A Users’ Perspective on Corruption: SMEs in the Hospitality Sector in Kenya

ABSTRACT Although the phenomenon of corruption has been studied intensively, most studies focus on causes and effects of corruption at the societal level. There is a dearth of studies looking at what corruption means to individual actors. This article investigates how managers of small and medium-sized enterprises in the hospitality industry in Lamu, Kenya engage in and cope with corruption. Our findings suggest an approach that models corruption choices as rational individual decisions is likely to fall short, as corruption-related costs and benefits depend on group membership, which for some (but not for all) to a certain extent is a decision variable itself. Our study shows that ‘on the ground’ corruption is a complex phenomenon that can only be fully understood by taking the particularities of the social context into account.

Very small firms can often remain 'under the radar' of corrupt officials. Larger firms tend to attract attention from officials, but are in a more equal bargaining position than SMEs. Moreover, the absolute amount spent on bribes does not seem to increase proportionality to firm size (Kimuyu 2007). Hence SMEs, which form the largest part of the private sector in many developed and developing countries, are particularly exposed. This is an important issue, as these enterprises are also considered a main driver of economic development, and institutional failures such as corruption seem to impede their growth (Beck & Demirguc-Kunt 2006).
The literature on corruption however gives us little insight into how managers of SMEs cope with corruption in their everyday lives (Mezias & Mezias 2010). In fact, 'for all that has been learned about corruption in recent years, the question how firms should manage in the face of corruption continues to be among the most important and elusive' (Rodriguez, Siegel, Hillman & Eden 2006: 9). In this study we aim to further our understanding of what corruption means to managers of SMEs in a developing country, and how they cope with it. For this purpose we performed a qualitative study, combining open interviews and non-participant observations during a three-month stay by the first author on the island of Lamu, off the coast of Kenya. Our purpose was to find out what the managers of firms in the tourist sector (the most prominent economic activity at Lamu) considered to be corruption, how they dealt with it, and what impact it had on their day-to-day business. The tourist sector on the island of Lamu provides an ideal micro-cosmos to study corruption and SMEs, as all the firms are independently owned, some by locals, others by investors from outside of Lamu, and again others by foreigners. This is important as corruption can have different effects on local and foreign firms, increasing the 'liability of foreignness' (Mezias & Mezias 2010). Hence our study aimed at exploring the experiences of the 'users' of corruption (the bribers), and the effects these have on their businesses, in contrast with the focus on corrupt officials and institutions, and the effects at the societal level, as in much of the literature.
The article proceeds as follows. In the next section we briefly review the literature on corruption, focusing on the individual user perspective. We then describe our research context and methods. The empirical results are presented in the following section. We conclude with a discussion of the implications of our findings, limitations of the study, and suggestions for future research.

Background
Although corruption is easy to observe in daily practice, it is difficult to define. A very simple definition of corruption is that of the World Bank: the abuse of public office for private gain (Abed & Gupta 2002). This is one of the most commonly used definitions of corruption within the public domain. Unfortunately, this definition does not take into account the benefit one may seek beyond one's immediate personal gain, for example benefits to one's party, class, tribe, friends or family. Moreover, corruption is not exclusively associated with dishonesty, a lack of integrity and abuse in the public domain. A more general definition is provided by Vito Tanzi: 'Corruption is the intentional noncompliance with arm's-length relationships aimed at deriving some advantage from this behavior for oneself or for related individuals ' (1995: 172). Both definitions still leave plenty of room to discuss which acts should be considered to be corrupt and which not. In order to understand corruption more fully it should be perceived as a subsystem that is embedded in the ethics, economy, politics and culture of a society (Abed & Gupta 2002). This means that approaches that abstract from the environment in which corruption takes place are bound to fall short.
There are also many different types, forms and categories of corruption. The forms most often mentioned in the literature are extortion, nepotism, fraud, embezzlement, bribery, cronyism, patronage, and graft (Aidt 2009;Abed & Gupta 2002;Shleifer & Vishny 1993). George Abed and Sanjeev Gupta (2002) propose a series of categories involving every individual form of corruption, be it bureaucratic or political; cost-reducing or benefit-enhancing; briber-initiated or bribee-initiated; etc. That corruption can assume so many different forms convolutes the discussion. To simplify things Boris Begovic (2005) proposes three basic types of corruption, which form a typology in which the different forms of corruption encountered by managers can more easily be classified. These three types of corruption are 'rule-bending', 'rule-breaking', and 'state capture'.
