The perceived impact of agency control on the performance of Zimbabwean small-to-medium enterprises

% of them fail to survive beyond the first 3 years (Mudavanhu et al. 2011). Business performance, especially amongst SMEs, is often attributed to factors such as government support and management competencies, ignoring other factors such as agency relationships prevailing in the businesses; yet, they create a unique business environment and a complexity (Kallmuenzer 2015) that has the potential to compromise business performance. Agency relationships are ubiquitous in business operations because they are a fundamental concept upon which business relationships are built (Eisenhardt 1989). Agency relationships were conceptualised in the seminal work by Jensen and Meckling (1976) as a contract whereby one or more individuals (the principal) employ another individual (the agent) to execute some task on their behalf, involving the delegation of decision-making to the agent. Agency relationships can also be understood as legal relationships in which the agent is authorised by the principal to perform on their behalf and is empowered to do what the principal could lawfully perform in person (Baze 2009). Background: Agency control has been studied in the context of large public listed businesses and is linked to business performance. However, very limited studies have validated such a role in small-to-medium enterprises (SMEs), especially in a developing economy. Aim: The current study aimed at ezstablishing the perceived impact of agency control on performance, as measured by return on investment (ROI) and innovation in SMEs in Harare Province, Zimbabwe. Setting: Primary research was conducted amongst owners or managers of SMEs in Harare Province, Zimbabwe. Methods: The final sample consisted of 106 owners or managers of SMEs. The study was purely quantitative and adopted the survey design. Structured questionnaires were used for data collection. Results: The study established that the most used control mechanisms were frequent monitoring and face-to-face deliberations with subordinates. Weak and statistically insignificant relationships were found to exist between agency control and business performance measured by both ROI and innovation. Conclusion: It is best to solicit expert advice to coach owners or managers on how best agency control mechanisms could be formally instituted and managed to leverage on them.


Introduction
Small-to-medium enterprises (SMEs) dominate the international business platform numerically (Ayyagari, Demirgüç-Kunt & Maksimovic 2011;Ramukumba 2014). They are the major types of businesses that contribute about 60% of the gross domestic product (GDP) of the countries, providing employment to about 70% of the population, and create value beneficiation in the range of 50% -60% (Organisation for Economic Cooperation and Development 2017). Japan has the biggest share of SMEs amongst the developed countries, making above 99% of total businesses (Economist Intelligence Unit 2010), and in India, SMEs account for approximately 45% of output in the manufacturing sector and 40% of the total exports (Ghatak 2010). In Zimbabwe, SMEs comprise 70% of all businesses; employ 60% of the country's labour force; and contribute above 50% to Zimbabwe's GDP (Dhliwayo 2019). However, about 60% of SMEs in Zimbabwe perform poorly and often fail during their first year, whilst 25% of them fail to survive beyond the first 3 years (Mudavanhu et al. 2011).
Business performance, especially amongst SMEs, is often attributed to factors such as government support and management competencies, ignoring other factors such as agency relationships prevailing in the businesses; yet, they create a unique business environment and a complexity (Kallmuenzer 2015) that has the potential to compromise business performance. Agency relationships are ubiquitous in business operations because they are a fundamental concept upon which business relationships are built (Eisenhardt 1989). Agency relationships were conceptualised in the seminal work by Jensen and Meckling (1976) as a contract whereby one or more individuals (the principal) employ another individual (the agent) to execute some task on their behalf, involving the delegation of decision-making to the agent. Agency relationships can also be understood as legal relationships in which the agent is authorised by the principal to perform on their behalf and is empowered to do what the principal could lawfully perform in person (Baze 2009).
Although issues of corporate governance and agency control, in particular, have been studied in the context of large public listed businesses and linked to business performance (Bendickson et al. 2016;Schulze et al. 2001;Yahya, Ali & Ghazali 2016), few studies have confirmed their role in SMEs, especially in a developing economy (Yahya et al. 2016). Yet, sound corporate governance plays a key role in improving the financial performance of SMEs (Mazikana 2019). Proper implementation of agency control mechanisms has the potential to create a business framework of attitudes and processes that not only add value to the business but also assist in building its reputation, thereby ensuring continuity and success of the business in the long run (Radebe 2017). According to Magaisa, Duggal and Muhwandavaka (2013), a trend exists amongst many Zimbabwean SMEs where the owner or manager is in control of everything, making it difficult for the business to have a board of directors who run separately from the stakeholders and are capable to account for the running of the business. Chrisman, Chua and Litz (2004) and Yahya et al. (2016) also show that agency relationships and their associated information asymmetries may exist between principals and agents in small businesses, making it problematic for principals to regulate the agents' behaviour, subsequently affecting business performance. Furthermore, the seminal work by Chrisman et al. (2004) underscores that separation of ownership and management (as is likely to prevail in most non-family-owned SMEs) generates costs that may be nonexistent where ownership and management are combined, such as in family-owned SMEs. The separation of ownership and management creates a conflict of interest that earlier proponents like Ross (1973) formalised as a principal-agent problem, in addition to obvious agency costs. It is against this background and also compounded by the fact that limited comprehensive studies on corporate governance and agency control, in particular, amongst Zimbabwean SMEs have been conducted (Mazikana 2019), that the current study is premised.

