The effect of banking services on the business performance of bank agents in Kenya

Abstract Agency banking is a relatively new banking concept introduced in Kenya in 2010 by the Central Bank of Kenya (CBK), the main objective being to increase financial services outreach and promote financial inclusion of the unbanked and under-banked population. The current study, based on agency theory, sought to assess the effect of banking services on the business performance of bank agents in Kenya from an entrepreneurial perspective. A quantitative research survey design was adopted for the study, and questionnaires were used for data collection. A sample of 384 respondents from three constituencies in Nairobi (Roysambu, Kasarani and Kamukunji) participated in the study. Both descriptive and inferential statistics were used to analyse data. Findings revealed that while there was growth in business turnover per month for agents after signing up for agency banking, this was only true for those with a turnover of above Kshs. 150,000. The study recommends that other entrepreneurs should consider being banks agents, while the existing ones should increase their capital to reap the most from agency banking.


Introduction
Agency banking means providing limited scale banking and financial services to the underserved population through engaged agents under a valid agency agreement, rather than a bank teller or cashier (Kumar, Nair, Parsons, & Urdapilleta, 2006;Rahman, 2016). In agency banking, it is the owner of an outlet who conducts banking transactions on behalf of a particular bank, hence mimics real banking (Hawkins, 2012). Globally, retailers are increasingly utilized as important distribution channels for financial inclusion where shop retailers and post offices are used as agents (Bangladesh Bank Order, 2009;Mlachila et al., 2013). The agents type range from post offices in the outback of Australia, where clients from all banks can conduct their transactions, to rural France where the Bank Credit Agricole uses corner stores to provide financial services, to small lottery outlets in Brazil where clients can receive their social payments and access their bank accounts (Katy, 2005).
In Brazil, private and state-owned banks deliver financial services through retail agents, including small supermarkets and pharmacies, post offices, and lottery kiosks (Kumar et al., 2006). These agents are called banking correspondents (BCs). In January 2006, India's central bank issued a circular permitting banks to use post offices and specialized microfinance institutions (MFI), including non-profit organizations (NGO), cooperatives, and for profit companies as retail agents. The circular calls these agents "Business Correspondents" (Central Bank of India, 2016). Models in these markets, contrast that of South Africa, where although branchless banking is permitted, the wide discretion and strict regulatory requirements for agents and banks to comply have resulted in the exclusion of smaller establishments that are more likely to be located in low-income areas. Bank agents are thus larger retail network/supermarkets with the resources to satisfy the regulator and bank's internal control and/or audit standards (CGAP, 2010).
Agency banking is not new in the world and it is growing fast. In Brazil, Agency banking was rolled out in year 2000 and within 5 years, it had over 90,000 new correspondent banking networks. These networks were from typically retail commercial outlets ranging from lottery kiosks, pharmacies, post offices, and construction goods stores. The agency network facilitated 12.4 million new bank accounts in the same period (Kumar et al., 2006). Peruvian banks started in December 2005 are the fourth in a number of agents worldwide after Brazil, the Philippines, and South Africa (Mas, 2008). Other widely known agency banking pioneers are in Kenya, Mongolia, South Africa, and the Philippines (Kumar et al., 2006).

