An analysis of factors affecting access to credit in Lesotho's smallholder agricultural sector

Abstract

The agricultural sector in Lesotho is underperforming mainly due to the inability of smallholders to move from traditional agriculture to a more scientific and technology-based one. Among the challenges inhibiting the ability of smallholders to make the step up is access to financial services, especially credit. The purpose of this study was to examine the factors that may influence the ability of smallholders to access finance by making use of a logistic regression model within the principle component regression framework. The results revealed that the ability of smallholders to access finance, and the potential to make the transition towards a more scientific and technology-based agriculture sector, is influenced by the level of farm and non-farm income, remittances and pension, farm size, availability of family labour, land ownership, savings and repayment ability. The results present important information in terms of guiding institutional arrangements needed to improve credit availability in Lesotho.

Keywords:

• agricultural credit accessibility
• smallholders
• logistic regression model

1. Introduction

Agriculture is the most important contributor to Lesotho's economy and supports the livelihoods of a large part of the population. In addition, agriculture forms a major source of economic growth in the country. Although being the mainstay of the economy, the sector is underperforming. This performance could mainly be attributed to the inability of smallholders to move from traditional agriculture to a more scientific and technology-based one. Among the challenges inhibiting their ability to make the step up is access to credit. Access to credit can be defined in many ways; however, a similar definition to the one used by Manganhele (2010) will be adopted for this study. Therefore, access to credit is defined as situations in which smallholders have the ability to tap into particular sources of capital that will enable them to obtain optimal inputs, such as improved seed cultivars or genetics, fertiliser, machinery, and so forth. Experiences in many developing communities demonstrated that access to credit could accelerate the adoption of new technologies (Manganhele, 2010). The aim of this study is therefore to examine the factors that influence the ability of smallholders to access credit in Lesotho.

The contribution of the agricultural sector to Lesotho's gross domestic product has declined from 25% in the 1980s to less than 10% in recent years (Central Bank of Lesotho, 2012). Dry-land crop producers are responsible for 70% of the agricultural gross domestic product, with the remaining 30% being made up by the livestock sector. The crop production system is inefficient and is characterised by low-input low-output traditional farming systems. Agricultural practices and machinery are old-fashioned with livestock that remains central in providing draught power for cultivation and the preparation of land. Similarly, the applications of locally produced inputs and family labour are common practice. In addition to draught power, livestock are essential for smallholders to maintain food security. The bulk of crops and livestock are grown in small rural remote villages that are poorly linked to markets. Most smallholders are of subsistent nature; however, land tenure in Lesotho is characterised by farmers who own their farmlands. This may be attributed to the fact that land is inherited in Lesotho; therefore, it is rare (but not unlikely) to find farmers farming on rented or communal lands.

From the above, it is evident that smallholder agriculture in Lesotho is traditional. As mentioned, among the factors inhibiting their efficiency or ability to move from a traditional to a more scientific and technology-based system is the lack of access to credit. The challenge of credit is not unique to smallholders in Lesotho. Kuhn et al. (2000) argue that the lack of access to credit inhibits agricultural development in less developed countries. The authors elaborated by arguing that, apart from the efforts of governments to ensure that smallholders have access to credit, the provision of financial services to these smallholders has generally been stagnant and has even declined in some developing countries because of the risks involved in dealing with smallholders.

The situation is no different in Lesotho, with the number of financial services having actually declined over time. The financial sector of Lesotho is characterised by a formal financial sector, the absence of a sizable micro-finance sector, and a strong informal financial sector. Although Lesotho has a strong informal financial sector, it is full of ambiguities, which makes it difficult to determine the extent and role that this sector plays in terms of providing rural agricultural finance. Similarly, despite being small, the existing micro-credit platforms in Lesotho are limited in scope and the contribution is also influenced by operational discrepancies. For instance, lending conditions, scope, target groups and interest rates charged are notably different between different micro-finance programmes.

The formal financial sector in Lesotho, like many other developing countries, is characterised by government intervention. These interventions are justified on grounds of market failure to deliver the needed financial services to the rural people to support their development initiatives. Past interventions by the government, which were aimed at improving rural finance to fulfil the national objectives of increased production, included among others the establishment of the Lesotho Agricultural Development Bank (LADB) with the objective to serve as the leading organisation for rural savings and a central source for agricultural credit. Under the auspices of group lending, Lesotho's government also established a block farming programme, whereby small tracks of land from different farmers are grouped together into larger, more economically viable and productive blocks, with the government providing a 30% subsidy and a 100% guarantee to commercial banks who offer loans to block farmers. Lately, emphasis has also been placed on project appraisals coupled with more relaxed collateral requirements and the charging of close-to-market interest rates. Most of these interventions, practices and operational procedures are geared toward the interest of borrowers (Spio, 2002).

