Impact assessment of microfinance programmes, including lessons from Khula Enterprise Finance

Abstract

This article discusses approaches to impact assessment of microfinance programmes through a survey of empirical literature and findings of an impact study of Khula Enterprise Finance, a South African wholesale finance institution that facilitates access to financial services by small, medium and microenterprises. The article notes that impact assessment has now opened up to the needs of a mix of stakeholders and a wide range of purposes. There has been a shift from a donor-driven approach to a practitioner-led approach that emphasises learning and improving practice. Nevertheless, findings of an impact study of Khula Enterprise Finance, a characteristically donor-driven type of study, indicate a positive impact on the beneficiaries of microfinance, especially women in rural areas where they are specifically targeted. Furthermore, the impact study shows that lower-income communities in rural areas have benefited less than their not-so-poor counterparts in the urban areas, an observation that is consistent with findings in other studies.

1. Introduction and background

Microfinance institutions (MFIs) that provide financial services to the poor include, among others, donor-supported non-profit non-governmental organisations (NGOs), cooperatives, community-based development institutions (e.g. self-help groups and credit unions), and commercial and state banks. Generally, NGOs and other non-bank financial institutions have provided leadership in developing credit methodologies for the poor and means of reaching large numbers of the poor. Throughout the 1980s and 1990s, their microfinance programmes have increased the range of financial services to the poor. Furthermore, they have shown that the poor can repay their loans and are willing and able to pay interest rates that allow MFIs to cover their costs.

Evidently, microfinance programmes have become one of the most important interventions in developing countries' efforts to reduce poverty. There has been a huge growth of MFIs in terms of numbers and size of organisations, numbers of clients and provision of subsidised donor funding in many developing countries. A large proportion of these organisations include poverty reduction in their mission or objectives. Hence, there is a need to understand and improve the impact of MFIs as a key premise to successful poverty reduction.

Until recently, however, most MFIs did not measure the impact of their work, nor did they establish whether there were ways in which they could improve the impact they had on the clients they sought to serve. Where impact has been measured, it has been donor driven, usually for the purpose of justifying funding. Pitt & Khandker (1996) observe that most comparative studies avoid calculations of poverty impact, arguing that the fact that small loans are being made is proof enough that the poor are being empowered, and the fact that loans are being repaid is proof that incomes have increased. Thus, the market is used as an indicator of impact. However, empirical evidence has demonstrated that financial performance does not measure change in people's lives. Experience in the real world has shown that, in practice, a number of factors lead to seemingly ‘irrational’ decisions by clients. For instance, indebted clients may repay loans even where their businesses have failed because of complex social factors (Cohen & Sebstad, 1999). Therefore, using market proxies would mask the range of client responses and benefits to the financial services.

Recent approaches to impact assessments recognise that MFIs seek to meet the needs of multiple stakeholders and to fulfil a number of different objectives, including the need for credible information to influence donors and other external stakeholders. Necessarily, most impact assessments would include a mix of objectives and approaches. Impact assessment can be done in different ways, depending on the type of information that is required and the purpose for which it is needed (Hulme, 1997; Simanowitz et al., 2000). It is noted that there is no blueprint approach to impact assessment, and there is a place both for simple client monitoring systems and for ‘high-end’ studies that seek to prove causality. Notably, there is a shift from donor-driven impact assessment (usually periodic with a narrow focus) to a practitioner-led approach with a wider scope, which emphasises learning and improving practice.

With reference to an impact study of Khula Enterprise Finance, a South African wholesale finance institution that facilitates access to finance by small, medium and microenterprises (SMMEs), this article reviews empirical evidence and new insights into the evolving state of the art of impact assessment of microfinance programmes. The rest of the article is divided as follows: Section 2 reviews literature and practices, Section 3 discusses a case study of an impact assessment of products of Khula Enterprise Finance, and Section 4 offers some concluding remarks.

2. Literature review

In many developing countries, MFIs have become an increasingly important component of strategies for promoting micro- and small enterprise development in order to reduce poverty. However, empirical evidence of their impact has been inconclusive and controversial. Essentially, there are three strands of evidence. The first strand, considered to be extreme, argues that microfinance has beneficial economic social impacts (e.g. Schuler et al., 1997, among others). The second strand, which is also considered extreme on the opposite end and is associated with Adams & Von Pischke (1992) and Rogaly (1996) and others, cautions against too much optimism by alluding to the negative impacts brought about by microfinance. The third strand is associated with Hulme & Mosley (1996) and Mosley & Hulme (1998). It takes the middle-road stance by acknowledging beneficial impacts, but stating that microfinance does not assist the poorest.

