The Opportunity Costs of Coastal Land-Use Controls: An Empirical Analysis

Transcription

1 University of Delaware From the SelectedWorks of George R. Parsons August, 1991 The Opportunity Costs of Coastal Land-Use Controls: An Empirical Analysis George R Parsons Yangru Wu Available at: 56/

2 The Opportunity Cost of Coastal Land-Use Controls: An Empirical Analysis Author(s): George R. Parsons and Yangru Wu Source: Land Economics, Vol. 67, No. 3 (Aug., 1991), pp Published by: University of Wisconsin Press Stable URL: Accessed: :35 UTC JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at University of Wisconsin Press is collaborating with JSTOR to digitize, preserve and extend access to Land Economics

3 The Opportunity Cost of Coastal Land-Use Controls: An Empirical Analysis George R. Parsons and Yangru Wu I. INTRODUCTION We use a method previously used by Edwards and Anderson (1984) and Shabman For more than a decade demographers and Bertelson (1979). First, we estimate a have been documenting the migration hedonic of the price regression for housing using U.S. population from interior to cross-sectional coastal data from a developed states. A belief that this trend will coastal persist market similar to that where landuse housing controls are being considered. Included has led to concern about increased and commercial development in coastal in this areas. Added development brings water housing pol- in the coastal area-view of the regression are amenities unique to lution and reduces the natural cover of the water, frontage on the water, and nearness coastline. Many states have responded to to the water. Next, we predict the number this trend with land-use controls that limit of new houses that would have been built new residential and commercial development on land adjacent to coastal water. built elsewhere with controls. These are in the coastal area without controls but are Controls such as these have essentially "displaced" houses. Last, using the hedonic regression, for each displaced house three economic efficiency effects. First, on the positive side, is the preservation of we predict the lost value of no longer having amenities unique to the coast. Summing coastal open space and reduction of water pollutants-benefits enjoyed by residents these lost values over all displaced houses and visitors. Second, and on the negative gives the total loss. In Shabman and Bertelson's terminology this is lost "development side, is decreased residential and commercial proximity to the coast-fewer households and businesses can locate near the The reasoning of the method is that con- value." waterfront. Third, and also on the negative trols reduce households' implicit consumption in- of certain coastal amenities by the side, is a potential loss of amenities at land locations. With greater development number in of displaced houses, and the value inland areas following controls, there may of each lost amenity may be approximated be increased housing density and less preservation of inland natural sites. Another markets. by its hedonic price in existing housing possible effect, which may be positive orthe two efficiency effects that we do not negative, is a change in infrastructure-- address-added open space and inland highways, sewage services, police services, externalities-are ignored for methodologiand so on. If these services, on net, change or the cost of providing them changes, there is yet another efficiency effect. Parsons is assistant professor, College of Marine We estimate the cost of the second efficiency effect-displaced residential devel- Delaware, and Wu is a graduate student, Department Studies and Department of Economics, University of of Economics, Ohio State University. opment or lost access to coastal amenities. We thank A. Myrick Freeman, III, William We analyze controls recently established infischel, Mary Jo Kealy, Eran Feitelson, and two Maryland for the Chesapeake Bay-the anonymous referees for comments; Richard Sacher for Critical Area Program which limits new de-programminvelopment in a 1,000 foot buffer zone abut-sions; and Jimmy and Anne Talkington for helping us assistance with the Box-Cox regres- obtain our data set. The research was funded by ting the water. We consider one county, NOAA Office of Sea Grant, Department of Commerce, under Grant No. NA86AA-D (Project Anne Arundel, and ignore displaced commercial development. No. SG87 R/CB-2). Land Economics - August (3):