Agency banking in Kenya
In Kenya, there are 43 licensed commercial banks, 12 Micro Finance banks and 1 mortgage finance company (Central Bank of Kenya [CBK], 2015). As of 30 June 2015, Central Bank of Kenya had authorized 17 commercial banks to offer banking services through third parties (agents). Since 2010 to June 2015, the 17 banks had contracted a total of 38,297 agents. These agents had undertaken over 175.4 million transactions valued at Kshs. 930.1 billion (CBK, 2015). An amendment to the Kenya Banking Act 2009 allowed banks to start using agents to deliver financial services (EFInA, 2013). Before the agency banking model was introduced in Kenya, there was a heavy cost of including low-value bank accounts and providing the (physical) "brick and mortar" banking infrastructure to unbanked regions.
Under the CBK regulations, agents can offer a number of banking services, including cash deposits and withdrawals, fund transfers, bill payments, loan payments, payment of benefits and salaries, and collection of account and loan applications. However, agents are limited to cashonly transactions and cannot access other banking application systems. Agents earn a commission from banks of Ksh. Twenty per deposit, and Ksh. Seventy for a withdrawal, split evenly between the bank and the agent (Kenya Commercial Bank [KCB], 2015). The CBK regulations require that agents have secure operating systems capable of carrying out real-time transactions, generating an audit trail, and protecting data confidentiality and integrity, which is all driven by technology. Transactions can be made via mobile phone, a Point of Sale (POS) system, or Internet banking, and must be reflected immediately on the bank's side in their core banking system (CBK, 2010).
The growth of agency banking in Kenya suggests a healthy uptake of the model with 38,297 agents and 17 authorized commercial banks by June 2015 (CBK, 2015). Widespread agency network of the model has helped to reduce transaction cost and extend services to far and remote rural areas (The Association of Community Development Finance Institutions, 2012). With financial inclusion of Kenya being at 70% (having been driven mainly by mobile banking such as Mpesa which has taken the largest share) there is still an opportunity for further growth of agency banking to improve the rating of financial inclusion (Villasenor, West, & Lewis, 2015).
Some studies have been undertaken on agency banking in Kenya, as scholars interrogate this relatively new model in the Kenyan market. Some of these studies include: one on impact of agency banking on customer satisfaction (Mbobua, Juma, & Musiega, 2013); a study on technology adoption and the banking agency adoption in rural Kenya (Nganga & Mwachofi, 2013;Saropa, 2013); and, some on analysis of the utilization of agency banking on performance of selected banks (Aduda, Kiragu, & Ndwiga, 2013;Musau, 2010). Though these studies are of great benefit to the banks (sponsor) and the end consumer (customer), and could benefit the regulator, no research seems to have been carried out to assess the effect of the bank agency business opportunity on the agent (entrepreneur), especially in relation to their business performance.
The main objective of the current study was to assess the effect of agency banking services on the agents' business performance. The two specific objectives of the study were to: (i) Examine the factors that motivate entrepreneurs to register for agency banking, and (ii) Evaluate agents' business performance as a result of offering agent banking services.
These objectives translated to the following research questions: (i) What factors motivate entrepreneurs to register for agency banking? And (ii) How is agents' business performance as a result of offering agent banking services?
This study was limited to agents licensed by the Central Bank of Kenya (CBK) and operational at the time of the study. The geographical area of study was the three constituencies of Nairobi County that have the highest adult population, namely: Roysambu, Kasarani and Kamukunji with 202,284, 200,984, and 211,991 people, respectively (IEBC, 2012). The three constituencies also lie to the eastern side of Nairobi, and have similar demographic and socio-economic characteristics.
3. Agency theory and its application to agency banking Jensen and Meckling (1976, p. 308) defined agency relationship as "a contract under which one or more persons [the principal(s)] engage another person (the agent) to perform some service on their behalf. This principal-agent relationship also involves delegating some decision-making authority to the agent." According to Lan and Heracleous (2010), under law, an agent is a person who acts on behalf of another (known as a "principal"). This theory, therefore, focuses on the relationship between the principal and the agent, which in this study refers to the bank and their agents, respectively.
The principal-agent model framework occurs in any social, political, or legal situation where two parties align themselves to fit a situation where the principal with which authority to act originally "hires" or delegates some of this authority onto the agent (Ross, 1973). This relationship is extended to persons or entities that make use of agents to deliver their business objectives, in this case, banks (principal) and banking agents (agent). Such relationships take the form of a contract specifying the terms and agreed consideration, as well as the costs arising and conflicts resolution mechanisms between the principal and the agent. Through regulation, the limits on the role agents can play in providing financial services are set. Some regulators consider different categories of agents based on the services offered with less stringent eligibility standards for those agents offering only basic services, such as cash-in and cash-out services (Tarazi & Breloff, 2011).
Agency theory can also be used to understand the first objective of this study, which is an understanding of the factors that motivate entrepreneurs to engage in agency banking. While intrinsic factors (such as profit-making) can be suggested as reason for taking up agency baking (Venkatesan, 2017), many "agent"-related factors, for example bank expansion, desire to reach the unbanked, regulations that allow the bank to engage the agent, among others, are key in the engagement of individuals for agency banking (CBK, 2010;Ndungu & Njeru, 2014). Agency theory is thus instrumental in examining the factors that motivate entrepreneurs to engage in agency banking.

