Pro-poor microcredit in South Africa: cost-efficiency and productivity of South African pro-poor microfinance institutions

Abstract

This article compares the performance of selected South African microcredit non-governmental organisations (NGOs) that have a poverty-alleviation focus against various benchmarks drawn from the MicroBanking Bulletin. Donors, governments and many analysts regard sustainability as the benchmark of microfinance institutions' (MFIs) performance. However, the most relevant question is whether microcredit NGOs are doing as well as they can in their context. Of particular contextual importance is income inequality in a society. South Africa has the world's second worst income inequality, after neighbouring Botswana. This creates a situation in which microcredit NGOs must recover ‘First World’ costs, particularly salaries, from revenues based on clients who can only afford loans on a par with Third World countries. Compounding this situation are structural obstacles to microenterprise in South Africa, as well as obstacles to productivity in microcredit NGOs. Taken together, this creates a ‘salary burden’ for South African microcredit NGOs, which is the highest in the world according to relevant benchmarks. South African MFI managers face significant obstacles to improving productivity to compensate for the divergence between staff and client living levels. These include an inadequate skills base, the small scale of the market, rapid labour turnover, and limited resources for capacity development. South African MFIs face the options of moving upmarket (which many have done), adopting methodological innovation or new product development, or closure. Of these, there is a strong argument to be made for supported savings and credit approaches as an alternative to NGO-based microcredit. Such an approach has the advantages of greater voluntary input and social capital formation.

1. Introduction

The question of ‘sustainability’ in microcredit is a subject of ongoing debate. The dominant view is what some call the ‘Ohio State’ school of thought, from the university of the same name in the United States. Broadly, this view espouses a market-led, full cost-recovery approach to microcredit, with no subsidies. It holds that sustainability is essential for two reasons: first, the goal of microcredit practice should be to extend the reach of commercial financial markets to the poor and/or excluded. This requires that microfinance institutions (MFIs) perform well enough to be able to access commercial wholesale finance, preferably sooner rather than later. Second, sustainability is necessary to prevent MFIs from concealing bad practice with ongoing subsidies. (For the purposes of this article, sustainability is defined as: Coverage of administrative cost + Loan loss + Cost of funds + Inflation + Capitalisation for growth from operating income.)

It is certainly reasonable to ask that the social cost-benefit ratio of resources directed to microcredit interventions should be at least as good as if those resources were applied differently, or even given away (Schriener & Yaron, 2001). There is little point in subsidising MFIs if the returns to doing so are not at least the same as for alternative uses – over an appropriate period and including ‘externalities’, i.e. non-financial impacts on clients and their communities.

For such calculations, absolute sustainability can be established empirically and easily – an MFI either is or is not sustainable. In most cases, however, the question is relative: how sustainable is an MFI compared with other MFIs? Is it headed towards or away from sustainability, and why? Most importantly, for our purposes, is the MFI doing as well as can be expected given the circumstances?

This last question is rarely raised. In particular, discussions about MFI sustainability performance often ignore national-level specifics. Donors, governments and analysts routinely compare MFIs in different countries to each other, telling MFI managers what norms they ‘should’ be able to achieve. Yet, as not all variables affecting sustainability are under the control of MFI managers; we should be prepared to ask whether an MFI is doing the best it can in its context.

Such a perspective is needed because any microcredit model involves a number of variables, both internal and external. For example, an MFI might vary with the size of a loan group, interest rates, or its incentive policy (internal variables). Or, it may be constrained by national laws, economic and labour market conditions, or political instability (external variables). Some aspects of comparative performance against international benchmarks are under an MFI's control; others may be determined by external factors that are not. There is also interplay between the two. MFI managers' decisions about things they can control may be shaped decisively by contextual factors they cannot control.

In assessing MFI performance towards sustainability, it is particularly important to take into account the level of income inequality in a society. In some contexts, it might be difficult or impossible to deliver poverty-oriented microcredit services, because the socially determined costs – principally personnel costs – of running a competent MFI are excessive relative to the income levels of client microenterprises and therefore their borrowing capacity.

