DRAFT ONLY, PLEASE CONSULT AUTHOR BEFORE CITING
Paper for presentation at SANREM CRSP conference, Athens, GA, Nov 28-30 2001
It takes a village to raise a Pigovian tax…or does it take more? Prospects for devolved watershed management in developing countries
Ian Coxhead, University of Wisconsin
1. Introduction......
2. Issues in the design of institutions and policies......
Defining jurisdictional boundaries......
Spatial and sectoral variation in abatement costs......
Fiscal and capacity constraints on local administrations......
3. Watershed management in the Upper Manupali basin......
4. What projects and other external agencies can do......
5. Conclusions......
References......
Abstract
After years of failed attempts at centralized control over watershed management (WSM), the conventional wisdom among the donor ‘community’ and many LDC governments themselves has now turned decisively in favor of local approaches. In many parts of the developing world, this shift coincides with the decentralization of numerous other government functions.
Can local governments do a better job of WSM than central governments? Some advantages are clear; local administrations can be expected to have specialized knowledge of environmental and economic conditions, and therefore should also have the ability to fine-tune policy. But there are disadvantages as well. Of these, the main one is that jurisdictional boundaries typically do not internalize watershed processes, and the sources of environmental problems are spatially and sectorally diverse. There is scope for policies applied to one sector to cancel out the effects of policies applied in another. A related problem arises because the basin-wide costs of abatement of a particular pollution problem are likely to vary across sectors and across spatial units, raising questions about the most economically efficient mix of policies. Moreover, local jurisdictions, especially those in relatively poor upland areas of developing countries, may face secondary constraints related to their capacity to conduct analysis, make and implement policy, and raise funds.
Most contributions to the small analytical literature in this area find that environmental linkages make it very difficult in practice to design decentralized WSM policies, or that “nature ‘trumps’ most of the conventional arguments for devolution”. Can local governments design and impose Pigovian (i.e., socially optimal) environmental policies? What can projects and other external interventions do to assist them? The goal of this paper is to explore these issues from an analytical perspective, drawing on lessons from recent Sanrem experience in Southeast Asia.
1. Introduction
Deforestation and agricultural production in uplands of developing countries creates significant downstream problems, as is evidenced somewhere in the humid tropics virtually every monsoon season. Contributing factors include the absence of property rights (forests and steep lands are treated as though they are open-access resources) and externalities associated with water runoff and soil erosion. Though regulations may exist, enforcement is often very costly due to institutional or other failures. For all of these reasons, the depletion of forests, the removal by erosion of upland soils, and consequently the degradation of watershed function, remain the most important environmental problems facing developing countries in the humid tropics (Jha and Whalley 1999).
Meeting the watershed management challenge in such settings is a problem that is as yet unsolved. After years of failed attempts at centralized control, the conventional wisdom has now turned decisively in favor of devolved approaches, in which central government agencies act in partnership with, or even under the leadership of, communities and local governments (e.g. World Bank 2000). The trend towards devolved NRM has been welcomed in principle by many development specialists, especially as it coincides with and is reinforced by a general trend towards democratization, especially at the sub-national level. Justifications advanced for devolution make use of both efficiency and equity arguments (although the latter are probably more robust). Not all central government NRM agencies agree, however, and the devolution of watershed management (WSM) and other natural resource management mandates has been noticeably slower than that in other areas of public policy and administration. Moreover, and in spite of much rhetorical support for community-level WSM, there is now some momentum emerging for the creation of broader WSM jurisdictions, notably at the level of the river basin or catchment area rather than any smaller unit. The introduction of a new level of government, presumably with powers that override those of local administrations, seems to run counter to the idea of community-based management and even, perhaps to take back some of the powers and responsibilities devolved (or anticipated to be devolved) to them.