The first type of corruption, rule-bending, is corruption within the limits of existing regulations. One possible form is accelerating attainment of some specific right that the citizen or firm is entitled to anyway (Begovic 2005). This is a type of corruption that entitles the briber, the user of corruption, to preferential treatment or interpretation of certain laws and regulations, without the corrupt party actually operating outside its legal boundaries. Andrei Shleifer and Robert Vishney (1993) call this form 'corruption without theft'. The most common occurrence of this type of corruption is allowing someone to skip the queue when applying for documents or licences to which they are legally entitled. There is no legal barrier for these documents or licences to be issued, faster attainment of these through corrupt dealings is therefore a mild form of corruption. This type of corruption is usually considered a form of greasing the wheels of the system (Leff 1964). It operates within the confines of the law and is similar to customer preferential treatment in the private sector, which also imply the payment of a certain amount of money in order to get faster or improved service.
The second type of corruption, rule-breaking, does in contrast violate legal rules, or at least presents a very biased enforcement of the rules (Begovic 2005). This type is also referred to in the literature as administrative corruption and is the most well-known and researched type of corruption (Habib & Zurawicki 2002;Abed & Gupta 2002;Tanzi 1995;Ake 1996;Leff 1964). Corruption of this kind may not only involve bribing but also fraud and extortion. Lastly, there is 'state capture', a type of corruption with the primary goal to change the existing rules and regulations in such a way that they favour the interests of the corruptor. The concept of state capture is primarily used to describe phenomena in transitional economies with weak institutions.
The diversity of types of corruption makes it difficult to make general statements. The effects of corruption can be positive or negative, depending on the type considered, and the perspective assumed. For instance, the manager of a firm that receives a licence quickly because of corruption may see this as positive, while the same manager may frown upon laws that seem to be made in order to favour particular competitors. In order to better understand corruption we can assume the user's perspective, and look at the question why an individual, for example the manager of an SME, would be willing to engage in corrupt activities. The obvious reason is that they assume the net effects will be positive, which gives them an incentive to do so.
Jin-Wook Choi (2009) proposes a simple formula modelling the individual's decision regarding corruption: Choi's formula states that individual utility is determined by the expected benefits from corruption minus the disutility from the probability of being caught multiplied by the severity of ensuing sanctions (Choi 2009). The formula does not take into consideration any moral obligation a manager could have to oppose corruption, nor the embeddedness of corruption in the social, cultural, economic or political environment, it purely focuses on utilitarian considerations. Clearly corruption can yield a number of positive effects for those who engage in corrupt behaviour, otherwise the phenomenon would not be so persistent. As a reward for bribing one gains a specific advantage over those who do not bribe. If these incentives are larger than the money spent corruption is beneficial. Leff (1964) is seen as the most prominent advocate of the controversial idea that corruption might be a facilitator of economic exchange and that therefore corruption has a positive effect on business and national economic development. From the perspective of managers of firms a strong incentive to engage in corruption is time gains. Bribing saves time by speeding up the various processes of the bureaucratic system. Such 'greasing the wheels of the system' (Shleifer & Vishny 1993) is usually associated with the rulebending type of corruption. Time is saved by avoiding the usual delays in obtaining licences and other documentation. Managers have a monetary incentive if the bribe paid is low relative to the chance and consequences of getting caught (Choi 2009). Managers might also consider paying a bribe to avoid unnecessary bother from government officials (Ashforth, Gioia, Robinson & Treviño 2008).
However, corruption also has negative consequences. From a normative perspective even mild forms of corruption are outside legal or normative boundaries and therefore considered wrong. Moreover, there are utilitarian reasons not to engage in corrupt activities. Even if the negative effects corruption may have on a society are left out of consideration, managers themselves may experience negative effects when engaging in corrupt behaviour. First of all, corruption opens the door for opportunism and denunciation within the implicit agreement between the two parties involved. Corruption must be based on trust, since legal redress is not an option (Lambsdorff 2002). Therefore, it creates an opportunity for those in power to abuse the agreement by threatening the bribee. There is also the possibility that corrupt officials or people in power will artificially create delays or obstacles in order to elicit a bribe (Svensson 2005). In that case, something which is seen as a positive effect of corruption is merely the circumvention of a distortion put in place deliberately to attract bribes.
At first glance the benefits to the user of corruption may often seem to outweigh the potential disadvantages or dangers. As indicated above, however, the calculus of costs and benefits of engaging in corruption are moderated by trust. Members of a conspiracy of graft or corruption cannot simply assume others are trustworthy (Uslaner 2008). Users of corruption have to trust their partners in a corrupt transaction follow through on the agreed upon benefits and will not expose them to denunciation. Corrupt officials on the other hand also need to feel sure the users will deliver on their promises and will not denounce them in their turn. Trust can thus determine the perceived likelihood of a corrupt transaction being beneficial or harmful to the user. Economic action is embedded in structures of social relations (Granovetter 1985). Therefore, pre-existing social relationships lay the foundations for economic exchange, including corrupt deals, by providing the required safeguards against opportunism. This means that the crucial question is who is sufficiently embedded in one's network of social relationships and who is not, i. e. who is an insider and who is an outsider (Lambsdorff 2002).