1.1 Purpose of the article

With this in mind, the article addresses the matter of context in South Africa. It investigates three issues:

1.	First, the efficiency indicators of four South African microcredit non-governmental organisations (NGOs), all of which target poor and very poor households with solidarity group-lending methodologies, are compared with relevant international benchmarks drawn from the MicroBanking Bulletin. (‘Poor’ households are regarded as those living at or less than the unofficial but most commonly used South African poverty line, which was about US$67 per person per month at the time of the research in 2002. ‘Very poor’ households are those that have access to half or less than this.)
2.	Second, it identifies contextual factors that undermine the ability of South African MFIs to match such benchmarks.
3.	Third, it considers implications for South African MFIs, the government and donors.

1.2 Methodology

The MicroBanking Bulletin (MBB) has provided useful benchmarking tables on the global microcredit industry for some years now, and is updated biannually. The definitions underlying the MBB data have been used to gather comparable data on South African MFIs. The main point of comparison is with other African MFIs, particularly those that are small scale and serving a low-income target market. Comparisons are also made with MFIs elsewhere by size (medium to small), methodology (solidarity group) and target market (low income).

2. The South African microcredit context

One of the most important external variables confronting South African MFIs is the country's extreme income inequality. Although South Africa's per capita gross national income (GNI) of US$2 8202 puts the country in the middle-income band globally, this conceals enormous variation in income distribution. The Gini coefficient is currently about 0,65, which makes South Africa one of the most unequal societies on the planet. The figure improved somewhat after the first democratic elections of 1994. Its relapse since then, however, is linked to a steep rise in unemployment and poorly paid employment. This has been driven by industrial and trade policies designed to improve global competitiveness, and a macroeconomic policy emphasising low inflation and a small government deficit in order to attract foreign capital.

2.1 South Africa's dualistic economy

The chief cause and manifestation of South Africa's radical income inequality is the dualism of the economy. An economically ‘advanced’ and globally integrated minority, black and white, coexists with a dependent and marginalised majority, almost entirely black. In South Africa, these are known as the ‘formal’ and ‘informal’ sectors. While the former enjoys a human development index comparable to southern Europe, the informal sector lives at a level comparable to that in South Asia.

The material basis of this dualism is both historical and structural. Unlike peasantries elsewhere in Africa, South Africa's rural poor lack access to basic means of production, such as land, because of unresolved issues of comprehensive settler dispossession. They live in crowded rural ‘villages’ squeezed between commercial farmland (no longer exclusively white) and tourist-oriented game reserves. In the urban areas, opportunities for self-employment are severely constrained by South Africa's manufacturing and retail sectors, the most advanced in Africa, which relegate small-scale trading and manufacturing to the margins.

Because of their lack of access to productive resources, South Africa's poor are almost totally dependent for their survival on the output of the formal economy. The things that sustain and enhance life are only available as commodities. The poor, however, are structurally excluded from access to the cash necessary to obtain these. One outcome of this situation is poor households' dependence on state transfer payments, such as pensions, disability and childcare grants, and inter- and intra-household transfers. This is especially marked in rural areas. Another result is a high incidence of predatory economic crime.

2.2 Microenterprise in South Africa

For most of South Africa's ‘informal sector’ poor, microenterprise means small-scale trading/hawking, personal services, and production of specialty items. However, the efficient manufacturing and retail sectors severely constrain opportunities to add value and accumulate capital in such informal activities. Small-scale traders capture only the tiny sliver of value arising from transport cost differentials and convenience purchasing, as most low-income households' requirements can be obtained from formal shops.

Informal clothing manufacturers add value by producing items required in small quantities, and which are thus unattractive to formal manufacturers, such as school uniforms. Some informal manufacturers in the ex-Bantustan areas take advantage of value vs volume transportation issues to produce bulky and/or heavy but low-value building materials, such as window and doorframes, or furniture, but rapid improvement in transport infrastructure is undermining this opportunity. Amongst the most rewarding (legal) informal occupations in South Africa are brewing traditional beer, best produced in small batches and consumed fresh; tavern keeping; and hairdressing.

