This section summarizes the content, implementation and impact of China's main poverty programs in the 1990s, and focuses on the investment programs which aim to promote economic　and income growth to lift the poor out of poverty on a sustainable basis. Relief programs, which　target a small segment of the poor, are also discussed (although they are usually regarded as　distinct from the main poverty program). Implementation of the programs has been a challenge　for the government, and it is believed that the efficacy of all of the programs could be substantially improved. The subsidized loan program has met difficulties in achieving its　objectives. Lending to enterprises has done little to reduce poverty, and directing loans to poor　households also has proved to be very difficult. These issues, compounded by low repayment　rates under the program, convinced the Government to experiment with micro-credit schemes, but　these have also encountered serious difficulties. The FFW program could be improved by more explicit targeting of the poorest areas, and by refocusing its work on the types of programs that bring the greatest benefits to the poor. The MOF grant program is also believed to have encountered a number of implementation difficulties, though there is scant objective information available for that program. The relief program should remain a key element of the Government's efforts, and retaining the funding and focus of this program is essential to the wellbeing of China's most destitute population. International researchers have undertaken quantitative analyses of the effectiveness of China's poverty reduction program, and have found some indications of a　significant impact on living standards.

The subsidized loan program is the largest of the poverty programs and is viewed as the　flagship of the Government's poverty alleviation efforts. The Government has invested about　Y 68 billion in the program since its inception, and has placed considerable emphasis on trying to　ensure the program reaches poor households and brings them clear benefits. However, this has　proven to be extremely challenging, and the Government has tried different approaches to try to increase the effectiveness of the program.
Following earlier emphasis on issuing loans to households, in 1989 poor county governments began using a large of portion of their available subsidized poverty reduction loans　to support the growth of county and township run enterprises in poor areas. In 1992 and 1993, about half of the subsidized loans were lent to industrial enterprises. Of this amount, more than 60 percent went to county run enterprises and the rest went to township run　enterprises. By the mid 1990s, this policy was recognized to have been less successful than originally envisioned. Many of the rural enterprises supported in the past with　poverty reduction funding have been capital intensive, loss making firms with minimal poverty　reduction impact. While the number of such firms is not known, it appears that poverty reduction　program funding for poor county TVE development may have even led to a net decline in　revenues in some poor counties. In addition, the use of government assistance for TVE　development has reduced funds and other resources available for household farm production and　off-farm production undertaken by small scale private enterprises, and may have contributed to　distorted patterns of development of enterprises in poor areas.

(II)

Recognizing the shortcomings of providing direct financial support, the government　recently decided to severely curtail poverty reduction program funding for poor county TVE　development. Instead, the more appropriate role for local governments is to establish a more　favorable environment for poor county TVE development. The experience of those TVEs in poor　areas which have better survived the current economic turbulence helps clarify what local　government can do to facilitate TVE development. Relatively small enterprises with a more　clearly defined ownership structure, operations based principally on local comparative advantage　(such as cheap labor and local raw materials), and relatively large enterprises combining local　resources with technology and market access provided by enterprises from more developed areas,　appear to have been most successful in weathering the current economic challenges. It is　therefore recommended that local governments support TVE development through (a) reforming　TVE ownership and management, (b) allowing those enterprises which have negative equity and　which have suffered from financial losses for a number of years to go bankrupt, and (c) encouraging local enterprises, which have operations consistent with local 　comparative　advantage, to establish joint ventures with enterprises from more developed areas. Past　experience indicates that local governments in poor areas are not well-equipped to successfully　establish and manage enterprises by themselves. These local governments should therefore　instead focus on simplifying the procedures for the establishment of private and collective　enterprises, reducing the taxes and other extra burdens on enterprises, improving local infrastructure, and providing more training for local accountants, auditors and technicians.
The strategy to target enterprises rather than households is in part responsible for reports　from many sources showing the very low share of loans that have directly benefited poor　households in the 1990s. Indeed, before 1990, some surveys indicated that 45 percent of the loans　reached poor households (and one survey taken in 1987 showed 92 percent of loans reaching poor　households), whereas most surveys taken during the 1990s showed much lower shares. Many　observers share the sentiment expressed by a provincial official that the failure　of loans to reach poor households has been the rule and not the exception. Park cites survey　results showing examples of counties where none of the loans had reached the poor, and of poor　villages where the majority of the loans went to farmers of average wealth and not to poor　farmers.