It could be argued that the seemingly conflicting empirical evidence could be attributed to definitional problems, the multifaceted objectives of MFIs, and the methodological approaches to impact assessment. The Imp-Act programme initiated in 2000 observes that impact assessment historically has been donor driven, whereby the emphasis has often been on justifying funding rather than helping organisations learn from and improve their work (Imp-Act, 2000). An Action Research Programme funded by the Ford Foundation recently proposed that the objective of understanding and improving the impact of development finance on poverty would provide new insights.2 This has revolutionised the practice of impact assessment. The questions remains as to what theoretical framework has underpinned this revolution.

Discussing the ‘state of the art’ in impact assessment, Hulme (1997) recommends that approaches to impact assessment should range from ‘proving impact’ to that in which impact assessment should be seen as a process of which the objective is to ‘improve practice’. The writer further argues that the two approaches exist in a continuum or spectrum along which the government, donors, practitioners and researchers can locate themselves, depending on their needs and interests at any particular period of time. This approach naturally renders itself to a multiplicity of interpretations. Conflicts of needs and interests of the different stakeholders tend to abound. Indeed, donors tend to be more concerned with sustainability and outreach to the poor, while practitioners would be more concerned with improving practice.

However, Hulme (1997) observes that the debate has been largely over the rationale, need, practicability and cost-effectiveness of carrying out impact assessment of the ‘proving impact’ variety. He proposes two approaches: the ‘intended beneficiary’ school and the ‘intermediary school’.

The ‘intended beneficiary’ school is basically the traditional project life cycle approach. Essentially, it is donor driven and seeks to justify further funding of programmes. It holds that the impact of aid-funded projects needs to be measured and attributed in order to justify the effect of the intervention through its direct impact on the poor. One way to establish this has been to use the assets and net worth of beneficiaries as categories for documenting the impacts of microfinance. Barnes (1996) examined 32 studies on the impact of microfinance programmes and observed that asset accumulation is incremental, and that successive loans lead to a build-up of enterprise and household assets. In a further study, Barnes & Keogh (1999) found that participation in microenterprise programmes results in improvements in the economic welfare of households, enterprise growth or stability, increased empowerment of women, and strengthened social networks.

Johnson (1998) observes that the ‘intended beneficiary’ approach views credit as a productive input in its own right, having an instrumental impact on improving livelihoods through a combination of raising incomes, ameliorating vulnerability, and reducing or alleviating oppressive credit relations. However, Adams (1988) has questioned the proposition that it is feasible to trace the effect of highly fungible credit through to particular beneficiaries. There is a strong tendency among poor people to use credit for consumption rather than for investment. Also, the ‘intended beneficiary’ school approach is further seen as deficient in the sense that it only considers the effect of credit provision on livelihoods without considering whether the existing informal financial markets are not being undermined by the new interventionist intermediaries undertaking delivery. Essentially, there is an assumption that the existing informal financial markets are exploitative of the poor.

By and large, the debate has centred on the question of causality and attribution, or whether it is possible to demonstrate that the particular intervention of the MFI has led to a specific change. Client livelihoods and the communities in which they live are complex. Clients may have multiple sources of income, and the credit provided by the MFI is fungible and is not necessarily used for the purpose for which it was requested. This makes it difficult and complex to attribute impact to an MFI, particularly given the difficulty of establishing effective control-group mechanisms that can establish a counter-factual (i.e. what would have happened without the MFI intervention). However, some researchers have been able to compare borrowers with a control group on a number of impact variables (Chen & Snodgrass, 1999).

Johnson and Rogaly (1997) observe that impact assessment of NGO-funded financial services places emphasis on measuring changes in income levels following credit programmes. This focus on income promotion is borne out of the definition of poverty as lack of income, vulnerability to income fluctuations, and powerlessness. However, they note that this approach has a number of methodological problems, which include establishing loan use in the presence of fungibility of cash; measuring change over time as a result of an intervention; and proving causality as other changes occur during an intervention.

As regards fungibility, Sebstad & Cohen (2000) observe that, in reality, resources within households are fungible and it is important to recognise that clients use microfinance services for a variety of purposes. Hulme (1997) acknowledges attribution and fungibility to be key problems for impact assessment studies, but argues that the problem of fungibility is not as serious an issue as many writers suggest. This view corroborates with the findings of some empirical studies (e.g. Dunn, 1996; Chen & Snodgrass, 1999; Sebstad & Cohen, 2000).

