluster reduces the measurement error compared to individual firm-level bribery level. Furthermore, another major concern of the analysis is the potential endogeneity of the bribe variable. To deal with this issue, we construct an IV for the bribe variable by obtaining the average bribery intensity in the country, location, and sectoral cluster (see e.g., Fisman and Svensson (2007), Şeker and Yang (2014), Mendoza et al. (2015) and Wellalage et al. (2020) for the use of a similar IV). Therefore, using an instrumental variable estimation tackles the endogeneity problem and reduces measurement errors. We use three general sectoral groupings: manufacturing (ISIC 15–37), services including retail (ISIC 51–52), and other service sectors like transportation, hotels, and restaurants, and construction services to obtain the instrumental variable and used a minimum of 20 observations in this cluster. Finally, we eliminate the countries that have less than 200 firms in the sample. The final data set consists of 78 developing countries with 20,343 firms. Each country has a different survey period, and we use the final survey year for each country covering the survey period between 2013 and 2017 (see Appendix Table A2 for the list of countries and the survey year).

Table 1 presents the descriptive statistics of the variables used in our analysis. It should be noted that the firm performance (sales growth), bribery, exports, external finance and crime variables are measured as percentages. On the other hand, the natural logarithms of the regulation, firm’s age, and manager’s experience are taken. Finally, the medium, large, foreign and sole ownership variables are dummy variables. It can be observed that 42%, 36% and 22% of the firms in the sample are classified as small, medium and large (firms that have less than 20, have 20–99 and have 100 or more full-time workers, respectively), respectively. Overall, there is a good variation in the real annual sales growth rates across firms, ranging between − 100 and 100%. Furthermore, we also observe a good level of variation in bribe and regulation variables.

Figures 1, 2 and 3 offer average real annual sales growth, bribery and regulation per firm across countries, respectively (see Table A3 for details for each country). On average, firms in some African, Eastern European and Central Asian countries experienced negative sales growth. For instance, the annual sales growth of an average firm in South Sudan, Tanzania, Uganda, Laos, Ukraine and Kazakhstan decreased by 41%, 32%, 28%, 25%, 15%, and 14%, respectively. On the other hand, an average firm in Tajikistan, Sudan, Uzbekistan, Morocco and Turkey experienced 14%, 13%, 12%, 11%, and 11% increases in their real sales growth rates, respectively. Similarly, bribery experienced by an average firm shows a significant variation across countries and the presence of bribery is relatively higher in some countries than others (Fig. 2). For instance, an average firm in Yemen, Cambodia, Kyrgyzstan, the Democratic Republic of Congo, Afghanistan, Bangladesh and South Sudan paid bribes for more than 35% of their transactions. On the other hand, firms in Georgia, Latvia and Estonia reported no bribery transactions, and only a limited number of firms in Slovenia, Israel, Bhutan and Slovakia paid bribes. Finally, the amount of time an average firm manager spends with regulations varies across countries (Fig. 3). Contrary to bribery, top managers of the firms in Latin America and Eastern European spent more time dealing with government regulations compared to firms in other regions. For instance, an average firm’s top manager in Uzbekistan, Poland, Croatia, Argentina, Tajikistan, Bhutan and Tunisia spent more than 20% of their time dealing with the regulations, while relatively less time is required for regulations in most African and Asian countries. The correlation matrix (Table 2) highlights that bribery (regulation) is negatively (positively) correlated with the annual growth rates, and there is a significant negative correlation between regulation and bribery. In other words, firms exposed to more bribery tend to spend less time dealing with regulations. Specifically, our data reveal that within-country correlations indicate Serbia, Poland, and Croatia display the strongest negative associations between bribery and firm performance. Conversely, Afghanistan, Dominican Republic, and Belarus exhibit the strongest positive correlations between regulation and firm performance. We also find Morocco, Nicaragua, and Czech display the strongest negative correlation between regulation and bribery.

The baseline model used for the empirical estimation is shown in Eq. (1)

where the dependent variable is the real annual sales growth of firm j at year t. The variable \({B}_{jt}\) is the percentage of bribery; \({REG}_{jt}\)is the natural logarithm of the percentage of senior management time spent in dealing with requirements of government regulation. The vector \({X}_{jt}\) contains the control variables reflecting firm characteristics that include firm’s age, exports, size, ownership, finance status, manager’s experience, and vulnerability to crime. The full set of yearly and sectoral dummies are also included. As each country has a different survey period, \({\lambda }_{t}\) is time dummies to account for time-specific factors. \({\eta }_{s}\) is sectoral dummies for manufacturing (ISIC 15–37), services including retail (ISIC 51–52), and other service sectors like transportation, hotels, and restaurants, and construction services.

