Let’s say you want to create the ideal program to mobilize finance for small and medium-sized agricultural businesses (agri-SMEs). What is the most efficient type of support for institutions lending to agri-SMEs? Is it:

1. Capacity support and Technical Assistance (TA) for lenders to better assess risk and understand opportunities?
2. Blended finance instruments to de-risk loans?
3. Ecosystem support mechanisms, such as value chain studies and policy advocacy, to improve the environment for finance?

Before we provide our answer (which you can also read using the report links at right), let’s first reflect on why this is such an important question.

Why agri-SMEs?

Growing the agricultural sector is one of the most effective ways to lift people out of poverty in developing countries, and at the backbone of the ag sector are small businesses. Producer groups, traders, processors and other SMEs play an outsize role in poor rural areas as sources of inputs and markets for farmers and employment for off-farm laborers. Agricultural SMEs tend to be particularly “inclusive” as they typically source from smallholder farmers and employ low-skilled workers, often women and youth.

All types of SMEs face severe credit constraints in developing countries, as banks see them as big risks due to their informal management and lack of collateral. As a result, only about 25-30 percent of businesses with 5-99 employees in the typical East African country have access to a bank loan or line of credit[1].

In addition, agriculture as a sector is highly underfinanced. In East Africa, the sector contributes 25-30 percent of GDP and often employs over half of the labor force. Yet in most East African countries, agriculture accounts for just 2-5 percent of domestic bank lending. For many banks, agriculture is not a priority due to risks that banks struggle to understand and manage (e.g. drought or price volatility), plus the high cost of serving rural customers.

Given the barriers to SME finance and the lack of finance for agriculture, it is no surprise that many agricultural SMEs in East Africa and elsewhere struggle to find appropriate financing for working capital and investment needs.

According to our research, the answer is…

So how best to close the financing gap for agri-SMEs? With a grant from DFID’s Impact Programme, along with funding from Ceniarth, Omidyar Network and Small Foundation, and with the support of the Council on Smallholder Agricultural Finance (CSAF), we gathered data from 28 lenders, including 11 social lenders with global portfolios and 17 domestic lenders from East Africa. Twenty lenders gave us various types of quantitative data on their agri-SME loan portfolio; we held in-depth interviews with another eight.

This analysis presented a complex picture of agri-SME lending profitability. Profitability varies based on lender type (i.e., deposit-taking bank vs non-bank financial institution vs impact-oriented social lender), loan size (roughly, from US$10K to US$1M) and lender strategy, organizational structure and human resources. Each lender faces distinct challenges.

We believe donor programs should be calibrated to the economic realities of lending to underserved sectors while still encouraging improvement and innovation. So, having analyzed the data for East Africa, our answer to the original question is a firm “all of the above.”

An overall view on the profitability of agri-SME lending

According to the data we collected, each type of lender has areas of strength and weakness – no model is profitable across the board. The chart below shows roughly how expected profit per loan varies with loan size for a “generic” lender of a given type in the markets we looked at. We’ll look at some of the drivers of these results in the next section.

Figure 1: Expected annualized net profit, by loan size and lender type

First, a key finding was that many agri-SME loans from commercial banks (the yellow line in Figure 1) are profitable even at sizes as small as $40-50K. Banks in our dataset had relatively low operating costs, driven by entirely domestic, lower-cost workforces and operating models optimized for volume. Also, as shown in Figure 2, they earn high interest yields – on average about 22-23 percent per year – on their loans. Even after adjusting for macroeconomic conditions in the countries where these banks operate, these local currency yields are 7-8 percentage points higher than the realized yields from hard-currency lending by global social lenders in our data.

Figure 2: Average realized revenue yields, annualized[2]

However, you will notice that the yellow line in Figure 1 stops at $100K. We know banks make some agri-loans above this size, but the agriculture and agri-SME business units we worked with did not make loans over $100K. Sometimes this was due to risk perception; other times it was because these loans are made out of general commercial banking units. We also know from industry engagement that the supply of credit over $100K is likely limited, and, as the Kenya Bankers Association noted last year, “the primary underlying reason for [low levels of bank lending to agriculture] is that the risk-adjusted returns to capital are too low to justify commercial lending to agriculture when other opportunities exist.”[3] Figuring out the economics of $100K+ bank loans to agri-SMEs remains a challenge on the research agenda and something we will tackle in a follow-on study later this year.

Turning to local non-bank lenders (NBFIs, the green line in Figure 1), we see a much higher size required to break even. Indeed, many of the NBFI loans we looked at were not yet profitable due to a combination of small size (and thus low overall revenue) plus higher operating costs than commercial banks. As we’ll see in the next section, these high operating costs were often due to the small current scale of the NBFIs we looked at; most were founded quite recently and are still growing into an appropriate scale for their fixed cost base. Without access to deposits, the cost of funds was also an issue for NBFIs.

