- Email: [email protected]
The effect of conservation policy and varied open space on residential property values: A dynamic hedonic analysis
Land Use Policy 73 (2018) 480–487
Contents lists available at ScienceDirect
Land Use Policy journal homepage: www.elsevier.com/locate/landusepol
The effect of conservation policy and varied open space on residential property values: A dynamic hedonic analysis
T
Linda Fernandeza,⁎, Monobina Mukherjeeb, Thomas Scottc a
Virginia Commonwealth University, United States Moulton Niguel Water District, United States c University of California, Riverside, United States b
A R T I C L E I N F O
A B S T R A C T
Keywords: Open space Wild habitat Conservation Policy Hedonics Valuation Spatial
We investigate open space value in an ideal setting for a natural experiment between Riverside County, with an open space conservation policy, and neighboring San Bernardino County without the policy. With spatial econometrics, time series and spatial data, this study accounts for both spatial and temporal variation of open space values. The novelty of our paper is that we combine an investigation of the effect of open space proximity on residential property value with an analysis of the effect of endangered species habitat preservation policy distinguishing between types of open space (wild habitat for endangered species versus developed parks) in a two county study. We find that proximity to open space has a positive and statistically significant influence on increased value of residential real estate, with some distinction among type of open space between the counties. Conservation policy for open space with wild habitats contributes to increased value of this amenity in Riverside County.
1. Introduction Open space may be protected or unprotected, public or private land (USDA, 2006). Public and private open space is highly valued (Irwin, 2002; Thorsnes, 2002). Open space includes parks, stream and river corridors, forests, and other natural lands within urban and suburban areas. The rate of open space conversion to urban development doubled in the late 1990s from rates of earlier decades for some areas (USDA, 2006). From such conversion, we are losing 6,000 acres of open space each day across the United States, at a rate of 4 acres per minute (USDA, 2006). This means losing public goods and services such as food, fiber, recreation, natural hazard mitigation, and habitat for endangered species (USDA, 2006). In order to address the loss of open space, it is important to understand the value of open space. This study aims to estimate the value of open space in the context of a formal conservation policy to preserve open space for habitat. This study’s estimated values help to evaluate the conservation policies that have been implemented in Riverside County, California for addressing endangered species habitat loss and open space protection. In this study, open space consists of the lot size aside from the structure size on a residential land parcel, and alternatively parks, forests and wild land for endangered species habitat. The objective of our analysis is to investigate how residential real estate
⁎
value is affected by proximity to a habitat and open space policy. A comparison of pre and post policy residential real estate price variation in neighboring counties of our study with and without open space conservation policy leads to a rigorous analysis of the impacts of the conservation policies on residential property values, controlling for similar characteristics in the two counties. Our results are particularly policy relevant because they apply to open space preservation on the wild land-urban frontier. This frontier is often where ecological value of open space is high because there is room to maintain contiguous habitat. Contiguous refers to uninterrupted habitat that is not broken up by development amidst habitat acreage. Our research suggests that there are statistically significant benefits to residential real estate values in addition to these ecological benefits. There is a history of effort for conservation and preservation of open space areas by Riverside County that follows a trend elsewhere of establishing habitat conservation plans for open space. The Riverside County Integrated Plan (RCIP) is a comprehensive, three-part, integrated program initiated by the Riverside County Board of Supervisors in 1999 to determine future conservation, transportation, and housing and economic needs in Riverside County (Riverside County, 1999). Protection of the natural environment by conserving endangered species habitat on open space through a Multi-Species Habitat Conservation Plan (MSHCP) is central to the RCIP. The MSHCP
Corresponding author. E-mail address: [email protected] (L. Fernandez).
https://doi.org/10.1016/j.landusepol.2017.12.058 Received 5 December 2016; Received in revised form 21 December 2017; Accepted 22 December 2017 0264-8377/ © 2017 Elsevier Ltd. All rights reserved.