Factors that motivate entrepreneurs to register for agency banking
In Africa and Asia, expansion of agency banking has been driven by the growth of mobile financial services, for which a dense network of cash-in and cash-out (CICO) points is needed to facilitate transactions (Aker & Mbiti, 2010;Ivatury & Mas, 2008;Mbiti & Weil, 2015). In Brazil, according to Sanford and Cojocaru (2013), regulatory and economic incentives spur banks (including public sector heavyweights) to form partnerships with existing retail outlets, and to develop their own networks of Banking Correspondents (BCs). In the span of a few years, a relatively small network of financial access points expanded to an impressive coverage of all but a few municipalities in the largest country in South America. Some of the factors that have contributed to bank agents are discussed next.

Regulatory factors
Regulatory factors are pivotal to the growth of bank agents. In 1999, the Central Bank of Brazil enacted a resolution that allowed banks to establish agreements with non-banking entities to provide bill payment services and distribute social transfer payments. In 2003, resolutions 3.110/ 03 and 3.156/03 modified the regulations for BCs to be less restrictive, and allowed for access to many similar services to those offered at bank branches, with few restrictions (Sanford & Cojocaru, 2013). Currently, nearly any retailer can become a BC, and authorization for each BC relationship by the Central Bank is no longer required. Although there is no regulation demanding exclusivity with correspondents, Caixa, Banco do Brasil, and Bradesco have all established exclusive contracts with their BCs, thus creating effective exclusivity in the market (CGAP, 2011). In both Kenya and Uganda, when mobile money was launched, it was common to find mobile money providers placing exclusivity clauses in their contracts with agents. For banking agents in Kenya, however, exclusivity clauses were prohibited when the CBK launched the banking agency model (CGAP, 2016).

Economic growth
Mexico's regulations to allow more types of financial institutions to operate through bank agents and allow the opening of savings accounts had a significant impact on financial inclusion and places Mexico among the leaders in agency banking in Latin America (Lee & Mexico, 2012). In Peru, agents carry out approximately 3 to 8 million transactions per month. There have been increased financial transactions in Peru, Colombia and India as well. For example, in 2010 in Peru, less than 50% of the total financial system transactions were conducted through traditional bank branches; ATMS and POS terminals accounted for 36% of total transactions (CGAP, 2010). Increased numbers of transactions reflect the increased economic activity of any economy, which suggests the need for more openings through bank agents. Following a boom of economic growth in the late 1990s and early 2000s, emerging sections of the Brazilian population had newly available disposable income, which created the need for banking services (Sanford & Cojocaru, 2013). The financial sector and retail players duly adapted to serve these new market segments. Maeri (2012) and Omumi (2010) study of Kenya Commercial Bank (KCB) and post-bank agents respectively, found out that agents benefit from the agency banking through the brand reputation association of their sponsors. Most of the commercial banks that have rolled out agency services are the larger financial institutions that command the greater banking sector, and have a significant influence in the region and in the global market. The agents also have free association brand advertisement through the financial institution brand marketing campaigns and reports. Mas and Radcliffe (2010) argued that brand strength affects mobile phone agents, and this is also likely to be the case with bank agents.