Problem background
The fact that SMEs are the main forms of business in Zimbabwe poses interesting agency relationship issues worth investigating. The separation of ownership from control in businesses may lead to the loss of appropriate monitoring by the owners on the managers and other employees who may use business assets for their private purpose to maximise their welfare (Panda & Leepsa 2017), with obvious implications on business performance. Monitoring agent behaviour is a core issue in the agency theory (Madison et al. 2015). Monitoring systems help the principal to systematically collect information on the agent's dealings. The agent is prone to serve the principal's interests, if the principal has the capacity to monitor agent behaviour (Boshkoska 2015). From its early formulation, Jensen and Meckling (1976) remarked that monitoring mechanisms were meant to control deviancy by the agent, which included controls such as budget limitations and operating guidelines -beyond merely detecting and assessing the agent's performance.
Monitoring involves the principals' ability to decide whether the agents have complied with the contract's provisions and that managers would not promote self-interest in decisionmaking (Boshkoska 2015). However, with relatively minimal or even inaccurate information because of asymmetrical information, the principal's ability to direct agent behaviour towards set agendas may be inhibited. For example, managers may take advantage of their principals' ignorance to advance their own interests by making false promises, on agreements or changing the terms of deals to benefit themselves. This creates 'unfavourable take off' -a sort of opportunism that is possible when the principal cannot identify the productivity of the employee until the conclusion of the contract (Popov & Simonova 2006). Related to the possibility of unfavourable take off is the fact that agents may simply not exert the agreed-upon effort (shirking) as they may attend to personal business during working hours.
The conflict of interest emanating from the separation of management and ownership makes the likelihood of theft and other kinds of opportunistic behaviour common in small businesses, which are likely to have less formal mechanisms of control than their larger counterparts (Chrisman et al. 2004). However, as the survival of such small businesses often depends on the actions and decisions of a handful of managers and owners, the effect of any opportunistic behaviour by them is proportionally higher (Chrisman et al. 2004;Martin & Butler 2017). Good examples of opportunistic behaviour are fraud, falsehood and larceny. Opportunistic behaviour by either party has the potential to impact parties outside the immediate principal-agent relationship, as well as direct and indirect stakeholders in society (Zardkoohi, Harrison & Josefy 2015) -but with farreaching consequences to the business. Besides the financial implications of opportunistic behaviours, opportunism may also affect mutual trust and commitment, thus creating serious threats in the area of quality of social embeddedness (Yaqub 2009).
Opportunism, if not put to check, is likely to cause reduced cooperation, reputation wear-outs and increased defections (Yaqub 2009). Opportunistic behaviours within cooperative arrangements may increase transaction costs as principals have to put in place expensive control mechanisms like complex and more explicit contracts to minimise opportunistic hazards. Examples of such costs include costs incurred when processing information to craft contracts, monitoring contractual promises, administering contractual provisions and legal and business costs incurred when coordinating and ensuring desirable behaviours (Yaqub 2009). Therefore, controlling the problem of either principal or agent opportunism becomes so crucial to the sustainable performance of SMEs and the strategic networks that come along with doing so. However, regarding Zimbabwean SMEs, studies (Magaisa et al. 2013;Musanzikwa 2014) established that there was rather a need to separate ownership from control of business affairs to ensure accountability. Another study on SMEs in the Manicaland province of Zimbabwe by Guruwo (2018) established a lack of proper governance frameworks amongst entrepreneurs. The study also found out that most owner-managed enterprises had family governance systems that promoted conflict of interest. It was also established that many SMEs had no management advisory board and that the division of responsibilities and authority was not clear as spouses and children were actively involved (Guruwo 2018). The above evidence provides a scope for further extending research on the perceived impact of agency control on the performance of Zimbabwean SMEs. Few studies on business performance and agency control (whose results are mixed) have been conducted to date. It should be noted that studies investigating performance (by using different measures of performance) present mixed results and conflicting opinions (Maalu et al. 2013;Sciascia & Mazolla 2008). Some of the studies (Chirico & Bau 2014;De Massis et al. 2015) were conducted in developed countries with different contextual realities than those in Zimbabwe. Research investigating on how agency relationships in Zimbabwean SMEs are linked to business performance is still in its infancy. Moreover, as principal-agent relationships apply in practice to relationships between shareholders and top management, as well as to that between managers and employees at all levels, it is interesting to explore how such relationships manifest themselves and influence agency control within the context of SMEs in Zimbabwe.