The perspective on social relations shifts the focus from the individual moral attitudes and incentives to relationships of corruption (Lambsdorff 2002). According to Eric Uslaner (2008) social inequality is key to understanding corruption. Inequality in a society leads to low generalised trust and pockets of high in-group trust, which sets the stage for corruption. The effect goes in both directions, as social bonds can also be further strengthened by corrupt activities. Partners to a corrupt agreement are 'locked-in' to their mutual relationship, because of the ever-present threat of mutual condemnation (Lambsdorff 2002).
This brief exploration of the literature shows that corruption comes in different forms, requires trust, can have both benefits and costs, and is both based in and constitutive of social relations. What is mostly absent in the literature, though, is an empirical view of corruption as it manifests at the micro level. Many studies look at macro-level outcomes of macro-level measures of corruption (for an overview see Pelligrini 2011). The few studies that do look at the impact of corruption at the micro level tend to employ rather simplistic measures. Thorsten Beck and Asli Demirguc-Kunt (2006) for instance use the World Business Environment Survey, in which corruption as faced by businesses is measured by the following item: 'How problematic is corruption for the operation and growth of your business: no obstacle (1), a minor obstacle (2), a moderate obstacle (3), or a major obstacle (4)?' Our study is motivated by the desire to acquire much more in-depth knowledge of how managers of SMEs are confronted by corruption, what their coping strategies are, and to what outcomes these lead. For this purpose we conducted an in-depth qualitative study.

Kenya and Lamu Island
Lamu Island, formerly known as Kiwa Ndeo, 'the Proud Isle', is part of the Lamu Archipelago of Kenya. The island initially prospered on the slave trade and became a local power after defeating Pate Island in the 19 th century, but declined after the British forced the closure of the slave markets in 1873. In 1890 the island became part of Zanzibar and remained obscure until Kenya was granted independence from Great Britain in 1963. Tourism centred on the 18 th -century Swahili architecture and traditional culture developed during the 1970s. At present Lamu mostly attracts visitors who want to relax after a safari on the mainland. The main industry of Lamu consists of beach resorts located in the small towns and villages along the coasts.
Although Kenya is one of the wealthiest countries in East and Central Africa it is still considered a developing nation. There are various causes for the persistence of poverty in Kenya, one of which is the high population growth rate. Over the past 30 years, the population has more than tripled, giving it the highest growth rate worldwide and greatly increasing pressure on the country's resources. Together with an increasing social inequality, this has eroded the effects of improvements in education, health, food security, employment and income. Population growth also brings opportunities: roughly 43% of the population of Kenya is under 14 years of age, which will pave the way for further growth because of an increasing workforce in the coming years. However, in 2008 the unemployment rates reached 40%. 1 Kenya's factor-driven economy relies heavily on natural resources as export products, the prices of which have remained low (Kimuyu 2007). Inequality in terms of the distribution of income and wealth has increased since independence and forms a pervasive problem in Kenyan society (Holtham & Hazelwood 2010). Partially, this inequality can be attributed to foreign aid as a disruptor of traditional ways of economic development. But for the main part corruption and institutional failures are to blame (Ake 1996).
The island of Lamu differs from the mainland because its economy relies mainly on tourism and not on agriculture. Respondents in our study estimated that 70% to 90% of the local populace relies on the flow of tourists to the island. The island is filled with beach houses, resorts and hotels and because of this Lamu enjoys a relative prosperity. However, this dependence on tourism also has its drawbacks: Lamu has been in economic decline since the political unrest in 2007 (Elhawary 2009). The unrest already drove away some tourists, but the most substantial blow to the local tourism industry was the kidnapping of a French woman in October 2011. 2 After the unrest the locals experienced a 20% slump in earnings, however after the kidnap, hotels had up to 90% of their bookings cancelled. Recently Kenya has yet again suffered a blow with the Westgate shopping mall attack. 3 Fortunately, the economic effects of this for Lamu have not been as severe as those of the kidnapping incident.
In response to the 2007 unrest, in 2010 Kenya adopted a new constitution that ensures the separation of powers between the executive, legislature and judiciary arms of government (Kramon & Posner 2011). Furthermore, Kenya is divided into separate counties with newly-formed county governments. Central power is devolved and more power is granted to the counties. The Kenyan devolution system still gives the president the power to suspend a county government under certain conditions (Kramon & Posner 2011). The devolution of power ensures the office of the president is not the only sought-after station during elections and offers more opportunities for a diversified representation of the population. Partially thanks to these constitutional reforms, Kenya now enjoys relative political stability, compared to its neighbouring countries. 4 Although the counties now have governments, governors and elections, the executive power to these governments has, as of yet, not been legislated. This transitional phase gives rise to potential abuses of power. The newly formed county of Lamu held its governor elections in March 2013. However, in the September the election was declared null and void after the constitution and electoral laws were found to be breached. Issa Timamy, the governor in question, was reinstated after the court overturned the nullification a month later. The elections on Lamu were described by our respondents as being stained by rigging and corruption, and have weakened the trust of the people of Lamu in the local governments. Basic requirements such as infrastructure, health and education are not provided by the local government but left to the local business managers and the community.