Contribution of banking services to agents' business performance
A business performance measurement refers to the use of a multi-dimensional set of performance measures for the planning and management of a business (Bourne, Neely, Mills, & Platts, 2003). In the current study, bank agents' business performance definition took considerations of the entrepreneurial motivation to grow their business. The study adopted measurements proposed by Parada (2013) on measuring performance in branchless banking. The following measures were therefore used: turnover per month, number of customers per day, number of new products, and number of transactions per day. These determinants were used to measure agents' business performance prior to, and the effect after taking up agency operations. The next discussion offers some of the specific ways agency banking can affect business performance.

Business turnover
Technology, and in particular the spread of real-time communications networks, permit banks to delegate, "the last mile" cash management and customer servicing functions to third-party retail outlets (agents), which transforms the agents' business turnovers through the low-balance volume transactions they undertake (Alexandre, Mas, & Radcliffe, 2010). In Colombia, for example, the agent banks performed more than 29 million transactions with a value of approximately US$27.0 million in 2009. In the same year, agents in Peru performed 67 million transactions, which documents double turnovers of the agents' business and the sponsor bank (The Association of Community Development Finance Institutions, 2012). The implication is that the agent banks have the pull of a large clientele to their stores. Consequently, banks are in a position to serve a broader number of customers if they have agency banking services in their stores, unlike when they do not have them.

Number of customers and new other products
The availability of banking services at agents' outlets other than the mainstream bank branches bring the new banking services customers to the stores who buy various goods and services (Ivatury & Mas, 2008). The new products demanded by the customers provide another business line for the growth of agency business. Generally, poor people find agency banking more convenient and efficient (Lyman, Timothy, Ivatury, & Stefan, 2006) and hence likely to take up more services from agents. In Brazil, customer traffic increased by around 35% to the various bank agents' outlets, affirming the observation of increased customers to these businesses (Sanford & Cojocaru, 2013). A case study for a computer firm revealed increased business volumes in its other lines of business, when the agency model was adopted. The study observed that more customers visit the store to seek various services they previously did not provide such as printing their bills from the Internet and other sources, and settling their bills right there (CGAP, 2010).

Number of transactions
In November 2009, the agent Nestor acting for Bradesco Bank in Brazil, carried out 1,397 transactions, constituting of largely cash withdrawals, bill payments, utility payments, accounts opening, and non-financial transactions, such as balance inquiries, etc. Most of the agents largely carry out all the transactions themselves hence minimizing the operation costs associated with labour (CGAP, 2011). In Kenya, The Equity Bank in Kenya acknowledges the strong indication of the potential increase in revenue streams to Kenyan banking agents whom they contract to offer their services. According to the Equity Bank, the first 9 months of 2011 banks pre-tax profit rose by 39% largely attributed to its push to deepen its access through branchless or agency banking, and which lead to growth in deposits and loans applications as well as disbursements. By 2015, Equity Bank posted 24,000 agents with 51 million transactions, creating a whole new business for the bank and commission income for agents who provide the service (CGAP, 2011;Equity Bank, 2015).
The relationships among the variables in the study have been summarised in the conceptual framework in Figure 1.

Method
A quantitative survey was adopted for this study as it sought to analyse the performance of agency banking in terms of the entrepreneur's business performance. Surveys have been found instrumental when collecting quantitative data (Wimmer & Dominick, 2013), which was the case in this study. Proportional quota sampling, a non-probability type of sampling, was used (Kothari, 2004). According to Sedgwick (2012), proportional quota sampling ensures that the resulting sample is a representation of the population. In this study, therefore, a sample size of 384 was determined using the formula adapted from Saunders, Lewis, and Thornhill (2012): Where: n is the minimum sample size required p% is the proportion belonging to the specified category q% is the proportion not belonging to the specified category z is the z value corresponding to the level of confidence required e% is the margin of error required With the proportion belonging to the specified category being 50% and the proportion not belonging to the specified category being 50%, using a confidence level of 95% and a margin of error of 5%, the sample size is 384. The minimum sample size calculation is thus given:  Based on population size, proportionate sampling was done with the following representations: Roysambu (33%), Kasarani (33%) and Kamukunji (34%). Specific information on the number of agents per constituency or country was, however, lacking as there was no published data on a number of agents in the geographical area of study by either the regulator or sponsor banks. Similarly, banks have only published the number of agents and the value of transactions on a national scale.