Research purpose and objectives
The agency theory, often applied to how the shareholders or owners (principals) of a business relate to their managers (agents), can also be applied to the relationship between managers and employees (Bartol 1999). According to the theory, when managers are remotely involved in the activities of employees and make no deliberate effort to ensure that the interests of a business are protected, agency problems tend to manifest themselves because of divergent interests between principals and agents. One of its assumptions is that both employers and employees are basically driven by self-interest. The theory perceives humans as being opportunistic and as utility maximisers (Hendry 2002) -implying goal conflict between principals and agents. Its second assumption pertains to the rationality exhibited by human behaviour (Selten 1998)known as bounded rationality. According to Simon (1997), the term 'bounded rationality' is used to label rational choice, which considers that rationality is bounded because of limits in ones thinking capacity, available information and time.
Incomplete contracting is optimal that can be attained because of bounded rationality; hence, contractual completeness would remain elusive. Contractual completeness is likely to remain elusive in SMEs. It may be impossible to completely regulate the whole relationship, as all relevant matters such as the employee's effort may not be foreseen and understood at the time of contracting. As a result of bounded rationality, many employment contracts are likely to be incomplete in that they do not specify accurately the duties of the contracting parties under all conceivable circumstances (Zheng & Deakin 2016). This is even made worse by the fact that for Zimbabwean SMEs in the manufacturing industry, the owner or manager does almost everything, making it difficult for the enterprise to have board members who are separate from the stakeholders (Magaisa et al. 2013).
The third assumption is that the agent is presumed to have reserved information accessible to the principal at a cost -known as information asymmetry (Balago 2014). Information is, therefore, perceived as a purchasable commodity. Information asymmetry is entrenched in that the employee knows his or her own skill, capabilities and their own behaviour (Cousins et al. 2008), as well as job-specific information, better than the manager. However, classical agency theorists (Daily & Dollinger 1992;Fama & Jensen 1983b) document that the mutual coordination and communication amongst family members within family businesses help to mitigate information asymmetry between the two parties.
Finally, the other assumption is that the agent is expected not only to be risk-averse but also to be work-averse (Baiman 1990). Agency problems could emanate from a minimal effort by the agent (e.g. agents may resort to tardiness, flubbing or other such shirking behaviours). Agents may simply not exert the agreed-upon effort (shirking) as they may attend to personal business during working hours. As it is virtually impossible to eliminate shirking regardless of the degree of supervision, the goal should be to minimise it to a level, which warrants that the principal's goals are achieved. This could be achieved through the use of Boards of Directors particularly designed to bring new ideas, which is meant to assist in both the crafting and implementation of policies, as well as accounting for the way the enterprises are run. However, this has been found to be lacking in Zimbabwean SMEs (Magaisa et al. 2013); yet, sound corporate governance practices have been found to play a crucial role in increasing the financial performance of SMEs in Zimbabwe (Mazikana 2019).
In view of the above, the purpose of the current study was to determine the perceived impact of agency control on the performance of SMEs in Zimbabwe measured by return on investment (ROI) and innovation (both process and product innovation), and is guided by the following objectives: • to examine the nature of agency control in Zimbabwean SMEs • to evaluate the impact of agency control on the performance of SMEs in Zimbabwe.

Agency theory
The agency theory was originally formulated in an impersonal context, specifically the fiduciary relationship between a business' distant shareholders (principals) and chief executive officer (agent), where there is a great geographical and emotional distance between the two (Cruz, GóMez-Mejia & Becerra 2010). This makes it very applicable to non-ownermanaged family and non-family businesses, where fiduciary relationships exist, and information asymmetry exists between principals and agents, and where it may be too hard for the principals to control agent behaviours. The theory applies in practice to both relationships between shareholders and top management as well to that between administrators and workers at all levels. The fact that the agency theory can be applied extensively (Ahmad, Farley & Naidoo 2012) and in a variety of settings (Kivisto 2007) makes it a leading theoretical perspective in the current study.
From its origins in information economics, the agency theory advanced along two strands, namely principal-agent and positivist strands (Eisenhardt 1989). These two strands have a shared unit of analysis, the contract that binds both the principal and the agent. The current study is more aligned with the positive-agency strand, because it has been widely used as a foundation for empirical and theoretical work by organisation theorists and management academics (e.g. Eisenhardt 1989;Jensen 1998;Tosi & Gomez-Mejia 1989). It is thus rooted in scientific realism. In view of this, unlike the mathematically complex principal-agent strand, it is the more practical orientation of the positive-agency strand (Shapiro 2005) that situates this view as the principal concern of this study.
It is appropriate to consider how the agency theory could be applied in Zimbabwean SMEs, where approximately 80% of the businesses are family-owned and are mainly ownermanaged SMEs without an advisory management board (Guruwo 2018;Mbetu 2016). Family-owned SMEs are a good example where the chief executive officer may be a member of the founding family, with most employees coming from the immediate and extended families. The assumption that the interests of family member employees are already aligned reduces the likelihood of both formal contracting and management systems (Brawley 2016). Although the family and business values may not always align, managers in family-owned SMEs may nonetheless prefer to use personalised, social forms of control, rather than official, objective methods (Brawley 2016). In addition, the emotional attachment coming from close family ties within familyowned SMEs may, but not always, put agent opportunism to check, especially when the chief executive officer is a member of the founding family and most employees are from the immediate and extended families (Cruz et al. 2010).

