|Objective: The pattern of land use and the changes in that pattern are thought by many to be the single most important factor affecting ecological systems. In developed countries and certainly in the United States, most land use change over the last few decades has resulted from the decisions of farm and forestland owners to sell their land for residential development. There is a second type of decision available to landowners, however, and one that is becoming increasingly prevalent in state and local government policy portfolios?that of enrolling property in a permanent conservation program. This involves the voluntary sale of the property's development rights to public agencies or to not-for-profit conservation trusts.
The major objective of this project was to develop a better understanding of how economic forces, locational characteristics, and policy features influence a landowner's farmland preservation decision. A secondary objective was to derive broader conclusions about the links between public policy mechanisms of this sort and environmental consequences. Although some environmental implications arose from the specific preservation versus development decision modeling, the secondary objective was addressed more broadly in a separate review of economic research and the land use-environment link (Bockstael and Irwin, 2000).
The chapter reviews the literature on two main themes. The first is the economics research on policies that affect land use change?growth control policies, policies to protect ecologically sensitive lands, and policies aimed at encouraging or discouraging agriculture. Agricultural preservation programs fall into all three categories. Although they provide a means to make continued agricultural activities possible in an increasingly urbanized landscape, they often are motivated as a means to control urban growth and/or protect ecologically sensitive areas from development?even though some types of agriculture are far from environmentally benign. The first conclusion drawn from this literature is that the interaction of the wide range of policies that affect land use decisions has had complex and indirect effects on land use change, often with unintended consequences. The second is that optimal resource allocation is very sensitive to location and pattern because the landscape is heterogeneous, both in its ability to provide ecological services and in its value in both agricultural and developed uses. It is especially challenging to integrate the effects of multiple policies, all of which affect the economic land use decisions of individuals and their ultimate effect on ecosystems, given that the spatial extents of ecosystems, political jurisdictions, and land markets rarely coincide. It argues for explicitly spatial economic analysis.
The second theme of the chapter is a review of spatial modeling of land use change. Space poses some particular challenges to economists, especially in understanding how the cumulative effects of many individual land use decisions combine over space and time to generate new spatial patterns. It is critical to remember, however, that the spatial distribution of land use and land use change are important to ecologists, hydrologists, and others, as many of their models depend on the pattern of land use as the essential input. As such, providing spatial models of land use change is a necessary task if economists are to enter into dialogues with other environmental scientists. While the specific research on agricultural preservation programs described below does not yet push the spatial dimension as far as we would like, the modeling approach incorporates some of the features that emerged in this literature review, and insights from this review have motivated the project's approach.
Summary/Accomplishments: In pursuing the specific research on agricultural preservation programs, we first recognized that these programs are typically implemented on the fringes of developed areas where growth pressures are leading to rapid increases in the development value of agricultural land. Development values of land can significantly exceed the agricultural use value in these areas, preventing new farmers from entering the industry due to the high acquisition cost of land. Development then becomes the only viable land use option when farm ownership turns over. In the absence of farmland preservation programs, the relevant land use question is not whether the land is developed, but rather when it is developed. By paying landowners for development rights that are permanently separated from the land, agricultural preservation programs offer an alternative to development. Landowners now face a complex decision that includes development or preservation at some time in the future. A characterization of the landowner's decision changes from simply an optimal timing of development decision to one that has two components: decisions about the optimal timing of development and preservation, and a discrete choice between these two land use alternatives. The policy advantage of the approach we have taken is that it attempts to reveal the features of programs that motivate landowners to preserve their land, using data that policymakers have readily available, rather than relying on survey data of individuals' characteristics.
We assume that landowners derive utility from net worth?the most significant component of which is the value of land or the proceeds from the sale of that land. In addition, landowners are assumed to derive utility from owning farmland, because they derive utility either from farming as an occupation or from holding land. Landowners also derive utility from avoiding farm debt. If the landowner knew all future prices, he could determine the optimal time to develop and the optimal time to preserve. The landowner could choose between these strategies to maximize net present value. However, perfect foresight is not realistic, and landowners are more likely to make their decisions in an iterative way, updating as more information becomes available. To capture this complex decision process, two different empirical approximations were conceived to account for at least some aspects of the multiple alternatives and the dynamic nature of the decision process.
One approach is to treat the observation period as one decision period, in which the landowner makes a discrete decision among three alternatives: preserve, develop, or postpone the land use change decision. The empirical counterpart to this model of the landowner's decision process is a polychotomous discrete choice model that allows for three alternatives: preservation, development, and postponement of the decision beyond the observable time horizon. An alternative approach is to assume that landowners update expectations in each period and ultimately choose the "terminal state" alternative (preservation or development) whose optimal transition time arrives first. This would be an accurate assessment if landowners do not possess a means to forecast the more distant future. This characterization of the problem describes the underlying assumptions of the competing risk model (specifically the Cox proportional hazards version of the competing risks model), which is a form of duration (or hazard or survival) analysis that is designed to analyze the occurrence and timing of events. The discrete choice and competing risks approaches differ in their implications for landowner behavior, and also in what the methods allow the researcher to capture in estimation. The chief drawback of the discrete choice approach is that it is a static representation of the decision process. The major drawback of the competing risks model is that it implicitly treats the optimal timing of development and the optimal timing of preservation decisions as separable.
The study analyzed decisions by landowners in four counties in Maryland?Calvert, Carroll, Frederick, and Howard. These counties were chosen because they are subject to heavy development pressure and have experienced considerable preservation activity. Landowners in all four counties can participate in the state-financed Maryland Agricultural Land Preservation Foundation (MALPF) program. In addition, Carroll and Frederick have "critical farms" programs that advance easement funds to farmers buying unpreserved farmland who wish to preserve the land through the MALPF. Howard has its own purchase of development rights (PDR) program, and Calvert has both a PDR and a voluntary transfer of development rights (TDR) program. The variation in circumstances helped to introduce variation in explanatory variables for estimating the model.
