A recent OED study analyzes the implementation of 23 rural development projects in Northeast Brazil--one of the Bank's most comprehensive rural development programs. It traces the influences on public sector performance in rural development and offers lessons on the implementation arrangements that perform best under different circumstances. The discussion of incentives for performance emphasizes the need to base project design on an understanding of local political forces and institutional structure.
Between 1975 and 1987, the Bank lent $1.4 billion for 22 integrated rural development projects and a region-wide land-tenure project in Northeast Brazil. The first generation of these projects had multiple components, ranging from agricultural credit and extension through feeder roads and electrification to health and education. The second generation were more streamlined, eliminating health, education, and roads components, and generally had longer implementation periods.
Most of the projects involved several levels of the Brazilian government--federal ministries, SUDENE (the Northeast regional development authority), semi-official banks, and state-level project units and executing agencies. The project-coordinating units set up in state departments of planning or agriculture were in charge of designing annual programs and supervising their implementation, but in general had neither executive responsibilities nor the formal power to grant funds or withhold them from the executing agencies. Municipal governments had no formal place in the projects but made some important unanticipated contributions.
Influences on performance
No projects, components, or agencies consistently performed better than others. Cases of successful performance were marked by narrowly specified tasks; measurable, conspicuous standards for performance; and clear penalties for not performing.
Performance depended less on agencies' inherent capabilities than on:
- Inherent ease or difficulty of tasks: For example, water agencies found rural water supply easier to design and install than irrigation, because the former required less analysis and was less dependent on decisions by agencies and users beyond the water agencies' own sphere of control. Sergipe's new rural water agency, for example, performed well in rural water supply and poorly, later, in irrigation.
- Full control over the quality and pace of project execution.
- Presence of outside pressure: Some project tasks were difficult or unsatisfying for some agencies--for example, redirecting services toward the poor, choosing less capital-intensive technologies for infrastructure, or reducing unit costs and reaching larger numbers of people. When performance was good, project management had often met clearly identifiable outside pressure to get things done, reach significant numbers of people, reduce costs, or be accountable in other ways. Pressure came from beneficiaries, governors, other state agencies, development banks, municipal governments, non-government organizations, and the Bank itself. It helps explain some surprising bursts of good performance. Withdrawal of pressure helps explain why agencies deemed strong suddenly performed poorly.
- Involvement of keenly interested local government: Some of the most successful cases combined unusually complementary action by state and local government. The local involvement helped reduce costs and delays, made state agencies more accountable, and encouraged use of local materials and labor.
When one of these variables changed significantly, performance went from good to bad, or vice versa.
Often, the better-performing activities departed from original designs.
Implementation in some cases was quicker than expected at appraisal. Examples include construction of wells and standpipes in rural communities, distribution of farm inputs, and--in Piaui-- acquisition of land for redistribution. Delays in execution in the first generation of projects had caused the Bank to lengthen the typical implementation period from five years to eight years in the second generation, to allow more time for institution building.
Goals became much narrower than envisaged at appraisal, with projects reduced to a few components--often rural water, community participation, and land redistribution. Examples occurred when supportive governors chose particular project components as their "signature" activity; when project managers gravitated toward their own favorite components; and when shortfalls and delays in the transfer of counterpart funds changed budgets enough to allow project managers to reshape projects in more manageable proportions.
Takeovers: Better-performing agencies took over tasks from the agencies originally charged with them. Most of the projects had been designed for execution by established agencies; project-coordinating units created from scratch were modest, without executive functions. But some of the best examples of performance came from agencies not established or specialized in a particular activity, and not originally meant to carry it out--as well as from dynamic managers of project coordinating units who wanted to play more than modest coordinating roles.
Successful managers who took over tasks from established agencies sometimes broke project rules and contracted out the work to public agencies other than the designated ones, or to private firms or non-government organizations. They obtained the best performance from other agencies not by coordinating their work but by contracting or forcing them to do what was needed. Their motivation was often more important than long experience; the takeover agencies liked the tasks that the established agencies disliked. And, because public-sector professionals flowed back and forth between agencies, the takeover agencies could draw on a large pool of expertise, often getting a specialist seconded to them from the taken-over agency itself.
