Evolution of the delivery of financial services in IFAD Projects

IFAD's projects in China have used two models (the PMO and RCC models) for the delivery of financial services. It is difficult to compare the two models by means of conventional performance measurement tools. They need, rather, to be analysed using a 'strengths and weakness' mode of analysis. The supply-driven credit delivery mode needs to be contrasted with the sustainable financial service delivery mode.

Comparison of financial services delivery models

Products

The IFAD projects using the PMO model offered only credit products, whereas the RCC model envisaged the integration of IFAD project funds into the financial system. Such arrangement was expected to develop the capability of RCCs to provide a full range of financial services to rural households, compared with the PMO model which provided only credit.

Targeting

The PMOs and RCCs established four primary criteria to select households for credit delivery: (i) potential loan repayment ability; (ii) skills or techniques to undertake an activity; (iii) households having designed and submitted a suitable plan for using credit funds; and (iv) ability to mobilize their own capital.

Review of targeting performance

The projects' targeting criteria would have contributed to effectively excluding some of the poor households. The projects did not devise systems to develop the capability of resource-poor households to receive credit. Three factors contributed to excluding the resource-poor: (i) projects did not envisage that the initial requirements of resource-poor were small amounts of funds to deal with household emergencies and/or sickness; (ii) household choice was limited since project loans were for productive purposes only; and (iii) project loans excluded households that received food subsidies from the civil administration and had inadequate labour.

The survey data show that households that had received project loans were in the better-off strata of society. The mission interviewed 96 households, of which half had received loans and half served as a control group. The mean per capita income of households that had not received project loans was about 45% below that of households that had. Only 19% of households that received project loans had a food deficit compared to 38% that had not. A higher percentage of households with project loans had access to other institutional finance; in contrast, as expected, a higher percentage of households without project loans used informal credit. The value of durable goods in households that had taken project loans was 18 times that of households that had not taken out such loans.

It is true that in the poor areas of rural China, many resource-poor households may have limited credit absorption capacity, particularly for productive investments. The real issue, however, is that local officials, including RCC and PMO officials, are more inclined to grant loans to better-off households. Such officials are more concerned about the ability of clients to repay their loans than with finding ways to reduce poverty. Poor households are usually to be found in remote mountainous areas and the cost of outreach, for delivering credit and monitoring such groups, is high.

Two lessons emerge here: (i) the perception that the poor cannot repay their loans is pervasive among project implementers; and (ii) the cost of delivery to poor households is very high. The targeting of poor households calls for a two-pronged strategy: loan products should match the requirements, absorption capacity and risk profile of the poor; and the financial institution involved should to be given incentives in the form of spread to deliver and recover credit.

Financial service delivery

Loan purpose. IFAD projects have placed emphasis on credit to improve agricultural productivity, usually medium-term loans for crop and tree planting and livestock-related activities. With the PMO model, loans have been provided as envisaged in project design, but have neglected the consumption needs vital to resource-poor households. In contrast, the RCC model not only envisages providing loans for productive activities using project funds but also involves lending to cater for consumption, using RCC funding.

Loan size and duration. The size of loans in the PMO and RCC models depends on project loan regulations and the activities for which loans are sought. Generally, projects provide uniform loans that are not necessarily in line with household requirements. However, RCCs can provide loans according to the needs of the households, repayment capacity and collateral. Both institutional models were designed to provide mostly medium- and long-term loans as stipulated in the project documents, but as a general rule RCCs provide only short-term loans.

Collateral requirements. RCCs generally seek collateral, and in practice this excludes the resource-poor farmers. The PMO model does not require collateral for loans. The RCC model does not envisage collateral for loans of up to CNY 3 000 but, in practice, RCCs also request collateral for loans of less than CNY 3 000.

Repayment period. The repayment period is dictated by the credit guidelines for both models and is fixed on the basis of the activity funded rather than the client's repayment capacity. Projects should not establish repayment schedules; this is better left to the financial institutions at the time of loan approval.