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
(2011). For a time horizon of three years in one city, Corona, CA, Yoo et al (2017) estimate value from developed urban park open space amidst commercial and industrial areas and residential real estate. Prior studies reviewed by McConnell and Walls (2005) did not have a spatial weight matrix to take into account spatial correlation in the dependent variable (i.e. house price in this case) and error terms. Past studies in the large literature review by McConnell and Walls (2005) include cross sectional hedonic pricing and contingent choice methods instead of the dynamic and spatial approach we use. Open space is clearly not a homogeneous good (Kroeger, 2008). We are able to distinguish between different types of open space where the distance change occurs to test the policy influence on the heterogeneous open space. With data of different residential real estate from 1996 to 2004, we conduct a spatial econometric analysis in this paper. The spatial weights in this analysis offer a useful framework for studying cross-sectional dependence with geographic distance changes (Anselin, 1988). The data used in our study include real estate sales information for residential properties in Riverside and San Bernardino Counties over a significant time horizon. These are two neighboring counties with similar characteristics except for the RCIP policy in Riverside County and not in San Bernardino County. The data in our analysis provides information useful for conducting a study over time and space, both of which together are not accounted for to a considerable extent in the past hedonic studies. For example, the data used in Brasington and Hite (2008) is only for one year in Ohio. Although their data covers the spatial dimension, it does not provide any information on the temporal dimension. Furthermore, in the study by Nordman and Wagner (2009), although they use a time series data for 1999–2005, they limit their study only to the town of Brookhaven in Suffolk County, New York. Therefore, the authors focus only on the temporal dimension but not on the spatial aspect of the hedonic study. Our study includes different zones across neighboring Riverside and San Bernardino counties sharing comparable demographic characteristics except for the RCIP policy, hence accounting for segmented markets even within counties. In this manner, we overcome the limitation that McConnell and Walls (2005) note about county level analyses in their hedonic pricing literature review. Estimates of open space values at a county level may lead to biased estimates as open space values differ by location and distance of open space from residential real estate (McConnell and Walls, 2005). Acharya and Bennett (2001) indicate the importance of the spatial scale to determine the value of open space variables. Our early view of small scale zones separately (Mukherjee and Fernandez, 2011) did not lead to a proper comparison across the entire two county dataset. The incomplete view lacks an evaluation with interaction terms to help gauge the change over time and space of residential real estate values between similar counties in all but habitat conservation policy occurring only in one. We analyze the influence of open space variables on residential property values before and after the RCIP policy is implemented in Riverside County versus San Bernardino County, without any conservation policy. In addition to the quasi-experimental setting, our analysis divides the two county study region into several zones. These zones are paired across counties based on similar population size and date of establishment of municipalities and census information. This pairing helps control for factors from the last sentence in residential real estate aside from open space policy in one county. As observed by Kuminoff et al., (2013), households "sort” across neighborhoods according to their wealth and their preferences for public goods, social characteristics, and commuting opportunities. The aggregation of these individual choices in markets and other institutions influences the supply of amenities and local public goods. Our policy analysis is an improvement over previous studies to address what Kuminoff et al. (2013) suggest in terms of providing a pseudo experiment of a control and treatment context with the conservation policy. Our analysis is at a finer scale compared to previous studies that have looked at open space
will provide habitat open space and protect watersheds and the environmental needs of the County. “Its (RCIP) objective is to retain and enhance the integrity of existing residential, employment, agricultural, and open space areas by protecting them from encroachment of land uses that would result in impacts from noise, noxious fumes, glare, shadowing, and traffic.″ (Riverside County, 2003). According to the Riverside County General Plan (2003), "The population of Riverside County and its cities is expected to double between the years 2000 and 2020, growing by approximately 1.4 million people. Efficient land use may have growth strategically located into existing developed areas, thus minimizing development pressures on rural, agricultural, and open space areas.