South Africa's informal traders, service providers and manufacturers are also constrained by the lack of cash in their communities. Most customers and clients of South African informal micro-enterprises are dependent on state transfer payments, inter-household transfers and informal microenterprise for their cash incomes. Cash cycles tend to be monthly, with a fresh influx at pension payout time. Formally employed persons in poor communities may spend some of their income on goods or services at microenterprises, but for everything above the most convenient and/or specialised purchases, there are formal supermarkets and shops reasonably close by.

This inauspicious context is illustrated by the fact that whilst nearly 40 per cent of employment in the South African retail trade sector is in microenterprises,3 the contribution of microenterprises to national retail trade output is only 2,3 per cent (SAIRR, 2001). Some 53 per cent of South African personal services employment is in microenterprises, but these contribute less than 10 per cent of the sector's output. Only about 8 per cent of microenterprises are involved in manufacturing. Overall, microenterprises provide nearly 20 per cent of South Africa's ‘jobs’ but contributes only 5 per cent to the gross domestic product (GDP). The national microenterprise income share is divided amongst nearly 8 million South Africans, which is 17 per cent of the population.

These factors lead to very low incomes in the microenterprise sector (SARB, n.d.; SSA, n.d.). The average annual microenterprise income is a little over US$1 000, which is about 46 per cent of the most commonly used South African poverty datum line (at US$1 = R8). Annual per capita income for persons in households whose income derives mainly from microenterprise is about US$250. This is well below the annual per capita poverty line of US$400, not to mention the ubiquitous ‘a-dollar-a-day’ regarded by some as the benchmark of absolute poverty. This is not to imply that microenterprise is the only source of income in such households. Many South African households dependent on microenterprise income would also receive some form of state grant via a resident pensioner or the childcare grant system.

2.3 MFIs in South Africa

South African MFIs straddle the country's formal and informal sectors. Although their clients are drawn from the poor communities and microenterprises described above, their staffs are solidly emplaced in a middle-class material environment little different from developed countries. This applies to all MFI staff, regardless of race. South African racial issues do have an effect on MFIs, however. In a peculiar but understandable paradox, South Africa's push for affirmative action and rapid black advancement means that skilled black MFI personnel are highly marketable, particularly in the state and private sectors, putting upward pressure on their salaries and leading to fairly rapid turnover. Yet, for reasons of equity and historical redress, donors and the government often disfavour available white personnel.

South African pro-poor MFIs are mainly rural, but their clients are not agricultural microenterprises. Life and work in remote rural areas, with client groups at great distances from one another, means an additional premium for MFI managers wanting to attract and retain good staff.

South Africa's educational system at all levels is woefully unprepared to produce the kind of skills and aptitudes needed by MFIs. Of the country's 30-plus technical colleges and universities, only two provide any microcredit-related training and both programmes are relatively new. Development courses at South African universities tend to be theoretical, general and geared to urban issues such as trade unionism. Because there are so few MFIs, there is no microfinance-specific labour ‘market’ and, consequently, most training takes place on the job. An ever-present problem is poaching of staff by other MFIs, NGOs, private firms and the state.

Overall, the ‘distance’ between South African MFI staff and their clients, both economic and social, is greater than in many countries, particularly in Asia. In India and Bangladesh, for example, it is not uncommon to find MFI clients with a fair amount of education and self-confidence. South Africa's low-income communities contain few people with ready-made skills to help manage microcredit solidarity groups. This places a greater burden on the MFIs to provide training and support for their clients.

3. The South African MFIs

Table 1 lists four South African NGOs that extend microcredit to poor and very poor households for microenterprise purposes. All these institutions use a group-lending methodology, although there are significant differences between them in this regard. All claim to be trying to reach very poor households, although only one employs a targeting methodology.