The cultural environment of Kenya and Lamu is very diverse and the various tribes and ethnic groups contribute to a variety of norms, values and practices. As is the case with many African countries, the borders of Kenya were drawn with no regard to customs, beliefs and ethnicities. Like many other countries on the continent Kenya has a diverse multicultural environment constituting not only different national identities but different tribes and sub-tribes. These tribes have different deep-seated views of societal norms and customs. However, tribalism is a relatively new phenomenon. A century ago the different tribes, especially those from the East and from the West of Kenya, had little contact with one another. Tribalism is often seen as a product of modern times arising from colonialism, urbanisation and the political culture that sprung up in independent Kenya (Ekeh 1975). Another reason tribalism is an increasingly important aspect is the politicisation of ethnicity. Voting is becoming more about which tribe the candidate is from than about issues he or she addresses (Kramon & Posner 2011).
At Lamu the dominant group is the Swahili, the majority of whom are Muslims who have a long history of interaction with the Arab world and other Indian Ocean societies. The Swahili culture is a mix of various traditions and local beliefs. Family is very important: families are large and centred on the male head of the family. Marriages outside the Swahili tribe are not common. Marriage serves the primary purpose of joining families, thus ensuring there will always be a group to turn to in times of need. There is a strong sense of preserving the Swahili way of life and the locals are cautionary to keep influences from outside as far away as possible. The Swahili see themselves as traders and consider employment disagreeable (Mazrui & Shariff 1994).
Tribalism influences corruption. Most types of corruption presented by Swahili respondents in our study are socially embedded in 'logics of negotiation, gift-giving and solidarity, making it hard for outsiders and themselves to see these acts as corruption' (Olivier de Sardan 1999). Tribalism is seen as being responsible for many general ills such as underdevelopment, rigging of elections and civil violence, as well as corruption. 5 Tribalism does not per se favour the Swahili in Lamu: when Kenyatta, a Kikuyu, became president his government started favouring fellow Kikujo tribesmen. This tribalism manifested itself in larger government funding for social infrastructure in Kikuyu areas, which Lamu is not a part of, and privileged right of entry to government jobs. 6

Research Methods
Our study focused on the firms in the tourism industry at Lamu Island where the first author stayed for three months, working for one of the companies while simultaneously collecting data. This prolonged stay allowed him to get acquainted with managers from the 22 member firms of the Lamu Tourism Association. Moreover, the local business manager for whom the first author worked introduced him to the owners/managers of the other firms, allowing him to establish a basis of trust. The aim was to study the entire population of firms associated with the Lamu Tourist Association, however four firms refused to cooperate. The sample for our study ultimately consisted of 21 owners/managers from 18 tourism companies at Lamu. 7 Data were obtained through semi-structured interviews and observations. Some key respondents were interviewed multiple times, and as our understanding of the phenomenon studied evolved we could adapt and refine the questions over the course of the study. This research approach was deemed important, as corruption is not a subject easily discussed with strangers. In surveys, managers are likely to understate their direct experience with corruption (Aidt 2009), and this may also be the case in interviews. By conducting the research in a small close community and by actively participating in said community, we aimed to avoid this danger as much as possible. The owners/managers interviewed were from nine different nationalities: nine were Kenyan, three were from the United Kingdom, two from Germany and the Netherlands, and one each from Belgium, India, Italy, Switzerland and the United States.

Data Analysis
Interviews were audio-recorded whenever the interviewee allowed this, and transcribed afterwards. When audio-recording was not permitted notes were made during the interview and written up immediately afterwards. Using the transcripts and notes the data were initially coded in first-order terms, according to the Gioia methodology (Gioia, Corley & Hamilton 2013). These first-order terms adhered faithfully to the words used by the respondents, or contained elementary, non-theoretical observations. Subsequently, the first-order concepts were grouped into second-order themes, more abstract dimensions corresponding to theoretical concepts developed in the literature discussed in the previous section. The next step was to distil 'aggregate dimensions' from the second-order themes. These aggregate dimensions reflect the main concepts in the study. The analysis was finished when theoretical saturation was considered to have been reached (Glaser & Strauss 1967). Figures 1 to 4 summarise our data structure. 8

Findings
In describing our findings we roughly follow the order employed in our discussion of the literature. We first look at what our respondents see as corruption, and the different forms  (1) with which they are confronted, afterwards we focus on the effects of corruption as described by the respondents. Subsequently, we look at the link between inequality and corruption at Lamu Island. We finish by discussing the different coping strategies described by the respondents.