Interest rates. Borrowers pay less interest than the normal RCC lending rate. Charging a higher rate of interest continues to be a contentious issue. The project authorities still maintain that a low rate of interest is an essential requirement for credit uptake by resource-poor households.

Flexibility to adjust lending rates. Neither model has sufficient flexibility to adjust lending rates to cover cost and risk. Microfinance projects have been allowed to charge more interest than regular financial institutions. A case in point is the Grameen Bank pilot project in Yixuan County in Hebei Province, where the lending rate is 16% per annum. Higher rates of interest mostly act as a selection mechanism for targeting because, for resource-poor households, access to finance is more important than the rate of interest.

Recovery performance. PMO officials reported a 70-80% recovery performance, but no records were provided to substantiate this and there was some doubt about the reliability of data provided by PMOs. Supervision reports suggest that inadequate performance at the county level has also been caused by delays in communications and instructions from townships to counties. As the RCCs have not kept separate accounts for the IFAD project loans, it was not possible to assess recovery performance. However, overall, the recovery performance for RCC loans to households was about 80%.

Credit risk. The local township government bears the credit risk in respect of loans granted by PMOs. The RCC model envisaged passing on the credit risk to the RCCs. In reality, even for projects using the RCC model, the credit risk has continued to rest with the local government. When RCCs do not have autonomy in selecting borrowers, they are understandably not prepared to take the credit risk. Since township governments are responsible for repaying IFAD loans to higher levels of government, they wish to exercise control over borrower selection and loan disbursement.

Training. IFAD projects have trained PMO staff in credit delivery and preparation of credit guidelines for the PMO model. In projects using RCCs, allocations for training RCC staff were not being used as the RCCs were unwilling to use loan funds for capacity building. Instead, IFAD has collaborated with the German Agency for Technical Cooperation (GTZ) to provide technical assistance to RCCs.

Earlier IFAD projects provided little training to households, but this problem has since been redressed by collaborating with the World Food Programme (WFP). One very positive finding is that training programmes have been effective, with close links observed between the purpose of the loan and the training involved. Almost 90% of all credit recipients were trained in relevant economic activities.

In addition, about 20% of households without project loans received training, most of which focused on technical issues and skills necessary for maintaining project-funded activities. Training programmes on general and reproductive health were also provided to women. Collaboration with WFP effectively enhanced credit utilization. The study team also found that there was little awareness of credit terms, such as interest rates, repayment frequency and periods, especially in terms of the PMO model.

The mission found no hard evidence to suggest that WFP collaboration had helped to improve resource-poor households' credit uptake and the provision of appropriate credit products. To some extent, increased attendance at meetings was due to the fact that food was distributed free at such gatherings. As a general rule, credit projects have by-passed the poor. Although the study was not specifically designed to test the impact of food distribution associated with training, it would be worthy of further evaluation.

Accounting system under the PMO model. The PMO model's accounting system is inadequate. The situation in respect of RCCs is better, as their management information system is superior. However, procedures related to asset classification, loan loss provisioning and accounting practices do not conform to international standards.

Revolving fund operation. The extent to which funds are being recycled with the PMO model remains a moot point. The design of IFAD projects incorporated two important safeguards to ensure such recycling. First, IFAD set out to disburse only incremental credit, but this was not found to occur in all the townships visited by the mission. The amount recovered is kept in a separate account, and its use remains unclear. In contrast, with the RCC model, monies recovered remain within the RCC system and are used for credit delivery.

The second safeguard was that IFAD funds are meant to remain at the township level for 15 years, the intention being to ensure that they are used for providing credit to households even after project termination. This is more likely to happen in the RCC model, while for the PMO model it would depend on current local government priorities.