In recognition of these failures and seeking to redirect investment directly to poor　households, the 1996 National Poverty Reduction Conference decided that 70 percent of　subsidized loans should reach poor households and 70 percent should be in agriculture (as opposed to industry). Early indications are that provincial and county governments appear to be making progress but have not fully achieved these targets. For example, expenditure data for Yunnan summarized in Table 3.2 below shows the balance between agricultural and industrial　spending did shift after 1996. However, sizeable lending for industry remained, and the　composition of spending did not reach the 70/30 stated targets. Agricultural spending increased from 27 percent to 41 percent and industrial spending decreased from 32 percent to 23 percent. The shift was most pronounced in use of central poverty loan funds (agricultural spending　increased from 31 percent to 51 percent), with other funding channels also registering an increase　in favor of agricultural lending. Moreover, the large increase in lending for agriculture in 1997　has not necessarily led to the intended concentration of lending away from large enterprises and　to households. Provincial officials indicate that large amounts of the agricultural lending went to　agricultural production bases and that loans to households comprise less than 30 percent of total　agricultural spending. However, a number of field investigations by Chinese and international　researchers indicate that the proportion reaching poor households is actually much lower, even as　recently as 1998, when the new policy should have been fatly established.

(III)

There are a number of factors that have contributed to the program's difficulty in　directing loans to poor households. The goal of the program is to both reach the poor and promote　economic development, and this has led to sometimes conflicting objectives for local officials　trying to carry out the program. For example, projects that will generate tax revenue have an　obvious pull. Lower level officials are aware that past performance will be a criterion for more　future loans, so they tend to pick projects that they perceive as bringing higher financial benefits　regardless of whether or not such projects will help reduce poverty. The common perception that　the poor are incapable of managing projects successfully also contributes to the preference for　lending to enterprises (and richer households).
Even when official policy succeeds in directing loans directly down to the household　level, there are a number of actors which contribute to diversion of loans to nonpoor households.　First of all, favorable loan terms (the annual interest rate is 2.88 percent) and large loan size leads　to strong competition from the nonpoor. Procedures for loan application and approval are　complicated, and this tends to discourage poor applicants. The problem of loan guarantees and collateral is also a significant factor. For example, to guarantee repayment of the loans, ABC　requires physical collateral which often leads to exclusion of poor households from the program.　And ABC's obligation to bear the burden of the repayment risk conflicts with its role as a　commercial bank, and many loans are diverted to commercial purposes and to richer households. Moreover, in 1998, ABC began passing down the repayment risk to local cadres and　governments, which in turn pass the risk down to village councils. Not surprisingly, village　councils often balk at lending to poor households. Inadequate monitoring and supervision capability of the local PADOs exacerbates the problems in implementing these programs.

In addition to the difficulties in targeting the poor, repayment of the government loan　programs has been an issue. The most extensive survey on repayment was carried out by the　China Science and Technology Commission between 1991-1993. That survey indicated that　repayment levels averaged 54 percent, which is well below sustainable rates. Overdue loans were　greater than 100 percent of all loans coming to term. Repayment of loans to enterprises was the　lowest at less than 40 percent. In addition to the lax supervision and monitoring efforts, low and negative real interest rates serve as a disincentive for repayment. Also, good repayment does not necessarily lead to additional loans (Park, 1998). Official data from Yunnan shows only a 39 percent repayment rate in 1997, with overdue loans representing over 100 percent of newly issued loans.

(IV)

Microcredit Programs. In response to the growing realization that government poverty　loan programs were not reaching the poor and to the low repayment rates of the programs, and　awareness of successful microcredit programs in China and abroad, the government began　experimenting with using the subsidized poverty loan funds for microcredit activities. The　Government's use of subsidized poverty loan funds for microcredit has expanded rapidly. In　1997, the program included more than 80 counties, and had invested more than Y 100 million. By　1998, it was reported to have reached 200 counties, and total investment was about Y 800 million. Most government funded microfinance programs have used variations of the Grameen　Bank approach, which has been employed successfully in other countries, and in some donor　programs in China. Initially, the program was organized by the PADOs, which lent the funds　directly to households.

Relatively high arrears rates in a number of the government's programs led to a recent　switch in the microcredit system, and regulations now require that microfinance experiments　using subsidized loans be undertaken by ABC, rather than the PADOs. Loan contracts are to be　signed between the ABC and households. The role of the PADOs will be to focus on organizing　households into groups and centers and facilitating loan repayment. This move addresses　weaknesses in financial management and supervision, which has been one of the central problems
in the government schemes. However, ABC's current institutional capacity is likely to hinder its　ability to undertake Grameen type microfinance experiments as many county ABCs do not have　agencies and staff at the township level. Moreover, the loan commission to be paid by ABC to the　PADOs for their work in organizing groups and repayment is most likely too small to sustain　their interest in microcredit.