Demand pressures are well known to affect public sector performance. But the Bank and other donors customarily take a "supply-side" approach to project design, concentrating on building up the inherent capacity of particular agencies. Experience suggests the value of:
- Designing projects flexibly, as a series of modules with short implementation periods. In many instances governors gave strong backing to the project components that most appealed to them, sometimes raising additional funding for them (see box).
- Channeling part of project funding through the users of services-- other public agencies, beneficiary groups, development banks, municipal governments--to help strengthen incentives for efficient performance.
Raising extra resources
Better-performing projects or components often elicited more resources than expected at appraisal--from governors, agency managers, state secretaries, mayors, banks, or beneficiaries themselves. This occurred at a time of extreme fiscal austerity in Brazil. Three examples:
- A state loan fund for works projects in municipalities resulted in a kind of informal municipal betterment levy in the form of land, materials, and fencing.
- A Bank-imposed ceiling on per-hectare costs for tubewell and riverine irrigation led to unanticipated land donations, for small-scale irrigation, by municipalities and private farmers in an innovative cost-sharing arrangement.
- A profitable spread between deposit and lending rates stimulated rural banks to mobilize deposits and increase their lending to small farmers.
Typically in Bank projects, counterpart funding must be committed before projects begin. But in these examples, the additional funds were not committed beforehand; the structure of incentives made it worthwhile for institutions and individuals to contribute after project implementation started, and in a way that did not add to inflation or the fiscal deficit. None of the incentives in these cases was intentionally provided by the projects, but there is no reason why they could not be in future.
Other recommendations for supplementing Bank loan resources:
- Apply some of the same approaches in rural development as in "intermediate-cities" projects in Brazil and elsewhere, using matching funds and other incentives as a way to draw on local resources and entrepreneurship, and so as to place responsibility for functions at the level where they work best.
- Link small-farmer lending more closely to the mobilization of rural savings. This may require free-standing projects for rural financial institutions, rather than credit components in agricultural development projects.
Land reform issues
The experience reviewed shows that land reform is likely to perform best where:
- It can take advantage of new technology. Many cases of successful land transfer and associated agricultural development occurred in already-settled regions where the market promised profits from intensifying the cultivation of small-farmer crops--tomatoes in Ibiapaba, oranges in Sergipe, irrigated vegetables in the Irece region of Bahia. This contrasts with the customary view that intensification of land use, and the increase in land values that accompanies it, are common features of already-settled regions, and make further large-scale land transfer less possible. Small-scale private irrigation associated with high-value agricultural production was notable in several of the cases reviewed.
- Land acquisition is followed by infrastructure investments (roads, water).
- Beneficiary groups share in decision making. Land markets worked better for small farmers when local organizations (cooperatives, labor unions, local government) and beneficiaries helped identify land and decide on land acquisitions and their prices. This more decentralized approach introduced some checks against collusion between large landowners and the state.
- Land is purchased, though with a credible threat of expropriation. Direct expropriation tends to be restricted to one or a few agencies, hedged around with size or other limitations; it often requires cumbersome review processes within the expropriating agency, and is liable to legal challenge and delay. Dedicated project managers were highly motivated to make land markets and other mechanisms yield land parcels for donation or purchase at low cost. To be within the law, land parcels less than 500 hectares could not be expropriated but had to be bought or acquired by donation. More project funding was made available for infrastructure investments and agricultural services than for land expropriation or purchase.
Many of the successes in agricultural research and extension were achieved through special campaigns against crop disease and pests--for example against the boll weevil in cotton-producing states, orange disease in Sergipe, and banana-root fungus in Paraiba. In these examples, defining tasks narrowly helped to produce good performance by agencies that did not do well with a broadly defined agenda:
- Attention concentrated on a single crop, or a single problem with that crop.
- Results were clearly measurable, penalties for poor performance were high, and performance was judged by outputs (e.g., reduced levels of pest incidence, number of diseased plants eradicated).
- Powerful interested parties demanded results--governors, directors of other agencies, mayors, farmer associations, and high-level officials who worried about the serious impact of possible crop loss on state tax revenues and on the region's agricultural economy.
- The campaigns had a clear beginning and end, usually within the four-year period of a governor's mandate and sometimes even within a one-year crop cycle--well within the five-to-eight year life of the Bank-financed projects.