The majority of variables included in the model capture parcel characteristics that are expected to affect agricultural returns, returns to development, and expectations on changes in those returns. Also included in the model is a time varying measure to capture changes in demand in the land market, which will affect expectations of changes in development returns and easement values. The explanatory variables also include a measure of the length of parcel ownership to capture differences in debt circumstances, because individuals who have recently purchased or inherited agricultural land may find themselves in a more untenable debt position than others and may be more likely to preserve sooner as a result. Other important factors affecting the easement payment are agency preferences for parcels with particular characteristics (e.g., the size of the parcel and its proximity to other preserved parcels). These parcel characteristics are included as are indicator variables capturing differences among preservation programs (agricultural district requirements, whether the state program is the only preservation option, and eligibility for bonuses).
Even though the competing risks and multinomial logit models are intended to mimic different theoretical approximations to the land use decision process, the empirical results from estimating these models are quite similar, perhaps because of the limited number of preserved and developed parcels appearing in this circumscribed data set. Where the explanatory variables increase the odds of preservation or development relative to waiting in the multinomial logit model, they bring forward the expected time of preservation or development in the competing risks model. The differences between the models are limited to the variables that capture the changing context of the decision. The competing risks model introduces more information (in terms of the ordering of decisions) and more accuracy in the measurement of time varying variables.
For the counties of Carroll and Frederick?where PDR programs are relied on?increases in expected returns from development increase the likelihood of preservation, but at a decreasing rate. Results also suggest that the state program is successful in preserving the most productive farmland in these counties. However, in Calvert County?where parcels are preserved under a TDR program?proxies for development returns are significant determinants, and the most likely parcels to be preserved are those with the least productive land. In such programs, developers buy development rights without regard to parcel characteristics.
The effects of time varying explanatory variables were most effectively captured in the competing risks framework. One of them is the debt position of the landowner, proxied by length of land ownership. In every county but Howard (where we now have reason to believe that some of our data are suspect), such landowners are more likely to preserve. Variables reflecting agency preferences for preserving farms also are time varying. For programs in which priority is placed on preserving in clusters, the expected time to preservation is sooner for parcels near already preserved farmland. However, prioritizing on farm size does not appear to lead to larger farms ultimately being preserved. Finally, a county's reliance solely on the state preservation program does not appear to adversely affect the rate of preservation decisions. The incentive bonus offered in Carroll County does, however, increase the likelihood that preservation will occur sooner for Carroll landowners who qualify for it.
At least in Maryland, growth control policies, including land preservation programs, are considered an important part of environmental protection strategies. Whether agricultural preservation programs successfully contribute to environmental policy goals depends on the ability of these programs to entice landowners to participate; the type of agricultural activities and spatial location of the land that is most likely enrolled; the type of development that is deflected; and the change in density and location of development that results. From a pollution loadings perspective, the answer depends in part on whether the agricultural activity to be preserved is cropland or pastureland/forest, whether best management practices are adopted, and whether the development that would take place is serviced by sewage treatment or septic fields. However, ecological damage tied to impervious surfaces is unquestionably increased by development rather than preservation, where as little as 15 percent impervious surface cover is needed for a watershed's biotic integrity rating (for fish and macroinvertebrates) to drop to "fair" or worse; at a 25-percent impervious surface cover, only a few hardy pollution-tolerant species persist.
The research in this project provides a means of predicting the spatial location of preservation/development outcomes?a task that will be undertaken as soon as updated preservation data are in place. Spatial pattern predictions are especially important as the spatial location can have differing environmental consequences?both because of its location relative to ecological features such as watersheds and because of its location relative to human provided infrastructure such as public water and sewer. Even without predictive mapping, some specific results with environmental implications have arisen from the work to date.
In the counties where preservation occurred primarily through PDR programs, agency priorities influenced the preservation outcome, because in each year agency budget constraints were binding: there was sufficient interest in the preservation option so that more offers were made to sell development rights than the agencies could afford to purchase. This suggests that preservation agencies can have an impact on the environmental features of the parcels that are preserved, when they use parcel characteristics as a means for prioritizing purchases of development rights. Because TDR programs operate without direct government intervention, the type of land likely to be preserved was found to be different from that preserved under PDRs. Purchasers of development rights (i.e., developers) in TDR programs negotiate purchases strictly on the basis of cost and are indifferent to the environmental features of the parcels and the location and type of land being preserved. TDR programs actually preserved land with poorer soils that generally require greater fertilizer and pesticide use.
In Maryland, participation in PDRs requires participants to adopt soil conservation and water quality plans. Statistical tests using variables intended to serve as proxies for the extent of necessary water quality management practices revealed no evidence that these requirements discouraged participation. However, the likelihood of preservation was found to increase with the percent of land in row crops. Although water quality and soil conservation plans are required to be implemented, this type of agricultural activity probably has more negative consequences on environmental quality than pastureland or forested land. Finally, counties whose agencies ration scarce preservation funds in PDR programs, partly on the basis of the parcel's distance to the nearest preserved parcel, have been successful in preserving land in clusters. This suggests that preservation programs can and do result in large tracts of contiguous, undeveloped land. If strategically located with respect to watersheds, these large contiguous areas can have important implications for stream ecosystems, because it has been found to be critical to preserve as much as 75-85 percent of a watershed to avoid compromising stream integrity.