The positivist strand
Researchers taking the positivist strand (Fama 1980;Fama & Jensen 1983a;Jensen & Meckling 1976) concentrated on detecting situations where the agent and principal are prone to pursue contradictory goals and then explain the mechanisms of governance that may put the agent's selfinterest to check (Eisenhardt 1989). Two main propositions summarise the governance mechanisms acknowledged in the positivist stream. The first is that, when contracts between principals and agents are outcome-based, agents tend to conduct themselves in the principals' interest. The second is that, when principals have evidence to verify agents' behaviour, agents are more prone to pursue the principals' interest (Eisenhardt 1989).
Regarding the first proposition, the argument is that contracts co-align the inclinations of agents to those of principals as rewards for both are subject to the same actions, and consequently, conflict of self-interest between agents and principals is minimised (Eisenhardt 1989). However, considering the obscured roles of principals and agents, especially in the family-owned SMEs, contracts might not be necessary (although they should be preferred to eliminate future disagreements), or may not be that enforceable. In this regard, most Zimbabwean SMEs are violating major provisions enshrined in the country's labour laws. Uzhenyu and Marisa (2017) established that a majority of those employed in the SMEs entered into verbal contracts of employment and that a lot of SMEs made no statutory contributions for pension, medical cover, income tax and insurance. Regarding the second proposition, agent opportunism is mitigated as they (agents) cannot cheat the principal because of information systems, which serve to update the principal on the actions of the agent. In the case of family businesses, the compassion of family owners towards their family members always shows in their being placed in senior managerial roles -and as a way of returning the favour, family managers often manifest exceptional loyalty and commitment to the business (Zhang & Cao 2016). This allegiance is likely to alleviate agent opportunism.
Agent opportunism and conflict of interest between principals and agents are the main assumptions of the positivist strand of agency theory. The positivist strand therefore assumes conflict of interests between principals and agents such that principals wish to increase their fortune subject to risk constraints, and agents strive to maximise their personal gain as they attempt to reduce personal risk and effort (Cuevas-Rodríguez, Gomez-Mejia & Wiseman 2012). This is often the case in non-family-owned businesses where there is a separation of ownership and management. The separation may not solicit altruistic behaviours, which may well be the basis of competitive advantage in their familyowned counterparts (Madison et al. 2015). Altruistic behaviours reduce information asymmetries and promote communication, fostering a commitment to family and a sense of belonging to the business (Eddleston, Kellermanns & Sarathy 2008). In addition, promises of familial succession and/or future ownership in family businesses are prone to encourage continued support from family employees and board members and thus help in minimising agent opportunism and risk-aversive behaviours (Eddleston et al. 2008;Pagliarussi & Costa 2017). On the contrary, the performance of Zimbabwean family businesses has been affected by difficulties attributable to a lack of formal organisational and corporate structures and includes the http://www.sajesbm.co.za Open Access complexities of family business dynamics -such as ownership, succession, family harmony and solidarity (Sikomwe et al. 2012). Interestingly, some studies (Bammens, Voordeckers & Van Gils 2011; Siebels & Knyphausen-Aufseβ 2012) even caution that altruism may be causal to self-discipline challenges that could further expose family businesses to even more types of agency costs. These issues, however, lie beyond the scope of the current investigation.

Agency relationships in small-to-medium enterprises
The conflict of interest caused by the separation of ownership from management makes the likelihood of theft and other kinds of opportunistic behaviour common in small businesses, which are likely to have less formal mechanisms of controls than larger businesses (Chrisman et al. 2004). In addition, most SMEs are closely run and managed by owners who have direct insights into the internal processes of the business (Visser & Chiloane-Tsoka 2014). The control function of the board may not be necessary, and many of their boards exist on paper only (Visser & Chiloane-Tsoka 2014). The business would therefore be indistinguishable from the owner-manager, upon whom its failure or success depends (Stokes & Wilson 2010).
Monitoring agent behaviour is a core issue in the agency theory (Madison et al. 2015). Monitoring systems help the principal to systematically collect information on the agent's dealings. Agents are prone to act in the interests of principals, if the principals have the capacity to monitor the agent's behaviour (Boshkoska 2015). From its early formulation, Jensen and Meckling (1976) remarked that monitoring mechanisms were meant to control deviancy by the agent, which included controls such as budget limitations and operating guidelines -beyond merely detecting and assessing the agent's performance. Monitoring involves the principals' ability to decide whether the agents have complied with contract provisions, and that managers would not promote self-interest in decision-making (Boshkoska 2015). However, given the relatively minimal or even inaccurate information because of asymmetrical information, the principal's ability to direct agent behaviour towards set agendas may be put to check.
The above literature suggests a relationship between agency control and SME performance, and consequently, the following hypotheses are proffered: H1: There is a statistically significant relationship between agency control and business performance as measured by innovation.