Table 1. Selected South African MFIs, May 2003

Name	Age	Offices	Staff	Active borrowers	Women borrowers	Average first loan	Average loan balance	Average balance/ GNI per capita
Beehive	8	3	54	5 892	75%	US$125	US$111	3,93
FINCA	3	1	50	1 386	96%	US$125	US$151	5,34
Marang	2*	19	145	15 836	95%	US$120	US$76	2,69
SEF	11	11	100	17 242	98%	US$67	US$82	2,89

Note: *Marang was formed out of two pre-existing MFIs, so aspects of its operation are older.

From the table it can be seen that:

1.	The average age of the South African group is six years, compared with 5,6 for all MFIs, but eight years for African MFIs.
2.	The average client base of 10 096 is comparable to the global average of 11 698, but significantly lower than the African average of 18 640.
3.	The average South African portfolio is US$867 348, compared with a global MFI average of US$3 859 273 and an African average of US$3 168 759. This puts the South African group on the boundary between the MBB's definitions of ‘small’ and ‘medium’ in the African context.
4.	The average balance outstanding of US$105 per client is considerably lower than the global average of US$453, but closer to the African average of US$181. The more relevant comparison, however, is to the MBB's ‘Africa small/low’ peer group, with an average of US$54.
5.	The average loan balance as a percentage of GNI per capita is 3,7 per cent, compared with 15,3 per cent for the Africa small/low group. However, this is not so much indicative of outreach performance as of South Africa's high per capita GNI. Thus, a small – even exceptionally small – microloan in the South African context is double the size, in absolute terms, of those given by the Africa small/low peer group.

4. South African MFI benchmark performance

The MBB's benchmarks include: outreach, profitability and sustainability, income, expense, portfolio quality and efficiency, and productivity measures. In this article, we are interested in outreach, expense, efficiency and productivity. Tables 2 to 6 compare the results for the South African MFI group against five categories of MFIs:

1.	Africa all: All African MFIs, regardless of size, target market, methodology, region, etc.
2.	Africa – small/low: African MFIs with a loan portfolio of US$800 000 or less and an average loan balance of US$150 or less
3.	World all: All MFIs, regardless of size, target market, methodology, region, etc.
4.	World – solidarity: All MFIs using solidarity group-lending methodologies
5.	World – low end: All MFIs with an average loan balance of US$150 per client

Table 2. Scale

Item	South Africa	Africa all	Africa small/low	World all	World–solidarity	World–low end
Age in years (average)	6	7	6	8	8	8
Offices	8,5	82	11	15	41	110
Active clients	10 096	18 640	11 678	11 698	47 884	42 520
Staff	90	120	47	101	278	222
Total assets (US$)	1 259 494	5 147 848	804 756	5 735 499	12 267 063	5 369 487
Outstanding principal balance (US$)	867 348	3 168 759	488 053	3 859 273	9 131 991	3 725 355

Table 3. Outreach

Item	South Africa	Africa all	Africa–small/low	World all	World–solidarity	World–low end
Average balance outstanding per client (US$)	105	181	54	453	371	135
Percentage of women clients	91%	77%	86%	61%	73%	86%
Average balance as a % of GNI per capita	3,7%	55,3%	15,3%	45,3%	46,0%	16,0%

Table 4. Expense

Item	South Africa	Africa all	Africa–small/low	World all	World–solidarity	World–low end
Total expense/total assets	101%	39%	50%	30%	31%	39%
Operating expense/total assets	84%	31%	37%	19%	22%	28%
Financial expense/total assets	11,7%	2,6%	2,5%	4,5%	3,0%	4,9%
Loan loss provision/total assets	4,8%	2,5%	3,6%	2,2%	2,3%	2,8%
Personnel expense/total assets	52%	15%	19%	11%	13%	15%
Non-staff administration expense/total assets	32%	16%	19%	9%	9%	14%