″ The MSHCP addresses permanent opportunities for habitat through land use policy. The RCIP does strive for the balance of its three parts: open space conservation, transportation and housing with some limits to insure open space is not less of a priority that the other two parts (Riverside County, 1999). The RCIP designates permanent public land habitat through county acquisition leading to no threat of land conversion in the future. Our primary objective of this study is to analyze the RCIP policy and its impact on property values. This study’s analysis assigns Riverside County as the treatment group where the RCIP has been implemented, and San Bernardino County as the control group where the RCIP policy has not been implemented. This study also compares residential property values for Riverside County before and after the county’s involvement in the RCIP program. We use a hedonic model with spatial econometric techniques for the valuation of open space. These econometric models will be discussed further in the methodology section. 1.1. Literature review The novelty of our paper is that we combine an investigation of the effect of open space proximity on residential property value with an investigation of the effect of open space preservation policy with a distinction between types of open space (wild habitat for endangered species versus developed parks). We investigate the effect of a county level endangered species habitat preservation policy with a basic premise supported by the literature of open space that is nearer to residential property is worth more than further away. We are able to show the marginal negative effect of greater distance from residential property to wild open space is absolutely larger in both a county with a preservation policy and one without. Amidst the large literature on economic value of open space, few papers (for example, Cho et al., 2009) account for simultaneous changes in time and space with respect to open space and residential real estate values and others have not included a policy effect with their hedonic pricing investigation of value of open space. Cho et al. (2009) find the marginal effects of different measures of land amenities (lower housing density, greenways, parks and water bodies) on property value were statistically positive and significant. There is a statistically significant increase in the time periods they include (1989–1991 and 1999–2001). In their study, while the marginal effects of lot size and proximity to golf course were also positive and significant, they decreased from the first to the second time period. Brander and Koetse (2011) estimate the value of urban open space using meta-analyses of contingent valuation and hedonic pricing methods of existing literature including Anderson and West (2006), Geoghegan et al. (1997), Irwin (2002), Acharya and Bennett (2001), Poudyal et al (2009) who include a percentage of urban open space within a given buffer distance and Wu et al. (2004) who include explicit distance to open space. Brander and Koetse (2011) focus mainly on forests and urban parks for their open space variables. Increase in distance from open space has a negative and statistically significant impact on house prices. The time series data used in our paper help capture changes in value of open space over time for a longer continuous time horizon than Cho et al. (2009) for open space that supports endangered species habitat unlike the open space in Cho et al. (2009) and in Brander and Koetse 481
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
2. Material and methods
The sales information on residential real estate of Riverside and San Bernardino counties were obtained from Dataquick data. The Euclidean distance measurements were computed using GIS.
2.1. Data
2.2. Methodology
The applied econometric estimation utilizes approximately 148,000 residential real estate sales observations for Riverside and San Bernardino counties, the two largest counties in southeastern California. The Inland Empire has a population of about 3.8 million (Bluffstone et al., 2008). This region is useful to study due to the neighboring counties similar in many aspects except the RCIP policy in Riverside County. We number zones in the data in pairs in order to arrange a test zone and control zone between the two counties within each pair of zones that share three similar characteristics including population size, date of establishment of municipalities, size of the municipalities. This arrangement helps to compare the value of open space in a zone that has a conservation policy (RCIP) with its value in a zone without the policy, controlling for other relevant factors for statistical robustness. We collected those three characteristics from each municipality’s city planning department in the two counties. The characteristics and the zones help include fixed effects in the analysis. The history of urban development in both counties and the similarities they share in terms of demographics and timing of urban development of cities in both counties helps the process of establishing zones in our study region along with the data we collect from the cities and counties through an academic consultant for Riverside County’s acquisition of habitat preservation properties (Thomas Scott, personal communication). Zone 1 consists of Murrieta and Temecula of Riverside