Data collection and analysis
The respondents were business owners or outlet attendants of the agencies where the attendant is the outlet business manager. Structured questionnaires were researcher-administered to the business owner or shop manager. Data collected were coded and entered into the Statistical Package for Social Sciences according to each variable of the study for analysis. This study made use of both descriptive and inferential statistics.
Pearson correlation analysis was used to test the association between two categorical variables. In addition, Wilcoxon signed-rank test was used to determine if there were differences on a dichotomous dependent variable between two related groups (Adedokun & Burgess, 2012). This test is considered to be similar to the paired sample t-test, but for a dichotomous rather than a continuous dependent variable − in this case the study dichotomous variables being before and after signing up for banking agency.

Validity and reliability
To ensure that the quality of the research was enhanced (Saunders et al., 2012), the questionnaire was designed according to the research objectives. A Pilot test was undertaken among 30 respondents from constituencies not included in the study, hence ensuring that both content and construct validities were taken care of. Content validity has been defined as the extent to which a measuring instrument provides adequate coverage of the topic under study, while construct validity refers to how well an instrument measures what it purports to measure (Kothari, 2004, p. 74;Salkind, 2010).
Reliability refers to a measure of the degree to which research instruments yield consistent results (Mugenda & Mugenda, 2008). In order to ensure reliability, consistency of a measurement technique in the questionnaires was upheld. Since the questionnaire employed Likert questions, the Cronbach's alpha was used to measure the reliability of the research tool. A Cronbach's alpha of about 0.7 is considered an acceptable measure of a reliable tool (Saunders et al., 2012).
A pilot study (with the sample respondents) was carried out to determine the reliability of the questionnaires. Reliability analysis was subsequently done using Cronbach's Alpha that measured the internal consistency by establishing if a certain item within a scale measures the same construct (Table 1). Gliem and Gliem (2003) established the Alpha value threshold at 0.7, thus forming the study's benchmark. Cronbach's Alpha was established and shows that motivation had a reliability of (α = 0. 812), which exceeded the prescribed threshold of 0.7. The tool was thus reliable.

Ethical consideration
In the study, participants were assured of the confidentiality of any personal material that they shared. They were also assured that identifying information would not be made available to anyone who is not directly involved in the study, and, that the study was for academic purposes only. The researcher also sought informed consent from the respondents after sufficient information had been presented (in an understandable language) so that the respondents made an informed judgment about participation. Finally, the study was submitted to the Strathmore University Institutional Review Board and approval granted ref. SU-IRB0092/17.

Findings
Out of the 384 questionnaires given out, 361 questionnaires were returned representing a 94% response rate. The majority of the respondents (61%) were female while a minority (39%) were male. Four major banks, according to study findings (Equity Bank, Kenya Commercial Bank (KCB) Co-operative Bank (CO-OP) and Family Bank) had market presence in the geographical area of study. However, Equity and KCB combined had a dominance of the market. Equity Bank had the highest respondents (83%) while the Family Bank had the lowest respondents (4%). Some agents were however found to operate multiple agency arrangements.

