World Development, January 2001, vol. 29, iss. 1, pp. 39-62. Available From:
This study evaluates the early impacts of three microfinance systems in China. It compares governmental and nongovernmental models of microfinance programmes that aim to provide financial services to individuals ‘traditionally excluded from the banking system, especially women’. In terms of outcomes, it looks at programme targeting, financial and operational performance, and the impact of each programme.
The three programmes under evaluation differ in the degree of local government involvement. The first programme is run entirely by an NGO (Funding the Poor Cooperative – FPC; 223 households), the second is run by a mix of government and NGO staff (FPC and government; 163 households), and the third is run fully by the government (63 households). The NGO programme provides assistance in the flood plain, while the other two are in mountainous regions. The full sample consists of 449 households from 18 villages with 305 microfinance programme members and 144 nonmembers randomly selected. The authors quantified the impact of the programmes as changes in income and used household survey data from 1997 for the evaluation.
The study uses probit models to estimate the targeting performance of the programme and ordinary least squares estimates to measure impact of the different programmes, controlling for observable household characteristics and eligibility criteria in treatment and control villages. It uses an instrumental variables approach to limit selection bias. The instruments it uses are those that influenced programme participation but were not correlated with income. The first instruments it uses are consumer durables and housing because these do not influence income but are predictors of wealth and eligibility. The second instrument it uses is eligibility itself to account for eligibility rules that are not based on income effects and to calculate the average programme effects on participants.
In terms of overall targeting performance, the study excludes the very rich, but among those eligible, richer and poorer households are equally likely to participate. For the NGO programme, owning a house, livestock or cultivated land predicts nonparticipation. In the mixed programme, consumer durables (−0.137), housing (−0.0273) and high quality land (−0.0506) predict nonparticipation. Having fixed assets (0.0207) and livestock (0.0472), however, increases the likelihood of participation. Therefore, participants tend to be poorer households that engage in activities other than cropping. In the government programme, participants are generally nonpoor and not engaged in cropping. Across almost all indicators of group performance—such as selection of group leaders, regular meetings, and so forth—the NGO group performed best, with the mixed organisation second and the government-run programme last. For example, in the NGO programme, participants meet regularly 89 per cent of the time, while participants in the mixed programme meet regularly 67 per cent of the time and participants in the government programme meet regularly 42 per cent of the time. The study finds a positive programme effect of 0.394 (p < 0.1) for the NGO programme and 0.144 (p < 0.05) for the mixed programme. The study provides no data on programme effect for the government programme.
This paper shows that government microfinance programmes have performed poorly, ‘casting doubt on the prospects for recent large-scale government microfinance initiatives’. The authors also suggest that some of the country characteristics may indicate that microfinance will not be successful in all areas. This is because China’s poor are concentrated in remote, mountainous regions where agriculture is most common. They also state that microfinance movements in China should be partnered with an overall strategy to improve the rural financial sector by building up a diverse institutional framework. Governments and donors should openly discuss the trade-offs between targeting and sustainability, and they should acknowledge that only a small percentage of the poor will utilise microfinance availability and that many are in need of other government services.