Wildlife damage and agriculture: a dynamic analysis of compensation schemes.

The two principal threats to African wildlife are agricultural expansion and hunting. Increasing human populations are associated with greater conversion and fragmentation of wild habitats, and more intense hunting pressure on remaining wildlife stocks. Increased human encroachment in formerly wild habitat also sets the stage for conflicts between humans and wildlife, with casualties on both sides. Perhaps the most common cost imposed on humans by wildlife is damage to agricultural output, where a significant share of agricultural production near the extensive margin of human-nature interface can be destroyed by wildlife (e.g., Deodatus 2000). Predator species routinely take local livestock, and various other species have made a habit out of invading fields. In many areas of the developing world the conflict between humans and wildlife is tense, and indeed growing tenser over time.

The economic (and emotional) costs of this conflict can be quite substantial--from merely significant at the national or regional scale, to outright disastrous for individual households (Thouless 1994; Hoare 1995; Ngure 1995; WWF 2000). It is perhaps no surprise that outraged and frustrated farmers and pastoralists often seek revenge for such damages. Mishra et al (2003) mention a deep resentment among pastoralists against large c.arnivores in India and Mongolia. In practical terms, the risk of wildlife-imposed damage provides strong incentives for farmers to hunt in order to keep animal numbers and damages low (Bennett 2000). Hunting also yields bushmeat and other traded animal parts such as skins--commodities that are often highly valued locally or provide external income. (1) From a conservation perspective, the issue becomes particularly problematic when charismatic species like elephants, rhinos, lions, tigers, or snow leopards are involved. When wildlife damages are caused by these icons of the international conservation movement, and when nuisance killings contribute to their demise, international concerns and intervention often ensue.

To mitigate the incentives peasants have to kill wildlife, governments or nonprofit conservation organizations sometimes put in place compensation schemes whereby farmers are given money, seeds, or livestock to cover a portion of the losses imposed by wildlife. In developing countries, such efforts have been met with mixed success. (2) Lack of funds, transaction costs, heavy bureaucracy, fraud, corruption, and moral hazard problems have been identified as potential reasons why compensation schemes might fail to achieve their conservation objectives (e.g., WWF 2000). (3) But sobering experiences have not tempered all expectations. Initiatives to compensate farmers for wildlife damages continue to be popular both among public and private agencies, and are in existence to promote the conservation of many different species on different continents (e.g., elephants, rhinos, and lions in Africa, snow leopards, tigers, and antelopes in India and East Asia, etc.).

In this paper, we develop a dynamic model to examine the consequences of introducing a compensation scheme on the size of a single wildlife stock and on the welfare of local peasants. We focus on the compensation of victims--a topic that goes back to Coase (1960). Baumol and Oates (1988) demonstrate that such payments induce entry of victims and thereby trigger excessive damages. Our model adds the complexity that the polluter (wildlife) and victim (farmers) both require land, and that land use choices by farmers impact the wildlife stock.

Our general equilibrium model embodies both major threats to wildlife in developing countries: hunting and habitat conversion for agriculture. We model an isolated rural economy in the tradition of Brander and Taylor (1998), and in some of its details the model is also related to that of Bulte and Horan (2003). Other relevant prior literature includes Wirl (1999) who analyses the management of land that can be used for forestry or converted to agriculture. He demonstrates that cycles of forest conversion and re-growth may be optimal. Swallow (1990; 1996) studies situations where habitat degradation is irreversible. Finally, there is also a literature on compensating property owners for the taking of private land for public purposes (prominently to protect the natural habitat of endangered species--see Blume, Rubinfeld, and Shapiro 1984; Innes 1997; Polasky and Doremus 1998; and Smith and Shogren 2002). To our knowledge, however, the economics literature has not yet considered the issue of wildlife damage compensation in a dynamic general equilibrium setting. This is our focus.

Our principal finding is that compensation schemes aimed at reducing hunting mortality can actually reduce the wildlife stock and have ambiguous welfare effects for local people. Hence, although compensation programs are well intended, they could lead to the most disastrous outcome of all: compensation that is costly for the sponsoring agency could result in a reduction in the wildlife population and a fall in local welfare (a loss-loss-loss outcome). The intuition for this result is quite straightforward. Compensation distorts relative commodity prices, increases the returns to agriculture, and encourages agricultural expansion. Thus, while compensation reduces the incentive to hunt for wildlife, the net conservation effect is ambiguous when the negative impact of increased habitat conversion is accounted for.

In the next section, we describe a dynamic model of wildlife damage, compensation and land use conversion in a small rural economy. We proceed by analyzing the consequences of compensation for land use, wildlife stocks and local welfare. Then we highlight the consequences of a transition from autarky to trade, which may be facilitated by compensation, and we reflect on the relevance and robustness of our theoretical results. In the penultimate section we highlight some policy implications.

The Compensation Model

In this section we outline the outline the model, focusing on both the economy (production and consumption) and the ecological features of the system. We look at dynamics as well as steady states.

Production

The economy is made up of myopic households with open access to both land for agriculture and wildlife for animal products. Labor can freely flow from one activity to another in response to profit differentials. The assumption that property rights over land and wildlife are not enforced (or even defined) implies that we are considering the context of a less developed country, where conflicts between wildlife and farmers are most profound. The assumption that households respond in a myopic fashion to incentives facilitates the analysis but is not necessary for most of the results that follow.

We assume that land and wildlife are: (1) biologically interconnected, so that the capacity of the land to support wildlife is reduced as habitat is converted to agricultural land; and (2) economically interconnected, in that the opportunity cost of time spent growing crops is the foregone return from harvesting wildlife (and vice versa). This is consistent with the observations of Noss (1998, p. 166), who notes that hunting in an area of the Central African Republic is declining because of the "growing dependence on agriculture and the necessary time investment in clearing, planting, tending and harvesting fields" (see also Hill 2003 for similar observations). In our economy, it follows that at any point in time, the proportion of land devoted to agriculture and the labor choice of households are endogenous. The (opportunity) cost of hunting effort is therefore endogenous also.

The model extends work by Bulte and Horan along two important dimensions. First, the model explicitly models wildlife damage and is used to analyze the consequences of introducing a compensation scheme. Secondly, we derive the microfoundations for macro behavior by analyzing a general equilibrium model over time, rather than postulating demand curves for key commodities. Within this framework, we allow the wildlife stock to change in response to households' labor allocation and land use decisions.

Consider a small economy with a fixed human population endowed with an amount of land L and a time endowment T. A portion A(t) of this homogenous land is used by villagers to grow crops while the remainder is left to be used as wildlife habitat H(t). Land not used for agriculture is assumed to be immediately suitable as wildlife habitat regardless of previous use. Thus, at any point in time, the following land constraint holds identically (where the time index is suppressed to simplify the notation):

(1) A + H [equivalent to] L.

Households divide their productive time between agricultural labor, W(t), and hunting effort, E(t), (4) constraining the economy to

(2) W + E [equivalent to] T,

where T is the aggregate time endowment. Thus, the model recognizes two sectors of production. An agricultural commodity such as maize or grains is produced with a combination of land and labor; and products derived from wildlife harvesting are obtained from labor and a wildlife stock, the size of which we will denote by X(t).