4 67(3) Parsons & Wu: Land-Use Controls 309 cal reasons. Policy analysts access amenities certainly may be approximated would by want such measures to make rational decisions. Unfortunately, our data set limits usamenities in the coastal housing market equation [1]-the revealed price of the to a narrower question, but one that is still without controls noting that the Critical of relevance to policy. Area extends.2 miles from the water. (See Edwards and Anderson [1984] and Shabman and Bertleson [1979] for a similar mea- II. THE OPPORTUNITY COST OF DISPLACEMENT sure of lost access amenity value.) If N such houses are displaced by the Our measure of the cost of displacement is based on a hedonic price analysis controls, 1, of v = the 1, the d = loss d') is - pj(x, simply f j'=l = 0, [pj(x, v = f 0, = Anne Arundel County housing market. d =.2)]. Since We houses may be displaced estimate a hedonic regression using for crosssectional data from that housing total market discounted lost amenity value is many years following the controls, the prior to the introduction of controls. The hedonic includes several structural and lo- LOSS = cational features of houses that will be defined in the next section. For now, we de-[p,(x,note these features by the vector x. We also 1,v,= 1,d =d') t=0 j=l include three coastal access amenities in - ptj(x,f = 0, v = 0, d =.2)]/(1 + r)t[2] the hedonic: a dummy for frontage (f = 1 if a house has frontage on the coast, 0 if where N, is the number of houses displaced not); a dummy for view (v = 1 if a house in year t, t = 0,..., T with the first year has a view of the water, 0 if not); and nearness to the coast (a linear measure of dis- following the controls being t = 0, and r is the rate of discount. LOSS is in terms of tance, d, in miles from the water). Our hedonic then is p(x, f, v, d) where p is the "year 0" dollars. Equation [2] may also be written as price of a house. Implicit prices of coastal access amenities are easily calculated using LOSS = this hedonic price function. For example, T the implicit market price of having a house SNE[pt(x,f = 1, v =1, d = d') near the water with view and frontage versus having a house with no view or frontage - p,(x,f = 0, v = 0, d =.2)]/(1 + r)' t=o and being.2 miles from the water is [2'] Ap = p(x,f = 1, v = 1, d = d') where E[pt(x,f = 1, v = 1, d = d') - p,(x, - p(x,f = 0, v = 0, d =.2) [1] f = 0, v = 0, d =.2)] is the mean value of the N, displaced houses in year t. The where d' (<.2) is how far the house is located from the coast. Ap is a measure of If controls displace some coastal houses subscript j has been suppressed. the discounted future implicit market value that would not have had frontage or view of these coastal access amenities. (houses that would have been near the water but were not so close as to have frontage Now consider how coastal land-use controls displace houses. Numerous houses or view), equation [2'] may simply be broken into groups-each group with different that would have been built on the coast (with frontage and view) had there been lost no amenity values. controls are instead built inland (without In our analysis we consider three frontage and view and at some distance groups: (1) houses that would have had from the water). For each of these displaced houses coastal access amenities are have had view but no frontage, and (3) view and frontage, (2) houses that would lost. Structural and other locational attributes are not. The implicit value of the lost miles from the coast but with no view houses that would have been less than.2 or T N,

5 310 Land Economics August 1991 frontage. Equation undeveloped [2'] portions is of the now coastline wri to three pieces. current markets, we believe this is a reasonable assumption. LOSS = Third, Ap does not capture the mitigating effect of inland amenity substitutes. If T Z NeE[ptl(x,f= 1,v = 1, d = d') new houses constructed inland following t=0 controls are built near parks, rivers, and - Ptl(x,f = 0, v = 0, d =.2)]/(1 + r)t T S Nt2E[pt2(x,f= O, v = 1, d = d') t=0 T we miss this offset. -Pt2(x,f= 0, v = 0, d =.2)1/(1 + r)t + N,3E[Pt3(x,f = 0, v = 0, d = d') t=0 where N,t, Nt2, and Nt3 correspond to the number of displaced houses in each of the groups. E[ptk(') - Ptk(')] is the mean value of the Ntk houses in group k (= 1, 2, or 3) displaced in year t. We estimate LOSS in the following section and divide it into 4 parts: LOSS = LOSS LOSS LOSS96 + LOSSo0105. [4] open spaces that substitute for lost coastal access amenities, this will offset the loss. By virtue of holding x fixed in equation [1] Hence, we qualify our estimates with the assumptions of a small and open market and future coastal access amenities being similar to current amenities. And, we inter- - Pt3(x,f = 0, v = 0, d =.2)]/(1 + r)t [3] pret it as a measure that misses the mitigating effects of households finding inland amenity substitutes. III. THE DATA AND EMPIRICAL ANALYSIS We analyze single-family houses sold in Anne Arundel County, Maryland in Anne Arundel County is located on the western shore of the Chesapeake Bay ap proximately 35 miles east of Washington D.C. Its major urban areas are Glen Burni LOSS86_0o is present value of lost amenity and Annapolis. In 1985 there were 141,000 values due to houses displaced in the houses years in the county and it had a populatio 1986 through 1990, LOSS91_95 is for of 412, through 1995 and so on. We present The the county has 432 miles of shoreline present value of losses for an average including year land immediately abutting the in each of these five-year increments. Bay and land abutting three major rivers Under reasonable assumptions, LOSS that feed is the Bay. The coastline is one of a defensible estimate. First, if the restricted the most intensely developed on the Chesapeake. We estimate that approximately 80 coastal area is "small and open" in the sense defined by Polinsky and Shavell percent of all housing in the county is located within one mile of the shore. Never- (1976), Ap captures the full value of lost coastal access amenities. The area is theless, more than half of the coastal land "open" if there is perfect migration within between the affected coastal area and other Area is undeveloped. 1,000 feet of the water in the Critical housing markets in the region. It is "small" Our sample is a random draw of 1,435 if it has an insignificant effect on the overall houses located less than six miles from the supply and demand for land for housing coastline that sold in (The Commission to was announced in December 1983 the region. Insofar as the controls apply only a 1,000 foot strip of land by the water, and established in 1984.) Any observation we believe the small and open assumption with missing data on characteristics, tha is reasonable. were not market sales, or that we could no Second, using Ap assumes coastal ac-locatcess amenities in current markets are simimately 5 percent of the data. Table 1 de- on a map were deleted-approxilar to what these amenities would be in future markets. Given the similarity of the structural characteristic data are from fines our variables. The sale price and the