H2:
There is a statistically significant relationship between agency control and business performance as measured by ROI.
The above indicators of performance (innovation and ROI) were chosen because, generally, the measures to be used to either compare or assess the performance of different small enterprises should be balanced through the inclusion of both financial and non-financial measures (Bititci, Firat & Garengo 2013). Return on investment becomes a very important measure that entrepreneurs can use, as it tells them the extent to which their investment has been a success (Lee 2012). Return on investment, a crucial balance sheet ratio, therefore helps in assessing the efficiency of the use of resources of a business, as well as in providing an indication of how effective one's investment in the business is (Lee 2012). In addition, because of the daunting challenge linked with gathering and interpreting financial data gathered from privately owned SMEs, researchers have little choice but to contend with subjective information on business performance (Dekker et al. 2015). Consequently, the overwhelming evidence regarding the importance of innovation in enhancing an enterprise's ability to adapt to dynamic business environments (Shouyu 2017) contributed to its choice as a subjective measure of performance in this study.

Research design
The study employed the survey design. The chosen design allows for the collection of sizeable data from a significant population in a very economical way (Saunders, Lewis & Thornhill 2015). Quantitative data are usually collected by using a survey, and such data can be analysed quantitatively by using either descriptive or inferential statistics. Furthermore, data collected by using the survey strategy are useful when suggesting possible reasons behind certain relationships between variables and for producing models of the relationships (Saunders et al. 2015). Through the use of sampling, the survey design can also generate findings representative of the whole population at a very low cost.

Research approach
The two general approaches, qualitative and quantitative, are associated with subjectivity and objectivity, respectively (Hughes 2006). The current study adopted a quantitative approach for the reason that it has the potential to deal with questions about relationships between measured variables with the purpose of controlling, predicting and explaining phenomena (Leedy & Omrod 2005). It is more suitable for examining the magnitude of an issue, phenomena or problem (Kumar 2014).

Research participants
As Harare Province has a higher concentration of SMEs (Madzivanzira 2011), SME owners or managers in the province were chosen. However, because of lack of a comprehensive database of SMEs in the province, it became impracticable to come up with a comprehensive sampling frame of the enterprises. Consequently, convenience sampling had to be used. Convenience sampling is sometimes called opportunity or accidental sampling. It entails choosing the nearest individuals to serve as participants and continuing the process until the required sample size has been attained (Cohen, Manion & Morrison 2011). For non-probability sampling methods, the issue of sample size is unclear, and, http://www.sajesbm.co.za Open Access unlike probability sampling, no rules exist (Saunders et al. 2015). A well-ordered non-probability sample often gives satisfactory results, such that researchers may not even consider probability sampling (Cooper & Schindler 2014). To mitigate the weaknesses related to convenience sampling, a reasonably large sample was used to increase its statistical power. A total of 106 questionnaires were completed.

Research instruments
The researcher used a self-constructed forced-choice questionnaire to collect data for the period 2015-2016. Items on such a questionnaire are usually easier and quicker to answer, as they involve less writing (Saunders et al. 2015). It is also easy to compare responses as they have been preset.

Ensuring reliability and validity
Reliability enables the researcher to estimate error, the larger the reliability, the smaller the error, and conversely, the smaller the reliability, the larger the error (Punch 2005). Reliability is concerned with whether a questionnaire produces similar results on repeated trials and under different conditions (Saunders et al. 2015). As internal consistency, reliability demonstrates that the instrument is run once only through the split-half method (Cohen et al. 2011). Cronbach's alpha measured split-half reliability by estimating the average correlation that would be obtained by considering every possible way to split the test in half (Gravetter & Forzano 2016). It produces values between 0 and 1.00, and a higher value shows a better degree of internal consistency or reliability. A value above 0.7 suggests good reliability (Hair et al. 2014). Table 1 shows the results of the reliability statistics.
A Cronbach's alpha of 0.816 (α = 0.816) is a good reliability measure. It means that all the items were internally consistent and reliable.
The questionnaire was also subjected to a pilot test on five SMEs that did not take part in the study. Pilot testing provided some idea on the questionnaire's face validitythat is, whether the questionnaire appeared to make sense (Saunders et al. 2015). Responses from pilot testing provided the researcher with an idea of both the reliability and suitability of the items, allowing for necessary amendments to be performed.
The content validity of a measuring instrument is the extent to which it provides adequate coverage of the investigative questions guiding the study (Cooper & Schindler 2014). If the data collection instrument adequately covers the topics that have been defined as the relevant dimensions, it can be concluded that the instrument has good content validity (Cooper & Schindler 2014). To ensure content validity, questions were given to HR experts and experts in SMEs to be examined for bias, sequence and clarity.