Table 5. Financial efficiency

Item	South Africa	Africa all	Africa–small/low	World all	World–solidarity	World–low end
Operating expense/loan portfolio	142%	56%	72%	27%	37%	50%
Operating expense per client	US$161	US$58	US$35	US$89	US$93	US$56
Personnel expense/loan portfolio	91%	27%	35%	15%	22%	26%
Average personnel expense as a multiple of per capita GNI	2,3	11,6	7,4	5,5	7,0	5,1

Table 6. Productivity

Item	South Africa	Africa all	Africa–small/low	World all	World–solidarity	World–low end
Borrowers per staff member	101	198	247	128	155	227
Borrowers per loan officer	150	413	462	308	356	682
Loan officers as a % of personnel	69%	42%	41%	45%	49%	49%

The following is clear from Table 2:

1.	South African MFIs are at the bottom of the scale in terms of average number of clients and the number of offices serving them. However, in both respects they are closer to the global average and to their African peer group.
2.	The South African group operates from a much lower asset base than all other categories, except their African peer group.
3.	The South African group carries a staff complement on par with the global average, but nearly double that of their African peer group.
4.	The South African group carries a much lower absolute loan portfolio on average than all categories of MFIs, except their African peer group, which is a little over half the size of the South African group.

Table 3 shows that:

1.	The average loan balance per client for the South African MFI group is on the low end of the scale, even in African terms, except for their direct peer group of small African MFIs targeting the very poor.
2.	As noted above, there is enormous disparity in terms of average balance per client as a percentage of per capita GNI. The South African MFIs are the lowest of any category – the only group in single figures – and only one quarter of the level of their African peer group.
3.	The South African group has the highest percentage of women clients.

In Table 4 we see the following:

1.	In every expense category, the South African MFI group is significantly out of line with other categories of MFI. Total expenses, operating expenses and non-staff administrative expenses as a percentage of total assets are roughly double those of the African small/low peer group.
2.	Financial expense as a percentage of total assets is also significantly higher than other MFI groupings, reflecting South Africa's high real interest rates.
3.	Personnel expense as a percentage of total assets is the most seriously inflated ratio in the case of South Africa, being five times the world average; 3,4 times the African average; and nearly three times that of the African peer group.

From the table the following can be seen:

1.	Unsurprisingly, given their relatively small scale, their inflated staffing and expense ratios, and the low average loan balances in proportion to per capita GNI, operating expense ratios in the South African MFI group are radically out of line with all other categories of MFI.
2.	The category most divergent from global and African norms is personnel expense as a percentage of the loan portfolio. The South African figure is six times the global average, 3,5 times the African figure and 2,5 times the norm of their African peer group.
3.	What is striking, however, given its personnel expense ratios, the South African group performs better than any other category of MFI in terms of average personnel expense as a multiple of per capita GNI. The South African group is less than half the global average, 20 per cent of the African average, and one third of the African peer group average. Compared with other MFIs, South African MFIs pay relatively low salaries in terms of the local economy. Again, this is due largely to the country's high average GNI per capita.

Table 6 illustrates the following points:

1.	The South African MFI group performs poorly against other groups in terms of borrowers per staff member, and performs particularly poorly against its African peer group. This is driven by especially poor performance in terms of borrowers per loan officer – half the global average and only one third of the level of the African peer group.
2.	Offsetting this somewhat, the South African MFIs are less top heavy than most other MFIs when considering the ratio of loan officers to total personnel.

5. The ‘salary burden’

One of the challenges of benchmarking analysis using performance ratios is to disentangle the relationships between various numerators and denominators. For example, the relationship between South African MFIs' expenses and their loan portfolios is poor compared with other MFI groups. Is this situation due to inefficiency, a high-cost environment, a high degree of societal income inequality, or all three?