6 67(3) Parsons & Wu: Land-Use Controls 311 TABLE 1 DESCRIPTION AND MEAN VALUE OF VARIABLES USED IN THE 1983 REGRESSIONS Standard Variable Description Mean Deviation PRICE Market price of a house 91,555 55,284 BD Number of bedrooms BATH Number of bathrooms DINED Dummy variable (1 = formal dining room) BASED Dummy variable (1 = full basement) AGE Age of a house (years) HISTDUM Dummy variable (1 = historic neighborhood) GARAGE Dummy variable (1 = garage or carport) AIRCON Dummy variable (1 = central air conditioning) FRPL Dummy variable (1 = fireplace) SF Interior area of house (square feet) 1, LOTSZ Area of lot (square feet) 23,755 12,305 MONTH Month the house was sold DISTANCE Linear distance (1 to the = nearest January, point on the Bay or = Decem tributary (miles) DISTCBD Distance to central business district (miles) FRONTAGE Dummy variable (1 = water frontage) VIEW Dummy variable (1 = water view) ED Percent of blockgroup over 18 years old with 4 years high school education or more %NWH Percentage of blockgroup classified as non-white HHINC Median household income of blockgroup 25,827 6,571 County Board of Realtors. The signs and locational are statistically significant in each variables are census (block group) regression. data For a and similar focus on coastal our own measures of distance to the Central Business District (CBD) and coastline. wards and Anderson (1984); Shabman and amenities in a hedonic regression see Ed- The FRONTAGE and VIEW variables are Bertelson (1979); Milon, Gressel, and Mulkey (1984); and Brown and Pollakowski from the County Board of Realtors which we verified for a random draw from our (1977). Using these results we estimate the value of lost coastal access amenities used sample. Our estimates of the hedonic regression (the p(x) to be used in equation [3]) are given in Table 2. We estimated three functional forms: Linear, Double-Log, and Linear Box-Cox. In the Linear Box-Cox, 0 transforms price and X transforms explanatory variables. In both the Double-Log and the Box-Cox a one is added to AGE, DIS- TANCE, DISTCBD, and %NWH because some observations have a value of 0 for these variables. The dummy variables are, of course, not logged in the Double-Log or transformed in the Box-Cox. For all of the regressions our chosen set of attributes explain a considerable portion of the variation in housing prices and for the most part have estimated coefficients with expected signs. The coefficients on frontage, view, and distance have expected in equation [3] for each of our three housing groups. We assume E[Ptl(x,f = 1, v = 1, d = d') - p,t(x,f = 0, v = 0, d =.2)] =(1/M1) [pi(x,f = 1, v = 1, d = d') - pi(x,f = 0, v = 0, d =.2)] for all t for group 1, E[Pt2(x,f = 0, v = 1, d = d') - Pt2(x,f = 0, = 0, d =.2)] M2 = (1M2) [p(x,f = 0, v = 1, d = d') - Pi(x,f = 0, v = 0, d =.2)] for all t for group 2,