Forms of Corruption
One of our respondents offered a definition of corruption that clearly reflected the 'rulebending' type: 'Corruption is: Trying to get a bribe so you can perform your duties which you are supposed to do freely, but instead want to be paid for so that you can serve the person' (Respondent E). Respondent K gave another example when she tried to get approval for a building plan which she was legally entitled to receive: 'To get the plan approved it can take you two years the official way. But if you know the right person, who knows the plan, you can give him some extra money to approve the plan immediately.' Another form of corruption within this type is the preferential interpretation of laws without actually breaking them. This was not a common form of corruption in our study, probably because it requires vague legal formulations to be effective, however we did come across a clear example on Lamu: the alcoholic drinks licence. The licence fee is formally dependent on the type of environment of a firm: urban or rural. However, firms on Lamu that have this licence claim they need to pay either the high or the low fee (30.000 KES or 15.000 KES 9 ), without there being any distinct difference in their locations. This does not constitute a breach of the law, since the island can be interpreted to be urban as well as rural, it is however a discriminating interpretation of the law.
The most common occurrence of 'rule-breaking' corruption on Lamu appears to be tax fraud and tax evasion, although rather than incriminating themselves our respondents typically pointed at others: Only two respondents admitted they broke the law and partially evaded their taxes: We did not pay our taxes, because they never check and they don't do anything with our money so why pay? … The tax man came around and said we did not pay taxes and said we had a big problem, they always say that when they want something. So we made a deal, I pay him half of taxes owed and the problem is no more. (Respondent F) The respondent of a second firm confirmed having evaded taxes after a review of their accounts, though they claimed during the interview they were not breaking the law. Harassment and intimidation from government officials is also mentioned as a common occurrence.
I am very much bullied by certain people. (Respondent A) They try to intimidate you! (Respondent G) They were threatening me, immediately, a few days later. (Respondent O) This form of harassment is used by public officials to scare people into paying a bribe or to create incentives to pay a bribe to make the problem of harassment go away. Officials will repeatedly ask managers for a bribe until they get it, or eventually back off.
Respondent O was threatened with nonexistent regulations in the hope that said business owner would pay 150.000 KES in order not to be fined.
There seems to be no evidence of 'state capture' in the tourism sector on Lamu, but we must take into account that state capture is the most difficult form of corruption to detect. Respondents replied negatively to questions regarding possible state capture, about state regulation change and its presence: We are too small to change it. (Respondent K) We don't have any say. (Respondent A) Our people will have less power to do all those things. (Respondent B) During the interviews examples of people changing laws or having the power to influence policy were not mentioned, however much government, both national and local, was distrusted. However, the ability not to follow certain laws, the room to interpret existing laws in a favourable way, and a general failure of institutions were pointed out.
These observations seem to suggest that state capture is not an option for the managers and the government officials on Lamu Island. First, their firms are too small to exert much pressure; second, they do not feel the need to work with the government or influence policy, as it seems to be easier to bend or break the existing laws than changing them; third, the fast-changing political environment in present-day Kenya might not present the right opportunities for the long lasting relationships necessary to be successful in state capture; and fourth, the pervasiveness of institutional failures mentioned by our respondents may imply that even if laws could be changed these might be implemented in an unpredictable way.
The different types of corruption discussed in this section share a few common causes. Rule-bending corruption relies heavily on the ability to resourcefully interpret laws and on failures in the bureaucratic system. Rule-breaking on the other hand relies on failures in the judiciary and executive systems and uncertainty in rules and regulations that can be exploited to threaten business managers with non-existing laws, or a very biased enforcement of rules and regulations. Ironically, existing failures in institutions help prevent state capture in this case, and these failures also provide some deterrence against corruption. The general distrust in institutions reduces the potential value of state capture.

Motivations to Engage in Corruption
According to our respondents positive effects of engaging in corruption include saving time, reducing costs, and avoiding emotional stress. Respondent U claimed it took a long time, almost a year, to get his licences, while Respondent B received his licences in one day: 'I know a lot of people who work for the government and it is always easy for me to get those things.' This speeding-up can save days, months and even years of time, but it comes at a price. A stark example concerned the zoning office where two respondents tried to get a plot and a building plan approved. The first one, Respondent H, did not bribe and explained he had to submit 23 different forms and applications, with the whole process taking over 15 years, starting in 1998 and only just being completed at the time of our study. The other, Respondent K, in contrast said: 'It didn't take long to get the approval because we know the guy and we know how it works so we go to his office and agree on a price.' In this case paying a bribe made the difference between waiting a few days versus more than a decade.