These approaches (policies), however, invariably resulted in distortions in the financial markets and reduced the number of financial products and services to which smallholders have access. The reasons for distortions and a decline in the number of financial products and services are quite complex and include, among others, the fact that: fully-fledged commercial banks are subsidiaries of foreign banks and their main business is not only to provide financial services to companies operating in Lesotho, but also to companies in neighbouring countries; the repayment culture of the locals (Basotho) is not very pronounced, and many individuals and companies borrow from parasternal credit institutions without being forced to pay back their loans; local infrastructure to avoid over-borrowing and double-borrowing, such as credit bureaus and the issue of identity cards to all citizens, are not in place; and the absence of a functional commercial court with accelerated proceedings and the rapid execution of court decisions against debtors (Finmark Trust, 2003).

Moreover, agricultural credit from the formal sector used to be provided mainly by the LADB, which has since been closed. According to Maili (2003), liquidation of the LADB had a significant impact on domestic economic productivity. The vacuum that is left by, among others, the closure of the LADB and other institutional shortcomings, as discussed, makes it crucial that an appropriate institutional framework is developed in order to address the provision of financial services for small-scale farmers. As mentioned, the inability of farmers to access financial services, and in particular credit, is negatively affecting the economy of the country. Moreover, it is believed that accessibility to credit can help reduce poverty and food insecurities by increasing rural incomes through improved agricultural production. Before an institutional framework can be developed, decision-makers need to understand the factors that influence smallholders’ access to credit. In other words, the focus of this study will be to determine the factors that influence access to credit and not to develop an institutional framework for credit delivery in Lesotho.

The following section provides a literature review on the theoretical framework followed by the methodological approach, results and, finally, conclusions and recommendations.

2. Literature review

Recent theoretical and empirical work in economics has established that credit markets in developing countries work inefficiently due to a number of market imperfections (Bell et al., 1997). According to Foltz (2004) and Carter (1989), these imperfections include interest rate ceilings, monopoly power, inflated transaction costs and moral hazard problems. Often, a combination of these imperfections combines to ration farmers out of the loan market.

To date, no previous studies have attempted to determine the relationship between access to credit and agricultural production in Lesotho; however, similar studies to this one have been conducted in many countries around the world. For instance, Foltz (2004) studied how access to capital affects agricultural profits and investment in Tunisia. Using secondary data, econometric estimates in the form of a switching regression model framework were employed to determine the impact of access to credit on profitability and investment levels in the agricultural sector. Similarly, Nuryartono et al. (2005) used both a probit model and a switching regression model to investigate the impact that access to financial services and credit constraints will have on the level of agricultural production in a rural region of Indonesia. A study by Spio (2002) also employed a switching regression model framework to determine the impact of credit on agricultural production and a logistic regression model to determine the accessibility of credit to smallholders in the Northern Province of South Africa. Mokoena et al. (1997) used a probit model to identify factors that determine women farmers’ access to credit in South Africa. On the same note, Eze et al. (2009) investigated women's access to credit from selected commercial banks in south-east Nigeria by making use of a logistic regression model.

Moreover, studies by Mohamed (2003), Subbotin (2005) and Kohansal & Mansoori (2009) also employed logistic regression models to study factors that influence access to credit by smallholder fisherman in Zanzibar, access to credit for corporate farms in Russia and factors influencing the repayment behaviour of farmers in Khorasan-Razavi, Province of Iran, respectively.

In all of the above studies, explanatory variables include human capital (education levels, age or experience), capital formation (land ownership or tenure security, farm size, average yield, machinery, other household assets, non-farm income, remittance, pension and savings), socio-economic variables (availability of family labour, gender), and the credit history of applicants (previous loans, default history, awareness of credit facilities). The significance or influence, if any, of each of these variables, however, differs between the different countries and regions. This will therefore form the basis of the empirical analysis in this study. It should be noted that studies focusing on access to credit and related issues are not confined to the above mentioned; however, these variables are closely related to the research question of this particular study.

Moreover, as can be concluded from the above-mentioned literature, the methodological approach used to determine the relationship between a dependent variable (access to credit) and independent variables (human capital, capital formation, etc.) is dominated by regression model frameworks.

3. Methodology and data

3.1 Data used

The study used primary data. The data were collected by means of a household survey conducted during the first quarter of 2008. The survey included a wide range of questions, firstly capturing demographic data and secondly on-farm and financial information that may influence the ability of smallholders to access credit. The questionnaire was completed by means of personal interviews. In order to improve willingness to participate among smallholders, selected smallholders were contacted in advance either directly or through the area extension officer.