Research procedure
Trained research assistants were enlisted for data collection. They assisted in hand-delivering questionnaires. This raised response rates because of the individual contact and because research assistants did not disturb participants during working hours (De Vos et al. 2011). Data for the years 2015 and 2016 were collected.

Statistical analysis
The Statistical Package for Social Sciences (SPSS) version 20 was used for both descriptive and inferential statistics. Bivariate correlation analysis in the form of Spearman's correlation coefficient was used to examine the perceived impact of agency control on business performance (Statistics Solutions 2020). Spearman's correlation has the advantage that unlike Pearson's correlation, it can evaluate a monotonic relationship between two ordinal or continuous variables (Minitab 2019).

Demographic details of respondents and small-to-medium enterprises
The demographic details of the respondents were as follows: 86.8% of the SME owners or managers were male, whilst 13.2% of the respondents were female. Of the SMEs involved in the survey, only 6.6% were medium-sized (employing between 31 and 75 employees), whilst the rest were small enterprises (employing between 6 and 30 employees Ninety-three per cent of the SMEs had no HR departmentthe owners, managers or owners who doubled as managers  had to put on many hats. It was also established that 69% of the respondents had either primary or secondary education qualifications, and the rest had either diplomas or degrees and that, on average, most SMEs had been operating for 7 years.
The nature of agency control Table 2 shows the nature of agency controls prevailing in the SMEs.

Impact of agency control on performance
Spearman's correlation coefficient analysis was applied to examine the relationship between agency control and business performance measured by innovation. The following hypothesis was tested: H1: There is a statistically significant relationship between agency control and business performance as measured by innovation. Table 3 shows Spearman's correlation on agency control versus performance measured by innovation.
The results displayed in Table 3 show a weak positive relationship of 0.117 between agency control and business performance measured by innovation in the SMEs. Furthermore, the relationship is insignificant because of a p-value of 0.234, which is above 0.05 (i.e. 5% significance level). This suggests the acceptance of the null hypothesis, which states that there is no statistically significant relationship between agency control and business performance measured by innovation.
Spearman's correlation coefficient was again used to examine the nature and strength of the relationship between agency control and business performance measured by ROI. The test was conducted under the following hypothesis: H2: There is a statistically significant relationship between agency control and business performance as measured by ROI. Table 4 shows Spearman's correlation on agency control versus performance as measured by ROI.
According to the results displayed in Table 4, there is a weak positive relationship between agency control and business performance as measured by ROI indicated by a coefficient value of 0.184. The relationship is statistically insignificant at 5%, as a p-value of 0.059 is >0.05. The results suggest the acceptance of the null hypothesis, which states that there is no statistically significant relationship between agency control and business performance as measured by ROI.

Discussion
The nature of agency control Table 2 show that owners or managers agreed with the items AR4, AR5, AR 6, AR11 and AR12. Item AR 4 enquired whether performance-related pay helped align employee job behaviours with owner or manager goals. The agency theory makes prescriptions regarding the  circumstances under which fixed pay (salary) and variable pay (e.g. commission and bonus) should be used to foster efficient alignment of principal and agent interests (Bartol 1991). Variable pay such as performance-related bonuses does not increase base pay permanently and must be reearned to be received again. It is in this sense that the performance-related pay helped align employee job behaviours. However, this runs contrary to the stewardship theory perspective, which holds that managers are stewards whose performance is built around an inherent aspiration to serve the business and will, as a result, naturally support the interests of their principals (Hernandez 2008;Zahra et al. 2008).

Results in
Item AR5 asked whether frequent monitoring of delegated enables owners or managers to keep track of operations, whilst AR6 solicited if frequent face-to-face talks with subordinates over delegated tasks kept them in control. The agency theory assumes that both the agent and principal are self-regarding and boundedly sensible, resulting in individual utility-maximising behaviour if appropriate incentives and controls to harmonise the goals of the agent with the principal are not enacted. Owners or managers agreed on the need for frequent monitoring. From a structural viewpoint, the agency theory suggests governance mechanisms that monitor and control agents to prevent agent opportunism and better align the goals of both principals and agents. Performance management is one such tool that allows for goal alignment between the two parties, thereby arresting one assumption of agency theory, which claims that both principals and agents pursue personal objectives (Van Puyvelde et al. 2013).
Owners or managers also agreed on items AR11 and AR12. Item 11 enquired whether clear documentation of tasks and outcomes were essential for effective delegation, and Item 12 enquired whether entrepreneurs train first before considering whom to delegate to. Delegation of monitoring is more likely to generate commitment. However, to avoid agency slippage (which may stimulate the agent to adopt a position different from that of the principals) as a result of the delegation structure (Da Conceição 2010), principals often resort to training first before the delegation. An analysis by Kräkel (2004) shows that delegating decisions to managers may be favourable despite the managers' inclinations over the company's resources and despite their risk-aversive behaviours. The strategic advantages from delegation may be realised if the preferences of managers over resources and their moderate risk-aversive behaviours do not offset the benefits from the delegation.
Regarding AR3 (owners' or managers' family members value relationships at the expense of performance), a mean of 2.43 suggests disagreeability. According to Chrisman et al. (2012), particularism emanates from the personalisation of authority and derives from the fact that the control rights of family cause a personalistic application of authority that permits family members to pursue varied objectives other than pure profit or business value maximisation. It is the result of the discretion of a family business to act as the family members see fit. Whilst managers in family-owned SMEs are expected to implement rational-calculative decisions, family control rights authorise the family to interfere with the matters of the business to replace other 'particularistic' benchmarks of their choosing (Carney 2005). Alternatively, the fact that family SMEs are likely to be entrenched in family relationships, such as the parent-child relationship (Schulze, Lubatkin & Dino 2003), parents may be excessively liberal to their kids, resulting in children taking advantage of the liberality by free-riding or shirking (Dawson 2012;Eddleston et al. 2008). Even though the current study did not separate family from non-family SMEs, its results could not sustain the aforementioned tendencies. In fact, in line with a recent study by Adjei et al. (2019), spouses and children showed a positive relationship with business performance.