The second incentive, apart from time gains, is the financial benefits one might receive from corrupt transactions. Although this type was less often admitted by respondents, some examples do show that large financial gains can sometimes be made. A first example was the tax evasion scheme of Respondent F discussed above. When confronted by a tax inspector this respondent was able to pay him off with 50.000 KES, while the taxes due amounted to 100.000 KES. Respondent F claims to have benefitted from the deal since no further inspections have taken place and he could save half the money owed.
Another example was given by Respondent C, who told the story of multiple business owners in the neighbourhood who were taken to court for not following the 2011 Tourism Act, which states the rules and regulations firms in the hospitality industry have to follow and the consequences for not doing so. The managers in question escaped prosecution by bribing the judge: 'They paid quite a lot of money, 600 euro. It's a lot of money to bribe an officer not to go to court' (Respondent C). These high sums were paid because of the penalty stipulated in the Act (section 112): 'A person who … commits an offence and is liable, on conviction, to a fine not exceeding one hundred thousand shillings or to imprisonment for a term not exceeding eighteen months, or to both'. A payment of 65.000 KES (600 euro) might therefore be considered an attractive alternative.
The third motivation concerns a reduction in stress and inconvenience for managers. Refusing to pay a bribe may cause much inconvenience to managers compared to the amount they would have to pay to avoid it: 'I just pay a bribe and then I can go out, I don't have to deal with all that paperwork' (Respondent T); or 'you find yourself being interrogated for a long time' (Respondent E). Bribing is facilitated by a lack of shame or secrecy, it is considered a generally accepted way of conducting business in certain circles and can be used in everyday practice. 'They are not ashamed; they do not do it in private. They do it in the open' (Respondent O).
As for negative effects: corruption is based on trust, the confidence that both parties will commit to the illegal transaction. This trust is even more important in corrupt deals than in legal deals. Corruption can backfire if one of the conspirators does not honour the deal. A public official can take a bribe but might still choose not to help and simply ask for more money at a different time. As Respondent K mentioned: 'You have to pay and they will come back to get you for all the money you have.' Furthermore, there is the threat of detection. In the case of Lamu it seems the chance of detection is low: 'We did not pay our taxes, because they never check' (Respondent F). Even if it is detected the sanction will not always take place: They went to the houses which didn't follow the law. Those houses get arrested … but the people don't want to go to court. So somehow they bribe them. They did a bribe … and the people came back and started following the law. (Respondent C) It is therefore apparently likely that, when a manager on Lamu perceives any positive effects or benefits can be obtained through a bribe, he or she will in fact engage in corrupt behaviour. However, the payoff of engaging in corruption always remains uncertain. Respondent E mentioned: 'If they find a simple mistake they might make it a bigger problem to you.' Respondent D stated something similar: 'The tax man will always find something to charge you on, whether you are straight or not.'

Inequality and Corruption
At first glance the benefits to individual bribers seem to outweigh the potential threats. There is, however, a moderating factor that influences the effects of corruption: the social group one belongs to. At Lamu Island there are two distinct groups and people, those who belong to an inner circle and those who do not. The managers within the inner circle experience more benefits and at the same time suffer fewer threats. This inner group consists of managers who are predominantly local Swahili men who seem to conduct both more and more effective corrupt transactions. Whereas local Swahili men primarily form this inner circle, there are a few exceptions: people who are married to a local who is part of the family and therefore member of the inner circle. Through gifts and close social interaction one might also become part of the circle but this will take a long time and one must thoroughly immerge in the community and culture to be successful. Entrance into a corruption network is not easy: 'It's like the mafia. If you come from the outside you need to pay a price' (Respondent O). The answers from our respondents suggest the boundaries of the inner circle are defined through nepotism, favouritism, and tribalism. Respondents explain why members of this inner circle on Lamu benefit more from corruption than the outsiders: The most influential factor seems to be tribalism. Tribalism describes the phenomenon where members of a tribe are more loyal to their own tribe than to any other social group or category. 'Tribalism is an important reason; they value their own tribesmen higher than anything else' (Respondent S). Tribalism provides a powerful structure of social relations which enables trust and commitment between fellow tribesmen. This commitment in turn provides protection and enables one to reap better benefits and to be exposed to fewer threats from corruption.
At first sight the inner circle might be misunderstood as being based on Kenyan birth. Most respondents do not mention Swahili per se, the difference between being Kenyan and being Swahili is therefore hard to perceive. Unfortunately there was only one manager in the sample who was Kenyan but not Swahili, making the distinction hard to verify. The manager in question said, however: 'I'm considered an outsider in the community' (Respondent E). He is not part of the inner circle even though he is Kenyan and we suspect this is the case because he is not Swahili. Respondent K mentioned that all the owners who are Swahili are part of an inner circle on the island. Respondent O answered 'It's because they are Swahili' to the question why some managers get an unfair advantage. This does not provide hard evidence that it is just being Swahili and not Kenyan that counts, however our observations and the answers from respondents indicate a stronger sense of tribalism than nationalism.