Well designed microcredit programs can avoid many of the pitfalls seen in subsidized credit programs around the world, and which have plagued China's poverty loan program. In　designing any future role for microcredit in the Government's poverty program, a number of　points should be emphasized. First, microcredit on its own is unlikely to meet the needs of the　absolute poor, and should be combined with other types of interventions in the poorest areas.　Second, improved financial management, monitoring, supervision and internal auditing, backed　up by intensive staff training, are key to the success of any microcredit program. Lax or non-existent systems made government experiments with microcredit vulnerable to spiraling　repayment problems, and financial mismanagement. Third, programs should avoid the tendency　to become top-down, and may wish to experiment with devolving response stability for implementation to grassroots organizations. It probably would be advantageous for the PADOs to　play a role in monitoring and supervising microcredit, but to have the actual programs managed　by organizations outside the government structure. At a minimum, staff should be hired by open recruitment and not be appointed by county or other officials. One positive aspect of the Green model is that its participatory methods promote initiative among the poor by encouraging them to　form groups, choose group leaders, and decide their own investments. If programs side step this　"bottom-up" approach, this aspect is lost, and targeting and repayment can also suffer. Fourth,　government programs have tended to charge rates of interest well below the level that would　allow them to cover operating costs. This can lead to leakage of funds to the non-poor, and also threatens the sustainability of the programs. Finally, the current group based models can be quite　costly, both to the poor and in terms of administrative costs (for example, through frequent group　meetings and repayment). Impact assessment should be undertaken to better understand the costs　and benefits to the poor of existing programs, and to assess whether variations on the current　models might be better suited to some areas.

(V)

In the longer term, mechanisms to provide credit for the poor could take many forms.　Developing savings services could be even more important for poor farmers than credit services,　but no savings should be organized until prudential regulation and supervision has been　established. Reforming existing financial institutions in poor areas may be the most efficient way　to reach large numbers of poor and near poor, but this should not preclude a future role for　informal financial organizations that now lend to the poor. Some of the current temporary microfinance program offices could possibly be converted more permanent institutions, if they　are able to meet appropriate performance and regulatory standards. Small, grass-roots　organizations have the potential to play a role in piloting innovative techniques and their strong　outreach is invaluable.

The FFW program, implemented by the State Development Planning Commission,　provides funding for infrastructure construction in poor areas. The program originally made use　of left over commodity stocks which were distributed to counties, which used a coupon system to　pay for inputs and labor done under the program. Funding from the program in the early years went almost entirely to rural roads, water, and land improvement. After large increases in the　program beginning in the early 1990s, the program has diversified and includes investments in a　much larger number of areas including water conservation, and even commerce, education and　health. Central government funds are supposed to be matched by provincial and county funds. However, in most cases, matched funds are inadequate. This has meant that central funds are　entirely used for construction inputs, and the shortage in funds is offset by the use of voluntary　labor contributions from villagers. (In most areas of rural China, villages operate a "work day　contribution system," with each villager obligated to work a certain number of days annually. FFW labor contribution is deducted from this annual requirement). Zhu and Jiang (1996) estimate　that 40 percent of labor used for FFW construction is free labor. In general, the types of　investment (for example, water, roads, land improvement) are made by national and provincial governments. County governments select the villages to participate in the program, and village　committees make decisions on the allocation of project investments (for example, which areas　should be terraced) and labor contribution. There have been few systematic evaluations of the　FFW program, but most reports indicate that the program has done a relatively good job in　constructing a good deal of infrastructure that has benefited poor areas.

From analysis based on data from FFW projects in the early 1990s, it　appears that villages with more favorable economic conditions, those in more remote areas, and　those with higher populations have been more likely to receive FFW projects (Zhu and Jiang, 1996). The poorest villages are less likely to receive these projects. In some cases it may be a　rational economic decision to omit some of the poorest villages from FFW projects based on　evaluation of economic returns. Building a road or providing electricity in a remote, sparsely populated village, for example, would in many cases not be the most efficient use of poverty　reduction funds. Nevertheless, it would appear that the impact of the program on the poor could　be improved by greater efforts to select poorer villages. Indeed, in the mid 1990s, more remote　areas were said to be targeted for FFW projects, but results from these efforts have not yet been　assessed.　

(VI)

Village committees are generally charged with selecting participants for labor work under　FFW, and most assessments indicate this is done in an equitable manner. In general, each　household is assigned a certain number of days of FFW work, based on the household's labor　supply. However, because labor provided under FFW schemes is in many cases, unpaid labor,　households with higher cash earnings sometimes choose not to participate and contribute an　equivalent sum of money. Some observers have criticized the free labor contribution as a "labor　tax" which falls disproportionately on the poor who are most likely to contribute labor to the scheme. While this tax may not be high, since typically the labor takes place in the off-season　when there is limited alternative work to be done, the program would probably bring higher　benefits to the poor if all work was paid, even if this reduced the total amount of infrastructure　built. (Zhu and Jiang, 1996, Park, 1999). International experience has shown that public works　schemes, using wage rates below the current wage, are very effective poverty alleviation　measures. FFW would do well to experiment with such schemes in order to better reduce poverty and provide some insurance that those who have escaped poverty don't fall back below the　poverty line. As such, using FFW as part of a public works scheme could form an important part　of China's emerging social safety net.