In contrast, the ‘intermediary school’ approach addresses issues related to the health of the financial intermediary delivering microcredit. It assesses sustainability – both organisational and financial – and evaluates the social benefit of intervention in terms of its outreach to the poor, using the criteria of numbers of users and their poverty profile. Measures of subsidy dependency and some standard banking-type measures of profitability and portfolio analysis are applied in assessing institutions supplying microfinance. Zaman (1998) argues that one of the most difficult aspects is achieving the twin goal of moving towards financial sustainability of an MFI and making an impact on poverty alleviation, especially by empowering women. He observes a growing emphasis on achieving sustainability in the long run, away from subsidies to self-financing.

As regards the empowerment of women, a study by Osmani (1998) on Grameen Bank women beneficiaries shows that access to credit brought only partial improvement in women's relative well-being. He attributed improvement being only partial to cultural conditioning (i.e. exposure to income-earning activities in a single generation may not wipe out centuries of non-exposure) and to low absorptive capacity (i.e. limited ability to utilise large volumes of credit) in poor communities. However, programmes that specifically target women generally have a positive impact (Barnes & Keogh, 1999).

Cross-sectional studies by Mosley & Hulme (1998), which examined the impact of 13 MFIs in seven developing countries, found that the impact of lending on the recipient household income tended to increase at a decreasing level as the recipient's income and asset position improved. Also observed was the low impact on the poorest recipients and a higher impact on the not-so-poor. Furthermore, impact varied with the design of the institution, with well-designed schemes achieving higher impact than ill-designed schemes. However, Simanowitz (2002) argues that it is not the poverty level of potential clients that determines access and impact, but the design of the services provided. Whilst there are trade-offs between social and financial objectives, MFIs can reach and impact on the poorest and achieve financial self-sufficiency.

Emphasising the growing importance of the intermediary approach to impact assessment, Mosley & Hulme (1998) conclude that financial sustainability correlates with the charging of market interest rates; availability of savings facilities; the frequency of loan collection; and the existence of material incentives to borrowers and staff of the lending agency to maximise the rate of repayment. Hulme (1997:22), however, concludes that there is no ‘optimal’ approach to impact assessment. Best practice suggests ‘achieving fit’ in meeting specific objectives ‘at an acceptable level of rigour that is compatible with the program's context [and] that is feasible in terms of costs, timing and human resource availability’.

New developments in impact assessment pioneered by the Imp-Act Programme3 seek to blend different approaches with a view to mould best practices that foster a greater focus on internal impact monitoring, learning and improving practice by MFIs. In this regard, Simanowitz (2001) documents methodological approaches that are a more ‘practitioner-oriented’ in that they feed into management and product design processes, allowing MFIs to improve their services and impact on their clients. The methodology for impact assessment recommended would depend on the purpose.

The Imp-Act Programme recommends that if the objective is to prove impacts to both an external audience and staff and management, the use of rigorous scientific methods that give statistically valid results is advised. On the other hand, if the objective is to improve impact, a qualitative participatory process is recommended. In this instance, impact assessment is viewed as an internal learning process, where the process improves knowledge about services, as well as an information management process to improve the practice of the organisation, its staff and clients. However, for most organisations, the approach would be somewhere in-between these extremes, and the objectives of the impact assessment would greatly influence the information needs and approach. The general consensus is that balancing the priorities of different stakeholders is an important consideration in impact assessment (Copestake, 2000; Hulme, 1997; Simanowitz, 2004; McGregor et al., 2000).

In summary, the approach observed in empirical literature is as follows:

1.	When impact assessment is a one-off, cross-sectional study whose objective is to prove impact so as to justify donor funding, external consultants are assigned to carry it out. Its scope would be mainly business, and the methodology recommended is a quantitative survey.
2.	When impact assessment is a long-term process whose objective is to provide management information to improve practice, it becomes a longitudinal study where clients, staff and other stakeholders are all involved. Its scope becomes wider and includes business, individual, household, community and wider impacts. The methodology recommended is a participatory process that is largely qualitative.

