Impact of designated preservation areas on rate of preservation and rate of conversion: preliminary evidence.

Concerns about farmland loss and suburban sprawl motivated 124 government entities to institute preservation programs at an overall cost of $3.723 billion (American Farmland Trust 2005a, 2005b). In addition, 38 states have implemented land conservation programs for ecological reasons. Critics suggest these programs do not prevent land conversion,(1) do not maximize social benefits, and do not prevent land fragmentation (Pfeffer and Lapping 1995; Daniels and Lapping 2005; Lynch and Musser 2001). Concentrated preservation may provide greater benefits if threshold impacts or economies of scale exist. For example, a critical mass of contiguous farms may be needed due to economies of scale in support industries (Lynch and Carpenter 2003) and to avoid conflicts with non-farm neighbors. Large blocks of undeveloped land may be needed for ecosystem services provision such as wildlife habitat and water quality.

Although Maryland has a plethora of land preservation programs, contiguity of preserved parcels has not been achieved. To prioritize contiguity, Maryland introduced the Rural Legacy (RL) program in 1997. The cornerstone of the RL program is to designate RL Areas, which receive special funding to preserve farm, forest, and ecologically important resource lands in a contiguous fashion. The special funding should result in more preservation in these areas but also might interact with the existing preservation programs. Thus, we examine the following preservation outcomes. (a) The rate of preservation in the RL area increases due to the extra funding itself and to its catalyst effect on the other preservation programs. If economies of scale or threshold impacts exist, other programs may receive higher benefits from each additional acre preserved in the RL area relative to a non-RL parcel. (b) The RL program's effort crowds out other programs' preservation. As competition between programs increases the cost of preserving in the RL area, other programs may select parcels outside the RL area. (2) We test to see if the rate of preservation measured as both number of parcels and number of acres has increased within the RL areas since 1997.

We also examine if the RL area designation affects the rate of conversion. Geoghegan, Lynch, and Bucholtz (2003) and Roe, Irwin, and Morrow-Jones (2004) find that preservation efforts can generate positive amenities for adjacent homeowners in the form of guaranteed open-space. Increased demand for housing near preserved parcels can create higher returns for housing construction and thus accelerate the rate of conversion. We test to see if the rate of conversion, measured as the construction of new houses or new subdivisions, is lower within the RL areas since 1997.

Background on Rural Legacy Program

Maryland has been a leader in agricultural preservation in terms of acres preserved and innovative programs but has had little success in preserving large contiguous blocks of land. The RL Program requires contiguous tracts of resource land be identified for preservation and provides extra funding for these areas. In addition, the program permits point-based payments to landowners that reward ecological and other attributes such as large parcels, wetlands, endangered species and other wildlife habitat, and swamps, which are often discounted in an appraisal approach. State-wide, the program has encumbered $156 million through 2006 and preserved almost 52,000 acres. RL areas are approved based on the agricultural, forestry, and ecological values (including economic values) and the level of conversion pressure.

[FIGURE 1 OMITTED]

Our three-county study area in Southern Maryland has five approved RL areas for a total of 84,102 acres. Three of the five are used in this analysis; two did not have data available because they were only recently approved. The Calvert Creeks area (20,527 acres) seeks to protect water quality and wetland habitat. The RL has protected 1,660 acres at a cost of $6.8 million with an average price of $4,003 per acre. The Huntersville area (8,357 acres) in St. Mary's County includes shoreline with endangered species habitat, wetlands, historic structures, agricultural and forest land, and archeological sites. This RL program has preserved 2,720 acres at a cost of $8.8 million; average cost was $2,693 per acre. The Zekiah Watershed area (31,000 acres) in Charles protects farms, forests, wetlands, and historic and archeological sites. This RL program has preserved 2,007 acres at a cost of $6.5 million at an average cost of $3,082 per acre.

