The effect of microcredits on business performance of small business owners in rural Tanzania

(1)

The effect of microcredits

on business performance of small business

owners in rural Tanzania

Master’s thesis in partial fulfillment for the award of

Masters of Science Entrepreneurship

submitted by Kathrin Borner

at the Universiteit van Amsterdam (11605855) and Vrije Universiteit Amsterdam (2628297)

Supervisor: prof. dr. Enno Masurel, Vrije Universiteit Amsterdam

Co-Supervisor: dr. Joeri Sol, Universiteit van Amsterdam

Application Date: April 1st ₂₀₁₈

(2)

Preface

The copyright rests with the author. The author takes full responsibility for the contents of this thesis and declares that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Universiteit van Amsterdam and the Vrije Universiteit Amsterdam are responsible solely for the supervision of completion of the work, not for the contents.

Front cover photo: Kathrin Borner.

The photo was taken at one of the food markets just outside Mzumbe University in May 2018. The photo shows a typical fruit market stand in rural Tanzania, it belongs to one of the fruit vendors who was a research subject for this master’s thesis.

(3)

Acknowledgements

This thesis, the final project for my Masters in Entrepreneurship has been a great learning ex-perience for me. Along this journey I have received support from several people whom I now want to thank.

First of all, I would like to thank my supervisor team. Thank you, Enno, for making this thesis possible, for sharing your contacts and experience. Your supervising style was the right mix between providing freedom and guidance. Thank you Joeri for co-supervising me, your feed-back on the questionnaire and the data analysis was very helpful; it gave a sense of certainty when needed the most.

Next, I would like to thank all experts who contributed with their knowledge to refine this work. Emiel, Lennard, and Rael (graduates) from the VU; Gera and Judith from Triple Jump; Rafael from the UvA; Anusiatha and Laurent from SUBO; Musabila, Ernest and Dr. Kato from Mzumbe University; Abdulrazak and Hamisi from the local government of Changarawe vil-lage; Kasanda one of the small business owners; as well as Eugene and Iris from the Dutch embassy.

‘Asante sana’ to Subi and Musabila for creating our start in Tanzania as comfortable as possible. You were always there for us and also made sure that we enjoyed our time in and around Mzumbe.

Furthermore, I would like to thank Irene, Mery, and Laurent for having been great research assistants, and also, thanks to all the small business owners for taking the time participating in our study.

Thank you, Sebastian, Ellie and Carmen for commenting, proofreading and spell-checking on a previous draft of this work despite your tight schedules! And thank you, Guido for coming to Tanzania with me!

(4)

Abstract

This thesis investigates the effect of microcredits on business performance of small business owners in the context of rural Tanzania. The qualitative pre-study interviews 18 experts who helped develop the basis to conduct a quantitative survey with 118 small business owners in the Mvomero district. Using t-tests and regression analyses, the results show that microcredits pos-itively and significantly influence business performance, specifically profit, business invest-ments growth and a total business performance indicator. Furthermore, we find that the size of the microcredit positively and significantly affects profit and income growth but were not able to assess the impact of microcredit duration. Our overall findings suggest that microcredits only impact business performance to a minor extent. Therefore, other determinants for small busi-ness growth in the informal economy should be further explored by scholars.

(5)

List of Abbreviations

BP Business Performance

GEM Global Entrepreneurship Monitor

GDP Gross Domestic Product

H# Hypothesis #

LDC Least Developed Country

IMF International Monetary Fund

M-Pesa Mobile Money (Pesa, the Swahili word for money)

MC Microcredit

MFI Microfinance institution

MSE Micro and small-sized enterprise

NMP National Microfinance Policy (of the United Republic of Tanzania)

ROSCA Rotating savings and credit association

SACCO Savings and Credit Cooperative

SUBO SUBO Financial Consultant and General Supplies Ltd (MFI)

TAMFI Tanzania Association of Microfinance institutions

(8)

List of Figures

Figure 1 The theoretical model. ... 23

Figure 2 Model of the impact chain. Adapted from Hulme, 2000. ... 25

Figure 3 The theoretical model including business performance measures. ... 34

Figure 4 Testing H1, the results... 47

Figure 5 Testing H2, the results... 51

(9)

List of Tables

Table 1 Operationalization of key constructs. ... 36

Table 2 Treatment and control group: The socio-demographics. ... 38

Table 3 Treatment and control group: The household characteristics... 38

Table 4 Treatment and control group: The business characteristics. ... 39

Table 5 The microcredit specifications. ... 40

Table 6 Financing market ... 41

Table 7 Factor loadings of Business performance measures. ... 42

Table 8 Business performance, the scores. ... 43

Table 9 Business performance, the scores. ... 44

Table 10 Regression models: MC obtained. ... 46

Table 11 Microcredit duration: The business performance measures. ... 48

Table 12 Regression models: MC duration. ... 50

Table 13 Microcredit amount: The business performance measures. ... 51

Table 14 Regression models: MC size. ... 53

Table 15 Capital spending. ... 54

(10)

1 Introduction

Today, around one billion people are living in the worlds least developed countries (LDCs). These LCDs are distinguished by three United Nations classified characteristics: low income, weak human capital stock (e.g. a low adult literacy rate) and economic vulnerability (United Nations, 2018a). Most of these LDCs are located in Africa (World Bank, 2018a; United Na-tions, 2018a). In the last 30 years, poverty in Sub-Saharan Africa has even increased (Collier, 2008).

The United Nations (2018c) is aiming for poverty alleviation as one of their sustainable devel-opment goals to end poverty in the next 15 years. An institutional perspective argues that in or-der to reach sustained economic growth and end poverty, good governance policies are neces-sary, such as strong and credible inclusive (financial) institutions with transparent rules and governments which are free of corruption (Acemoglu & Robinson, 2013; Moyo, 2009). How-ever, we must consider that these institutions will help realizing opportunities, but not create them (Collier, 2008). This gap is where the small business owners must step in. Hence, to alle-viate poverty, change must come from within the LDCs through self-empowerment of the peo-ple (Collier, 2008; Landes, 1998).

Access to finance is key enabler for creating small business growth, leading to economic growth and thus for development and against poverty. However, until now, the poor are largely ex-cluded from formal financial systems (Brau & Woller, 2004). It is estimated that in total, 1.7 billion people are unbanked, while most of them live in LDCs (Demirgüç-Kunt et al., 2018). Accordingly, in Sub-Saharan Africa approximately 67.2% of all people do not have a bank account at a financial institution (World Bank, 2018c) and in rural Tanzania only 8.6% of the working population are banked, compared to 32.1% of the urban population (NMP, 2017). Microcredits have the reputation to fill this gap by providing banking to the poor (Yunus, 2003). These microcredits provide capital to entrepreneurs who are then able to create new firms and expand existing ones (Banerjee, 2013). This can lead to small business performance growth ultimately resulting in economic growth (Low & McMillan, 1988) and serves the overall aim to alleviate poverty (Brau & Woller, 2004; Bruton et al., 2013; Collier, 2008; Khavul et al., 2009; Moyo, 2009; United Nations, 2018d). Accordingly, the Tanzanian government has

(11)

intro-duced in October 2017 a new tyear National Microfinance Plan which describes that en-hancing access to micro-finance is seen as a way to creating employment, economic growth, and reducing poverty (AllAfrica, 2018; NMP, 2017).