A simple random sampling technique was employed, which covered 10 villages, representing approximately 30% of the total number of villages in the target areas − the two largest agro-ecological zones of Lesotho – the Lowlands (both northern and southern) and the Highlands regions. Stratified random sampling was employed to select borrowers and non-borrowers for the study, and this entailed dividing the entire farmer population into mutually exclusive strata according to the targeted ecological zones and then randomly selecting units from each stratum. Random sampling was applied within each stratum, because this often improves the representativeness of the sample by reducing the sampling error (Babbie, 2001). In total, 100 farmers were interviewed (see Table 1).

Table 1: Distribution of borrowers and non-borrowers (n = 100)

Region	Borrowers	Non-borrowers	Total
Lowlands	19	41	60
Highlands	13	27	40
Total	32	68	100

3.2 Model

When analysing factors that limit the ability of small-scale farmers to access credit, the dependent variable is either binary or categorical. This implies that farmers have access to credit or that these farmers do not have access to credit. Normally, this is represented by one or zero, where one represents access to credit and zero represents no access to credit. In addition to the logistic regression modelling framework, several other modelling frameworks can be used to model the relationship between a categorically dependent variable and a number of independent variables. These include probits, tobits or even ordinary least square or discriminant function analysis; however, a probit analysis is normally used when the dependable variable reflects an underlying quantitative variable. This implies that a probit analysis is inappropriate when the dependent variable is qualitative. This statement is also supported by the theory of normal probability distribution (Montshwe, 2006). For the same reason, a tobit analysis is not appropriate to analyse factors that limit small-scale farmers’ access to credit (Spio, 2002).

The logistic regression modelling framework is more general compared with the above mentioned, provided that the independent variable is not restricted to a categorical dependent variable or limited to a single independent variable (Montshwe, 2006). Therefore, the logistic regression model was considered the most appropriate modelling framework. The estimation of the logistic regression model was guided by the dependent variable, considering that the purpose is to determine the factors influencing small-scale farmers’ access to credit. Independent variables include human capital, capital formation, as well as socio-economic and financial variables. The list of specific variables and the expected outcomes are portrayed in Table 2.

Table 2: Data specifications: credit status equation (logistic regression model)

Variable	A priori expectation
Dependent variable 1 = Access to loan 0 = Otherwise
Independent variables
Age of the household head in years	Age is expected to negatively affect the probability of being a borrower, because comparatively older farmers are not as active in their farming activities.
Farm income (previous year)	A high farm income may reduce the demand for credit. On the other hand, it may increase the farmer's creditworthiness and in some cases create a demand to expand production. Hence, the effect of farm income is indeterminate.
Non-farm income (previous year)	Non-farm income is expected to reduce demand for credit and can be used to purchase cash inputs. The coefficient is expected to be negative.
Financial assets (savings)	Savings can influence the supply and demand size differently; its sign therefore is indeterminate.
Remittances and pension	This variable is expected to reduce the demand for credit. Its sign is expected to be negative.
Farm size in hectares	Farm size is expected to positively affect the amount of the loan, as there is greater need for variable cash inputs, and it is expected to increase capital access.
Family labour stock	The effect of family labour stock is indeterminate.
Land ownership (dummy: 1 = title deed, 0 = otherwise)	Ownership as opposed to rental and other forms of access to land is expected to increase the long-run investment incentives and the collateral value of the land to lenders. Its sign is expected to be positive.
Awareness of credit facilities in the area (dummy: 1 = yes, 0 = no)	Farmers’ awareness of credit channels available in their area is likely to have a positive bearing on their accessibility to credit.
Gender (dummy: 1 = female, 0 = male)	Males are expected to have greater access to credit than females; hence the gender sign is expected to be positive.
Education (dummy: 1 = formal education, 0 = otherwise)	Literacy status is expected to positively influence farmers’ access to credit institutions because literate farmers are assumed to have a better technical knowledge of and information about markets and facilities provided by financial institutions.
Repayment (dummy: 1 = good repayment record, 0 = bad repayment record)	A good repayment record is expected to positively affect borrowing. The coefficient is expected to be positive.

Similar to most modelling frameworks, logistic regression models are also subject to weaknesses, among which is multicollinearity. Principle component analysis is considered relevant to solve the problem of multicollinearity (Leedy, 1994). The logistic regression model was therefore estimated within the principal component regression (PCR) framework. This implies that the variables included in the logistic regression model are subjected to principle component analysis in order to reduce the variables into a few uncorrelated principle components. After the principle components were calculated, those with the smallest eigenvalues were eliminated and the PCR was fitted using standardised variables to improve the assessment ability of the logistic regression model.

