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

Propensity scores are estimated with a Logit model using Stata software. Table 1 reports the estimated coefficients for predicting a parcel's inclusion in a RL Area. Having a large number of preserved acres nearby, having waterfront property, having a high percent of estuarine area, having a higher percent of a floodplain, and land-uses of cropland and pasture all increased a parcel's likelihood of being within a RL area. Parcels with zoning densities permitting more houses per acre were less likely to be included within the area.

Two matching methods were used: kernel and local linear techniques. The ATT varied very little by matching technique. When comparing preservation outcomes, we restrict the sample to parcels greater than ten acres since these are the most likely to be preserved. When evaluating conversion outcomes, we match treatment parcels to control parcels within the full sample of all parcels three or more acres.

Balance tests are conducted using the standardized difference test; a t-test for equality of the variable means in the matched treated and control groups. For the preservation outcomes, we find that the treatment and control groups (ten+ acres) have equal means except for the permeability and waterfront variables. Surprisingly, the matched RL parcels have a higher mean permeability (64% compared to 58%) and a higher likelihood of having waterfront property (7% compared to 5%) than non-RL parcels. Theoretically, parcels with higher permeability levels and waterfront would be more likely to be developed rather than preserved. The quality of the waterfront may vary between parcels; for example one may have an inaccessible area, another have a picturesque riverbank. We also find permeability does not balance for the full sample (3+ acres) with RL parcels having slightly lower mean values (56% compared to 59%).

Results

Our matching estimates of the impact of the RL Areas appear in table 2. These include the bias-corrected average treatment effect (BC ATT) and the bootstrap standard errors computed from 500 independent draws. We report BC ATT for both time periods: Post-1997 (post-RL beginning) and Pre-1998 (pre-RL).

Two different outcome variables are used to determine whether the RL designation affects the rate of preservation: number of acres (breadth of outcome) and number of parcels (decision-making unit). The number of acres preserved in the RL area is greater than outside the designated area following RL designation. In essence, the results show that each parcel preserved has 8.7 to 8.9 more acres on average if within the RL.4 Before 1997, acres within and without were equally likely to be preserved. Therefore, the RL program has had a positive impact on the number of acres retained. In addition, parcels within the RL area were more likely to be preserved than those outside RL areas. In this case, we find that this was true even before RL designation. However, the rate of preservation increases 50% from 4.7 % to 7.1%. This indicates that the designation is having an impact on preservation outcomes on two levels: a higher percent of parcels and more acres/larger parcels.

Similarly, we look at two outcome variables to evaluate the impact on conversion: the number of acres and parcels that have a new house or a new subdivision within the RL area compared to those outside the RL area. We find that RL designation has no impact on the number of acres or the number of parcels being converted. Conversion is occurring on approximately 20% of the parcels both within and without the RL areas. Thus, the RL designation and special funding does not appear to be having an impact on housing construction or subdivision creation as measured. Due to time lags in construction, some of these homes may have been planned before the RL program began. Therefore, a re-evaluation at a later date may prove otherwise.

Conclusions

Many counties designate farmland retention areas, but these areas are often more than 50% of a county's land and too large for the financial resources available. Programs may also have equity concerns, specifically in regards to spreading the budgeted resources over the entire state or geographic area. Both these may result in inefficient outcomes especially if economies of scale and/or threshold levels exist. For example, Wu and Skelton-Groth (2002) show that parcel selection based on political equity concerns may lead to the lowest possible benefits in the presence of threshold impacts.

In an attempt to encourage more contiguous preservation, Maryland introduced the RL program to preserve large contiguous blocks of land with high social value. Local entities designate preservation areas and become eligible for special funding. Overall, the designation appears to have a positive impact on acres retained and on probability of preservation for the identified area. The increased preservation appears to be primarily due to the new RL funding rather than from attracting more preservation from other programs. While the raw data suggests crowding out, no evidence of it was found. The program has enrolled more acres and larger parcels due to the extra funding and the new payment schemes based less on market appraisals. In Charles County, the RL designation resulted in preservation of a completely new area.