3. The Khula impact case study

3.1 Overview of the products of Khula Enterprise Finance Limited

Khula Enterprise Finance Limited, hereinafter referred to as Khula, is an agency of the Department of Trade and Industry established in 1996. It is a wholesale finance institution that facilitates access to credit by South African SMMEs through various mechanisms and institutions. These institutions include commercial banks, retail financial intermediaries (RFIs), microcredit outlets (MCOs) and private equity funds. In addition, the agency provides mentorship services to guide and counsel entrepreneurs in various aspects of managing a business. As a wholesale finance institution, it means that entrepreneurs do not get loans directly from Khula but through the various independent institutions, which collectively are termed its ‘products’. Their characteristics are briefly discussed below.

3.1.1 Microcredit outlets

These collectively constitute the KhulaStart Programme, which is an intervention strategy to facilitate greater access to microcredit by rural and peri-urban communities. The programme targets historically disadvantaged communities, particularly women, and has a specific objective that 70 per cent of microloans should be granted to women. MCOs are initiated through existing NGOs or community-based organisations that are already involved in some form of SMME support activity in the community, such as business training and advice.

The KhulaStart Programme employs the group solidarity methodology whereby individuals group themselves into groups of three to ten members. Microloans are disbursed on an incremental basis and can be used to start or expand any type of legal business activity, though the bias is towards small-scale manufacturing, crafts and tourism.

3.1.2 Retail financial intermediaries

These can be NGOs and other legal structures which, under South African law, engage in offering financial services to SMMEs. RFIs that meet eligibility criteria can obtain loans from Khula for onlending to SMMEs. In addition, Khula can advance interest-free seed capital or grants for new RFIs, and also offers capacity-building services for both existing and new RFIs. However, Khula's subsidised business loans generally cannot be used to finance the operational cost of the RFI.

3.1.3 Commercial banks

Where commercial banks require guarantees in order to lend to SMMEs, Khula furnishes credit guarantees to approved applicants. Three Khula credit guarantee products are offered: individual, institutional and portfolio guarantees:

1.	An individual credit guarantee is given where an individual small entrepreneur has been unable to furnish adequate security to obtain credit, and the bank applies to Khula for additional security.
2.	An institutional credit guarantee arises when RFIs wishing to fund their lending operations in the SMME market require guarantees to secure a loan from a commercial bank. Khula then provides the bank with the appropriate guarantees for approved RFIs.
3.	A portfolio guarantee is sought when an RFI considers an SMME portfolio to fall beyond its current lending practice, in which event Khula would indemnify the RFI for up to 80 per cent of its irrecoverable loans. Generally, credit guarantee schemes are for medium-level loans rather than micro- and small loans.

3.1.4 Business support services

Khula has a mentorship scheme known as the Thuso Mentorship Scheme, which is a programme set up to guide and assist the small business community in managing and administering their enterprises in order to improve sustainability and success. The mentorship consists of pre-loan services (e.g. development of a business plan) and post-loan support services (e.g. management support). Mentorship is invariably a requirement for qualification of a Khula credit guarantee.

3.1.5 Regional private equity funds

These were established in July 2000 to address the lack of equitable access to risk capital by emerging entrepreneurs in Mpumalanga, KwaZulu-Natal and Limpopo, which are some of the poorest provinces in the country. The strategic role of Khula-supported private equity funds is to create a vehicle for delivering risk capital to emerging enterprises that create wealth, asset ownership and employment opportunities in the previously marginalised communities of South Africa.

3.2 Objective of impact assessment

In 2001, Khula commissioned the Bureau of Market Research of the University of South Africa to conduct an impact assessment of its products, focusing on:

1.	Job creation
2.	Gender empowerment
3.	Rural versus urban breakdown of beneficiaries
4.	Income and wealth effect
5.	Household support
6.	Financing of new versus existing businesses

Typically, the impact assessment could be categorised as a one-off, cross-sectional study whose objective is to prove impact so as to justify donor funding contracted to external experts. The donor in this instance is the government. It can be argued that the objectives of the impact assessment are twofold: first, to prove impact by Khula's management in order to justify further funding by the government; and, second, to prove impact for advocacy purposes, possibly for political reasons by the government to prove to the wider community that its efforts to empower the poor and alleviate poverty are working.

3.3 Methodology

The methodology adopted for impact assessment was largely based on the following four lines of enquiry (see BMR, 2001:44–56, for a detailed methodology):

1.	Qualitative enquiries using semi-structured interviews, and discussions with stakeholders (intermediaries and clients) were held to determine the impact of the intervention.
2.	A questionnaire-based survey was carried out to elicit beneficiaries' views on the impact on their lives and communities of Khula products.
3.	Aggregate data from the survey were quantified where possible and analysed. (Tables 1 to 5 summarise the quantitative features of the results.)
4.	Personal visits to microfinance providers and the beneficiaries' business sites were conducted.