This methodological approach is not novel. As a result, the specification and estimation of the model will not be discussed in detail. A detailed discussion on the specification and the estimation of the logistic regression and PCR model, which is similar to the application in this study, can be seen in Montshwe (2006).

4. Results and discussion

The results of the model are presented within the PCR framework. The use or non-use of credit sources is explained by using the analysis. The logistic regression model can directly estimate the probability of an event occurring. The model predicts whether or not an event will occur, and identifies the variables useful for making this prediction. Table 3 presents the results of the estimates.

Table 3: Logistic regression estimates

Variables	Coefficient	Standard error	t ratio	Probability
Constant	−1.8869	0.4711	−4.0051	0.0001
Gender	0.1763	0.3492	0.5048	0.6149
Age	−0.3049	0.2553	−1.1941	0.2356
Farm income	1.0021	0.2189	4.5785	0.0001***
Non-farm income	−0.0064	0.3779	−0.0169	0.0865*
Remittances and pension	−0.2188	0.4056	0.5393	0.0910*
Savings	−0.5657	0.3332	−1.6978	0.0930*
Awareness	−0.1528	0.2562	−0.5962	0.5525
Educational level	−0.1661	0.2538	−0.6546	0.5144
Tenure	−0.6586	0.2025	−3.2523	0.0016***
Farm size	0.7784	0.2070	3.7605	0.0003***
Household size	−0.4063	0.3619	−1.1228	0.2645
Family labour	−0.3271	0.2677	−1.2218	0.0249**
Repayment	−0.8569	0.2090	−4.0999	0.0001***
Log likelihood	−23.7282

Notes: ***Significant at 1%.

**Significant at 5%.

*Significant at 10%.

The variables used to explain the household credit constraints are a response to both demand-side and supply-side circumstances. As mentioned, access to the loan variable was regressed on age, farm income, non-farm income, financial assets (savings), remittances and pension, farm size, family labour, land ownership, credit awareness, gender, education level and repayment ability (see Table 2).

In the model, the coefficients of only eight out of 13 possible explanatory variables are significant, at a minimum significance level of 10%. In other words, eight out of the 13 explanatory variables included are likely to influence the chances of individuals accessing credit from formal and non-formal credit sources in Lesotho. These include farm income, non-farm income, remittances and pension, farm size, family labour, land ownership, savings and repayment ability. The following sub-sections will elaborate on these findings.

4.1 Income levels

Non-farm income, remittances and pensions, as a proxy for welfare status, confirm that increasing the total income of households reduces the probability of households being credit constrained. These variables have the expected negative signs and are both significant at 10%. These results show that households have less demand for loans because of their own equity capital accumulated through past income earnings and may use this income to purchase cash inputs.

This finding is consistent with the pecking order theory, which states that farmers will choose from a hierarchy of preferences in deciding on the source of finance to utilise (Spio, 2002). Laper et al. (1995) state that the choice is based on the ‘safety first principle’, with internal funds being the safest among the choices. The authors further state that the more assets these farmers have, the more likely it is that they will not seek external funds but will utilise internal resources to operate their farms. The results validate this statement. The other reason, however, might be the poor repayment rates; most farmers might have been denied loans because they had previously defaulted. The significant negative coefficient of the repayment variable validates this statement. Lenders consider the welfare status of clients or potential clients before signing contracts to provide loans. Farm income, on the other hand, is positive and significant at 1%, confirming that a higher farm income may improve the creditworthiness of farmers and in some cases may create a demand to expand production, thereby increasing the demand for credit.

4.2 Farm size

Farm size has the expected positive sign and is significantly different from zero at 1%, with a coefficient of 0.7784, suggesting that a unit increase in the size of farms is more likely to increase the chances of farmers to obtain loans, and further suggesting that the bigger the farm size, the more likely it is that farmers would obtain loans. Sial & Carter (1996) support this hypothesis by stating that larger farm sizes affect the amount of loans needed through a greater need for variable cash inputs; consequently increasing the need for credit. The results are further supported by Mbowa & Nieuwoudt (1999) and Binswanger et al. (1993), who pointed out that transaction costs associated with many small loans act as a disincentive and that the cost of credit to smallholders is more likely to increase, thereby discouraging farmers from applying for loans. In the presence of fixed transaction costs, the cost of borrowing in the formal credit market is therefore a declining function of the farm size. This result is consistent with other results (Kashuliza & Kydd, 1996; Mokoena et al., 1997; Spio, 2002).