Although tribalism is the main mechanism defining the inner circle, being Swahili is not the only way to enter, and conversely might not be sufficient to become part of the protection network. The second and easier method is by marrying into one of the big families of the group. For a manager this might be an interesting option, since it will allow you to get easier access to the benefits of corruption. Respondent M is married to a Swahili and he therefore gains the protection benefits of being part of that family: 'My family is Swahili and is quite an extended one so I get protection from that side … I usually don't have a problem … I just gain time and people will get informed that he better leave me alone or he will get a problem.' Respondent K claims the same protection: 'I'm not at a disadvantage because I'm married to a local person'. Respondent O was advised the following upon starting her business: 'You should marry a Swahili, that would make everything much easier.' The last method to enter the inner circle is to try to make friends and connections within the community. This method takes a long time and is considered the least effective since it does not provide the same protection, and at a larger cost: After many years in this country I can move amongst friends all the time … particularly now, after 15 years here, you're in the system. (Respondent H) Especially with the community and all the politics which go on in Lamu and Shela … you've got to listen to what people say and take note and make people feel important … Shela can be quite tough for people. You really got to play, play the game. (Respondent N) 10

Coping with Corruption
In this section we will look at how managers at Lamu Island cope with corruption. Three main options appear to be open to managers. One way is to choose the corrupt option, and by bribing and illicit negotiating try to obtain what one wants or to get rid of harassment by a public official. A second approach is to ignore corrupt officials and try to maintain business as normal, without either confronting corruption or indulging in it. Lastly, there is the option of actively pursuing justice and upholding moral values, trying to fight corruption.
Bribing a public official has the most direct effect and provides instant gratification at a price around 2.000 to 20.000 KES, depending on the issue: The amount they have to pay to bribe is only 10.000 shilling and for that they usually get away with anything. (Respondent F) The only way of succeeding here is bribing, if you don't bribe you cannot succeed. And you have to do the bribing, that is the usual thing around here. (Respondent B) As mentioned in the previous section only members of the inner circle can fully reap these beneficial effects. Someone who is outside of the group will encounter relatively little protection against corrupt officials who may either not back-up the deal or may simply come back for more after a few weeks or even days.
Just ignoring an official is also considered an option by several respondents, though the problem will not go away immediately and it can take quite a while before a corrupt official backs off. Respondent L: 'You have to play innocent … it can take months before you get anything done, but you will get it in the end, even though it is quite tiring.' This option is most commonly taken by those who do not want to risk getting caught or who do not want to be caught up in any corrupt scheme. The possibility of employing this coping strategy relies heavily on having everything in order. 'We are less vulnerable because we behave properly' (Respondent M). When a manager is clearly breaking the rules he or she cannot ignore the consequences, even if they can refuse to pay bribes. But managers who do follow the rules are still confronted with threats and manipulation, and in these instances character and being steadfast in ignoring is deemed to be important. Corrupt officials will eventually let go, but it takes an effort not to give in. This option seems to be the optimal coping strategy for those who are not considered part of the inner circle. Though one cannot reap the benefits of corruption, one can employ an agent to deal with public officials. This does not exempt the business manager from the consequences in terms of delays and financial penalties, but it tends to keep harassment to a minimum: 'You need an agent, you can't possibly do it yourself' (Respondent H); or 'You should not go yourself to the city council. I have someone for that, you will never see me there' (Respondent G).
Fighting, the third coping approach seems to have the same as or even less effect than ignoring, while at the same time causing more stress and emotional exhaustion: I have been battling them for seven years and you really get tired. (Respondent S) Of course, if you want to face corruption you pay more and will never win. Because everywhere you go there is one of them, corruption. They try to protect each other. (Respondent C) Hence this option is not preferred by most managers, for those in the inner circle it is better to give in and for those who are not it is easier to ignore. Most managers think corruption is undesirable, but they are not motivated to actively try to solve the problem as long as their competitors continue to reap the benefits. Corruption is part of a larger system and to fight corruption means to fight the system. The central government does not seem to have any interest in fighting corruption, because they too stand to gain, according to Respondent C, or because they do not have effective means, according to Respondent J. It is therefore unlikely that corruption will be effectively combated on Lamu in the foreseeable future, and effective coping strategies remain essential for managers of SMEs.

Discussion and Conclusions
In this final section we discuss our findings, address the limitations of our study, and formulate promising directions for future research.