Table 1. Total number of beneficiaries and sectoral distribution of respondents, 1996 to March 2001

			Credit Guarantee Scheme
Item	Khula-Start MCOs	RFIs	Individuals	Portfolios	Thuso Mentorship Scheme	Regional equity funds
Total number of beneficiaries	11 875	89 354	1 420	762	677	282
Number of respondents	566	407	232	56	181	30
Economic sector of beneficiaries (%)
Retail	59,9	46,4	34,5	73,2	35,9	6,7
Wholesale	0,5	2,7	4,3	0,0	8,2	0,0
Construction	0,5	4,2	6,9	0,0	3,9	6,7
Manufacturing	21,4	16,2	14,7	3,6	12,1	20,0
Catering	2,1	7,9	9,5	12,5	4,4	3,3
Motor vehicle repairs	0,7	3,7	6,9	0,0	1,7	0,0
Agriculture	7,1	0,5	0,0	0,0	1,7	0,0
Personal services	4,6	9,3	8,2	7,1	13,8	10,0
Art and culture (handcraft)	1,6	1,0	2,2	1,8	4,9	13,3
Repairs and household goods	0,7	1,0	0,4	1,8	0,0	0,0
Cleaning services	0,2	1,0	1,7	0,0	0,6	0,0
Crèches and after-school centers	0,7	1,5	2,2	0,0	0,6	0,0
Security	0,0	0,5	0,9	0,0	2,2	0,0
Information services	0,0	1,0	3,4	0,0	1,7	16,7
Transport services	0,0	2,5	1,7	0,0	4,4	13,3
Bookkeeping	0,0	0,0	0,4	0,0	1,1	6,7
Accommodation	0,0	0,0	0,4	0,0	0,0	1,7
Other	0,0	0,0	1,3	0,0	1,1	3,3
Total	100,0	100,0	100,0	100,0	100,0	100,0

Table 2. Loan features

			Credit Guarantee Scheme
Item	Khula-Start MCOs	RFIs	Individuals	Portfolios	Thuso Mentorship Scheme	Regional equity funds
Percentage of beneficiaries with only a Khula supported loan	97,7	95,1			–	–
% of beneficiaries to secure a loan	19,1	29,7	15,1	10,7	–	3,3
Source of assistance (%)
Khula mentorship	44,4	37,2	28,6	16,7	–	–
NGO	26,9	28,1	14,3	0,0	–	–
Friend/family/colleague	25,9	28,1	2,9	83,3	–	–
Other	2,8	6,6	54,3	100,0	–	–
Application of loan	30,2	52,6	67,0	26,8	–	56,7
Start-up of business	69,1	41,8	23,7	17,9	–	36,7
Consolidate credit	0,9	2,2	2,2	0,0	–	0,0
Continuation/maintenance of business	0,0	5,4	5,4	32,1	–	0,0
Other	0,2	3,6	3,6	25,0	–	6,6

Table 3. Employment features

			Credit Guarantee Scheme
Item	Khula-Start MCOs	RFIs	Individuals	Portfolios	Total
Average employment size
Male	0,34	4,90	8,17	1,95	–
Female	1,58	3,44	3,84	1,50	–
Total	1,92	8,34	12,01	3,45	–
Full-time	1,85	7,19	10,68	2,98	–
Part-time	0,07	1,15	1,33	0,47	–
Total	1,92	8,34	12,01	3,45	–
Estimated total employment*
Male	4 038	437 835	11 601	1 486	454 960
Female	18 763	307 378	5 453	1 143	332 737
Total	22 801	745 213	17 054	2 629	787 697
Full-time	21 968	642 455	15 166	2 271	681 860
Part-time	833	102 758	1 888	358	105 837
Total	22 801	745 213	17 054	2 629	787 697
New jobs created by Khula loans	15 489	109 505	12 847	721	521 307

Note: *Gross jobs created, maintained and leveraged from March 1996 to March 2001