The program is administered in a fairly top down manner, and in some cases the actual　project investments are contrary to villagers expressed needs. For example, Zhu and Jiang document a case where villagers voiced a clear need and preference for more roads, but were　instead told to build terraces by county officials. In addition, both domestic and international　observers have suggested that infrastructure built under the program would have a greater effect if　it were integrated with other investments that benefit the poor.　

Finally, there are also concerns that the expansion of the program in the mid 1990s, has　diluted the effectiveness of the program and its impact on the poor, and led to some laxity in　funds management. Zhu and Jiang point out that some of the investments included in FFW in the　1990s have no direct relation to poverty reduction (for example investments in rural post offices). Also, the large number of sectors and line ministries now　involved means that the matching funds shortage has become more acute, and many projects are left unfinished, and project funding is often too diluted to have sufficient impact. Supervision of　funds usage has also become more complicated and there have been reports of funds diversion.　 for poor areas is provided through earmarked grants for education,　a revolving fund, and also tax incentives and budgetary subsidies.

MOF grant funds have grown very slowly,　decreasing in real terms from the late 1980s and not recovering their 1986 value until 1998. MOF　grant funding has not been carefully investigated, and there is little published evaluation of its　efficacy. There are a number of implementation problems with the grant schemes including difficulty in raising counterpart funds, diversion of funds to pay administrative costs, unclear authority over fund use, slow and late delivery of funds not synchronized to project　needs, and dispersion of funds which precludes economies of scale in project investments.

(VII)

China's rural relief system, implemented by the Ministry of Civil Affairs and provincial　and county Civil Affairs bureaus, is independent from the government's poverty relief program.　Nonetheless, many recipients of the program are members of China's poor population. The bulk　of the program targets those affected by natural disaster, such as earthquakes, flood and drought.　A much lesser share of the rural relief program funding is directed to those considered truly　destitute, who lack any means or potential to support themselves. Members of this second group are guaranteed food, clothing,　housing, medical care, and funds for burial, and education (for orphans). These are referred to as　the "five guarantees" in Chinese. Provincial governments provide most of the budgetary funds for　the program, but the system also receives donations from other sources. Township level　governments use funds allocated from the county Bureaus to purchase necessities (including　grain, clothing and blankets) for relief recipients. It is generally prohibited to give money directly　to relief recipients.
Snapshots of the program in Ningxia and Yunnan indicate that the program has increased　in size over time, and that its functions have been maintained. In Ningxia, interviews with　Jingyuan County officials indicated that most of the funds are used for disaster relief, and funds　used for the "sanwu" group accounts for only a very small proportion of the total budget. In 1996,　for example, the county had a total budget of about Y 1.22 million yuan for relief purposes, of which, Y 1.10 million was used for disaster relief, and only Y 0.12 million used for the "sanwu"　group. In 1998, Y 1.01 million was used for natural disaster relief out of a Y 1.27 million total　budget. Most funds for natural disaster relief money go towards food consumption. The Jingyuan　County officials reported their belief that the effectiveness of the program has increased over　time.

In Yunnan, interviews with county officials also indicated that the program has expanded　in step with economic growth. The size of the provincial rural relief program budget increased　from Y 46 million in 1990 to Y 132 million in 1998. As in Ningxia, most funds were used for　disaster relief. In Luquan County, for example, the total budget for the relief program was Y 1.21　million, of which, Y 0.94 million was used for disaster relief. In 1998, the pattern was similar　with Y 2.22 million used for disaster relief out of a total Y 2.99 million budget. However, provincial officials claim that an increase in the frequency of natural disasters has offset the　budget increases. For example, in Luquan County, the total area affected by natural disaster　increased from 39,000 mu in 1991 to 143,000 mu in 1998. In Yunnan, natural disaster relief　activities mainly focus on food, clothing, housing and basic medical care.

Historically, the relief system has functioned well in providing a safety for the most　destitute and preventing starvation during the most severe circumstances. It appears that concerns　that reform of the grain marketing system might undermine the effectiveness of the system have　not come to pass, and that funding (at least in the provinces cited above) has been maintained in　real terms. It is important that the focus and the intensity of the program continue. Recent　government pronouncements on intentions of setting up a stronger safety set, that would seem intended to cover rural as well as urban areas, are a positive sign.