In the South African case, three things stand out:

1.	The extremely low average loan balance relative to per capita GNI
2.	A low average personnel expense as a multiple of per capita GNI
3.	The poor productivity of loan officers, which leads to inferior overall physical productivity per staff member

Christen (2000) employs a useful measure that encapsulates all three factors. This is the ‘salary burden’, computed as follows:

This measure exposes the proportion of the MFI's portfolio each employee ‘represents’ in terms of the national economy. The higher the figure, the higher the proportion of an MFI's portfolio and operating revenue that is consumed by its personnel costs. The qualification ‘in terms of the local economy’ is critical. While it is useful to know the absolute level of average salaries and loan balances, what is really important is the relationship between the two in any given context. By comparing the two controlled for per capita GNI, ‘salary burden’ is a way to see the effects of societal income inequality and/or poor productivity.

Compared with other MFIs, average South African MFI salaries in terms of per capita GNI (i.e. the local economy) are low (Table 7). Yet, average South African loan balances per client in terms of per capita GNI are exceptionally low. Combined with poor physical productivity, this produces a situation in which, even though South African MFI staff are paid relatively poorly in local income terms, a low relative portfolio income base still makes it difficult to cover personnel costs.

Table 7. The ‘salary burden’

	South Africa	Africa all	Africa–small/low	World all	World–solidarity	World–low end
Average salary as a multiple of GNI per capita	2,3	11,6	7,4	5,5	7,0	5,1
Average loan balance per client/GNI per capita	3,7%	55,3%	15,3%	45,3%	46,0%	16,0%
Average clients/staff member	101	198	247	128	155	227
Salary burden	61%	11%	20%	9%	10%	14%

In considering the salary burden, which is more important in the South African MFI context – a low average loan balance or poor physical productivity? Table 8 considers two hypothetical scenarios. In one, the average loan balance per client in relation to per capita GNI is adjusted upwards to the Africa small/low peer group level. This produces a salary burden of 15 per cent. The second holds loan balances constant while varying productivity upwards to match the African small/low peer group. This produces a salary burden of 25 per cent. This suggests that the contribution of South Africa's small average loan balance per client is the greater of the two, but only marginally; and the two are more or less equally problematic.

Table 8. Salary burden scenarios

	South Africa actual	South Africa with adjusted loan size	South Africa with adjusted productivity	Africa–small/low	World–low end
Average salary as a multiple of GNI per capita	2,3	2,3	2,3	7,4	5,1
Average balance per client/GNI per capita	3,7%	15,3%	3,7%	15,3%	16,0%
Average clients/staff member	101	101	247	247	227
Salary burden	61%	15%	25%	20%	14%

An average South African MFI salary burden of 15 to 25 per cent would still be high in global terms, but much closer to the African peer group and to the global low-end microcredit sector than at present.

6. Obstacles to productivity in the South African context

The MFIs surveyed are all committed to reaching the very poor in the South African context. Therefore, we can assume that they would resist increasing their average per client loan balances and would rather concentrate on improving productivity. What are the obstacles to improved productivity in the South African context?