Table 4. Other socio-economic features

			Credit Guarantee Scheme
	Khula-Start MCOs	RFIs	Individuals	Portfolios	Thuso Mentorship Scheme	Regional equity funds
Spatial composition (%)
Urban	30,0	60,4	86,6	42,9	76,8	93,3
Rural	70,0	39,6	13,4	57,2	23,2	6,7
Dependency on Khula-supported business
Percentage of owners deriving more than 50 per cent of household income from business (excluding “don't know/cannot say”)	26,6	32,8	39,5	15,1
Estimated number of people financially benefiting from business
• Average per business	6,23	15,86	17,95	9,04	–	–
• Estimated total for all businesses	73 981	1 417 154	25 489	6 888
Client/customer base in terms of turnover
• Businesses	5,9	32,9	44,5	13,4	–	–
• Individuals	94,1	67,1	55,5	86,6
Image of Khula or its agents	88,0	94,3	84,5	76,8	80,6	35,7
Percentage beneficiaries recommending Khula or its agents (excluding “don't know”)

Table 5. Gender issues of beneficiaries

			Credit Guarantee Scheme
Item	Khula-Start MCOs	RFIs	Individuals	Portfolios	Thuso Mentorship Scheme	Regional equity funds
Gender of respondent (%)
• Male	14,3	46,2	74,6	71,4	71,8	93,3
• Female	85,7	53,8	25,4	28,2	28,2	6,7
Gender of employees (%)
Full-time
• Male	17,8	59,1	69,0	60,5	–	–
• Female	82,2	40,9	31,0	39,5	–	–
Part-time
• Male	14,3	56,5	59,9	30,8	–	–
• Female	85,7	43,5	40,1	69,2	–	–
Gender of individual clients (%)
• Male	35,1	52,1	52,1	41,6	–	–
• Female	64,9	47,9	47,9	58,4	–	–

The methodology employed, which relied heavily on historic and qualitative information collected through semi-structured interviews and focus group discussions with users, has been criticised as arbitrary. According to the new paradigm espoused by Imp-Act, the Consultative Group to Assist the Poorest (CGAP) and Assessing Impact of Microenterprise Services (AIMS)4 suggest the use of scientific methods for proving the impact. However, they acknowledge that where the purpose is advocacy, the methodology need not necessarily be scientific, but simply credible enough to support a strongly held predetermined position.

3.4 Findings and discussion

The Khula Impact Report (BMR, 2001) documents a multiplicity of impacts arising from Khula interventions, which are discussed in four broad categories below. In the first instance, findings are summarised in Tables 1 to 5, sourced from the said report.

3.4.1 Impact on participants' businesses and employment

Table 1 shows the estimated total number of beneficiaries and the sectoral distribution of respondents. Note that the majority of beneficiaries are serviced by RFIs, followed by the KhulaStart Programme and commercial banks through the credit guarantee scheme. Private equity funds have not so far benefited a wider community. (They are possibly too new, as they were less than one year old when the impact study was conducted.)

Outreach to the poorest in the rural areas serviced by the KhulaStart Programme, which is a major concern to donors (the government in this case), is outstripped by about eight times by the outreach to the not-so-poor communities in the areas serviced by RFIs. Also, the Thuso Mentorship Scheme with its business support services is being accessed by an insignificant number of beneficiaries, despite the fact that SMMEs generally lack support in this regard.

As regards sectoral application of loans, the overwhelming majority of loans are allocated to the retail sector, which consists of spaza (tuck) shops, hawking, fruit and vegetable sales, buying and selling of clothes, and the like. The second significant category that receives loans is the manufacturing sector – baking, confectionery, sewing, metalwork, woodwork and dressmaking. Of significance, MCOs finance substantial agricultural activities such as poultry and piggeries in the rural areas. The other sector that significantly receives loans is the personal services sector (catering, hairdressing, etc.).

Table 2 shows that nearly all beneficiaries depended on a Khula-supported loan, indicating the limited participation of the formal financial sector in servicing SMMEs. Interestingly, relatively few businesses and applicants make use of any help in preparing their application to secure a loan. Worse still, a smaller percentage (19,1 per cent) in the poorer rural segment seeks assistance, compared with those in urban areas (29,7 per cent). For those few who seek assistance, it is mostly obtained from Khula's mentorship scheme (the highest percentage), NGOs and friends and/or relatives. Thus, the mentorship scheme is highly regarded by those loan applicants who need assistance, although it is generally underutilised.

For the KhulaStart Programme, most microloans (69,1 per cent) are used for extending existing business, whereas in urban areas serviced by RFIs most loans (52,6 per cent) are used to start up a business. The implication is that people who were already in business are accessing MCOs more than new entrepreneurs.