7 312 Land Economics August 1991 TABLE 2 HEDONIC REGRESSIONS, 1983 Linear Variable Linear Double-Log BOX-Coxa INTERCEPT -42,742(5.8) 4.8(16.5) 4.6(297) BD -11.6(0.0).06(2.3).009(3.1) BATH 6,742(3.1).1(4.3).01(5.1) DINED 7,674(4.0).05(4.1).007(3.9) BASED -1,056(0.5).003(0.2).001(0.9) AGE 3.3(0.1) -.06(10.7) -.004(9.8) HISTDUM 46,237(4.0).6(9.3).08(8.5) GARAGE 9,841(4.7).08(6.0).01(6.4) AIRCON 5,298(2.3).05(3.4).007(3.6) FRPL 3,053(1.4).08(6.2).01(6.5) SF 29.3(16.4).4(14.6).007(14.1) LOTSZ.08(10.5).1(17.6).001(17.1) MONTH 338(1.2).03(3.4).003(4.0) DISTANCE -2,326(1.7) -.07(4.2) -.008(3.8) DISTCBD - 1,053(3.7) -.06(7.7) -.005(6.8) FRONTAGE 66,880(17.7).4(18.1).06(18.1) VIEW 7,418(2.3).07(3.5).0096(3.4) ED 789(5.1).06(3.8).01(7.1) %NWH 220(2.9).02(4.2).002(3.8) HHINC.6(3.0).2(8.1).002(5.7) R F-Statistic Observations 1,435 1,435 1,435 Note: t-statistics are in parentheses. abox-cox parameter estimates are K =.26 a E[Pt3(x,f = 0, v = of 0, lost access d = amenities d') for group 1 houses is $96,672/house. These are houses that - Pt3(x,f= 0, v = 0, d =.2)] lose frontage, view, and.2 - d' miles of access. For group 2 the average loss is = (1/M3) [pi(x,f= 0, v = 0, d = d') $6,553/house. These are houses that lose - Pi(x,f= 0, view v and =.20, - d' miles. d = And,.2)] for group 3 for all t for the average group loss is $447/house. 3. These houses only lose.2 - d' miles of access. The set i = 1,.. We. use, these M, implicit is values houses to estimate the in sample that are less cost of displacement. than five years o have frontage and view. The sets i = Next, we predict the N,,, N2t, and N3, 1,..., M2 and i = 1,..., M3 are thefor t = 0,..., T-the number of displaced counterparts for groups 2 and 3. We use houses in each group for years t = 0 new houses because all displaced houses through T. To do this we use an estimate are new. Appendix Table 1 presents means of the total number of houses expected to and standard deviations for the housing be built in Anne Arundel County from 1985 characteristics in each group. to Then, we consider two scenarios In these equations, f = FRONTAGE, to predict the proportion of these houses v = VIEW, and d = DISTANCE in the that would have been built in the Critical hedonic. The estimated values are presented in Table 3; the calculations for the more than 50 percent of the coastal area in Area had there been no controls. Because Box-Cox regression are in the Appendix. In Anne Arundel County is undeveloped and the Box-Cox regression the average value in desirable locations for housing, we do