RL designation, however, seems to have done little to discourage land conversion relative to non-RL areas. We do not find any impact on the rate of conversion or on the number of acres converted in the RL area. On average, 20% of parcels were converted post-1997. However, more time may be needed for a conclusive test of the effectiveness of targeted preservation area on preventing conversion.

This analysis examines a limited number of questions and focuses on the initial years of the program. Further evaluation is needed to determine the impact of and the overall benefits of targeted preservation areas. County-level analysis may be illuminating given differences in local preservation programs and local planning but technically this may prove challenging. Future work will examine how the preservation and conversion patterns have impacted fragmentation and/or contiguity issues. The targeted RL areas are of different sizes--one has only 8,000 acres; another 30,000 acres. What is sufficient? What is the trade-off between size of area and contiguity of preservation assuming limited resources? Does the spatial configuration of the area matter? In addition, understanding the threshold levels for different objectives, i.e., how much area of forest or farmland is needed for habitat purposes or at what percent of impervious surface area (roads, houses, other structures) does water quality impacts begin, would be of interest. Further exploration of the change in the cost per acre of preservation will also be conducted to determine if crowding out is an issue. Currently, the RL funds are shared between all the RL areas within the states based on submitted proposals of willing landowners within each areas. Another key question is whether the state would be better off preserving all of one RL area before shifting resources to another RL area.

Support was provided by the Harry R. Hughes Center for Agro-Ecology, Inc., and the National Center for Smart Growth. We thank Karen Palm, Jay Harvard, Sabrina Lovell, and Seth Wechsler for their research assistance. We also thank Virginia McConnell, Elizabeth Kopits, and Margaret Walls for assistance with Calvert's TDR data. Any errors remain the authors' responsibility.

References

American Farmland Trust. 2005a. Fact Sheet Status of Local PACE Programs. Northampton, MA: American Farmland Trust Farmland Information Center.

--. 2005b. Fact Sheet Status of State PACE Programs. Northampton, MA: American Farmland Trust Farmland Information Center.

Daniels, T., and M. Lapping. 2005. "Land Preservation: An Essential Ingredient in Smart Growth." Journal of Planning Literature 19(3):316-29.

Geoghegan, J., L. Lynch, and S. Bucholtz. 2003. "Capitalization of Open Spaces." Agricultural and Resource Economics Review 32(1):33-45.

Heckman, J., H. Ichimura, and P. Todd. 1997. "Matching as an Econometric Evaluation Estimator." Review of Economic Studies 64:605-54.

Liu, X, and L. Lynch. 2006. "Do Agricultural Preservation Programs Affect Farmland Conversion? Evidence from a Propensity Score Matching Estimator." Department of Agricultural and Resource Economics Working Paper 06-08 2006, University of Maryland, College Park.

Lynch, L., and J.E. Carpenter. 2003. "Is There Evidence of a Critical Mass in the Mid-Atlantic Agricultural Sector Between 1949 and 1997?" Agricultural and Resource Economics Review 32(1):116-28.

Lynch, L., and W.N. Musser. 2001. "A Relative Efficiency Analysis of Farmland Preservation Programs." Land Economics 77(4):577-94.

Lynch, L., K. Palm, S.J. Lovell, and J. Harvard. 2007. Expected Cost of Tripling Maryland's Preserved Acres. Wye Mills, MD: Center for Agroecology.

Pfeffer, M.J., and M.B. Lapping. 1995. "Prospects for a Sustainable Agriculture in the Northeast's Rural/Urban Fringe." Research in Rural Sociology and Development 6:67-93.

Roe, B., E.G. Irwin, and H.A. Morrow-Jones. 2004. "The Effects of Farmland, Farmland Preservation, and Other Neighborhood Amenities on Housing Values and Residential Growth." Land Economics 80(1):55-75.