PMO and RCC Models Reassessed

In the absence of healthy and willing rural financial institutions, the PMO model has been instrumental in infusing much-needed capital into the rural economy. From a purely credit disbursement perspective, the PMO model appears more efficient. The coverage of households within localities with the PMO model (Sichuan Livestock Development Project) was about 66% compared to 20% with the RCC model (Southwest Anhui Integrated Agricultural Development Project).

The PMO model is a short-term measure, however, and cannot be further strengthened to deliver sustainable financial services. This model can function as an allocation mechanism only for as long as the project funds remain at the township level. The PMO model represents a supply-driven credit-delivery model, and it suffers from the known limitations of this approach. Funds are allocated for disbursement with little attention to loan appraisal, repayment capacity, client credit history, and guarantee and collateral requirements. Moreover, it does not incorporate microfinance methodologies to assist the resource-poor households to 'graduate,' i.e. to progress from initial small loans with short repayment periods and higher frequency of repayment, to ultimately access larger loan amounts with longer repayment periods for agricultural investments.

Borrower selection is based on an annual planning process with no flexibility to provide loans to households if needed during the middle of the year. Activity selection, repayment terms (period and frequency) are insufficiently flexible for household requirements, determined as they are by the requirements of the Staff Appraisal Report and credit manual specifications. Two very important weaknesses in the PMO model, constraining development of sustainable rural financial services, are: (i) inability to mobilize savings and recycle them for investment; and (ii) potential moral hazard, since borrowers may perceive that loans from government departments can be simply written off.

Despite their current limitations, RCCs are the mainstay of the rural financial sector in China. They have established a large network and are the only formally authorized financial institutions to serve rural households. Their contribution in mobilizing and deploying savings for investment is substantial, but their problems derive from a skewed policy environment with regulated interest rates that makes it difficult to reach out to dispersed rural households. The RCC network also suffers from performance deficits caused partly by historic burdens of bad debts and partly by ongoing operating weaknesses. But with the efforts of the PBC and through its direct supervision, RCCs are slowly but steadily being restructured. To increase the effective yield of the funds employed, microfinance methodologies are being built into the credit delivery systems of selected RCCs on a pilot basis.

The RCC model is intended to pass on the credit risk to RCCs by allowing them full authority to accept or reject loan applications forwarded by the PMO. RCCs mobilize savings and are under Central Bank supervision. This model is the way to proceed to integrate the project-directed credit component into the financial system. This is in line with a State Council directive to stop all lending activities within government units.

IFAD is the first international financial institution (IFI) to start mainstreaming financial services into the financial system of the country. Even the most recent World Bank projects use ABC as the agent for delivery of credit, and it was found that efforts for complete integration of project credit delivery into ABC had yet to begin. Within the IFAD project framework, attempts are being made to integrate RCCs for delivery of financial services, although this is difficult to achieve.

Factors impacting on the involvement of RCCs in IFAD projects

Procedural limitations pertaining to use of IFAD loans

The MOF has placed severe procedural limitations on the use of IFAD funds. These funds are passed as loans to the township level, and township PMOs are ultimately responsible for repayment. This makes it difficult to bring about the necessary interaction between PMOs and RCCs at the township level. The relevant regulations [see paragraph 9.1 of main report] constitute a twofold challenge to IFAD projects. First, MOF or finance departments at the provincial level cannot assign IFAD loans to PBC, RCCs or other financial institutions. Second, the township governments may not use IFAD funds for activities other than credit, as such funds have to be repaid. It does not appear to be a realistic proposition to expect governments of poor townships to repay the funds used for the development of rural infrastructure, training and other activities.

The PMO model has been a convenient tool in the context of current MOF procedures, as the borrower of IFAD loans. IFAD loans are passed on directly to the provinces and are repayable in 30 years after a ten-year grace period, and with associated exchange risk. The provincial government bears the exchange risk and passes the loan on to the subsequent lower level, repayable in 15 years after ten years' grace. The township government, which ultimately receives IFAD loan funds, has to repay the borrowed amount to the higher levels of administration.