County. Zone 3 consists of Beaumont, Calimesa and Banning of Riverside County. Zone 5 consists of Moreno Valley, Riverside and Mira Loma of Riverside County. Zone 2 consists of the cities of Montclair, Rancho Cucamonga, Ontario, Fontana and Upland, which falls within San Bernardino County. Zone 4 consists of Yucaipa, Redlands and Loma Linda of San Bernardino County. Zone 6 consists of San Bernardino, Highland, Grand Terrace, Colton and Bloomington of San Bernardino County. The zones are depicted in Fig. 1. The pairing of zones (for example 1 and 2) is based on similarity between them across the two counties in terms of the average time of establishment of the municipalities, population size and municipality size. We are able to control for these factors through the zone fixed effects in our model. We investigate the temporal variation around the RCIP policy through regressions that compare 1996–99 and 2000–2004, prior and post RCIP implementation. The data contains residential property sales information for cities of Riverside and San Bernardino counties from 1996 to 2004, useful for conducting a study over time and space. The data provide information on all variables in the econometric regressions including: distance of the houses from forest and parks, residential lot size, structural area of house, age of house, percentage of wild habitat areas and percentage of urban density within 2 kilometers (km) of the residential real estate parcel. The data are also geo-referenced with latitude and longitude coordinates of each residential property that has been useful for our spatial analysis with digitized measure of distance from each residential property to each type of open space. The following sentences indicate variable names and definitions of these variables for the analysis. “Lot size” is the area of the residential lot aside from the residential structure itself measured in square feet. “Distance from park” is the Euclidean distance of the residential property to the nearest park in meters. “Distance from wild habitat” is the Euclidean distance of the residential property to the nearest wild habitat in meters. “Percentage wild habitat” is the percentage of density of wild habitat within 2km of a residential property. “Percentage urban” is the percentage of urban density within 2km of a residential property. “Age of the house” was calculated based on the years between the time it was built and sold.
With pooled data across two counties, our spatial econometric hedonic model helps estimate both policy and different open space value results. We analyzed three kinds of models - log-log Ordinary Least Square (OLS) regression model, log-log spatial autoregressive model (SAR) and log-log spatial error model (SEM). This paper presents results of the SEM model as it is the best fit to account for the spatial distribution of data observations and omitted variables within the study area depicted in Fig. 1. Since the SEM allows for uncertainty regarding an exact model to include independent variables such as the RCIP policy and there are no spatial lags of the dependent variable that corresponds with a SAR model, SEM is more appropriate for our analysis. Neglecting to account for error caused by spatial dependency could result in biased and inconsistent estimators. We do detect a significant spatial correlation in the error terms and hence, the fit of the model improves significantly with SEM model. The SEM provides more robust estimates compared to OLS that assumes data observations are independent of space or SAR.1 The R square of the spatial autoregressive model turned out to be much lower (∼0.4) compared to the R square (∼0.7) of the Spatial Error Model. The estimates in the SEM model had a higher statistical significance than the SAR model and standard error of estimates in the SEM model were lower than the standard error of estimates in the SAR model. Based on these goodness of fit statistics we chose SEM model over the SAR and the OLS regression models for our analysis. Our regression equation in logarithmic form accommodates different units of measure of variables, enabling us to estimate the coefficients as elasticities. The logarithmic form enables comparison in terms of relative size of estimated coefficients. In general, log transformation has improved fit over the raw scale (Hao and Naiman, 2007). The log-log coefficients are typically interpreted as elasticities with a 1% increase in the independent variables (Wooldridge, 2006).
values at a county level such as Doss and Taff (1996).
2.3. Spatial econometric model specifications In the spatial analysis performed in this paper, the weight matrix consists of the inverse of the Euclidean distance calculated based on the geo-reference latitude and longitude coordinates of the residential property and open space variables [in Eq. (i)]. Thus, the weight matrix assigns higher weights to the property that are situated closer to each other compared to those that are situated at a farther distance from each other. A Matlab program with the parallel processing toolbox has been used in this analysis to construct the spatial weight matrix faster by dividing the task into different processors of the computer. A spatial weights matrix provides the “structure” of assumed spatial relationships.