Impact of agency control on performance
According to Table 3 and Table 4, there was a weak positive relationship between agency control and business performance measured by both ROI and innovation in the SMEs. Furthermore, the relationships were found to be statistically insignificant because of p-values >0.05 in both cases. This suggested the acceptance of the null hypotheses, meaning that no statistically significant relationships existed between agency control and business performance as measured by both ROI and innovation. Such atypical results could be explained by the fact that agency control has a bearing on the professionalisation and formality, or a lack thereof obtaining in the businesses (Chu 2009). The fact that 93% of the SMEs involved in this survey had no HR department and that 93.4% were small enterprises (employing relatively fewer employees) suggests a lot of informalities on how human resource issues are handled. Whilst such informality is often praised (as it allows decisions to be made with minimal red tape), the same informality could lead to inconsistencies in the manner in which people are treated. This is likely to have a bearing on performance. As enterprises grow, informal systems cannot be relied on, as more structure will be required. Legal issues also emerge, as businesses need to comply with employment laws as they add staff (Lorenzet, Cook & Ozeki 2006). The other reason for the results could be that employees in smaller businesses often have to perform a greater variety of tasks than those in larger businesses, and specialists are less likely to be found in these businesses (Dalotă &Grigore 2010). This could lead to poor performance despite the level of agency control prevailing in the enterprises.
Regarding agency control and business performance, there is a dearth of studies in Africa and Zimbabwe in particular, and hence, there is a need for more research. Despite the aforesaid dearth of literature, Panda and Leepsa (2017) observe that, generally, the separation of ownership from control in businesses may lead to loss of appropriate monitoring by the owners on the managers, who may use business assets for their private purpose to maximise their welfare, with obvious implications on business performance. Opportunistic behaviour by either party has the potential to impact parties outside the immediate principal-agent relationship, as well as direct and indirect stakeholders in society (Zardkoohi et al. 2015) -but with far-reaching consequences to the business. Besides the financial implications of opportunistic behaviours, opportunism may also affect mutual trust and commitment, thus creating serious threats in the quality of social embeddedness (Yaqub 2009;Zardkoohi et al. 2015).
A study by Li, Armstrong and Clarke (2014) provided fresh indications that the corporate governance bundle (measured by constructs emanating from the literature on big businesses) obstructs the financial performance of small businesses. These findings confirm results from the current study. Furthermore, the aforesaid study is related to the current one because it admits that the agency theory functions as the basis for corporate law and for regulations and principles of corporate governance. The study also established that prevailing theory failed to acknowledge differences between large businesses and their small counterparts, and that small businesses are yet to include governance practice for them to perform better financially if regulatory bodies are able to design small business-tailored corporate governance-guiding principles (Li et al. 2014). The study also found out that the same governance, like the board of directors, may serve different roles in small businesses compared with their larger counterparts.
The seminal work by Chrisman et al. (2004) established that agency problems in American small family businesses were less severe than in their non-family counterparts. The study established that agency relationships obtaining in a business may either increase or decrease performance because of agency costs. Contrarily, although the current study did not take a comparative approach, and that its conceptualisation of an SME could be different from the American one, a weak relationship between agency control and business performance measured by both ROI and innovation in the SMEs was established. Their results also suggested that, excluding the effect of agency cost control mechanisms, family and non-family businesses investigated showed comparable economic performance when measured by short-term sales growth (Chrisman et al. 2004). Thus, in line with the arguments of founding agency theorists, their results suggest that the involvement of family has the capacity to reduce overall agency problems (Fama & Jensen 1983a;Jensen & Meckling 1976). However, this is an area that requires more research because the family setting provides a variant to the impersonal setting implied by the agency contract (Kotlar & Sieger 2018;Madison et al. 2015), thus presenting confounding scenarios.
Utilising agency theory, an enquiry by Hussain and Hadi (2018) found that ownership concentration had a considerable negative relationship with the performance of businesses. The study also concluded that the mechanisms of corporate governance had a significant effect on the SMEs' performance in the Construction Industry Development Board (CIBD) in Malaysia. These results disagree with those from the current one in that no statistically significant relationships existed between agency control and business performance. The differences could be explained by the fact that the study focussed on CIDB-registered businesses, most of which (above 99.9%) were run by a board of directors; yet, only 3% of the medium enterprises in the current study had a board of directors. The control function of the board may not be necessary, and many family-owned SME boards exist on paper only (Visser & Chiloane-Tsoka 2014). The business would, therefore, be indistinguishable from the owner or manager, upon whom its failure or success depends (Stokes & Wilson 2010), but whose competency in all areas pertaining to the business may be questionable. However, it should be noted that research on agency control and its impact on the performance of SMEs is in its infancy, and results are mixed.

