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

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.