1.	Distances and mobility: Most South African pro-poor MFIs operate in rural areas, with client groups separated by large tracts of commercial farmland and game reserves. This is very different from the situation in densely populated parts of Asia, particularly Bangladesh, whose MFI sector skews global benchmarks through sheer size. The South African scenario is not so different from other African countries, although the influence of urban MFIs in the African figures must be taken into account. One important factor is the lack of independent transport for many South African MFI loan officers, who rely on generally inadequate public transport, taxis and their feet to reach their clients.
2.	Penetration levels: In a situation of relative remoteness, it would be logical for South African MFIs to try to develop as many clients in each village as possible. However, this is difficult because of both the dearth of opportunities for microenterprise and the similarity of opportunities. One MFI that has considered this issue carefully concludes that it can sustainably reach no more than one in five households in very poor communities. Thus, in a (large) village of a thousand households, even if a single loan officer reached every possible household (200), he or she would still be below relevant global and African measures of loan officer productivity (350 to 450 households).
3.	Need for greater client training input: A factor rooted in South Africa's apartheid past is the low level of literacy, business skills and general self-confidence in the rural communities. Women are particularly disempowered, given traditional patriarchal social structures, and small enterprise is not as much of a tradition for them as in other parts of Africa. This means that MFI loan officers spend a significant amount of their time assisting very poor women to develop the basic self-confidence and skills to run their businesses – not necessarily though training but through general encouragement and social empowerment.
4.	Skill levels and attitudes to work: South African MFI loan officers are typically right out of high school or have a few years of post-secondary education. Many are drawn from the ranks of unemployed teachers. They tend to be young, and many lack life skills and a mature work ethic, considering their MFI jobs neither as a career nor as particularly ‘developmental’. The hard work involved in rural microcredit and the attractions of city life lead many to jump at the first opportunity for alternative employment.
5.	Labour relations: The previous factor is related to South Africa's generally combative labour relations environment. Trade unions helped overthrow the apartheid regime and some MFIs are unionised. Even those that are not must comply with restrictive employment legislation that makes it difficult to fire underperforming staff. There have been a number of strikes at South African MFIs.
6.	Management inexperience: South African MFI loan officers share responsibility for their productivity with management, who decide operational, human resource and strategic planning issues. Given the country's small MFI sector, there is a very small pool of experienced top-level MFI managers and many mid-level managers are still learning the ropes. Management turnover is particularly damaging in such a context, because most replacements must learn by doing rather than bringing pro-poor microcredit skills with them to the post.
7.	Under-management: South African MFIs have a significantly higher ratio of loan officers to management and administration staff than other MFIs, but this may contribute to low productivity because of insufficient supervision of loan officers. Low productivity might tempt MFI managers to employ more loan officers to increase their portfolios, thus reinforcing the under-management problem.
8.	Relative salary levels: Are South African MFIs compensating for poor productivity and relatively high non-loan officer salaries by keeping loan officer salaries low compared with the salaries of management and administration staff? There is some evidence – but this is ultimately self-defeating, as income inequality within an MFI would tend to undermine loan officer morale.
9.	Lack of appropriate support: At present, South Africa lacks a coordinated source of capacity-building resources for the NGO microcredit sector. The state has yet to grasp the nettle of capacity development as a precondition for the emergence of a successful pro-poor microcredit sector, and microcredit NGOs are left largely to their own devices.

7. Analysis and implications

7.1 What is the significance of income inequality?

Some might question whether loan size as a percentage of per capita GNI is a meaningful measure in the South African context. It is self-evident that this figure will be low compared with that of most developing countries. As South African MFIs are not lending to the middle class, it might seem more useful to compare loan balances to average incomes in the communities where their clients live rather than an average for the entire society.

This objection is only relevant, however, if we are interested in assessing poverty outreach, by comparing clients of South African MFIs with clients of MFIs elsewhere, which is not the purpose of this study. Here, we wish to understand how income differentials between South African MFI staff and clients affect MFIs' ability to attain sustainability.

All other things being equal (including average salaries and interest rates), the smaller the average loan, the more clients are required per MFI employee to cover personnel costs. AS other things are almost never equal, per capita GNI is useful to compare such relationships across different economies. South African MFI salary levels are contextually low compared with the selected benchmarks, but average client loans are, again contextually speaking, extraordinarily small. This leads to a very high salary burden. To be very poor in South Africa means to have an absolute income similar to – even below – that of very poor people in other developing countries, while South African MFI staff members are drawn or pushed towards relatively higher income requirements than their foreign colleagues. The average South African MFI loan size as a percentage of per capita GNI is indispensable to assess the client side of this equation.

7.2 Salaries and productivity

South African MFI managers, wanting to reach the very poor but under pressure to achieve break-even, are in an unhappy situation. They could compensate for the salary burden by offering comparatively low salaries (as some do), but this would tend to reinforce poor productivity, poor labour relations and high turnover. They could try to increase productivity – specifically the number of clients per staff member – to exceptional levels, with salaries to match. However, this is unlikely in South Africa's young microfinance sector, with inexperienced staff, a limited market for replacements and strong competition for suitable staff from other MFIs, the government, NGOs and the private sector. In such a situation, highly productive staff would tend to have their salaries bid up. Put differently, MFI salary levels are an exogenous variable, beyond the control of MFI managers.