Conclusions and managerial implications
This study aimed to establish the perceived impact of agency control on SME performance as measured by innovation and ROI in SMEs in Harare Province, Zimbabwe. It was established that the most used control mechanisms were frequent monitoring and face-to-face deliberations with subordinates. A weak relationship between agency control and business performance measured by both ROI and innovation in the SMEs was established. No statistically significant relationships were found to exist between agency control and business performance measured by both ROI and innovation.
Given the number of items owners or managers showed indecision (five in total), it is recommended that expert advice be enlisted to coach owners or managers on how best principal-agent relationships could be managed. It is also recommended that apart from formalising principal-agent relationships, monitoring mechanisms, incentives and strategic planning systems be formally instituted. It is again recommended that medium enterprises, because of their size, may need to run under governance structures like oversight boards because compromised corporate governance or not adhering to its codes can result in fraud, misuses and the poor performance of the businesses (Solomon 2010). In view of the above, agencies like the SEDCO need to be recapitalised through deliberate government funding for them to offer mentorship and learnership to owners or managers.

Limitations and directions for future studies
Apart from the accuracy of the agency theory and that it has been the dominant paradigm in corporate governance for the past 40 years or so, considering it alone in this study gave the current study a narrower view to corporate governance, which is but a broad area (Armstrong & Sweeney 2002). Other predominant theoretical perspectives such as the stewardship theory or the resource-based view (RBV) could be used to support agency theory, as none on its own could successfully address all the complexities associated with SMEs and their various modes of ownership. Whilst the study made use of cross-sectional data, forthcoming studies should strive to gather longitudinal data involving nationwide samples to validate and enhance the generalisability of findings of the current study. The other challenge arose from the reluctance by owners or managers of most privately owned businesses to share objective financial data. This challenge was mitigated by including a subjective measure (innovation) in the measurement of performance. Finally, given that the parameters of generalisability for convenience sampling are negligible (Cohen et al. 2011), more robust samples could be used in a nationwide survey in future.

Contributions of the study
This study validated the agency theory and the relationship or a lack thereof between agency control and SMEs' profitability and innovative potential. Furthermore, studies linking agency relationships to performance have been studied in the context of large public listed businesses (Bendickson et al. 2016;Yahya et al. 2016). Very limited studies have explored the role played by agency in SMEs, especially in a developing economy (Yahya et al. 2016). This study is one of the very few contributing to the growing literature in the area and could be the first of its kind in Zimbabwe. In addition, the study employed a hybrid approach to performance measurement, making use of both an objective and more subjective measure -a departure from the thinking that prefers the use of objective measures only (Yildiz & Karakas 2012). Therefore, the current study's contribution in the area of performance measurement lies in its ability to adopt balanced measures of business performance by including a more recent, intangible and externally focussed measure (innovation) to complement a more traditional financial measure (Bititci et al. 2013).
Research ethics help prevent research abuses by placing emphasis on the sensitive and humane treatment of participants, who may be exposed to some hazards by research processes (Bless, Higson-Smith & Sithole 2013). Typically, participants were informed about the research purpose and its basic procedures, the researcher's identity, the sponsor and how the data might be used (De Vaus 2012). They were also provided with a description of the likely benefits of the study, a description of how they were selected, an offer to answer any questions, a statement that participation was voluntary (this was contained in the cover page) and that each participant could pull out at any time or decline to respond to any particular question (De Vaus 2012). The researcher had to seek for clearance from the Faculty Research Ethics committee as well. To ensure confidentiality, data were stored in cabinets, and all information that could lead to the identification of respondents was removed.

Funding information
This study was funded by the Research Board, Great Zimbabwe University, and the University Research and Innovation Committee, Central University of Technology, Free State, South Africa.

Data availability
The author confirms that the data supporting the findings of this study are available within the article.

Disclaimer
The views and opinions expressed in this article are those of the author and do not necessarily reflect the official policy or position of any affiliated agency of the author.