Table 3 reflects a somewhat impressive employment creation, which favours urban areas serviced by RFIs over the rural areas by a wide margin. However, the figures should be treated with caution, because SMMEs often experience high mortality rates and the figures are inclusive of existing employees prior to Khula's intervention.

The Khula Impact Report (BMR, 2001:58) observes that the majority of businesses (mainly new businesses) do not generate sufficient cash flow to proceed with activities after the expiry of a loan disbursement period. More than half the new businesses could not sustain themselves further after the discontinuation of further loans. However, this observation may not take into account that microfinance clients are exposed to frequent and wide-ranging risks. It has been argued that an understanding of client strategies for dealing with risk and the role of financial services should enable better design of products and improve impact and sustainability (Cohen & Sebstad, 1999; Dunn, 1996).

Table 4 shows that at least 1,52 million people directly and indirectly benefited financially from Khula's intervention from 1996 to 2001. The breakdown of the estimated total number of people financially dependent on Khula-supported business is given as follows: 73 981 for the KhulaStart Programme, 1 417 154 for RFIs and 32 377 for the credit guarantee scheme. RFIs that serve mainly urban areas financially benefited more people than MCOs that target the rural areas, by a margin of about 1,3 million. In addition, the utilisation of the mentorship scheme by rural communities (23,2 per cent) is very low, compared to urban communities (76,8 per cent).

Taken together, Tables 3 and 4 show that concentration in employment and associated benefits rather accrue to urban areas (mainly serviced by RFIs) than to rural areas (serviced by MCOs). Although this disparity could be explained by the lack of favourable infrastructure in rural areas, this observation lends support to the findings by Mosley & Hulme (1998) that higher-income households, which are largely urban dwellers, on average tend to experience a higher programme impact than households below the poverty line, the majority of whom live in rural areas.

3.4.2 Impact on women beneficiaries

In terms of employment, Table 3 shows that more women beneficiaries benefited from the KhulaStart Programme, where a target of 70 per cent has been specifically set. However, in the case of RFIs and credit guarantee schemes where women are also favoured generally, but with no specific target having been set, they benefit less than men. Considering women's low participation in credit guarantee schemes where larger loans from banks are involved, the observation is that women are more likely to qualify for small loans, while men are more likely to qualify for larger loans. This ultimately prevents women from moving from survivalist endeavours to profitable businesses.

Table 5 reinforces the observation that where women are not specifically targeted, they benefit less than men. The predominance of males as employees increases as the size of business increases. However, women as clients/customers predominate in the KhulaStart Programme, RFIs and the portfolio credit guarantee scheme.

In summary, Tables 3 and 5 show employment gender distribution that is in favour of women in the rural areas and in favour of men in urban areas, albeit with a smaller margin. The fact that the target of reaching at least 70 per cent of women in the rural areas through MCOs is being met, is taken as proof of the general empowerment of women. However, this may not be synonymous with empowered women, because the study does not use the theoretical framework of a bargaining model of households, as espoused by Amartya Sen. According to this model, the well-being of men and women depends on their bargaining power, which, in turn, is dependent on three factors: their breakdown positions (i.e. individual welfare in the event of a breakdown of cooperation); their perceived contribution to the family; and their perceived self-interest.

In the case of Khula, further research is required on women's well-being, focusing on the standard three sets of capabilities: the degree of autonomy with which women can live their lives; their ability to control decision making within the family; and their relative access to household resources, such as food, education, health care, etc. Notwithstanding, it can be argued that Khula intervention is indeed empowering women from a cultural context. Mayoux (1999) has argued that women in Africa are more widely involved in production and marketing, and are often responsible for household subsistence. Consequently, they are likely to use credit for their own enterprises and control the income.

3.4.3 Impact on households and the community

Attribution and additionality are now standard concerns of impact assessment. Table 5 shows an estimate of people financially dependent on Khula intervention both directly and indirectly, i.e. owners of businesses, employees and members of their households. The effect is more pronounced in urban areas served by RFIs. Also, commitment of risk capital in the form of regional equity funds is negligible in the rural areas, which only receive 6,7 per cent, as opposed to 93,3 per cent for the urban areas. Khula is therefore not achieving its stated objective of regional equity funds benefiting the previously marginalised community, with a bias towards women, the disabled and enterprises in rural areas.

Again, this supports earlier observations by Mosley & Hulme (1998) that impact is more pronounced in higher-income households than in lower-income groups. However, in the case of Khula's regional equity funds, it could be too early to conclude thus, because they had only been in existence for barely a year before the study was conducted.