4.3 Family labour

Family labour stock, on the other hand, has a negative sign and is significant at 5% with a coefficient of −0.3271. This result shows that a unit increase in family labour stock will decrease the demand for loans. On the one hand, the result suggests that larger farm families have a lower tendency to obtain loans. Family members may substitute labour for cash inputs, such as herbicides, and/or sell additional family labour on the market, and in return use off-farm income to purchase cash inputs, consequently reducing the need for loans. On the other hand, this result may mean that households with larger families tend to be poor and in most cases may not qualify for loans. As mentioned earlier, lenders consider the welfare status of clients or potential clients before signing contracts to provide loans. The result is consistent with other results – for instance Nuryartono et al. (2005), who reported that households with larger families tend to use family members for labour, but the larger the number of household members, the greater the probability of credit constraints.

4.4 Land ownership

Land ownership (tenure) as opposed to rental was expected to improve the ability of farmers to obtain loans due to the collateral value of land to lenders (FAO, 1996). Land ownership, however, is significant in this model, but does not have the expected positive coefficient. The negative relationship between land ownership and access to credit could most probably be explained by the fact that most borrowers were participants in the government's programme of block farming, where collateral was not asked and issues of land tenure and ownership were not considered in loan approvals and disbursement processes.

4.5 Savings

Savings are significant at 10% and have a negative coefficient, indicating a negative relationship between access to credit and savings. This shows that savings decrease the demand for credit and are expected to substitute credit. Saving accounts in the study area have little value to lenders as a source of informal collateral.

4.6 Loan repayment

A good repayment record is expected to affect borrowing positively. The repayment coefficient was expected to be positive; however, it has an unexpected negative coefficient, and is significant at 1%. The negative relationship between repayment and access to credit in the case of Lesotho could have been caused by the fact that most loans come from the Ministry of Agriculture and Food Security through its programme of block farming. These loans are mainly production loans. The government of Lesotho offers a 30% subsidy and a 100% guarantee on all loans offered to block farmers. This, as a result, fails to achieve the objective of changing the mind-set of farmers that loans are not a subsidy and therefore have to be paid back, even when yields are low.

5. Conclusions and recommendations

In Lesotho, the small-scale farming sector continues to battle with the task of moving from a traditional agriculture to a more scientific and technology-based one, and this consequently leads to the poor performance of the agricultural sector. Moreover, results from the empirical analysis indicate that the ability of small-scale farmers to access credit is influenced by the level of farm and non-farm income, remittances and pension, farm size, availability of family labour, land ownership, savings and repayment ability. Farm income, non-farm income, savings and remittances, and pensions confirm that by increasing the total income of households, the probability of households being credit constrained is reduced. This shows that better household situations could either reduce the demand for credit or could positively influence lenders to assist with funding specific farming endeavours. A lower credit demand due to good household situations does not necessarily affect the ability of smallholders to move towards modern agricultural practices.

In addition, farm size also has a positive relationship with the demand for credit, which suggests that a unit increase in farm size is likely to increase the demand of individuals for credit and that the bigger the farm size, the more likely it is that farmers would obtain loans. Land ownership, on the other hand, showed an unexpected negative relationship with credit. The negative relationship between land ownership and access to credit could be either that most borrowers obtained their loans from semi-formal or informal institutions, or that they were participants in the government's programme of block farming, where collateral was not asked, issues of land tenure and ownership were not considered, part of these loans was subsidised and guarantees were provided. The same accounts for the negative relationship between repayment ability and history, with the expectation that a good repayment history will be required to obtain loans; however, the rather low repayment culture among smallholders in Lesotho is also offset by the block farming programme. As a result, the block farming programme is synonymous with high default rates, which inhibit long-term sustainability. Moreover, the programme lacks a strategy and institutional capacity to improve business skills and to enforce mechanisms for timely loan repayments upon the recipients of the subsidised credit. In addition, political interference, the lack of a credit culture and discipline on the part of beneficiaries also contribute towards the high default rates.

Although it cannot be ignored that donor facilitation is needed in the development of rural financial markets, it is clear that it needs to be accompanied by the necessary capacity-building and training programmes. Moreover, training programmes must be complemented by evaluation and monitoring mechanisms that should also be practised to keep smallholder borrowers informed and reminded of their loan repayment obligations. It is also recommended that institutional arrangements which do not promote a ‘dependency syndrome’ are prioritised in policy-making.

Disclosure statement

No potential conflict of interest was reported by the authors.