The idea of Microfinance has likely existed probably informally for centuries, but until 1977, the foundation of the Grameen Bank, microfinance as a concept did not receive great attention. Now, the microfinance sector is forming into a global industry (Brau & Woller, 2004) with more participants than ever before. Indeed, worldwide over 3,000 MFIs offer microcredits to more than 200 million borrowers (Microcredit Summit, 2018a). This makes microfinance one of the largest development programs in the world, both in terms of financials and number of people targeted.

There is no doubt that the microcredit market is booming, but can microcredits really fulfill all those promising goals? Do they really increase performances of small business owners, lead to economic growth and lead people out of poverty? Or like Bruton et al. put it: “we seek to under-stand microfinance supported entrepreneurs” (2013, p. 687). Since entrepreneurship is still a young field of study (Davidsson & Wiklund, 2001; Zahra & Dess, 2001) with its research focus on North America and Europe (Bruton et al., 2008), there is still a clear lack of entrepreneurial research in the informal economy (Kolk and Rivera-Santos, 2018; Webb et al., 2009; 2013), especially in the African context (Kiggundu, 2002).

In order to achieve progress, we need a greater understanding about the impact of microcredits (Brau & Woller, 2004). Until now, there is no general agreement on microcredit impact studies yet, while early and recent research findings differ (Roodman & Morduch, 2014). Early studies primarily find a positive impact of microcredits while recent studies question this (see, for ex-ample, Armendáriz & Morduch, 2000; Banerjee, Duflo et al., 2015; Duvendack et al., 2011). In recent studies, scholars do not praise microfinance but rather judge microcredit impact to be “modestly positive” (Banerjee, Karlan et al., 2015, p.1; Brau & Woller, 2004; Karlan & Zin-man, 2009). However, results still vary from study to study, also because the impact of micro-credits depends on the context of the study (Brau & Woller, 2004; Van Rooyen et al., 2012). Also, most microfinance research was conducted in Asia, where the concept originated (Van Rooyen et al., 2012). Accordingly, Kolk and Rivera-Santos (2018, p. 430) call for more con-text-specific research and state “the potential of Africa-based research has still not been

(12)

ful-filled”. Moreover, in Sub-Saharan Africa “further research is clearly needed” (Van Rooyen et al., 2012, p. 2259). In fact, Van Rooyen et al. (2012) found for their literature review only 15 relevant microfinance studies conducted in Sub-Saharan Africa and only one of them in Tan-zania.

Although the microfinance sector in Africa is still relatively small, it is growing and research in this field provides an opportunity to shape policy-making (Van Rooyen et al., 2012). For example, Tanzanian policy-makers perceive microfinance as one contributor for transforming the country into a middle-income nation (NMP, 2017) with the help of small business owners (Khavul et al., 2009). However, those policymakers currently face a high level of information asymmetry and unreliable data which is why they ask for more research in the microfinance sector (NMP, 2017).

Hence, the central research question of this study is:

To what extent do microcredits determine the performance of small businesses in rural Tanzania?

With this study we follow the call of Bruton et al. (2013) to investigate the impact of micro-credits from a micro view with a quantitative approach while most studies have applied a macro view. Furthermore, this study meets the demand for more impact studies that also investigate how microcredits are used (Banerjee, Duflo et al., 2015; Wiklund et al., 2009). Examining this contributes to the broader literature and to our understanding of the impact of microcredits in rural Tanzania. Accordingly, we will develop theoretical and practical implications.

This study first summarizes the concept of microfinance and alternative financing options for small business owners. Next, we provide an overview about performance measures in the con-text of this study and continue with the impact of microcredits leading to three Hypotheses. In the following chapter, we explain the Methods used. This includes the sections research design, research context, operationalization of the key constructs and data analysis. Then, the results chapter shows the investigated effect of microcredits on business performance. Finally, we dis-cuss the results and provide theoretical and practical contributions accordingly. The paper con-cludes with limitations and avenues for future research.

(13)

2 Theoretical background

2.1 Microfinance and alternative financing sources

In the LDCs most people do not have access to a formal bank account (World Bank, 2018c). Likewise, businesses rarely have access to formal capital provided by banks as a consequence of operating within the informal economy and not having a legal status. However, access to capital is crucial, especially for small business owners in LDCs who constantly fight to over-come insecurities and scarce times (Demirgüç-Kunt et al., 2018). Taken together, capital can force small business owners to save money instead of spending it immediately; this can help managing financial risks and even encouraging starting a new business (Demirgüç-Kunt et al., 2018).

Financing options in LDCs reach from informal social support networks to formal banks (Brau & Woller, 2004). Since access to banks is limited; microfinance, savings groups, and social net-works play the major roles in accessing or managing capital. Further informal financing options in LDCs include saving cash at home; investing in livestock, jewelry or real estate (Demirgüç-Kunt et al., 2018) or borrowing from informal lenders (Battilana & Dorado, 2010). Traditional moneylenders were more popular before MFIs existed, they provide immediate cash and charge in return extremely high interest rates of often above 100% annually which is why they are also referred to as loansharks (Battilana & Dorado, 2010).

Microfinance is exploiting a ground-breaking idea: it sees the poor as customers (Brau & Wol-ler, 2004). This idea got popular in the 1970s when the 2006 Nobel Prize winner Muhammad Yunus started to hand out small amounts of money to the poor and based on its success found-ed the Grameen Bank. The Grameen Bank is a social enterprise which provides finance to the poor with the overall aim to alleviate poverty (Yunus, 2003). Yunus must have hit an unde-served need resulting in an ongoing demand for microfinance. Consequently, many nongovern-mental organizations followed and offered microfinance as a development tool (Battilana & Dorado, 2010). Since the early 1990s, when demand exploded, we discover more commercial MFIs joining the market (Battilana & Dorado, 2010) which offer products with modified spec-ifications (Karlan & Zinman, 2008). Initially, those for-profits had issues freeing themselves from government subsidies or donors (Brau & Woller, 2004).

(14)

Nowadays, microfinance institutions offer the same service types than formal banks, just in different scales and methods of delivery (Brau & Woller, 2004). In the literature, the term mi-crofinance refers to an umbrella term which includes, but is not limited to the products micro-credit, micro-saving, and micro-insurance. A growing body of literature has examined the mi-crocredit, which is also the most common product offered by MFIs and will be further explained in Section 2.1.1.

saving and micro-insurance are compared to microcredits rather new concepts. Micro-saving programs usually force customers to save a minimum amount each week. The accumu-lated savings then serve as a group collateral with strict rules about when the money can be accessed (Brau & Woller, 2004). Micro-insurance provides financial security for health-related issues to a small group of people who function together as a pool for all risks and resources. In fact, it is a market-based solution for people in LDCs who are still mostly excluded from health care insurance. This could indicate a great market opportunity, or also a lack of demand. One challenge that micro-insurance needs to overcome is to increase people’s value perception for making up-front payments for something that might benefit them in the future (Dror & Jacquier, 1999).

2.1.1 Microcredit

2.1.1.1 Definition, current development and issues

In the literature microcredits refer to small amounts of money given to poor business owners provided by MFIs or charities (Daniels et al., 2016). The money amounts are small so that the poor customers are able to repay them (Banerjee, 2013). Typically, MFIs provide only business loans, but some also provide consumption or emergency loans (Brau & Woller, 2004).