Wu, J., and K. Skelton-Groth. 2002. "Targeting Conservation Efforts in the Presence of Threshold Effects and Ecosystem Linkages." Ecological Economics 42(1):313-31.

(1) Recent evidence finds that preservation programs decrease the 5-year rate of county farmland loss by 3 percentage points; a 43% decrease from the average 5-year rate of 7.31% (Liu and Lynch 2006).

(2) Data limitations precluded an analysis of the programs' payments over this period.

(3) More details on data are available in Lynch et al. (2007).

(4) Liken this to each worker earning an average of $2.00 more per hour after a training program.

Lori Lynch is Associate Professor and Xiangping Liu is Graduate Student in the Department of Agricultural and Resource Economics, University of Maryland.

This article was presented in a principal paper session at the AAEA annual meeting (Portland, OR, July 2007). The articles in these sessions are not subjected to the journal's standard refereeing process. Table 1. Estimated Coefficient Estimates for the Propensity Score Logistic Regression Pseudo [R.sup.2] = 0.4643 Log likelihood = -3979.79 Dependent Variable In/out of Rural Legacy Area Independent Variables

Coef. Estimates Std. Err. Acres 0.0048 0.0043 Waterfront 0.8542 *** 0.1544 On public sewer 0.2139 0.3596 Miles to Washington D.C. -0.03442 0.0252 Zoning density -1.3419 *** 0.1743 Cropland_1997 (%) 1.1939 *** 0.2030 Pasture_1997 (%) 3.2922 *** 0.8071 Forest-1997 (%) 0.2419 0.1619 Wetland_1997 (%) -0.7950 0.5333 Soil flood susceptibility 1.3973 *** 0.4598 Soil slope medium + 0.1998 0.3555 Soil permeability medium + 0.2050 0.4243 Special habitat areas (%) 0.2051 0.8455 Estuarine (%) 4.5068 ** 1.8402 Preserved land within

1-mile index 0.9822 *** 0.3519 Acre * preserved within

1 mile -0.0008 0.0008 Acre * waterfront -0.0011 0.0015 Acre * wash. miles -0.00006 0.0001 Acres2(squared) 0.000002 0.000006 Wash D.C.miles2 -0.0002 0.0003 Preserved within 1 mile2 0.1555 ** 0.0667 Estuarine2 (%) -6.8795 4.1750 Zoningdensity2 0.1247 0.0825 Cropland_19972 (%) 0.0707 0.2248 Pasture_19972 (%) -1.6596 1.0676 Forest_19972 (%) 0.8342 *** 0.1590 Soil flood2 -4.4510 *** 0.5627 Soil slope medium+2 -1.1905 *** 0.3745 Special habitat areas2 (%) 0.0125 0.8485 Constant -5.3717 *** 0.6824 Number of obs 26,048 Note: Double asterisk (**) indicates statistical significance at the 0.05 level, and triple asterisk (***) indicates statistical significance at the 0.01 level. Table 2. Effect of Rural Legacy Designation on the Number of Acres Preserved, Rate of Preser-vation, and Number of Acres Conversion and Rate of Conversion Outcome Pre or Post # Treated Obs./ Variable RL Designation # Control Obs. Preserved acres Post-1997 696/6851 Preserved acres Pre-1998 683/6850 Preservation rate Post-1997 696/6851 Preservation rate Pre-1998 683/6850 Converted acres Post-1997 2153/23895 Conversion rate Post-1997 2153/23895 Outcome Kernal Normal Local Linear Variable BC ATT Biweight BC ATT

(Std. Err.) (Std. Err) Preserved acres 8.73 ** 8.97 **

(2.38) (2.39) Preserved acres 3.60 3.52

(2.79) (2.74) Preservation rate 0.071 ** 0.072 **

(0.018) (0.017) Preservation rate 0.047 ** 0.045 **

(0.016) (0.016) Converted acres 0.019 0.032

(0.28) (0.27) Conversion rate -0.01 -0.008

(0.013) (0.012) Note: Double asterisk (**) indicates statistical significance at the 0.05 level.