3.4.4 Wider impacts

McGregor et al. (2000) have argued that while the wider impacts of microfinance interventions are of importance, there is less clarity about how they can be effectively assessed. Moreover, evidence on them tends to be weak, circumstantial and not well researched. This study, however, demonstrates that wider impacts could be manifest through four channels:

1.	Through the labour market, as previously unemployed people become employed by expanding microfinance businesses
2.	Through the capital market, as the intervention of an MFI induces existing lending institutions (e.g. private moneylenders) to lower their interest rates in order to compete
3.	Through the markets for goods consumed by poor people, when microfinance induces an increase in supply resulting in the lowering of prices for poor people
4.	Through production linkages as an expansion of livelihoods (especially in the agricultural sector) stimulated by microfinance. This can induce the formation of rural industries and services, and further induces income changes among non-borrower households and clients' participation in social and political processes.

In order to assess the wider impacts of Khula's intervention, regard must be had to its overall strategy. Khula's overall programme can be considered a collection of state and donor-aided interventions addressing related constraints and problems through the agency of organisations – MCOs, RFIs and banks – rather than at the level of the beneficiary, as the agency does not directly advance loans to clients. A similar programme in Africa that has achieved significant progress through this approach is the British Aid to Small Enterprise (BASE) programme in Kenya.

BASE has operations with a stated programme objective to develop the capacity of private sector intermediary organisations to promote the growth of micro- and small enterprises (MSEs) that fall into three categories: the development of financial services; the development of business support, training and development services; and the development of an enabling environment for small enterprises in Kenya. In the first two categories, a number of mainly NGO microfinance, microenterprise and technology development and training projects are supported, while in the third category it has a deregulation project to improve the legislative and regulatory environment for MSEs through collaboration with the government.

According to Ferrand (1997), the success of the programme has been partly attributed to its recognition that constraints experienced by MSEs are interlocking and hence need to be alleviated with a coordinated approach. By and large, BASE's programme structure is similar to the Khula programme in that there is also a recognition that constraints faced by SMMEs are interlocking and hence the need for a coordinated approach. The notable exception is that Khula does not have, among its portfolio of projects, a deregulation project for the improvement of the legislative and regulatory environment for MSEs, as is the case with BASE. Notwithstanding, Khula's intervention is bound to have wider impacts through changes in the labour market, capital market, market for goods and services, and production linkages, especially as it is a national programme that covers the entire country.

3.5 Summary of findings

In summary, the intervention of Khula has produced the following impacts:

1.	A positive impact on employment and household welfare is observed as the targeted communities are becoming financially dependent on Khula-supported businesses. The impact is more marked in urban areas than in rural areas.
2.	The programme has greatly benefited women in the rural areas and peri-urban areas, where specific targets have been set for their empowerment.
3.	In the urban areas where women are targeted but no specific percentage target has been set like in rural areas, the programme has benefited more men than women.
4.	Generally, more men qualify for larger loans than women, as they are more likely to be eligible for credit guarantee schemes to obtain larger loans from commercial banks.
5.	There is low utilisation of the mentorship scheme in rural areas, compared with urban areas.
6.	Outreach of the programme is more pronounced in urban areas, signifying the widening divide between urban and rural.
7.	Households dependent on the programme are more pronounced in the urban areas.
8.	The majority of loans are used to start up businesses as opposed to use for existing businesses.

3.6 Limitations of the Khula impact study

By and large, attribution of impact relied on qualitative interpretation of evidence rather than statistical inference, as is required in studies of this nature to be considered rigorous. No regression analysis has been attempted to establish linkages. It could therefore be argued that the study can only be considered exploratory and indicative, but not conclusive.

4. Conclusions

The Khula impact study, albeit with a narrow, traditional donor-driven focus, shows that microfinance programmes lead to improvements in the economic welfare of households, enterprise development and increased empowerment of women. Notably, it emphasises the importance of targeting specific groups, as findings show that when this is not done, more men are likely to benefit more than women. Targeting should also be a consideration as regards the desired outreach, without which the not-so-poor urban communities tend to benefit more than their poor counterparts in the rural areas.

Notes

The website at http://www.ids.ac.uk/impact/stateart/stateart.html publishes the activities of the programme.

The website at http://www.imp-act.org/chronicles the new developments in this regard.

See http://www.imp-act.org/, http://www.cgap.org/ and http://www.enterprise-impact.org.uk/ for details of the new paradigm in impact assessment.