1. Amenity Value of Proximity to National Wildlife Refuges Timothy Hamilton North Carolina State University Camp Resources XVIII

2. National Wildlife Refuges  502 NWRs in the lower 48 states  Operations = $362m, Maintenance = $140m  ~ 40 million visitors each year  Most open to public  70% have hiking/walking trails  60% have visitor facilities  50% open fishing and hunting  70% offer educational programs  Habitat conservation

3. Management Problem  NWR establishment removes land from the tax base.  FWS pays $1.65 per acre  By comparison, Cape May refuge in NJ. Average home value/acre implies $1,732 per acre in property tax revenue.

4. Management Problem  However, NWRs provide local communities benefits:  Ecological functions  Recreation benefits to larger community  Localized property value impacts  Proximity to NWR may increase value of nearby homes  increases tax base  Quantification of these benefits continues to be a key challenge for federal agencies

5. Hedonic Valuation  There are dozens of hedonic estimates of the capitalization value of proximity to open space:  Permanent vs. agricultural and/or developable open space  Boyle, Paterson and Poor (2002); Neumann, Boyle and Bell (2009)  Case-study of four NWRs (one near an UA, two more remote)  Obtain transaction data  Hedonic analysis of sales data to determine value of proximity to NWR  4.8% increase for homes in the urbanized area; significantly less for non-urbanized area.

6. Hedonic Valuation  Broader programmatic approach desired  Census Microdata offers an opportunity to consider broad geographic models  Davis (2011)  Proximity to power plants  Use census microdata across the U.S.  Rabindran and Timmins (2011)  Proximity to superfund site  Use both census and transaction data

7. Study Area  Begin with GIS of all NWR boundaries  Overlay with GIS of all urbanized areas (UA)  Contiguous, densely settled census blocks/groups that meet minimum population density requirements (1,000 people/sq.mile, 500 in surrounding blocks)  NWR boundary must be within two miles of the boundary of an UA (188)

8. 8/7/11

9. Study Area, continued  Final sample must also have:  NWR established in 1999 or earlier  Final Sample: 90 NWRs  Northeast: 34  Southeast: 20

10. Link NWRs to Census Blocks  Census geography is defined as:  Block: smallest geography available  Typically defined by geography  Population taken into consideration  Census data links household to blocks

11. 8/7/11

13. Proximity to an NWR  Linear distance between census block centroid and NWR boundary is measured  Each census survey is linked to the block in which the house is located

14. Housing Data  Confidential Census Data: decennial census long-form with block identifier  One-in-six sample of entire U.S.  Housing characteristics  Housing type, age of structure, total # rooms, bedrooms, heating type, parcel size ( 10 acres)  Housing value  Owner-report, 24 categories.

15. Basic Model  X i = housing/household characteristics  Z b = block characteristics  K g = neighborhood/block group characteristics  D b = distance of block centroid to NWR  Fixed effects by NWR and tracts

16. Model Covariates  Housing Characteristics  # of rooms and bedrooms, lot size, age of house  Block Characteristics: GIS  Proximity of a block to:  Nearest Interstate Highway  Nearest national or state park  Coastline  Urbanized area boundary  Centroid of the nearest MSA

17. Model Covariates  Other open space  2001 National Land Cover Database  Based on satellite imagery in 30-meter pixels  Calculate % of each census block in each land classification calculated.  29 land use categories aggregated to:  % open water  % developed open space  % developed low, medium or high density  % forest  % shrub/grassland

18. Model Covariates  Neighborhood/Block Group Characteristics  Population density  Median family size  Median number of children 18 and under  Median number of adults over 65  Median household income  % Owner occupied  % Vacant for seasonal use  % Single family detached  % Apartment

19. Impact of Distance to NWR  Continuous Distance  d(D b ) = β 1 D b + β 2 (D b ) 2  Discrete Intervals  d(Db) = β 1 I[0,.5] + β 2 I(.5,1] + β 3 I(1,1.5] + β 4 I(1.5,2] + β 5 I(2,2.5]  Omit dummy for block > 2.5 miles from NWR

20. NE Results 3 miles to NWR, 10 miles to UA 3 miles to NWR, 8 miles to UA 3 miles to NWR, 5 miles to UA Distance -0.0430*-0.0744***-0.141*** (0.0219)(0.0274)(0.0457) Distance^2 0.0132**0.0209***0.0360*** (0.00639)(0.00802)(0.0132) Observations14,55110,7754,660 R-Squared.66.63.68

21. NE Results: Categorical Distance 3 miles to NWR, 10 miles to UA 3 miles to NWR, 8 miles to UA 3 miles to NWR, 5 miles to UA ≤ 0.5 miles 0.02320.0365*0.0976*** (0.0186)(0.0214)(0.0364) 0.5 to 1.0 miles -0.0115-0.0203-0.00491 (0.0159)(0.0183)(0.0294) 1.0 to 1.5 miles -0.0185-0.0151-0.0110 (0.0144)(0.0169)(0.0285) 1.5 to 2.0 miles 0.00430-0.006970.00919 (0.0128)(0.0150)(0.0242) 2.0 to 2.5 miles -0.00986-0.0180-0.00562 (0.0105)(0.0125)(0.0193)

22. SE Results 3 miles to NWR, 10 miles to UA 3 miles to NWR, 8 miles to UA 3 miles to NWR, 5 miles to UA Distance -0.171***-0.186***-0.221*** (0.0248)(0.0287)(0.0449) Distance^2 0.0442***0.0518***0.0569*** (0.00692)(0.00793)(0.0124) Observations18,13413,9266,970 R-Squared.57.64.61

23. SE Results: Categorical Distance 3 miles to NWR, 10 miles to UA 3 miles to NWR, 8 miles to UA 3 miles to NWR, 5 miles to UA ≤ 0.5 miles 0.0874***0.0686***0.101*** (0.0206)(0.0239)(0.0350) 0.5 to 1.0 miles 0.0171-0.008620.0377 (0.0171)(0.0202)(0.0294) 1.0 to 1.5 miles -0.00851-0.0420**-0.0352 (0.0146)(0.0168)(0.0250) 1.5 to 2.0 miles -0.0193-0.0375***-0.0339 (0.0123)(0.0143)(0.0210) 2.0 to 2.5 miles -0.00501-0.0191-0.0231 (0.0103)(0.0120)(0.0181)

24. NE Results: Illustrative Marginal Effects  Assume value of $250,000  Categorical model (3 miles to NWR, 8 miles to UA) :  Adjacent houses are valued $9,125 higher than those 2.5 - 3.0 miles away.  Continuous Model (3 miles to NWR, 8 miles to UA) :  Moving house from 0.25 to 1.25 miles decreases value by $10,775

25. Summary  NWRs appear to provide amenity values as expected  Amenity impact is highly localized (≤0.5 miles). Census track or even block-group data may not be refined enough to identify impacts. Access to block-level data important.  Amenity value tends to be higher for homes closer to densely populated area

26. Policy Implications  Programmatic approach to benefits estimation  Transactions data not available and case-study approach can be difficult for agencies to apply broadly in their programs  Additional tax revenue from (particular) NWR  Calculate average capitalized value across all homes  Using average acres per home, find average increase in tax revenue per acre

27. Policy Implications: Total Capitalized Value for Homes within 8 miles  NE: Total Capitalized Benefits of $55 million across 9 NWRs  Annual tax revenue increase of $69 per acre  SE: Total Capitalized Benefits of $100 million across 21 NWRs  Annual tax revenue increase of $40 per acre

28. Thank You Questions? Comments?