- The intense public-sector effort around a particular crop in a particular region for a limited period guaranteed the smooth supply of the improved inputs that was often problematic in more routine activities. Having a relatively assured supply of inputs in turn made adoption more attractive to small farmers.
- Agency staff were energized by having clearly defined and visible problems to work on.
The experience suggests that project components to support research and extension should favor single-crop or other highly-focused interventions, with a clear beginning and end, whose results can be measured by output. The broad-ranging support that is consistent with recent emphasis on farming-systems research is organizationally burdensome; it is more appropriate in free-standing projects dedicated to building up a single agency over a long period.
Funding for research and extension should be channeled partly through user groups, who place a higher value on applied work and dissemination than research agencies do.
Box 1: Good governors, hard deadlines
Many of the successful experiences owed much to support by energetic governors or state secretaries. Governors tended to see Bank-supported projects as offering a menu of components for their support. They identified more readily with simple activities that could be planned and executed within the four-year election cycle. Whole projects that had many components and long execution periods did not seem good vehicles for making a political mark; the resources they provided risked being used to meet short-term budget needs or pay off political debts.
Governors who supported pieces of Bank projects as "signature" activities in effect ensured that these activities worked well. The Governor of Sergipe refashioned the Bank-financed project around rural water supply: 140 simple systems, 2,500 cisterns to catch rainwater from roofs, and 600 small communal ponds; he also financed some 7,500 cisterns with funds raised independently. The Governor of Piaui supported the land purchase component of the state rural development project, because it produced quick results that could be celebrated in dramatic, highly publicized, ceremonies where landless peasants received their titles. The Governor of Pernambuco chose small farmer credit fashioned around actions to break bottlenecks in the production of certain crops; the state government supplied most of the funds from outside the project budget.
In such activities at their best, the constraints and pressures the governors faced caused them to demand good performance from the agencies they controlled. The governors provided crucial protection against pressure to hire or fire staff on political grounds, making technically undesirable choices, and shortfalls or delays in funding transferred from central government.
Box 2: General recommendations
Involve participating organizations as much as possible in project design, not only to improve the design, but to stimulate good performance by the implementing agencies and to mobilize additional resources. In most of rural Northeast Brazil, with little tradition of association at the community level, opportunities were good for stimulating group action for economic or social purposes.
Properly appraise the macroeconomic and political context of proposed new projects.
Design projects flexibly enough to be able to respond to the priorities of elected officials, when these promise real, sustainable improvements for the Bank's target groups.
Give a single agency sole responsibility for the project as a whole, whether the tasks are few or many, whether that agency is established or new, and whether it is an executing agency or a coordinating unit. For a coordinating unit, that responsibility should extend to the disbursement and withholding of funds.
Ensure the agency has the political and financial resources to carry out the project tasks itself or to contract them out--to other public agencies, private firms, or non-government organizations.
In planning implementation, distinguish between those interventions that work best with decentralized management and those where economies of scale call for management on a regional or state-wide basis.
Assign activities to implementing agencies with a view to their relative ease and difficulty. "Easier" tasks for relatively inexperienced agencies include campaigns to combat epidemics of crop disease and pests, installation of simple rural water systems, and some forms of land acquisition.
Consider placing activities outside their traditional bureaucratic homes--in "inappropriate" agencies or even new units--if the latter are more motivated by sympathy and outside pressures to do well. Depending whether technical or social skills are most needed, the appropriate implementing agency will differ markedly. Even established infrastructure agencies sometimes do quite poorly at tasks assigned to them by Bank projects; other agencies, with less experience or specialized expertise, sometimes do better.
Carefully appraise and monitor the performance of organizations that have not previously served the rural poor, which may have difficulty adapting to a new philosophy or organizational mode. Back this up with the right to cancel contracts.
Editor's note: For a description of recent changes in the approach to implementation of the Northeast Rural Development Program (NRDP), readers are referred to the Region's "Mid-term Review of the NRDP", SecR92-8, January 31, 1992. On rural development performance more generally, see "Annual Review of Evaluation Results, 1990", Report No. 9870, August 30, 1991, Chapter 5.