Can local governments do a better job of WSM than central governments or their designates, the catchment authorities? Some of the advantages are reasonably clear. Local administrations typically have specialized knowledge, and in principle are thus able to fine-tune policy to local conditions rather than applying ‘one size fits all’ solutions based on average conditions. But there are disadvantages as well. The disadvantage most frequently cited is that the boundaries of local jurisdictions usually don’t internalize environmental processes; this is the chief argument for the formation of special purpose jurisdictions at intermediate scales, such as the catchment. A second problem is that for any given pollution problem, abatement costs are likely to exhibit both sectoral and spatial variation, meaning they will vary both with the type of emitter and its location (an analogous argument applies to the costs of conservation, for example in the maintenance of biodiversity; for convenience, we will focus only on the ‘pollution’ problem). It often follows that responsibility for emissions policies is allocated to different agencies, whether at local or at higher levels of government—or both. For example, the Ministry of Agriculture may set crop-specific policies for farmers, while local government determines policies applied to rural agro-processing industries, and the Ministry of the Environment applies additional policies, regardless of industry, to a watershed. There is scope for policies applied by one agency to be canceled out by policies applied by another, and achieving a least-cost solution may require the creation of new mechanisms for inter-agency coordination. A third question is whether local jurisdictions have the fiscal, administrative and information-gathering capacity to formulate and implement decentralized WSM policy. A fourth problem is whether—even given the appropriate scale, administrative institutions and fiscal capacity—local governments have the political will to set socially efficient policy.
Advocates of the devolution of WSM powers have only recently begun to confront these issues. While there is a large economics literature on the interactions of policies of different types, that on the interaction between devolution and WSM is very small. Within this literature, there is not much analytical support for local WSM control—a striking contrast with the literature on purely fiscal devolution. A recent contribution by Smith et al (1999) finds, for example, that environmental linkages—ignored in most standard analyses of fiscal devolution—make it very difficult in practice to design decentralized WSM policies. They find that
“Nature ‘trumps’ most of the conventional arguments for devolution. That is, the linkages environmental resources create between activities taking place at different locations must ultimately limit the extent of independence in decision making (p.128).”
The goal of this paper is to explore these issues in a fairly specific developing country setting. Due to space limitations we focus mainly on the first two of the four problems just listed; however, we recognize that a complete solution requires that all four be addressed.
2. Issues in the design of institutions and policies
Many key NRM decisions are made by private “firms”, whether they are individuals, corporations or other groupings. Empirically, markets are highly influential in these decisions, even in relatively remote areas of developing countries. We may think of a typical NR manager in an upland or forest margin areas of a developing country as maximizing profits subject to the constraints of input availability, cultural norms, institutional and legal conditions, and in some cases household economic welfare. Policies and other forms of intervention by external agencies enter this problem indirectly through markets, as well as directly through command and control measures. In either case the problem arises of incentive incompatibility between agencies and NR managers. The two groups may be maximizing welfare over a different group of agents, may use different rates of time discount, or may apply different current-period valuations to non-marketed resources and products. Regulatory or policy-making agencies may make incorrect conjectures about the behavioral strategies and constraints of firms. In addition, we typically find that more than one agency has (or claims) a mandate over a particular set of environmental services. When there are gaps or overlaps (in general, inconsistencies) among mandates the likelihood of inefficient NRM outcomes increases. These inconsistencies, which may be horizontal (among local agencies) or vertical (between local and national), may result in the negation of one agency’s environmental management strategy by that of another. It follows that institutional and policy reforms which remove or resolve inter-agency inconsistencies in the area of NRM, or which remove incentive incompatibilities between private NR managers and society at large, will be welfare-improving.
Defining jurisdictional boundaries
A pure local public good is defined as a good for which consumption is both non-rival and non-excludable to a defined group of agents. The decentralization theorem (DCT) holds that if the unit costs of providing a local public good are not decreasing with size, its provision through decentralized rather than centralized means always yields greater welfare (Oates 1972). The DCT applies for public goods where these confer no benefits or costs on agents beyond the boundaries within which the good is supplied. Where the benefits from a public good “spill over” beyond these boundaries, there is an incentive to supply less of it than is required to maximize collective utility (Olson 1965; Oates 1972:9). Similarly, a jurisdiction larger that the area over which the public good is provided includes agents who may be taxed for its provision, yet receive no benefits; this too is suboptimal. It follows that for each type of public good there is an optimal size of jurisdiction: that which exactly internalizes all the costs and benefits associated with the provision of the good. This condition is known as “fiscal equivalence” (Olson 1969).
A pragmatic recognition of fiscal equivalence has resulted, in some countries, in a proliferation of single-purpose jurisdictions (fire districts, water districts, school districts and so on); in the U.S., there are now more than twice as many such units as there are counties (Deller 1998). In developing countries, however, the economic costs and practical difficulties of creating and administering such special purpose units may substantially reduce the benefits of their formation.