In the private sector, the options in such a situation are straightforward: innovate new production techniques; find new products to produce and sell; or close down. South Africa is an unlikely source of methodological microcredit innovation for the very poor. Indeed, aside from a few trailblazers, it has demonstrated remarkable slowness in experimenting with existing methodologies. Even fairly obvious adjustments (e.g. independent transport for loan officers) are adopted slowly and cautiously. Far more common than methodological innovation has been ‘mission drift’ towards better-off clients.

7.3 Is microcredit appropriate for poverty eradication in South Africa?

What about new products – ones that rearrange the cost-revenue relationship? If salary levels are exogenous and the South African pro-poor microcredit NGO sector is unlikely to innovate new microlending techniques, this is the only other option besides closure.

There are three broad forms of microfinance intervention available to South African NGOs with a poverty-alleviation focus: individual microcredit, group microcredit, and supported Accumulating Savings and Credit Associations (ASCAs). By contrast, deposit taking is illegal for non-bank institutions in South Africa. Individual microcredit is demonstrably too expensive for very poor households, and there are no examples of successful South African MFIs reaching the very poor this way.

South Africa has thousands of ASCAs and several initiatives that link them into larger networks, but to date there has been little serious work on their microfinance potential. The NGOs active in this area concentrate on savings and credit as a vehicle for social mobilisation rather than access to savings and credit services per se. As a result, their performance in providing access to small-scale credit for business, emergencies and consumption, based on intermediated group savings, is poor. Their main effectiveness as poverty alleviation strategies lies in the development of social capital in savings and credit groups and in their larger networks. In this respect, some South African savings and credit networks have been remarkably successful.

Nevertheless, it is notable that elsewhere in the southern African region, some NGOs have been successful in encouraging the formation and functioning of ASCA networks providing meaningful local-level financial services that directly improve the poverty situations of their members. The cost of supporting these groups is a tiny fraction of the cost of microcredit programmes, with comparative per-client cost ratios of 1 : 100 common. In other parts of the world (notably South and Southeast Asia), NGOs and parastatals have successfully provided external credit lines to functioning ASCA networks, enabling them to have a significant impact on poverty without using NGO-to-client microcredit.

As ASCAs have the potential to not only deliver microfinance services, but also create social capital amongst target communities – thus creating the possibility of a more radical approach to poverty that goes beyond reliance on market solutions – it may be time for South African NGOs to consider this approach more seriously.

8. Summary and conclusions

Because of the country's extreme income inequality, to be very poor in the South African context means to have a real income, and thus capacity for borrowing, more or less on a par with MFI clients elsewhere on the African continent. But to be a South African MFI staff member means to have socially determined expenses and thus income expectations on par with the developed world, or at least much higher than elsewhere in Africa and Asia.

The income and social inequality thus makes operating a microcredit business in South Africa unusually expensive relative to other developing countries, and there is little that South African MFIs can do about it. This is so even though MFIs are paying their loan officers much less, relative to the local economy, than most MFIs globally and in Africa. However, it is clear that South African MFIs could go much further to improve their physical productivity. Until this is attempted, it is impossible for them to say with certainty what special consideration they might need and/or deserve.

This problem raises the question of whether microcredit is an appropriate solution to poverty in South Africa. Substituting the voluntary input of savings and credit group members for the paid exertions of professional fieldworkers has the potential not only to improve microfinance performance, but also to create social capital that can be used to address poverty in a variety of ways.

Notes

At current prices, in mid-2003. Following MicroBanking Bulletin practice, this and other figures in this document have not been adjusted for purchasing power parity. For full details of the MicroBanking Bulletin approach to benchmarking, see http://www.mixmbb.org. (US$ 1 = R8 in 2002, when this study was undertaken.)

Because they are based on the number of employees rather than turnover, official South African microenterprise figures include small but high-value formal sector firms as well as informal sector microenterprises.