In comparison to traditional moneylending “microlending can be seen as an innovation in the technology of lending that leads to a large reduction in interest rates” (Banerjee, 2013, p. 489). On the contrary, compared to formal banks the interest rates of microcredits are relatively high (Daniels et al., 2016). Reasons for this are high uncertainties and large fixed costs (e.g. custom-er documentation, collatcustom-eral management, repayment monitoring) which need to be covcustom-ered by a relatively low loan sum (Banerjee, 2013; Battilana & Dorado, 2010). High uncertainties are

(15)

in-evitable when targeting poor business owners who are often exposed to “boom-and-bust cy-cles” (Battilana & Dorado, 2010, p.1423). Therefore, it is not surprising that MFIs, who often also aim for profits, charge high interest rates (Webb et al., 2013).

Previous research focused on default rates to judge the success of a microcredit program. Those are usually below 10% (Brau & Woller, 2004; Yunus, 2003; 2004), often even below 2% (Banerjee, 2013) and then speak for the success of the microcredit program. But looking at re-payment rates does not say anything about the business success, because until now we do not certainly know from what sources small business owners repay their microcredits (Brett, 2006). In resource scarce environments, personal networks provide access to finance (Elfring & Huls-ink, 2003) and also remittance payments from friends or family are known practices (Rai & Sjöström, 2004). This money could be used to create an illusion of microcredit programs being successful (Duffy-Tumasz, 2009; Montgomery, 1996).

2.1.1.2 Purpose of microcredits

In the traditional approach microfinance fulfills the purpose of alleviating poverty and empow-ering women.

(1) Poverty alleviation

Microcredits provide finance to the poor (Karlan & Zinman, 2008). Arguing from a construc-tionism perspective lets us believe that poverty exists, because of the system that we have cre-ated. Hence, the poor have the same abilities than the non-poor but live in a resource constraint environment which hinders them to change their situation. Yunus, for example, claims that making finance accessible to the poorest motivates them to invest in small enterprises and helps them to find their own way out of poverty (2003). Furthermore, he suggests that everyone is entrepreneur, and everyone (especially women) should be able to borrow money to create busi-nesses which then solve societal problems.

Anderson et al. (2002) argue that microcredits change the external environment in which small business owners operate by breaking constraints and generating new inclusions. In fact, micro-credits provide access to (additional) capital (Creighton & Omari, 2000) to help households starting or expanding small businesses (Banerjee, Duflo et al., 2015). Ideally, this creates extra

(16)

income and helps small business owners to provide a better living for them and their families (Microcredit Summit, 2018b).

(2) Empowerment of women

In the traditional approach, microcredits not only seen as a way to support business creation, but also enhance women’s empowerment. This is why until now many MFIs, and also the Grameen bank only provide microcredits to women. (Yunus, 2003; 2004). Accordingly, in 2013 82.6%, or more than 94 million microcredit customers were women (Microcredit Summit, 2018a). The social reason behind this is that until now, women (in Bangladesh) still have less rights than men (Yunus, 2003). Providing microcredits to women gives women the ability to control capital. This increases their bargaining power which ideally reduces gender inequality. Economically speaking, there are also reasons for supporting women over men. In Sub-Saharan Africa, where most of the LDCs are located, female owned businesses significantly show a lower performance level than men owned (Campos & Gassier, 2017). One reason for this under-performance is, that women are mostly held responsible for managing the household besides being the main business owner. Balancing family, business, and personal needs leads to trade-offs which then often leads towards neglecting economic goals (Dobbs & Hamilton, 2007). Furthermore, women are said to have characteristics that improve the impact of microcredits, such as the ability to manage scarce resources, a greater long-term vision, and a higher tendency to repay (Yunus, 2004).

2.1.2 Alternative financing sources for small business owners in LDCs (1) Social networks

LDCs show generally a lack of resources and weak institutions (Khavul et al., 2009). This is why the main source for capital in LDCs is the business owner’s social network, such as his family or friends (Banerjee, 2013; Daniels et al., 2016). Social networks offer support, bridge the lack of financial capital and formal regulations, and play an important role in encouraging entrepreneurship and economic survival (Adler & Kwon, 2002; Khavul et al., 2009).

Capital provided by social networks comes with fewer transaction fees and regulations com-pared to more formal alternatives (Edelman et al., 2016) and family members in LDCs have

(17)

usually more privileges, compared to Western families. Even distant relatives have the right to receive financial support, employment, or other resources such as a share of one’s property, if they ask for it (Smith, 2009).

However, relying completely on ones’ social network is not an ideal solution. It restricts small business owners to the financial means that are available in their close environment. Further-more, it forces them to spend lots of time to establish and maintain their relationships (Adler & Kwon, 2002). In addition to this, borrowing from relatives and friends can create tensions, and pressure, especially when a small business owner borrowed money and is unable to repay.

(2) Savings groups

Financial pooling systems, such as rotating savings and credit associations (ROSCAs) are a group of 10-14 people who decide to save money together. Saving groups function without an institution. The members contribute regularly (often daily) a small amount of money, which is collected by one of the group members and in rotation, one member at a time receives the collected money in form of a credit. The accumulated amounts are small, even by local stand-ards. This is why members of savings groups perceive this form of saving as a game, and also call it a “lottery”.

Saving groups exist, because they help the individual to save money instead of spending it immediately. The main advantage is that members keep their autonomy and can freely decide how to spend the accumulated money once it is their turn. Furthermore, participants can manage the commitment amount themselves and choose the right scale. Lastly, group members do not pay interest rates.

One of the disadvantages of savings groups is that they are based on trust between members; thus, the success depends highly on the participants. The participants themselves decide the rules, for example if and how much money is collected. The biggest risk for savings group members is that money collectors keep the accumulated money instead of paying it out. (Creighton & Omari, 2000).

(18)

2.2 Performance of small business owners

2.2.1 Defining the small business owner

Cantillon (1755) established the research field of entrepreneurship. In his research, he divides economic actors into land-owners, wage workers and entrepreneurs and sees the latter ones as individuals who engage in arbitrage and bear risk while having no influence on supply and demand. Kirzner (1974) gives entrepreneurs as the ability to recognize new economic opportu-nities and Schumpeter (1934) even sees entrepreneurs as the key figure in the economy by prescribing them the ability to create creative destruction, innovations and growth. In short, entrepreneurs identify, recognize and exploit opportunities (Shane & Venkataraman, 2000; Zahra and Dess, 2001) to create economic wealth (Schumpeter, 1934).

Subsistence entrepreneurs create innovations and welfare (Morrison et al., 2003) and therefore contribute to the development of LDCs. They act within the subsistence economy, which shows a general lack of resources, and consists of around 2.47 billion people who earn less than $2 per day (Bruton et al., 2013; Viswanathan et al., 2010). Subsistence entrepreneurs act within the informal economy (Van der Sluis et al., 2005) with the main goal to provide a living for their family (Khavul et al., 2009).

The informal economy is the main economic form in LDCs and shows “weak enforcement of formal laws and regulations” (Webb et al., 2009, p. 504). It consists of “the paid production and sale of goods and services that are legitimate in all respects besides the fact that they are unreg-istered by or hidden from the state for tax and/or benefit purposes” (Williams & Nadin, 2010, p. 363).