IFAD credit funds are used as intended for lending to the households, but MOF regulations prevent townships from using such funds for infrastructure development and other related activities. Instead, townships use them for providing credit to households, which can be recovered and repaid. This means that IFAD's intentions are thwarted: the projects provide funds for infrastructure development and other activities but these funds are not used as required. Township governments use 100% government funds for rural infrastructure, training and other activities, excluding credit. This is in conflict with IFAD's intention as allocations have been made for such components in the project documents. Instead, township governments claim reimbursement from IFAD for all activities. The reimbursements received from IFAD for rural infrastructure and other activities are again deployed for providing credit to households. This is in addition to funds allocated for credit.

The mission's discussions and fact finding suggest that the entire IFAD loan is used for the credit component, while the township governments use counterpart funds to undertake other activities. The rationale behind this is that the IFAD funds are loan funds and will have to be returned or repaid, so they need to be deployed as loans and not for infrastructure and other investments that cannot be recovered directly from private households as 'repayment.' For as long as MOF uses the current procedure of passing on IFAD funds as loans, local governments at the township level will find it difficult to absorb IFAD credit for infrastructure and other activities and repay the same at a later date. This conclusion is based on probing questions and discussions with project management staff. The success of RCCs is predicated on a set of complex institutional issues that range far beyond the apparently simple one of margin availability.

The township, county and provincial governments are unwilling to place the funds - which they will ultimately have to repay - with the RCCs and thereby relinquish control over their use. They see no role for themselves in IFAD projects if the funds allocated for the credit component are passed on to the RCCs. It will be impossible to integrate PMO and RCC activities within the IFAD project framework if MOF continues its current loan deployment procedure. Instead, one possible solution would be for township governments to receive IFAD funds as grants from MOF for rural infrastructure, training and other activities, together with loan funds for the credit component.

The MOF argues that the major constraint on a shift to the RCC model is the poor health of RCCs and the risk of non-repayment of funds provided to them. This argument is not valid. First, the poor performance of RCCs is not due to the above-mentioned problems of the institution alone. Second, macro factors linked to government policy - including governance issues, historic bad loans and the policy environment - militate against generating longer-term sustainable growth for RCCs. Third, to advance reforms and permit higher rural economic growth and employment, the Government will need to assist the RCCs to restructure, supported by relevant enabling policies and regulatory framework.

Limited incentive Sstructure for RCCs

The RCCs have limited incentive to operate within the current framework of cooperation with PMOs. Four factors contribute, namely (i) high cost of funds; (ii) perceived high credit risk; (iii) inconsistencies in lending policies and procedures; and (iv) perceived non-requirement of external funding.

Cost of funds

IFAD funds are channelled to the provinces by MOF at a very high cost, passing through five administrative layers before reaching the RCCs. It passes the funds with credit risk and exchange risk to the provinces, but it charges 4.5% to the respective provincial Finance Department as against the 0.75% charged by IFAD. It has, however, passed on World Bank project funds to the provinces at considerably lower rates. However, in 2000, IFAD secured more advantageous or reduced margin requirements in the transfer of funds to the provinces.

It may be argued that the cost structure for RCCs under IFAD projects unwittingly ensures failure of the RCC model. The interest rate spread for RCCs is too low for them to be interested in seeking IFAD funding. The rate of interest on loans to households varies from project to project. In the Sichuan Livestock Development Project, in the areas where the RCC model is being tried, the rate of interest charged to households is 12.06%, whereas in areas with the PMO model the rate of interest is 8%. The spread available to RCCs varies from 1.1% to 1.3% in the Southwest Anhui Integrated Agricultural Development Project. The RCCs have been promised an additional spread of 1.5% by PMOs. In practice, this would mean that the PMOs would have first to receive a 1.5% spread refund from MOF prior to passing this refund on to RCCs.