8 67(3) Parsons & Wu: Land-Use Controls 313 TABLE 3 AVERAGE VALUE OF LOST COASTAL ACCESS AMENITIES (Measured per House and in 1983 D Group 1 Group 2 Group 3 Houses losing Houses losing frontage, view view and Houses losing and (.2 - d') miles (.2 - d') miles (.2 - d') miles distance: distance: distance: p(x, f== 1, v 1, d= d') p(x, f= 0, v = 1, d = d') p(x, f= 0, v = 0, d = d') -p(x,f = 0, v = 0, d =.2) -p(x, f= 0, v = 0, d =.2) -p(x, f= 0, v = 0, d =.2) Regression Linear $74,763 $7,883 $233 Double-Log $82,883 $7,041 $524 Box-Cox $96,672 $6,553 $447 Note: The value of lost amenities for Gro houses less than 5 years old in Group i (= characteristics for each group in the samp not consider supply Under the 100% side Displacement const scenario ture development. 1,027 houses per year are displaced from Our 100% Displacement scenario assumes that 34 percent of all houses built in figure is only 211. After the turn of the cen to Under the CBF scenario that the county from 1985 to 2005 would have tury the number of annual displacements been built in what is now the Critical Area. under the 100 percent scenario is about 500 During the early 1980s, 34 percent of new houses per year compared to about 100 housing construction in the county went into what is now the Critical Area. Our Chesapeake Bay Foundation (CBF) scenario, based on a projection by that Foundation, assumes that 20 percent of all houses built in the county would have been built in what is now the Critical Area and In each scenario we assume that 20 percent of the houses are in group 1, 10 percent are in group 2, and 70 percent are in group 3. These are based on historic shares for houses per year under CBF. For the 100% Displacement scenario the present value (1983 dollars) of losses for an average year from 1986 to 1990 is $19.1 million. These drop to $5.9 million per year by the years 2000 to The decline is due to discounting and a declining rate of that with controls 65 percent of these would growth of housing construction. Under the still be built in that Area. This is a net displacement of 7 percent of all new houses. loss is $3.9 million in the earlier years drop- Chesapeake Bay Foundation scenario the Certain designated areas in the Critical ping to $1.2 million in the later years. The Area allow continued construction, and present value of the displacement cost of there is some grandfathering that allowsthe controls per county resident ranges controlled areas to have new construction. from $46/year in the early years to $14/year This scenario accounts for these aspects in of the later years in the 100% Displacement the controls. scenario, and from $9/year to $3/year per resident in the Chesapeake Bay Foundation scenario. IV. CONCLUSIONS these amenities for houses in the county. The expected growth in housing and ourwe have presented a method for estimating Ta-the value of lost coastal access ameni- displacement scenarios are presented in ble 4. Table 5 divides the displaced houses ties due to land-use controls. The estimates into groups of five-year increments. The are qualified by the assumptions of the are present value of lost coastal access amenities for an average year over the five-year and by current coastal access amenities be- affected by controls being small and open increments is presented in Table 6. ing similar to future access amenities. The

9 314 Land Economics August 1991 TABLE 4 HOUSING IN ANNE ARUNDEL COUNTY, Projected Number of Houses in Anne Arundel County:* Total 141, , , , ,400 Average Annual Growth Rate Over Preceding 5 Years 2.1% 1.6% 1.2% 0.8% Absolute Change Over Preceding 5 Years 15,100 12,200 9,800 7,300 Annual Average Change Over Preceding 5 Years 3,020 2,440 1,960 1,460 Average Annual Number of New Houses in Critical Area Over Preceding 5 Years: 100% Displacement Scenario Without controls - 1, With controls CBF Scenario Without controls With controls *Projections by Maryland Department estimates also of ignore this trade-off, a measure the of the offse opportunity costs of the controls. at inlan amenity substitutes displaced houses. The measure is one piece of the information required to assess the efficiency of con- These results have immediate use for policy. Coastal zoning authorities at state trols. Insofar as the measure is given per and local levels are showing increased house, interest in land-use controls as a policy for control schemes may also easily be com- the opportunity cost of alternative protecting coastal open space. Many types pared. We only need to know how the number of displaced houses varies between al- of controls are considered-usually density restrictions or moratoria on development. ternatives. An example might be comparing In designing such controls, communities alternatives with different density restrictions. This is a point often hotly debated in face the trade-off of lost coastal access amenities. Our estimate provides a measure designing controls and a point about which TABLE 5 AVERAGE ANNUAL NUMBER OF DISPLACED HOUSES IN FIVE-YEAR INCREMENTS Average Annual Ni, Years Group 1: Nit Group 2: N2, Group 3: N3, Total 100% Displacement Scenario: , Chesapeake Bay Foundation Scena