Throughout this paper we perceive the small business owner as “an individual who establishes and manages a business for the principal purpose of furthering personal goals. The business must be the primary source of income and will consume the majority of one’s time and re-sources.” (Carland et al., 1984, p. 358). These business owners operate MSEs which are “poorly managed, sometimes temporary, less productive, [if compared to small and medium sized en-terprises,] and undercapitalized” (Kiggundu, 2002, p. 248).

(19)

2.2.2 Performance of the small business owner

The dependent variable of this study is business performance. Until now, scholars have not agreed upon a standard measure for business performance (McKelvie & Wiklund, 2010; Van der Sluis et al., 2005; Wiklund & Shepherd, 2009). Van der Sluis et al. (2005) have for example examined 129 entrepreneurial studies in the context of developing countries and found that 70 of them use self-employment earnings as performance proxies (measured by earnings, income, or profit), 21 of them inputs (typically measured by employment) and 19 of them duration or survival of the firm.

To unravel the complex construct of business performance, we need to go back to Venkata-raman and Ramanujam (1986) who developed a model to measure business performance. This model suggests taking financial and nonfinancial (operational) perspectives into account when measuring business performance; consequently, scholars now largely perceive performance as a multi-dimensional construct (Murphy et al., 1996; Wiklund et al., 2009).

Growth (i.e. change in sales, change in employees, change in income) is besides efficiency and profit one of the most used dimensions to assess business performance (Murphy et al., 1996). Previous studies found that it is a very common (Wiklund & Shepherd, 2009) and the most reliable and valid business performance measure (Delmar, 2006). Likewise, business perfor-mance and growth are often used interchangeably (Wiklund, 1998).

The discussion how to measure growth is one of the core topics in entrepreneurship theory (Shane & Venkataraman, 2000). Accordingly, different perspectives exist about how to meas-ure it (Wiklund, Patzelt & Shepherd, 2009). Murphy et al. (1996) advise scholars to use multiple measures, because interrelationships between measures can exist. That is, one positive predictor of a growth measure can be a negative predictor of another growth measure (McKelvie & Wiklund, 2010; Murphy et al., 1996).

In addition to this, we need to take the data characteristics as well as the theoretical perspective into account (Delmar, 2006). Conducting an empirical study in the informal economy faces the constraint that small business owners usually do not keep accounts; but instead, they memorize their expenses (Creighton & Omari, 2000). Thus, data that tracks their performance is often not available or of low quality (Young, 2012) which means that traditional numerical indicators

(20)

such as growth rate, profit margin or number of innovations are not suitable performance measures in this context (see, for example, Wiklund & Shepherd, 2003).

Data on sales is for example usually underreported and of poor quality, whereas data on labor supply is relatively reliable (Dupas & Robinson, 2013). Since small business owners are said to create jobs (Dobbs & Hamilton, 2007), measuring growth via the increase of employment might be the right approach. But on the other hand, the likelihood that small business owners create jobs is rather low since they operate their business themselves with the initial goal to make a living (Khavul et al., 2009). They rather aim to increase sales than employment (Da-vidsson & Wiklund, 2000).

Evidently, the one, right measure that captures all the information does not exist. Therefore, most scholars recommend using multiple growth indicators to obtain richer information (Wiklund et al., 2009). This is the reason why we investigate business performance with six common growth indicators, derived from the literature:

1. growth in sales, which is seen as most common or effective measure for growth (Cam-pos & Gassier, 2017; Davidsson & Wiklund, 2000; Duvendack et al., 2011; McKelvie & Wiklund, 2010; Morrison et al., 2003; Wiklund et al., 2009)

2. employment growth (Campos & Gassier, 2017; Dupas & Robinson, 2013; McKelvie & Wiklund, 2010; Morrison et al., 2003; Wiklund et al., 2009)

3. profit growth (Campos & Gassier, 2017; De Mel, McKenzie & Woodruff, 2009; Van der Sluis et al., 2005)

4. business supplies growth, a common practice among small business owners is for ex-ample, to invest in ingredients after harvesting season when prices are low, and stock the ingredients until needed (expert interview Dr. Albogast Musabila, Appendix A). 5. business investments growth (Dupas & Robinson, 2013; Shepherd & Wiklund, 2009;

Stewart et al., 2010), and

6. income growth (Van der Sluis et al., 2005).

2.3 The impact of microcredits on small business performance

The findings about the effect of microcredits on business performance are still controversial. This goes in line with Hulme (2000) who investigated the effects of microcredits and found on the one site studies that find microcredits to have positive economic and social impacts while

(21)

others find negative impacts. Reasons for these contrary findings are probably the diversity of the sector (Banerjee, 2013; Duvendack et al., 2011) and the high dependency on the context (Brau & Woller, 2004). Pitt & Khandker (1998) were the first ones who evaluated the impact of microcredits and found a positive impact on household consumption. Both, supporting and opposing studies followed while previous studies have become more careful in their perceptions and found no, or only moderate impact (Berge et al., 2015; Demirgüç-Kunt et al., 2017). Arguing from the resource-based perspective a larger microcredit size increases capital access and small business growth (Davidsson & Wiklund, 2000; McKelvie & Wiklund, 2010). How-ever, Karlan & Zinman (2009) argue that expanding access to capital can also harm the small business owner because of overborrowing (i.e. borrowed more than is sustainable) or unpro-ductive lending (i.e. no increase in business output).

Moyo (2009) finds that longer lending periods of loans help to reduce poverty. This goes in line with Karlan and Zinman (2008) who found that maturity (the time to repay a loan) has a higher impact on microcredit demand than the interest rate. Shorter maturity means higher (monthly) repayment amounts while longer maturity means the repayments are lower, because they are spread over a longer period. Therefore, longer maturity is preferred by many borrowers which comes with a longer loan duration.

Duvendack et al. (2011) investigated 2643 microfinance impact articles, 58 of them in detail and found that overall microcredits increase business profits; hence, increased income and con-sumption. But they found also evidence for business losses because of inability to repay result-ing in a decrease of income and consumption.

Until now, the impact of microcredits in a rural Tanzanian setting is rarely investigated. Berge et al. (2015) gave capital (80 dollars) to microentrepreneurs in Tanzania and found only little, non-significant effect on business performance. They investigated men and women separately. For men they found negative influence on sales and living conditions and positive impact for profits and happiness. For women they find positive impact on all variables except short-term sales and profits. Overall, they find that loans improve business performance.

(22)

Studies that used an established research method, namely randomized controlled trials, failed to find proof that microfinance alleviates poverty (Banerjee, Karlan et al., 2015; Karlan & Zin-man, 2009) but promote the overall perception that microcredits have a “modestly positive” impact (Banerjee, Karlan et al., 2015, p.1; Brau & Woller, 2004; Karlan & Zinman, 2009). With this said, the hypotheses are as followed:

Hypothesis 1 (H1): Microcredits positively influence the business performance of small business owners in the informal economy in rural Tanzania.

Hypothesis 2 (H2): The duration of a microcredit positively influences business perfor-mance of small business owners in the informal economy in rural Tanzania.

Hypothesis 3 (H3): The size of a microcredit positively influences business performance of small business owners in the informal economy in rural Tanzania.

Figure 1. The theoretical model.