The average cost of RCC funds, as calculated by interest expenditure to outstanding deposits, has been declining steadily over recent years following PBC's downward revision of interest rates. The overall cost of funds declined from about 6% in 1997 to about 4% in 1999. The weighted average rate of interest for new deposit mobilization is much less than the RCCs' current average cost of funds. The latter ranges from 1.75% to 2.21% and is falling, with a declining proportion of time deposits of longer maturity.

It is definitely not a viable business proposition for RCCs to accept project funds at 6-7% when additional funds could be mobilized at a rate not exceeding 2.26% (excluding the cost of mobilization). In addition, the cost of deposit mobilization by RCCs is approximately 1.35% of the total deposit outstanding. The upper limit of total cost of funds is estimated at about 3.6%. To wit, PBC provides funds to RCCs at 2.9% interest. The high cost of funds has acted as a disincentive for RCCs, not only in terms of participating fully in IFAD projects but also with regard to taking the full credit risk. It can be safely assumed that RCCs would not be able to accept project funds at a cost exceeding 3-3.6%.

The design of IFAD projects has assumed that a spread of 4-4.5% should suffice to cover all costs of credit delivery, recovery and credit risk, and make for a reasonable profit. This hypothesis is also invalid. Based on the survey of RCCs in five provinces, it was found that the RCCs were able to employ a spread of 7% in 1995, but performance has not improved since then. The need for a larger margin remains, clearly indicating that the current incentive structure for RCCs does not encourage them to participate in IFAD projects.

Perceived credit risk

Even if a sufficient margin is provided to the RCCs to bear the full credit risk, they require complete independence in selecting the borrowers. Once the RCCs achieve complete independence, the RCCs will probably reject about 40-50% of all loan applications forwarded by the PMOs, since the RCCs rightly place emphasis on the credit history and the repaying capacity of the clients. This means a completely changed credit delivery scenario: the credit uptake will be far less than the current projections.

Differences in lending policies and procedures

There are three inconsistencies between the project-propagated and RCCs' own credit delivery systems. They pertain to: (i) loan size; (ii) loan duration; and (iii) the requirement for a no-due certificate. The size of the loans provided to households under the project varies from CNY 500 to CNY 3 000. Most loans for agroforestry plantation, irrigation and land development are about CNY 3 000, and loans for pig raising, pig breeding, duck raising, etc., are about CNY 500. In comparison, the sizes of RCCs' own loans to the households are usually smaller.

The RCCs provide short-term loans not exceeding one year. This helps RCCs to recycle funds at a faster rate and achieve a higher effective rate of interest on loans. The loans provided under the project are medium to long-term in nature. Even duck- and pig-raising, which yield returns within one year, are supported through loans with a two-to-three-year repayment period.

The RCCs insist on a no-due certificate from prospective borrowers. This means the households with outstanding loans from ABC or a PMO cannot obtain a loan from an RCC. This procedure does not apply in PMOs, and therefore it is an important issue as it potentially lowers the quality of lending through PMOs.

Limited demand for external funding

The mission visited some RCCs that felt that external funding was not required for their operations. This was more evident in Jiangxi Province, where the RCCs felt that adequate deposits could be mobilized to fund their lending programme. At the current lending rate, RCCs find it risky to provide loans to poor households. In contrast, empirical evidence confirms unmet credit demand and willingness to borrow at higher interest. New credit products with flexible interest rates and incentives for on-time repayment could enhance both credit demand and the incentives for RCCs to expand lending.

Current scenario

Given the constraints in implementing the RCC model as prescribed in IFAD project documents, the RCCs and PMOs in Anhui Province have reached a tactical arrangement to facilitate implementation. The PMOs continue to bear the credit risk and the RCCs deliver loans to all borrowers identified by the PMOs. This contrasts with the subsidiary loan agreements signed between the RCCs and the PMOs, where the PMOs stipulate that the full credit risk remains with RCCs and that the latter will have complete autonomy to select borrowers. According to PBC, this is an 'agent arrangement' between the PMOs and RCCs.