Following the threshold models of Hansen (2000), Caner and Hansen (2004), Kourtellos et al. (2016) and Seo and Shin (2016), to examine the potential nonlinear impact of firm characteristics on firm performance at different corruption and regulation levels, we extend Eq. (1) to be regime-specific, where the group is determined by the value of the threshold variable. The panel threshold regression model is of the form:

where \({\gamma }_{0}\) is the true threshold level and \({q}_{jt}\) is the threshold variable.

We consider the threshold variable to be bribery variable (B) and senior management time spent in dealing with requirements of government regulation (REG). Note that we can integrate (2) and (3) to have a compact form of the threshold model as follows:

where \(\gamma ={\left[{\alpha }_{0}^{2}, {\alpha }_{1}^{2},{{ \alpha }_{2}^{2}, {\beta }^{2}}^{T}\right]}^{T}\), \({\Omega _{{\rm{jt}}}} = {\left[ {1,{B_{jt}},RE{G_{jt}},X_{jt}^T} \right]^T}\), \(\delta = {\left[ {\alpha _0^1 - \alpha _0^2,\alpha _1^1 - \alpha _1^2,\alpha _2^1 - \alpha _2^2,{{({\beta ^1} - {\beta ^2})}^T}} \right]^T}\), and \(I\left(.\right)\)is the indicator function.

To control for endogeneity resulting from the bidirectional causality between \({g}_{jt}\)and \({B}_{jt}\), we use a generalized method of moment (GMM), see Caner and Hansen (2004) and Seo and Shin (2016), to estimate (1) and (4) with the moment condition being given by \(E\left({z}_{jt}{\epsilon }_{jt}\right)=\)0 where \({z}_{jt}\) is the instrument vector. We instrument \({B}_{jt}\) using the percentage of the sectoral-location-country average of the bribery variable.

Most of the existing literature on the determinants of firm performance examines the direct linear effects of the standard set of variables and ignores the possible complex relationships among firm characteristics. In this paper, using extensive firm-level data and a threshold regression methodology, we show that firm-level corruption and regulation play a vital role in the factors affecting firm performance. We demonstrate that bribery intensity affects firm performance negatively, but this effect is higher for firms that are exposed to lower bribe levels. On the other hand, we found that the regulation could improve firm performance if firms are exposed to lower levels of bribery. The existing literature argues that bribery could be helpful to ‘get things done’ if firms face rigid business-related regulations and therefore spending more time with regulations could be counter-productive if firms pay high bribes to ‘get things done’. However, if the firms are exposed to low levels of bribery, then business-related regulation could promote firm performance since regulation could promote firm innovation (e.g., Weiss et al., 2019; Tchorzewska et al. 2022).

We also find that regulation and bribery intensity are relatively more beneficial for certain types of firms. Firms that pay more bribes and spend more time with government regulations above a given threshold are more export-oriented and exports boost firm performance relatively more if firms spend more time with regulation and pay higher bribes. Our findings reveal that external finance becomes ineffective in improving firm performance if firms are exposed to high levels of corruption and regulation. However, if firms are exposed to lower levels of regulation and corruption, then access to external finance improves the firm performance. Overall, our findings highlight while high business-related regulation and bribery may improve the returns to exports, business-related regulation and access to external finance could boost firm performance if firms pay lower bribes. Our paper shows that the effect of firm characteristics on firm performance displays variation based on business regulation and bribery levels.

The findings of this paper have various policy implications. Firstly, eliminating firm-level bribery improves firm performance and allows firms to benefit from firm-level regulation. Therefore, developing countries should prioritize policies to detect bribery practices by government officials and eliminate such informal practices. Secondly, high bribery boosts firm performance through the increased positive effect of exports on firm performance. Given that bribery has direct and indirect negative implications for firm performance, the governments could increase the efficiency of export-related licensing and registration activities (e.g., improve the time required to carry out these activities) to limit bribery requests.

Overall, even though the previous literature has examined the role of different firm characteristics on firm performance, their complex relationships are harder to capture with linearly specified models. In this paper, we found that beyond their direct effects on firm performance, regulation and bribery intensity could alter the effects of different factors on firm performance. Future research could examine other possible heterogeneous effects of firm characteristics by using different firm characteristics as threshold variables.