3 Methods

3.1 Research Design

The concept of microfinance has existed since 1977 (Yunus, 2003). Since then, hundreds of articles have been published on this topic (Brau & Woller, 2004). Nevertheless, research on micro-insurance and micro-saving is still in a nascent state (Banerjee, 2013) while microcredits are the most common microfinance products that have also received most research attention in the past (Brau & Woller, 2004). This implies that, the subject of microcredits is no longer in a

(23)

nascent state. Accordingly, in 2004, Brau and Woller already argued that research on micro-credits had matured from a nascent state but has not reached the state of maturity yet.

The findings on the effect of microcredits vary from study to study, suggesting that they de-pendent on the context of investigation (Brau & Woller, 2004). Until now, there are only few microcredit studies in an African context available, which is why investigating the effects of microcredits in a rural Tanzanian setting is considered a nascent research field.

The informal economy receives research attention since the International Labour Organization introduced the term “informal sector” in 1972. Back then the informal economy was perceived as not very productive and micro- and small enterprises (MSEs) tended not to make any signif-icant contribution to enhancing economic growth (Van der Sluis et al., 2005). In 1980 this per-ception changed. Extensive household surveys unraveled that small businesses significantly contribute to household income, and that some are even able to scale-up (Van der Sluis et al., 2005). From this point on, researchers began to investigate the performance of small business owners in the informal economy (Van der Sluis et al., 2005).

Until now, no general consensus on the effects of microcredit has been reached and only minor research is available in the rural Tanzanian context. Therefore, we conclude that research on this subject is in an intermediate state. Consequently, this study applied a hybrid approach by combining a quantitative main study with a qualitative pre-study (Edmonson & McManus, 2007).

The purpose of the qualitative pre-study was to get a deeper understanding of the local context and shape the questionnaire for the main study accordingly (Edmonson & McManus, 2007). Therefore, we conducted semi-structured interviews with 18 experts chosen based on conven-ience sampling. The interviewees had expertise in the (micro-) finance sector, entrepreneurship, in the local context or in conducting qualitative studies with small business owners in Tanzania (for interview summaries, see Appendix A).

In order to assess the effect of microcredits on business performance, we used a control group method. As shown in the simple impact model in Figure 2, we investigated the business perfor-mance of small business owners who have received a microcredit in the past 12 months (treat-ment group) and compared the scores with the ones who have not received a microcredit

(24)

(con-trol group). This method allowed us to assess the effect of the microcredit because the micro-credit provided to the treatment group modifies their business behaviors and practices which then leads to decreased or increased business performance (Hulme, 2000). Ideally microcredit program participants only differ from non-participants in terms of having obtained a micro-credit, so that the difference in business performances (impact) can be explained through the micro-credit (Hulme, 2000).

Figure 2. Model of the impact chain (adapted from Hulme, 2000).

We used a non-experimental form of comparative design and opted for a large sample size of n>100 to achieve high power (Stevens, 2009). We targeted the owners who are typically also the managers (Dobbs & Hamilton, 2007), because they are the ones who create the business outputs and can therefore also report them best (Morrison et al., 2003). For our study popula-tion, we defined business owners as those whose business 1) is their primary source of income and 2) will consume the majority of their time and resources (Carland et al., 1984). The re-spondents self-reported their business performance on a five-point Likert scale between their current standing and where they were twelve months ago.

3.1.1 Challenges of the research method

Many studies use the control group method to assess the effect of microfinance, but this also brings some challenges that we must consider when assessing the results (Hulme, 2000).

Business owner Behaviors and practices over_{the past 12 months} _performanceBusiness

The difference between the outcomes is the impact of the microcredit Business owner Modified business performance Behaviors and practices over

the past 12 months Mediating Processes

Microcredit

Mediating Processes

(25)

Firstly, the method may cause a selection bias concerning the location. This means that the eco-nomic, physical, and social environment of the control group might not match with that of the treatment group (Hulme, 2000). In this study setting, we do not see this being an issue. The treatment and control group participants both operated in the same geographical area with their businesses next to each other oftentimes. Thus, there was no visible difference concerning the location and therefore, we exclude a selection bias concerning the location.

Secondly, the treatment group could possess a nonvisible attribute which the control group lacks and we neither observe nor control for (Hulme, 2000). For example, attributes such as entrepre-neurial motivation are likely to affect business performance (Eijdenberg, 2016) and could bias our results as they are neither observed or controlled for (Hulme, 2000).

Finally, the control group method loses its functionality when the treatment group does not use the microcredit for business expenditures, but rather for personal consumption. This would bias our results and is a challenge to control (Hulme, 2000). For this reason, we have included ques-tions about microcredit spending, but we cannot judge if all respondents told us the truth. As a result, we cannot exclude this bias.

3.1.2 Validity and reliability

For the quantitative main study, we developed a questionnaire to measure predetermined indi-cators and test the hypotheses (Appendix C). The questionnaire is based on the existing litera-ture (see Chapter 2) and refined with the suggestions made by the experts (Appendix A). With the help of research assistants, we translated the questionnaire from English to Swahili. To check lingual validity (equivalence of the questionnaire), the questionnaire was translated back to English again. Furthermore, to ensure that questions, instructions, and scales were clear and not too sensible to answer we conducted a pilot study (Saunders et al., 2009) with eight food vendors in a similar setting than the main study: the Sabasaba market in Morogoro town. Based on the feedback, we adjusted the questionnaire accordingly. Piloting showed, for example, that providing information about daily revenue was not considered a sensible topic, but some food vendors were not able to recall this information, while all fruit vendors were. As a result, we included two answering options that all food vendors remembered: ‘number of dishes sold per day’ and ‘amount charged per dish’ to our questionnaire and calculated the daily revenue based on this.

(26)

For this study, we applied a convenience sampling technique. Furthermore, we focused on one industry and one geographical region which is why external validity is limited. However, we reached out to all food vendors on the customer list of our partnering MFI SUBO (see, Section 3.2.2) which made our sample a quota sample showing higher internal validity than randomly selecting food vendors in the region.

To check reliability of the business performance scale, we calculated the Cronbach’s alpha in-dicating good internal consistency (see, Section 4.2).

3.2 Research context and data collection

3.2.1 Tanzania

Despite the known difficulties of collecting data in the African context (Kolk & Rivera-Santos, 2018), this study takes the challenge to investigate the impact microcredits in rural Tanzania to gap the lack of studies in this region.

3.2.1.1 Economic situation

Tanzania is one of the LDCs in the world. Currently, 68% of its 54 million inhabitants live be-low the poverty line (Central Intelligence Agency, 2018). Tanzania has an annual population growth of around 3% as well (World Bank, 2018b) which adds another challenging dimension when aiming for sustainable economic growth. Nonetheless, the overall economic outlook for Tanzania is positive. Collier (2008) even describes Tanzania as the most hopeful low-income country. This is in line with the Gross Domestic Product (GDP) growth in the past decade of averagely 6% per year and the many natural resources, such as gold and gas that the country has access to (UNESCO, 2018).

However, compared to the previous years, in 2017 Tanzania’s real GDP growth rate has slowed down. According to government data from 7.3% in Q1-Q3 2016 to 6.8% in the same period in 2017. Also, the inflation rate decreased from 4.2% in 2016 to 2.5% in 2017 (World Bank, 2018b). The reasons for the current GDP decrease are 1) on the supply side the slower growth of services, and 2) on the demand side the slower expansion of consumption and investments (World Bank, 2018b). Additionally, Tanzania still shows a lack of foreign investments because it has not managed yet to create macroeconomic stability (World Bank, 2018b).