10 67(3) Parsons & Wu: Land-Use Controls 315 TABLE 6 PRESENT VALUE OF AVERAGE ANNUAL LOSSES: OVER FIVE-YEAR INCREMENTS IN 1983 DOLLARS (Millions) Scenario % Displacement Chesapeake Bay Foundation Note: Each entry in this table is computed using the Box ((DISTANCE + 1).26-1)/.26 Cox implicit values in Table 3 and the estimated number of displaced houses in Table 5. For - example,.005 ((DISTCBD the + average 1).26-1)/.26 annual loss of coastal access amenities from 1986 to 1990 (Annual +.06 FRONTAGE VIEW average of LOSS86s9o in equation [4]) under the 100% Displacement scenario is [($96,672 * 205.4) + ($6,553 * 102.7) + ($447 * 718.9)1/ (1 +.03)3 = $19.1 million. This entry is shown in the table above. The increment is discounted from 3 years in the future. The increment is from 8 years, is from 13 years, and is from 18 years. We use 3 percent as our real rate of discount. there is little information concerning cost or benefits. The measure may have other applications. It may be used to diffuse detractors that insist opportunity costs approach zero with little evidence or detractors that insist costs are inordinately large with little evidence. Such battle lines inevitably form when controls are being debated. The measure may also be used to estimate compensatory payments that may be designed for landowners holding coastal land that is designated restricted for development by the controls. It may even be used in the development of payments for a transferable development right scheme. We believe our findings provide economic information that may aid in rational policy-making and does so in an area of environmental control where such information seems to be unusually scarce. APPENDIX CALCULATION OF IMPLICIT PRICES OF COASTAL ACCESS AMENITIES USING THE Box-Cox REGRESSION The Box-Cox hedonic with X =.26 and 0 = -.17 is p = {-.17[ (BD.26-1)/ (BATH )/ DINED BASED ((AGE + 1).26 _ 1)/ HISTDUM +.01 GARAGE AIRCON +.01 FRPL (SF.26-1)/ (LOTSZ.26-1)/ (MONTH.26-1)/ (ED.26-1)/ ((%NWH + 1).26-1)/ (HHINC.26-1)/.26] + 1}/-.'7.[A1] We rewrite [Al] as p = {-.17[C +.06 FRONTAGE where VIEW ((DISTANCE + 1).26-1)/.26] + 1}1/-'17 [A2] C = (BD.26-1)/ (BATH.26-1)/ DINED BASED ((AGE + 1).26-1)/ HISTDUM +.01 GARAGE AIRCON +.01 FRPL (SF.26-1)/ (LOTSZ.26-1)/ (MONTH'26-1)/ ((DISTCBD + 1).26-1)/ (ED.26-1)/ ((%NWH + 1).26 _ 1)/ (HHINC.26-1)/.26. Substituting the appropriate values for DIS- TANCE + 1 (d), FRONTAGE (f), and VIEW (v), our measure of Ap for a house in each group is Ap1 = {-.17[C, +.06(1) (1) -.008(d.26-1)/.26] + 1}'/-.7 - {-.17[C, +.06(0) (0) -.008( )/.26] + 1}"/-'7

11 316 Land Economics August 1991 APPENDIX TABLE 1 MEAN VALUES OF VARIABLES IN THREE HOUSING GROUPS Means for Means for Means for Group 1 Houses Group 2 Houses Group 3 Houses BD BATH DINED BASED AGE HISTDUM GARAGE AIRCON FRPL SF 2,372 1,655 1,764 LOTSZ 26,552 12,863 15,450 MONTH DISTCBD ED %NWH HHINC 29,220 26,883 25,537 DISTANCE PRICE 225, ,047 99,937 of Economics and Statistics 59 (Aug.):272- AP2 = {-.17[C2 +.06(0) (1) (d.26-1)/.26] + 1}1/-.17 Chesapeake Bay Foundation "An Analysis of the Impacts + of.0096(0) Housing Availability - {-.17[C2 +.06(0) -.008( )/.26] and Local Economy." + l}1/-.17 Maryland Chesapeake Ap3 = {-.17[C3 +.06(0) Bay Critical Area + Program,.0096(0) January (d.26-1)/.26] Edwards, S., + and l}1/-'17 G. Anderson "Land Use Conflicts in the Coastal Zone: An Approach for the Analysis of the Opportunity - {-.17[C3 +.06(0) (0) -.008( )/.26] + 1}1/-.17 Costs of Protecting Coastal Resources." Journal of Northeastern Agricultural Economics (Apr.) where Api refers to group i. Ci is calculate the values of all other explanatory variab Maryland Department of State Planning the relevant house in the subsample for g "Projections for Maryland's Subdivisions." In each group our subsample is the set o In Development Planning Series '85. October. houses (less than five years old) that sold Means of the explanatory variables Milon, W. J., J. Gressel, and D. Mulkey subsample for each group is given in Ap "Hedonic Amenity Valuation and Functional Table 1. Form Specification." Land Economics 60 A Api is calculated for each house in the subsample for group i. The mean value of Api across (Nov.): Polinsky, A. M., and Steven Shavell the subsample for each group is given in Table 3 and used in our displacement estimates. References Brown, G. M., Jr., and H. O. Pollakowski "Economic Valuation of Shoreline." Review "Amenities and Property Values in a Model of an Urban Area." Journal of Public Economics 5: Shabman, L., and M. K. Bertelson "The Use of Development Value Estimates for Coastal Wetland Permit Decisions." Land Economics 55 (May):