The typology of corruption into the three main forms of rule-bending, rule-breaking and state capture developed in the literature fits our empirical data reasonably well, even if state capture was not found at the level of the SMEs in our sample. However, more important than identifying these forms in our view was fleshing out the conditions and circumstances in which corruption comes about at the micro level. Based on our interviews and observations we see a condition of legal uncertainty and a lack of trust in institutions as the two main dimensions that lead to corrupt behaviour by SME managers. The legal uncertainty seems to be rooted in inadequacies of the system, exemplified by sudden changes of the law, legal regulations that are open to various interpretations, and uncertain levels of charges. This legal uncertainty may in itself already undermine trust in the functioning of institutions, but in the answers of our respondents this lack of confidence was mostly connected with the behaviours of those working in the administrative system: politicians surfacing through stacked elections, officials and even judges who take bribes, a government administration that fails to deliver, and arbitrariness of police officers (see Figures 1 & 2). Within the context set by this legal uncertainty and not-to-be-trusted institutions managers of SMEs see three types of incentives: saving time, saving money, and sparing themselves emotional stress (see Figure 3). Given these incentives it is no surprise that the dominant coping strategy we found at Lamu was to engage in corrupt behaviour and pay bribes when these help reach the desired benefits. However, we found that this coping strategy is not equally accessible to everyone. At Lamu there is an inner circle, mostly defined by ethnicity, within which engaging in corruption is an attractive and effective strategy (see Figure 4). However, for those outside of this inner circle the choice to be corrupt is very risky: it may open the door to more pressure and extortion. The dominant   Figure 4 Data structure (4) coping strategy for these 'outsiders' therefore seems to be to 'sit it out': stay within the law, refuse to pay bribes, and accept that things will probably be slowed down considerably. The third coping strategy, fighting corruption, was not a realistic option to the owners/ managers we interviewed.
Our findings to a considerable extent corroborate Choi's (2009) micro-level theoretical analysis. Although our interviewees occasionally lamented the ubiquity of corruption, their attitude predominantly reflected a calculus of costs and benefits. But Choi (2009) models an individual decision without taking group membership into account. As we have seen in the case of Lamu, group membership may influence the expected costs and benefits of corruption coping strategies in a decisive way. A conclusion could be that individuals deciding about corruption face different individual incentives and constraints, depending on their group membership. But this would neglect the fact that group membership, at least to a certain extent, is also a choice variable. Some business owners/managers at Lamu had deliberately chosen to become an insider. This choice itself also brings costs and benefits (apart from altering those associated with corruption), so these should also enter the individual calculus. In contrast, other owners/managers (foreigners) are categorically excluded from membership of the inner circle. Hence the calculation of costs and benefits may sometimes be much more complex than suggested by Choi (2009), and sometimes much more simple.
One could also pose the question, to what extent for members of the inner circle corruption is actually fully calculative behaviour. One might simply be expected to bribe/be bribed. If this is the case, Choi's fully calculative model would also fall short, but for a different reason. Our respondents belonging to the inner circle mostly talked about corruption in a calculative way, but it is not possible to say whether this is a reflection of their decision-making process or an ex-post rationalisation of an ingrained practice.
Our study is characterised by a number of limitations, and each of these points to fruitful future research. First of all, our study is based on a small sample within one particular context. Although we think it is highly likely that the forms of corruption, coping strategies, and inner-outer circle phenomena that we found at Lamu are also present in other social and geographical contexts, some of our findings may be location-specific. For instance, the Swahili culture attaches strong value to the extended family (Mazrui & Shariff 1994), this may work out differently in other cultural settings. For this reason it would be interesting to repeat the study in one or more radically different socio-cultural environments.
Second, we did not try to quantify corruption. Therefore, when we state that one form of corruption is 'more important' than another, this can only be taken to mean that it was more prevalent in our sample. It would be desirable in a future study to try to quantify different types of corruption. Maybe a single instance of rule-breaking is worth more in dollars than ten cases of rule-bending. Although some work has been done that quantifies the payment of bribes (Reinikka 2006), this typically does not make fine-grained distinctions between different types of corruption.
Third, our study was unavoidably biased in the sense that we only interviewed incumbents, i.e. those owners/managers who had selected to be active in the tourism sector on Lamu and had been able to survive (and in some cases prosper). This is a limitation because the inner-circle effect that we identified can be expected to deter investors from the outer circle (both foreigners and non-Swahili Kenyans) from entering this sector, or to withdraw once their competitive disadvantage becomes manifest. We can surmise, but not conclude, that conditions such as those found at Lamu discourage foreign investment, as is also found in country-level studies (for example, Habib & Zurawicki 2002). This is an important issue, as foreign direct investment may help developing economies not only to accelerate growth, but also to combat corruption (Robertson & Watson 2004). Hence, a better understanding of corruption and how it influences insiders and outsiders is of great practical importance.