(27)

These numbers give an indication for the current economic situation but should also be noted with care as official reported numbers in an African context come with uncertainties (Young, 2018; expert interviews, Appendix A).

Recently, Tanzania has significantly transformed to a market economy by switching focus from agriculture (23% of GDP) to service (48% of GDP) and industry (29% of GDP) (Central Intel-ligence Agency, 2018). The industrial sector still does not play a major role nor employ a sig-nificant portion of the population since it only deals with small values, such as food, beverages and tobacco and is threatened by increased imports, mainly from China (Lyons et al., 2014). As a consequence of limited wage employment opportunities, many Tanzanians own businesses (Campos & Gassier, 2017; Van der Sluis et al., 2005). Low-income sectors, such as food, com-merce and textiles are dominated by low-educated women (Morris et al., 2006; Smith, 2009).

3.2.1.2 Political situation

Tanzania is a presidential republic, with the latest elections in late 2015. These resulted in the

5th_{government with president led by Magufuli (Central Intelligence Agency, 2018). The current}

government plans to create a better business environment and therefore has been investing in infrastructure, providing increased access to finance and supporting education. This is captured in the second Five-Year Development Plan (2016/17-2020/21) aiming to transform the econ-omy through industrialization and human development. Furthermore, the government aims to strengthen public finance management to create a more attractive and predictable business en-vironment (IMF, 2017). This should attract more foreign investors and reduce their concerns about recent governmental moves. In 2016/17, for example, the government decided to increase liquidity and support the private sector with more credit (IMF, 2017).

3.2.1.3 Mvomero district

Around two thirds of the Tanzanians live in rural areas (Central Intelligence Agency, 2018), which explains why the study was conducted in a rural setting as well, at the villages around Mzumbe University in Mvomero district (Bigwa, Changarawe, Fire, Gulioni, Kikundi, Kipera, Konga, Mikongeni, Mkuyuni, Mlali, Mzumbe, Osterbay, Uswazi juu, Uswazi chini, Vikenge,). The Mvomero district in 2012 counted 312,109 inhabitants. Hereof, 276,447 lived in rural and 35,662 people in urban areas respectively (National Bureau of Statistics, 2016). The villages around Mzumbe university count according to the local government of the Changarawe village

(28)

around 6,025 in-habitants and 1,614 households; 822 of them are farmers and around 200 are food sellers (see expert interview, Appendix A). The economic situation is described by the authorities as “not that good” due to a lack of income sources, bad weather conditions and governmental changes in 2015. Furthermore, we found it remarkably that most of the villagers were not born in the region but instead moved there, attracted by the Mzumbe University and its increasing number of students, which represent the main customers group for the food ven-dors.

3.2.1.4 The food sector

This study selected a narrowly defined industry to achieve sensitivity to context (Davidsson, 2004); that is, the informal food sector in Mvomero district, Tanzania.

We chose to investigate small businesses in the food sector, because they are among the most present microcredit borrowers (Banerjee, 2013). In accordance, local experts recommended us to focus on businesses in the food sector also because these types of businesses occur most frequent in Mvomero district. As food vendors we considered two types of businesses: 1) the fruit and vegetable vendors, and 2) the so-called mama lishes who prepare and sell meals “to make a living and support the extended family” (Creighton & Omari, 2000, p. 334).

The mama lishe usually own basic cooking equipment, tableware, a bucket to collect water, a stove, and sometimes some chairs or a table for customers. They need money to buy water, fuel (charcoal or firewood) or ingredients (such as milk, meat, tomatoes, onions, tea, sugar, rice or beans). If there is not enough money, they might go to a seller they know and receive a credit until they are able to pay. Their main tasks include preparing and cooking food, childcare, serv-ing food, washserv-ing dishes, and cleanserv-ing up after customers have left. One of their daily chal-lenges is to decide how much food to buy and prepare. Having too much and not being able to sell it harms her income, as well as running out of food when there are lots of customers. (Creighton & Omari, 2000).

The fruit and vegetable vendors buy fruit and vegetables from different farmers and sell them at their market stall. Usually each farmer provides only one kind of fruit or vegetable. Some fruit sellers prepare the fruit before they sell it, they cut for example pineapple or mangos or open the coconut once requested by the customer, but most vendors solely resell without any additional service (see also expert interview Kasanda, Appendix A).

(29)

3.2.1.5 The financial sector

The financial sector in Tanzania has expanded in the past years and competition among foreign owned-banks (48% of banking industry’s total assets) increased efficiency and quality of finan-cial services (Central Intelligence Agency, 2018). The interest rates are still relatively high, reflecting high fraud risk (Central Intelligence Agency, 2018).

Also, the microfinance sector gains importance in Tanzania. In 2000, Tanzania launched its first National Microfinance policy (NMP) (NMP, 2017). One year later, in 2001, the Tanzania Association of Micro finance Institutions (TAMFI) was founded, which has now 99 member institutions, such as commercial, community and microfinance banks, NGOs, private MFIs and SACCOS who have together reached out to 1.2 million MSEs already (TAMFI, 2018).

The first NMP aimed to increase local and foreign microfinance service providers; subsequently also the 2017 NMP promotes the further growth of the microfinance sector but also provides regulations for customer protection. Furthermore, the NMP contributes to increasing the level of financial inclusion in the country, to “meet the real needs of the low-income population”. In addition, it states that the “microfinance sub-sector will immensely contribute to the generation of employment and raising the incomes of the low-income population” resulting in poverty reduction and economic growth (NMP, 2017).

In Mvomero district, microfinance is increasing in popularity. Our interview partners from the local government of Changarawe (Appendix A) estimate that around 50% of the villagers make use of microcredits; about 70% of the agriculture sector depend on microcredits and almost 100% of the food sellers. Despite large interest rates (10-25% per month), microfinance is a popular support for small businesses. The experts from the local government believe that many villagers are dependent on microcredits and would not operate their businesses without them, 30% of all microcredit borrowers take multiple loans.

3.2.2 SUBO Financial Consultant and General Supplies Ltd (SUBO)

We cooperated with the MFI SUBO Financial Consultant and General Supplies Ltd (SUBO), founded in September 2017 and located in Vikenge village (20 km northeast of Morogoro town). SUBO is a privately-owned for-profit MFI, operates only locally and counted until April 2018 205 customers, 28 of those have dropped out of the microcredit program. Since September

(30)

2017, SUBO has handed out a loan sum of 1,850,000 TZS, received a repayment sum of 2,330,000.00 TZS and made expected profits of 615,000 TZS. SUBO counted in April 2018 three full-time employees who managed all operational tasks and two founders who were mak-ing the strategic decisions. For this study, we interviewed two employees and one of the found-ers (see, Appendix A).

In Mvomero district, SUBO is the only MFI present with an office in Vikenge village. Other MFIs offer their services in the area as well, but have their offices in Morogoro town, located around 20 km northeast (for a list of MFIs in the area, see Appendix A). According to the mixed market website which lists MFIs worldwide, the four biggest MFIs in Tanzania are Access Bank – TZA, Akiba, Finca, and Brac (MIX Market, 2018). However, in rural areas 34.9% of the labor force are still excluded from accessing financial systems and penetration of banks and MFIs is low, because of poor infrastructure and low and irregular income of the rural population resulting from involvement in seasonal or low productivity agriculture activities (TAMFI, 2018).