Factors that motivate entrepreneurs to register for agency banking
The respondents were asked to indicate their level of agreement on statements about the factors that motivated them to register for agency banking. A scale of 1 to 5 was used, and assigned as: S.D = Strongly Disagree; D = Disagree; N = Neutral, A = Agree and, S.A = Strongly Agree. The responses were analysed with Std. = Standard deviation. The study findings are as shown in Table 2.
Majority of the respondents (97%) agreed or strongly agreed that the CBK regulations are favourable to the agency banking business model. This had a mean of 4.22 and a standard deviation of 0.49. Majority of the respondents (81%) agreed or strongly agreed that there were strong indications of potential growth in revenue streams for their business if they signed up as bank agents. This had a mean of 4.07 and a standard deviation of 0.80. Majority of the respondents (83%) agreed or strongly agreed that the availability of banking services will bring customers to the store to buy various goods and services. This had a mean of 4.01 and a standard deviation of 0.74.
Another large number of respondents (66%) agreed or strongly agreed that their business association with the sponsor bank benefited the image and reputation of their company/shop. This had a mean of 3.92 and a standard deviation of 0.86. A further analysis, however, found a good number of respondents (66%) who disagreed or strongly disagreed that fees paid for services by the sponsor bank are satisfactory for their business. This had a mean of 2.26 and a standard deviation of 0.83.

Changes in agents' business performance
12.1. Business turnover per month before and after signing for agency banking The respondents were asked to indicate their business turnover per month before and after signing for agency banking. The study findings are presented in Figure 2.
Thirty-seven per cent of the respondents had the highest turnover of between Kshs.100,000 and Kshs.150,000 before agency banking. However, those that were found to have the highest turnover after signing up for agency banking (50%) had a turnover of between Kshs.150,000 and Kshs.200,000. Interestingly, and perhaps contrary to research expectation, those who had a turnover of below Kshs.100,000 had a decline in their turnover after signing up for agency banking.

Comparison between gender and business turnover
Female-operated agents with a turnover of up to Kshs.150,000 were more (65%) before signing up as an agent and equally more (30%) after signing up as agents. This is in comparison to those operated by male (53%) before and (21%) after signing up for agent banking. Although male agents dominated the category of turnover above Kshs.150,000 in both cases of before (48%) and after (79%) signing up as agents, female agents in same turnover category significantly increased from 35% to 70% (as indicated in Figure 3).

Number of customers before and after signing for the agency banking
The respondents were asked to indicate the number of customers they served per day before and after signing up for agency banking. The study findings are presented in Figure 4, which shows that the highest respondents (98%) had up to 50 customers per day. Few respondents (0.8%) had above 100 customers per day before signing up for agency banking. Respondents who served 51 to 100 customers per day had positive change (5% increase) after signing up for the agency services.  The findings suggest that about 5% of the respondents in the 1-50 category had moved to the 51-100 category since the percentage difference is the same in the two categories.

Relationship between numbers of products offered before and after signing for agency banking
A Pearson product-moment correlation was run to determine the direction of the relationship between the number of products offered before and after the entrepreneur signed for agency banking as per the findings on Table 3. There was a strong, positive correlation between the two, which was statistically significant (r = .360, p = .001). This means that there was an increase in the products offered after signing up for agency banking.
A further analysis presented in Table 4 shows the output of a paired t-test (Wilcoxon signed test) carried out to determine if there was a significant increase in the number of products offered before and after the various entrepreneurs signed up for agency banking. A Wilcoxon signed-rank test showed that signing up for agency banking-elicited a statistically significant change in the number of products offered or in the demand (Z = −14.401, p = 0.001). Indeed, the number of products increased from 8 to 13, an improvement of 5 products.