The blue line in Figure 1 – global social lenders – appears mostly below break-even, but it’s important to realize that these social lenders fill an important niche in East Africa. A a survey we did showed only ~20 percent of their borrowers had loans from a commercial bank when they first came to a social lender. That’s because (unlike NBFIs or banks we looked at), they focus on loans of $200K - $2M+ to larger SMEs. Social lenders can serve larger borrowers profitably. We estimate the break-even loan for a “generic” social lender is around $1.2M, but some lenders break even as low as $750K. Low profitability among social lenders is driven by high operating costs – social lenders employ a mix of local and international staff and invest significantly in due diligence, given the large loan sizes – and volatile, but often high, credit losses. When combined with the low realized yields we mentioned earlier, margins are very tight.

Overall, our analysis of loan- and lender-level financial data revealed the need to address challenges including high operating costs, high and variable credit losses, and the high cost of capital, all of which were obstacles for one type of lender or another.

Operating costs and credit risk – key issues for donors to address

One driver of lower profitability on smaller loans for NBFIs and social lenders is their current small scale relative to their fixed costs. When we looked at the average operating cost per loan for these types of lenders, we found a steady downward trend as they grew; average costs typically dropped by 50 percent+ (see Figure 3). This is great news and implies that one donor intervention could be providing low-cost patient capital to small but growing lenders with innovative models, to support them during their growth phase.

Figure 3: Average operating costs (USD / loan) vs year of origination

It’s also worth looking at credit losses in our dataset. As Figure 4 below shows, typical credit losses experienced by lenders in our sample ranged from about 3-5 percent of the average amount at risk per year. Local banks had lower losses in this data (although we’d want a larger sample to confirm this), and we also noticed lower credit losses for social lenders as borrowers became larger and presumably more experienced and formalized.

Figure 4: Average annualized credit losses by lender type

We also noticed higher credit losses in lenders without specialized agriculture or agri-SME business units. While this sample is small and may not be generalizable, it hints at an interesting question for further research.

Overall, these credit loss figures – which are already higher than banks would be comfortable with in other sectors (despite banks preferentially selecting only the most “bankable” borrowers), and higher than social lenders experience in more mature agri-SME markets like Latin America – suggest donor interventions such as new types of credit guarantee schemes to absorb risk among newer or less formal borrowers, or capacity building for lenders with limited agricultural experience, to improve their underwriting and risk assessment.

What lenders themselves say

In addition to examining the data, we also asked representatives from eight domestic banks and NBFIs what their biggest challenges to increasing agri-lending were and which of five hypothetical support models they would rank highest. Figure 5 shows the results.

Figure 5: Results of Dalberg interviews with 8 local banks and NBFIs

Overall, what jumps out is the spread of answers. Domestic lenders did not have one single preferred mode of support, which is one reason we say “all of the above” is our preferred answer! However, they did express a unified concern with risk, suggesting the strongest need for innovation is in risk management. This is not surprising given the loss figures discussed in the last section.

While there are many partial risk-sharing facilities out there, typically covering 50 percent of losses on a qualifying loan, lenders expressed a desire for new types of risk-sharing that would offer more certainty in untested market segments. That’s one reason we are especially excited about a first-loss guarantee program. This type of scheme, if designed with appropriate qualification rules to avoid moral hazard, could allow donors to achieve significant leverage, with each dollar of grant funding supporting $10 or more in lending to underserved segments.

Conclusions and what we’re doing next to fill the agri-SME financing gap

Overall, our research suggests that donors may want to explore a full range of interventions to encourage agri-SME lending. At the very least, we recommend interventions such as:

• Cash incentive payments to defray the operating costs associated with reaching rural SMEs and underwriting loans in new value chains. These incentive payments would be a small portion of the overall loan size, again offering a high leverage ratio on donor capital.
• First-loss credit guarantee schemes to cover the first portion of losses on high-impact segments that are also higher risk, such as new borrowers or SMEs working in staple crop value chains.
• Technical assistance to lenders to improve their agricultural knowledge and risk management practices, as well as TA for borrowers to make them more “bankable” by improving financial management, governance and business processes.
• Patient and low-cost capital to allow innovative financial services providers with an interest in agri-lending to scale up.

We’re already collaborating with one exciting initiative – the Prosper Africa initiative led by CSAF and the Global Development Incubator – to roll out some of these ideas in East Africa. Prosper Africa plans to use an iterative process of testing different forms of support, gathering more benchmarking data from participating lenders, and gradually refining and expanding. We think this is the right way to balance the desire for data-driven aid deployment with the urgent priority of increasing finance for agricultural SMEs across Africa.

Please read our Summary Report for more details on the points above or the full version for even more information about the methodology, key findings and recommendations for the donor community. And keep watching Agrilinks for further research on the economics of agri-lending in East Africa and elsewhere, as we try to close the knowledge gap and help donors make data-driven decisions about how to catalyze lending to this important sector.

[1] Enterprise Surveys (http://www.enterprisesurveys.org), The World Bank. Indicator: “Percent of firms with bank loan/line of credit;” various years 2006-2013 depending on country.

[2] “Realized yield” means actual interest and fee income received, divided by the dollar-years of lending outstanding. Yields shown here exclude losses of principal, but reflect economic losses due to foregone interest, delayed payments or restructuring.

[3] Kenya Bankers Association, Realisation of Full Potential of the Agriculture Sector: Is Commercial Financing a Core Missing Cog?, 2018, p. vii.