3.2.3 Microcredit information

To receive a loan from SUBO, candidates must fill in a loan application form. Most small busi-ness owners are not used to fill in forms which is why they often take the forms home or request help from a SUBO employee. Filling in the form then takes around two hours. The form states the loan conditions and a repayment plan while asking for personal information, collateral, the contact details of up to four guarantors, and an explanation for how the money will be used. The application form is only valid with a signature and a finger print. In addition to this, it requires a passport photo, and a confirmation from the local government which states that the specified collateral really belongs to the owner. According to SUBO, around 80% who apply for a microcredit also receive one. The decision if and when people receive a microcredit is mostly dependent on the willingness of SUBO to increase its capital rather than on the candi-dates’ characteristics. Only some applicants get rejected because they had in the past not repaid on time or had too little collateral to show (Appendix A).

Officially, SUBO states in its lending policy and procedure manual that loan duration can be one month, 3 months, 6 months or up to one year and the amount that is borrowed can vary from 50,000 to 1 million TZS. However, in practice, most customers of SUBO receive a loan

(31)

sum of 200,000 TZS which converts to 88.34 USD (07/2017) and need to repay 270,000 after the duration of three months (for repayment schedule, see Appendix A).

3.2.4 Data collection

The data collection took place in April and May 2018. Because most food vendors have diffi-culties with filling in forms; thus, would require a long time to fill in a questionnaire (see, expert interviews, Appendix A), we worked together with local research assistants who were fluent in Swahili and English and orally surveyed the respondents one to one while capturing the re-sponses on the paper-based questionnaires.

The response rate was very high (97%), 118 out of 122 participated in the questionnaire. Rea-sons for the non-participation were insecurities, not understanding the purpose of research, fac-ing sensitivity issues on the financfac-ing part or havfac-ing been too busy durfac-ing the time of observa-tion. We got the perception that for most business owners the curiosity of participating in an interview with a western researcher compensated for spending 30 minutes to one hour in par-ticipating in the study. Nevertheless, to compensate for the time, we handed out a postcard for each survey that was completed.

The main challenge of the data collection was finding sufficient microcredit users. From the experts we learned, that microcredit customers might be too embarrassed to admit borrowing money (see, Appendix A). For this reason, we worked together with the MFI SUBO. From SUBO’s list of 205 customers, we first sorted out the ones who do not operate in the food sector. In fact, we read through all paper-based loan application forms one-by-one resulting in having 41 customers left who operate as food vendors (while this was still the biggest customer group). Next, we bundled the customers based on their residence and organized meetings with them. In the first round, we called the ones who live nearby SUBO’s office and asked them to come into the office. This worked well only for some respondents. Reasons for others to not make it to the office, was having been too busy, not having money or means for transportation or the heavy rainfalls that we faced every day. As a consequence, we had to find the remaining customers at their workplaces or houses which was only possible with the help of a SUBO employee. The pitfalls of working together with an employee of the MFI were that it can cause biased answers about loan usage and repayment. Nevertheless, for some questions, we found it beneficial to have an MFI employee with us; they checked for example if the amount borrowed and amount that needs to be repaid was stated correctly.

(32)

In addition to reaching out to current, active customers, we followed the call from Karlan (2001) to also include previous customers who dropped out of the microcredit program to our respond-ents. This allows us to receive a more accurate picture of the microcredit effects and lowers biases for a number of reasons. Firstly, assessing the effect of microcredits based on respond-ents who remained in the program would lead to overestimating the effects. The current pro-gram participants most likely benefited from the microcredit while the ones who have not ben-efited have probably left the program already (Karlan, 2001). Secondly, we included dropouts because those could be the highest performing business owners having left the microcredit pro-gram because of better financing options. Thirdly, by including dropouts we aim to have equal control and treatment groups. The control group consists of both customer groups, the ones who will remain and the ones who will drop out. Hence, to have equal groups, the drop out customers must be included in the treatment group as well (Karlan, 2001).

Our control group consists of small business owners who have never applied for a loan and who have applied for a loan but not received yet. We received the latter ones as respondent group with the help of SUBO. Just before the SUBO employees paid out the first microcredit to a number of new customers, we conducted our survey.

In addition to working with SUBO, we randomly selected food vendors on two food markets nearby the campus of Mzumbe University, and one market in the village of Changarawe. To balance the sample, we asked for socio-demographics, firm characteristics, and microcredit specifications in the questionnaire, because we noticed for example that the small businesses close to Mzumbe University performed better than the businesses in the rural villages.

3.3 Operationalization

The challenge of measuring performance lies in unraveling its complex nature and operation-alizing it in appropriate, measurable terms (McKelvie & Wiklund, 2010; Morrison et al., 2003; Wiklund et al., 2009).

This study uses growth as performance indicator. Growth is a process that changes over time (Davidsson & Wiklund, 2000); it is usually expressed in a growth rate and measured as the outcome between two different points in time (Levie & Lichtenstein, 2010; McKelvie & Wiklund, 2010). Therefore, measuring growth requires to define an appropriate time frame.

(33)

In the literature there is no agreement over a standard time frame (Wiklund & Shepherd, 2009), while the most common time frames are a 1, 3, or 5-year period (McKelvie & Wiklund, 2010). We had to consider that 1) SUBO handed out its first loans in September 2017, 2) small business owners must be able to recall the information over the chosen period, and 3) the impact of microcredits must be visible within this period. To combine these three aspects and also in-crease comparability to previous studies which often used a one-year period to measure micro-credit impacts (see, for example, Banerjee, Duflo et al., 2015; Dupas & Robinson, 2013) we decided measuring growth compared to 12-months ago.

Performance can be measured based on basis of the business or the individual, the small busi-ness owner (Davidsson & Wiklund, 2000; Van der Sluis et al., 2005). This implies that one must decide if following the firm or the individual when measuring growth. In small businesses, the owner often gets support from family or friends (Khavul et al., 2009). This makes a distinc-tion between the owner’s contribudistinc-tions and those of the household members that make up the performance of the venture very difficult which is why we focus for this study on the perfor-mance of the small business rather than the individual.

To increase robustness and best assess business performance, this study uses several growth measures (see also Section 2.3) and also include control variables (Murphy, 1996). In a nutshell, the growth measures are 1) sales growth, 2) employment growth, 3) profit growth, 4) business supplies growth (e.g. food ingredients), 5) business investments growth, and 6) income growth.

Figure 3. The theoretical model including business performance measures.

Microcredit (MC)

Business performance (growth) Sales Employment Profit Business supplies Business investments Income Total BP score Duration MC Size MC H1 H2 H3

(34)

As controls serve the socio-demographics (Morrison et al., 2003), household, and business characteristics. For the socio-demographics, firstly, we included gender, because previous stud-ies recommend to study men and women differently since they face different constraints and act on different opportunities (Van der Sluis et al., 2005). Secondly, we included age, because it showed the biggest effect in explaining small business performance in rural Tanzania (see, for example, Eijdenberg & Borner, 2017). Finally, as a third control variable, we added educa-tion, because it influences performance (Dobbs & Hamilton, 2007). We measured education by the help of a dummy variable which indicates the highest schooling level obtained. The reason for using this measure compared to years of schooling, was that for the small business owner this was easier to recall (to increase comparability to other studies, Appendix B describes years of schooling for each schooling level). Further controls include the current revenue on a daily basis, to control for the current performance level and firm size (Davidsson & Wiklund, 2000). Lastly, the age of the business was measured by asking the time of business start-up (Davidsson & Wiklund, 2000).