The state program, Maryland Agricultural Land Preservation Foundation (MALPF), has preserved agricultural land in all three counties. The program has protected 4,263 acres in Calvert, 5,872 in St. Mary's, and 3,474 in Charles. Payments in 2002 were $7,714 per acre in Calvert, $2,480 in St. Mary's, and $3,024 in Charles. Transfer of development rights (TDR) programs are also available. Calvert has protected almost 13,000 acres through its TDR program; St. Mary's has 221 acres, and Charles has 1,554 acres. Private conservation groups have also preserved land.

Almost 30% of the RL areas have been preserved compared to only 5.6% of the non-RL areas. Clearly, there exists a predisposition to preserve in the RL area without considering parcel characteristics. Figure 1 delineates the RL areas and the level of preservation under the different programs geographically.

Since 1997, significant preservation activity has occurred from all the preservation programs. RL has preserved almost 6,500 acres in these three counties since 1997. The MALPF program has preserved almost half a of its acreage since 1997. The TDR programs s in Calvert and Charles have preserved 40% of their acres since 1997. The raw data suggest a crowding out phenomenon: the TDR programs had preserved 42% of its acreage within the RL area pre-1998; but after 1997, the TDR program preserved only 30%. Similarly, MALPF preserved 28% of its acreage in RL areas before designation but only 4.5% t after 1997.

Methodology

If parcels in RL areas have characteristics that make them less likely to be developed and/or more likely to be preserved, then assigning the impact to the designation of a preservation area would be incorrect. One cannot construct the proper counterfactual, i.e., one does not know if a particular parcel would have been preserved and/or developed if not included because it can not be in the two states simultaneously, i.e., cannot be both within and outside of a designated preservation areas at the same time. To overcome this, a propensity score matching (PSM) developed by Rosenbaum and Rubin (1983) is used to estimate the treatment effect, i.e., does being targeted for preservation and within the designated area change a parcel's likelihood of being preserved? We assume a conditional mean independence (CMI) criteria E[[Y.sub.0] [absolute value of D = 1, X] = E[[Y.sub.0]] D = O, X] = E[[Y.sub.0] [absolute value of X] P(D = 1] X) [member of] (0, 1) to estimate the average treatment effect on the treated (ATT) (Heckman, Ichimura, and Todd 1997).

A propensity score is computed for each parcel as to its likelihood of being within a RL area and then is used to explicitly match treatment parcels (those within the RL areas) with control properties (those outside the RL areas). Benefits of PSM include that the matching protocol ensures comparison of RL parcels with the most similar non-RL parcels in terms of characteristics. Parcels with different characteristics and outliers will have no/little influence. In addition, PSM does not impose a specific functional form.

Data Sources and Included Variables

Data was collected on agricultural and forest parcels greater than three acres in three Maryland counties. We collected information on variables that affect RL area designation and preservation outcomes. The primary data set, the MDPropertyView 2002 Database, provided parcel-level information such as parcel size, zoning density, waterfront access, public sewer availability, housing construction date, subdivision designation, and geographic coordinates. Geographic Information Systems (GISs) were used to extract 1997 land use data from satellite and aerial photography, and the percent of cropland, pasture, forest, and wetlands were computed for each parcel. Soil data were extracted and grouped by slow versus medium or rapid permeability, flat versus medium or greater slope, and susceptibility to flooding (Maryland Department of State Planning; 1973). Distance to Washington, D.C. in miles was computed using U.S. Census Bureau road networks. For the ecological values of each parcel, we computed the percent of Maryland's Sensitive Species Project Review Areas, which include rare, threatened, and endangered species and rare natural community types (Maryland DNR 2003), and Non-tidal Wetlands of Special State Concern (Department of the Environment and DNR 1998). The percent of the parcel in estuarine wetland status was also extracted from the Maryland Wetland Map (Maryland DNR 1998). (3)

Preservation data was collected from the state-wide preservation program MALPF parcels through 2005, from the Maryland Environmental Trust through 2004, from the Calvert Transfer of Development Rights program through 2004, and from private conservation groups (Nature conservancy, private land trusts) through 2005. Maryland DNR provided information on RL area designation and approved landowners who were matched to specific parcels. The number of preserved acres within a one-mile radius was also extracted for each parcel.

Propensity Score Estimation, Matching Methods, and Balancing Test