Where environmental management is concerned, the DCT has been shown to hold even when environmental standards are manipulated to achieve other policy objectives. Thus inter-jurisdictional competition for fixed capital investments such as factories, in which environmental standards may be ‘traded away’ in return for local employment or wage gains, is found always to be efficient so long as policies reflect the preferences of the median voter and there are no cross-border spillovers (Oates and Schwab 1988; Levinson 1997). Equally clearly, when spillovers do occur, local control over environmental standards will always provide too little environmental protection by the standards of society as a whole. In such cases the DCT indicates that on efficiency grounds, policy should be set at a higher level of government. In the case of surface water resources, the optimal size of jurisdiction is the river basin. Devolving policy powers to smaller units will result in socially inefficient policy—that is, in higher emissions and lower conservation investments than desired by society at large.
Spatial and sectoral variation in abatement costs
Defining the appropriate NRM jurisdiction is a challenge in institutional design. Even supposing that this challenge can be met, however, the task of designing efficient NRM policies remains a substantial one where water resources are concerned, due to the high costs of acquiring information about polluters and pollution damages. Environmental economists usually refer to this as the non-point source pollution problem. This problem arises when it is technically or economically infeasible to identify individual water take-off rates or contributions to downstream pollution. The non-point source problem means that policies cannot easily be tailored to specific polluters or users of water resources. In the uplands of developing economies there are many water users and metering is impossible; polluted runoff comes from many small farms and other enterprises, and cannot be measured at source. A national agency, or even one operating at the scale of a river basin, will be unable to gather all the data necessary to tailor policies exactly to the problem. As a corollary, it may be infeasible to design a policy based only on market-based (MB) instruments (the most efficient mechanisms under the DCT).
One common solution in this setting is to suppose that all producers sharing some set of characteristics (including location, land use and so on) have similar propensities to use water and to release pollution loadings. For example, if all farmers grew only one crop and used similar technologies, then it might be possible to assign abatement costs by location alone. In uplands, however, land use is also heterogeneous within locations. Some farmers, for example, might cultivate only perennials and contribute little in the way of soil erosion, while their immediate neighbors grow annual crops with much higher erosion rates. In general, abatement costs are likely to exhibit both sectoral and spatial variation, meaning they will differ both with the type of emitter and its location. Thinking about this problem in a fairly formal way can generate some insights into the challenge of efficient policy design, even when the jurisdictional boundaries are drawn so as to internalize all externalities.
For the most general statement of the problem, we imagine an economy in which production generates an unwanted output, pollution. Consumers derive utility from both private goods (x) and a public good, environmental services (–q). Pollution diminishes these services. For each consumer in the affected area, marginal willingness to pay for the public good (improved environmental services) in terms of diminished consumption of private goods depends on the marginal rate of substitution associated with the utility function ui(xi, q):
.
In equilibrium, with all n consumers facing the same relative prices, we know from Samuelson (1954) that the optimal reduction in x is that which satisfies:
i=1,… n.
If the public good is financed by a non-lump sum tax, then we have the familiar expression
where MCF is the marginal cost of public funds, a measure of the utility losses associated with the imposition of a distorting tax (Fullerton and Metcalf 1997). As Smith et al. (1997) have pointed out, use of this relation helps generate a taxonomy of cases relating to local vs. federal provision of public goods. Different taxes (local and federal) generate different levels of excess burden; if jurisdictions can freely trade among themselves then the setting of taxes influencing technology (the MRT) should be federal; if preferences are diverse, then there is a case for a variety of locally-provided public goods.
This standard result, however, relies on strong assumptions, including that of no environmental or other non-market linkages, and that of zero transactions costs. Relaxing these assumptions raises a problem of institutional design relating to the optimal definition of spatial and sectoral “control areas” (Sen 1972). For the moment, suppose that an optimal delineation of such control areas (i.e. jurisdictions) could be defined in terms of the parameters of the model. Then we have a basis for evaluating the existing allocation of mandates between different levels and agencies of government in relation to this optimal delineation.
To analyze the policy design problem, we adopt a much more restrictive setup. Suppose (for simplicity) that the jurisdiction consists of many industries producing a single common pollutant, qj, and that pollution has no direct impact on consumers. Each firm produces yj with technology ƒj(x). Although there may be many sectors, each with its own technology and emissions, we can explore the key issues by assuming all firms to belong to a single sector. The impact of each firm’s emissions, measured at a common receptor point, is modified by a transmission parameter dj, so that aggregate pollution measured at the receptor site is . If the policy goal is to reach a specified target quantity of pollution q*at the receptor point, the efficient solution is found by:
, (1)
where cj(•) is a cost function inclusive of abatement expenditures. The solution to this problem is characterized by:
(2)