Another item in the questionnaire, was meant to serve as control, namely since when the par-ticipants make use of microcredits. Since this question did not deliver enough valid responses it was not used as a control. Also, we collected data on loan usage, repayment and financial sources to understand if borrowers really invested in their business, if overborrowing is an issue or if microcredits provide the owner access to additional capital (Karlan, 2001).

(35)

Table 1. Operationalization of key constructs. Ob je ct iv es / r es ea rc h q u es ti o n / h y p o th es es Co n ce p t In d ic a to r It em Qu es ti o n n a ir e Or ig in o f it em Pe rs o n al c h ar ac te ri st ic s So ci o d em o g ra ph ic s Ge n d er Q1 Va n d er S lu is et a l. , 2 0 0 5 Ag e Q2 Ba n er je e, D u fl o e t al ., 2 0 1 5 Hi g h est co m p let ed ed u cat io n Q4 Ba n er je e, D u fl o e t al ., 2 0 1 5 ; Va n d er S lu is et a l. , 2 0 0 5 Ho u se ho ld i n fo rm at io n Ho u se ho ld co m po si ti o n Pe o p le w h o d ep en d o n i n co m e Q5 Ba n er je e, D u fl o e t al ., 2 0 1 5 ; We b b e t al ., 2 0 1 3 Ro le i n h o u se h o ld Q6 Bu si n es s ch ar ac te ri st ic s Bu si n es s sp ec if ic at io n s Own er o f th e b u si n ess Q7 Ca rl an d e t al ., 1 9 8 4 , p . 3 5 8 ; Ba n er je e, D u fl o e t al ., 2 0 1 5 ; Nu m b er o f b u si n ess es o wn ed Q8 Ba n er je e, D u fl o e t al ., 2 0 1 5 Ag e o f b u si n ess Q9 De M el , M cKe n zi e & W o o d ru ff , 2 0 0 9 , p . 1 9 Bu si n es s re g is te re d (i n fo rm al e co no m y ) Q1 0 De M el , M cKe n zi e & W o o d ru ff , 2 0 0 9 Bu si n es s si ze Nu m b er o f em p lo ye es Q1 1 , Q1 2 Bo so , S to ry & Ca d o g an , 2 0 1 3 ; V an P ra ag , 2 0 03 Da il y r ev en u e Q1 3 Ba n er je e, D u fl o e t al ., 2 0 1 5 Te st in g t h e h y p o th es es Re se ar ch q u es ti on : T o e x p lo re th e re la tio n sh ip b etw ee n m ic ro cr ed it s an d bus in es s pe rf or m anc e of s m al l bus ine ss ow ne rs i n rur al T anz ani a. Mi cr o cr ed it (M C ) Þ bus ine ss pe rf or m anc e (H 1) Du ra ti o n MC Þ bus ine ss pe rf or m anc e (H 2) Si ze MC Þ bus ine ss pe rf or m anc e (H 3) Mi cr o cr ed it sp ec if ic at io n s Di st in ct io n c o n tr o l/ o b se rv at io n g ro u p Q4 0 Ka rl an , 2 0 0 1 Si ze o f m ic ro cr ed it Q4 1 Ba n er je e, D u fl o e t al ., 2 0 1 5 ; A n d er so n e t al ., 2 0 0 2 Ty p e o f m ic ro cr ed it (g ro u p /i n d iv id u al ) Q4 2 Br au & W o ll er , 2 0 0 4 Du ra ti o n o f m ic ro cr ed it Q4 3 Ka rl an a n d Z in m an , 2 0 08 ; M o y o , 2 0 0 9 In te re st ra te o f m ic ro cr ed it Q4 4 Ba n er je e, 2 0 1 3 ; Br u to n e t al ., 2 0 1 3 In te re st ra te s p ec ifi ca ti o n s (c o n st an t/ n o n -co n st an t) Q4 5 Tr ain in g Q4 6 Ash ta , Kh an & Ot to , 2 0 1 5 ; B ru to n , Ke tc h en & I re la n d , 2 0 1 3 Ty p e o f tr ain in g Q4 7 Ex p er t in te rv ie w s Fi rm p er fo rm an ce (c h an ge of da il y bus ine ss pe rf or m anc e com pa re d to 12 mo n th s ag o ) Sa le s Q1 4 Ba n er je e, D u fl o e t al ., 2 0 1 5 ; D u v en d ac k e t al ., 2 0 1 1 Nu m b er o f em p lo ye d p eo p le Q1 5 Ba n er je e, D u fl o e t al ., 2 0 1 5 ; W ik lu n d e t al ., 2 0 0 9 Pr o fi ts Q1 6 De M el e t al ., 2 0 0 9 , p .2 1 ; Du v en d ac k e t al ., 2 0 1 1 ; Va n d er S lu is et a l. , 2005; Bu si n es s su p p li es ( fo o d i n g re d ie n ts ) Q1 7 Ba n er je e, D u fl o e t al ., 2 0 1 5 Bu si n es s in v es tm en ts ( as se ts ) Q1 8 Ba n er je e, D u fl o e t al ., 2 0 1 5 ; D u p as & Ro b in so n , 2 0 1 3 In co m e Q1 9 Du v en da ck e t al ., 2 0 1 1 Fu rt h er v a ria b le s o f in te re st Mi cr o fi n an ce e n v ir o n m en t Re je ct io n r at e A ppl ie d for a m ic roc re di t Q3 7 Ka rl an , 2 0 0 1 or : Re as o n s fo r n o t ap p ly in g or : Wh y not a ppl ie d fo r m ic ro cre d it Q3 8 Ex p er t in te rv ie w s Ef fo rt o f o b ta in in g cr ed it A ppl ie d at how m any ins ti tut ions Q3 9 Ex p er t in te rv ie w s Re q u ir em en ts f o r m ic ro cr ed it t o c re at e im p ac t Fi n an ci al i n cl u si o n Ac ce ss to fi n an ce : B o rro w ed m on ey fro m Q3 5 Ba n er je e, D u fl o e t al ., 2 0 1 5 Mo n ey s p en d in g Mo n ey s p en t o n b u si n es s v s. o th er opt ions Q4 8 Ba n er je e, D u fl o e t al ., 2 0 1 5 Re p ay m en t Fu rt h er i m p ac t m ea su re mi cr o cr ed it Re p ay m en t tr o ub le s Q4 9 Ex p er t in te rv ie w s; C re ig h to n & O m ar i, 2 0 0 0 Re p ay m en t tr o ub le s sp ec if ie d Q5 0 Ex p er t in te rv ie w s; C re ig h to n & O m ar i, 2 0 0 0 fu rt h er p erfo rm an ce m ea su re Re p ay m en t so u rc e (b u si n es s v s. o th er ) Q5 2 Br et t, 2 0 0 6 ; D u ff y -Tu m as z, 2 0 0 9 ; M o n tg o m er y , 1 99 6

The effect of microcredits on business performance of small business owners in rural Tanzania