Safety Nets

Site Index
Main Page
Issues in Program Design

Contributor: World Bank
Author: Decentralization Thematic Team
Contact: Jennie Litvack

Decentralization and Safety Nets

Safety nets protect a person or household in three types of situations (1) when there is chronic incapacity to work and earn (e.g., the severely disabled, elderly, young orphans etc.), (2) when there is an unpredictable "idiosyncratic" shock (e.g., sudden death or serious illness of bread-winner), or (3) when there is an unpredictable community "covariate" shock (e.g., economic recession, bad harvest, flood). Hence, safety net programs serve two important roles: redistribution for situations such as (1) (e.g.,income transfers, food supplement programs) and insurance for situations (2) and (3) (e.g., public works programs, drought relief). A key challenge in the design of safety nets is to maximize the benefits to the needy and vulnerable for a given cost. Costs include administrative costs, disincentive effects and political economy costs. For example, an important obstacle to improving the targeting of services and transfers to the poor is the high costs that can be involved in obtaining accurate information on their incomes and needs (Subbarao et al. 1997, Beasley and Kanbur, 1993, Van de Walle, 1995).

It has been suggested that decentralizing from national to local level poverty monitoring and the management of anti-poverty programs can reduce costs and improve targeting of intended beneficiaries. Local governments and administrators may be better informed about members of their community and thus better able to identify their poor.

There is some empirical evidence about this and the experience seems to differ depending on which level of local government is involved. For example, in Albania, where social assistance allocations are determined at the commune level (population of about 2500) local authorities appear to have more information than available through formal monitoring systems and this enables them to target the poor more effectively. District officials in Karnataka State, India have attributed a ten-fold increase in information flow from communities to decentralization. This has helped to increase the warning time before natural disasters, and has improved the government’s ability to respond and fend-off potential diseasters. In Vietnam, centrally funded transfers go more towards the poorer provinces but not necessarily the poorer areas and people within the provinces. This is because each province uses different criteria for distributing funds within their jurisdiction. In Argentina, wealthier provinces appear to be more effective at targeting federal safety net subsidies to their poorer areas than poorer provinces.

Many of the benefits of decentralization seem most apparent at the village level, and least apparent if decision making rests at the provincial level. Decentralization at the village level can lead to greater participation and more effective local development strategies and thus, enhance the delivery of pro-poor services. But the evidence is mixed. In some cases, there is also evidence of corruption at the local level (e.g., India).

In some countries (particularly the transition countries of the Former Soviet Union) national governments have decentralized safety nets with the goal of lowering the national deficit. But often expenditure responsibilities have been decentralized without adequate provisions made for revenue sources. Such unfunded mandates lead to a deterioration in the quality and quantity of services.

There are many different ways a safety net program can be decentralized. It is difficult to discuss the potential advantages or disadvantages of safety net decentralization, per se, since these will differ depending on what aspect of the program is decentralized. Different levels of government can be given responsibility for different aspects of program design and implementation (e.g., eligibility criteria, financing, and administration). The design of intergovernmental grants is also key to determining how each level of government will participate in the program. Each shall be briefly discussed:

Eligibility Criteria: Eligibility criteria can be set at the national level or decentralized to the subnational level. The main advantages of nationally consistent targeting criteria are (1) central funds can be distributed to the greatest number of nationally defined absolute poor or vulnerable people, regardless of where they live, (2) it is possible to evaluate the programs consistently across areas to assess their effectiveness, (3) clearly defined eligibility criteria set from the center may limit the ability of local elites to monopolize benefits. The main disadvantage to central eligibility criteria is that the identification of vulnerable groups will not reflect local values and preferences. (This need not be a problem if the locality can raise additional local funds to supplement the central allocations, but if they are very poor, they may not be able.) However, depending on local power dynamics, this may actually not be desirable. Local eligibility criteria may be appropriate for distributing locally raised revenues; however, nationally uniform eligibility criteria are generally recommend for the distribution of central poverty alleviation or safety net funds.

Project Implementation: Identifying Individuals according to predetermined eligibility criteria: Based on criteria set by the center or a subnational level, individuals or households can be identified for participation by central administrators in the locality, local officials from the provincial, district or village level, or members of local NGOs. Generally, project implementation is best undertaken at the lowest possible level, with monitoring by higher levels so as to ensure that standard eligibility criteria are being followed.


A well functioning social safety net is an important element of a braod-based inclusive growth strategy. Given the objectives of a safety net identified earlier, the central government has an important role in financing it. For example, areas which require the most assistance often have the least local revenues with which to support safety nets; subnational levels are often unable to deal with covariate risk. Nevertheless, subnational governments are also often concerned about household vulnerability and the need for safety nets and sometimes are able to raise substantial revenues to supplement central funds. Such co-financing of safety nets (between levels of government) can be encouraged through the design of matching intergovernmental transfers. (For example, in India, the centrally sponsored safety net schemes are generally matched by states).

Design of Intergovernmental Transfers:

Intergovernmental transfers can be either "block" or "specified." For block transfers, a sum of funding is provided to the local government to deliver a range of services but actual expenditures are determined at the local level (e.g., Latvia, Hungary). Since a minimum level of services is desired by the central government, it is usually recommended that safety nets not be funded through block grants. Instead, specified transfers which require the subnational level to spend the designated funds on safety nets are preferable. Specified transfers can be "matching" or "non-matching" grants. The central government can encourage subnational governments to raise local funds for safety nets by using matching transfers and they can vary the matching rates depending on local fiscal capacity. For example, a wealthier province might receive one central dollar for every three dollars it raises and spends on safety nets, while a poorer province might receive three central dollars for every one dollar it raises and spends on safety nets. While these methods, combined with centrally determined eligibility criteria can encourage local governments to deliver safety nets, another option is for it to administer the program itself through a representative in the locality. This may be desirable if the central government cares a lot about access to these services and has concerns about performance at the local level.

In short,"decentralization" of safety net financing can involve several different arrangements. First, the central government can devolve resources to the province to cover a range of expenditures, including safety nets. Each province then determines how much it will spend on safety nets. Second, it can set the criteria by which the province will select the poor areas to receive funds and create incentives which will reward provinces for allocating the funds to the poorer areas within their jurisdiction. A third option, is for the center to simply ask the province to "pass on" the central funds to the villages and households which it has identified as vulnerable. The second option is likely to be most effective.


To determine if funds are being targeted to the appropriate individuals and areas it is important to have information on program expenditures and household and village characteristics. Although there are many different design possibilities for decentralized safety net programs, ultimately, the central government, whether it is involved in financing safety nets or not, should monitor the programs. Information is best collected regularly and maintained at the subnational level.