Discussion
The aim of the present study was to assess the effect of agency banking services on the agents' business performance. As seen in the conceptual framework in Figure 1, the factors that motivate individuals to consider agency banking reflected those in literature (regulatory factors, economic growth and brand reputation) (CGAP, 2010;Omumi, 2010;Sanford & Cojocaru, 2013) and these were antecedent variables, or came before individuals took up the agency banking. Consequently, being a bank agent had an effect on the agent's business performance (dependent variable) as it led to increased turnover, increased number of products and customers, as well as increased transactions (CGAP, 2010(CGAP, , 2011Ivatury & Mas, 2008).  One aspect of the study that stood out and hence warranted a discussion was gender and business performance. Notably, both genders operated agency services according to the study findings. Female agents were however found to operate at a lower value turnover of up to Kshs.150,000 while male agents were more dominant in turnover of above Kshs.150,000. As most financial engagements in African countries are patriarchal (Mwobobia, 2012), this finding is not surprising. This suggests the need to have more female entrepreneurs encouraged and supported to increase their turnover for greater business growth in the future.
The study also identified turnovers of above Kshs.150, 000 as the most growing after signing up to agency banking services, probably owing to increased transactions (CGAP, 2011). It was interesting that those who had a turnover of below Kshs.100,000 had a decline in their turnover after signing up for agency banking. Most probably, clients would be discouraged whenever they would show up for transactions to find inadequate float (Dupas, Green, Keats, & Robinson, 2012). This discouragement is bound to have had a ripple effect as such clients not only failed to come back, but they also probably discouraged other clients from visiting the affected agents.
With regard to the sponsors, only four commercial bank sponsors were found to have active agents, namely, Equity Bank, KCB, Cooperative Bank and Family Bank. There are more than 44 financial institutions in Kenya, and by 2010, 17 commercial banks had reported to have started recruiting agents (CBK, 2010). According to CBK (2013), however, only 11 banks had agents and from the findings there is, therefore, need to encourage more commercial banks to venture into agency banking.

Factors motivating agents to offer banking services
A substantial number of agents had multiple sponsor agreements, which confirmed the CGAP (2016) findings that the CBK prohibits exclusivity of banking agents. The research also established that other favourable regulatory and economic factors motivate entrepreneurs to sign up for agency banking, in line with the arguments by Sanford and Cojocaru (2013). Majority of respondents (96%), however, agreed or strongly agreed that they experienced delays in the process of becoming bank agents because they were agents of a telephone company, confirming CGAP (2011) observations that such restrictions demotivate partner agency relationships.
Other factors that motivate agents include the opportunity for brand association with a sponsor to build their company/shop image and reputation (Maeri, 2012;. The finding that only four sponsor banks have rolled out agency banking in areas of this study contradicts Sanford and Cojocaru (2013) which documented a very fast growth of agents by banks across South America. This contradiction with the Kenyan reality could be due to the restrictions of MPesa shops that entered the market first, or the lack of motivation by banks who could have considered agency banking as a third party (CBK) initiative.

Contribution of banking services to agents' business performance
The study found out that more male than female respondents manage transactions above Kshs.150,000 per month. These higher turnover agents were also found to have grown the number of transactions after taking up agency banking services (CGAP, 2011;Ivatury & Mas, 2008). The Pearson product correlation analysis further demonstrated a positive business performance of five new products sold per day by agents who signed up for agency banking services. In addition, 98% of the respondents, and who serve 51 to 100 customers a day, had an increased number of customers per day.
The study used business performance determinants as defined by Parada (2013). The positive effect illustrated by the major determinants of turnover, customer number growth and new products also supports the study's conclusion that agency banking improves the business performance of entrepreneurs who take up the business (Ivatury & Mas, 2008).

Conclusion
This study, which sought to assess the effect of banking services on the business performance of bank agents in Kenya, concluded that several aspects of agency service motivate entrepreneurs to be bank agents. Performance of agents' business was found to improve (after adoption of agency banking) in several determinant aspects such as turnover per month, and number of customers attended to per day.
A key recommendation from this study is to encourage agents to invest in larger amounts (at least Ksh. 150,000) if they are to benefit from agency banking. This can be encouraged by banks when agents sign up for businesses with them, as well as for the agents in the field who can be encouraged to increase their float. Similarly, the delays experienced in signing up as a bank agent should be re-looked into so as to encourage more agents to sign up for agency banking.

Limitations of the study
While this study focussed on the largest constituencies in Nairobi County, Kenya, the 384 samples may not fully give the story of agency banking in the County or even the nation at large. Future studies could include more constituencies, or even more counties within the country.