⎡ 0 1/ d12 1/ d13 ⎤ W = ⎢1/ d21 0 1/ d23 ⎥ ⎢1/ d 1/ d 0 ⎥ 32 ⎦ ⎣ 31
(i)
In Eq. (i) dijindicates the distance between residential property in location i and location j. The diagonal terms in the matrix is zero as there is no weight for location i on location i. The weight matrix assigns higher weights to residential property situated closer to each other compared to residential property, which are situated farther away from each other. The spatial weights matrix is row standardized. The specification of the Spatial Error Model (SEM) model is 1 The results produced by the log-log OLS model and log-log SAR are not reported in this paper but can be made available upon request.
482
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
Fig. 1. Study Region Map.
wild*year)it+ β10 ln(distance from park*year)it + β11 (year dummies) + β12 (zone*year) + β13 (countydummy*year) + ε.
provided in Eq. (ii). In Eq. (ii) W is the spatial weight matrix from Eq. (i) and the parameter λ ρ is the coefficient for error correlation. The parameters βi indicate the statistical significance of the independent variable’s influence on the dependent variable ln(salevalue ) . The real estate sale value on left hand side of the Eq. (ii) is the real estate sale price of the residential property sold at time period t. The right hand side variables in Eq. (ii) constitutes of lot size (lotsize) of residential property, structural area (structure)of a house, Euclidean distance of a residential property from the nearest park (distance from park), Euclidean distance of a residential property from the nearest wild habitat area (distance from wild), percentage of urban density (percentage urban) and density of wild habitat (percentage wild) within 2km of a house, age of the house (age) when it was sold, year fixed effects (year dummies) and interaction of year and distance from park (distance from park*year) and interaction of year and distance from wild (distance from wild*year) and interaction of county with year dummies (countydummy*year) and interaction of zone with year dummies(zone*year) ε is a vector of error terms. The Spatial Error Model (SEM) used in this study is:
ε = λWε + μ
μ ∼ (0, σ 2In )
(ii) (iii)
The data are pooled for the regression of the SEM with relevant dummy variables for each zone assigned and also time dummy variables help in our comparative investigation. 3. Results The results in Table 2 from all observations pooled across two counties indicate the positive and statistically significant influence from onsite characteristics of house structure size and lot size on residential real estate sale price and a negative and statistically significant influence of house age on residential property sale price. The abbreviation of variables in Table 2 appears in parentheses in this section next to their complete name. The negative and statistically significant coefficient on distance from wild (DW) and distance from parks (DP) indicates a reduction in residential real estate sale price overall as distance to both types of open space increases. We find that distance to wild habitat area dominates in magnitude as the decrease in residential real estate sale price is higher due to a marginal increase in distance from wild habitat
ln (salevalue)it = β0 ln(structure)it + β1 ln(lotsize)it + β2 ln(distance from park)it+ β3 ln(distance from wild)it+ β4 ln(age)it+ β5 ln(percentage wild)it+ β6 ln(percentage urban)it+ β7 ln(distance from wild*zone)it+ β8 ln(distance from park*zone)it+ β9 ln(distance from 483
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
area compared to a marginal increase in distance from park. The positive and statistically significant coefficient on percentage wild (Pwild02) indicates that a homebuyer’s marginal value increases from more open space within a one mile circumference of the residential property. One mile is a standard distance for real estate market appraisers generating comparable values for residential properties. The negative and statistically significant coefficient on percentage urban (Purban02) indicates a reduction in residential real estate sale value from an increase in urban density within the one mile circumference. The interaction between distance from wild and each zone (DW*Zone in Table 2) shows a negative impact on residential real estate values in Zones 1, 3 and 5 with a marginal increase in distance to wild in Riverside County. The highest marginal value for proximity to wild habitat areas are observed in Zone 3 as a marginal increase in distance from wilds has the highest negative impact on residential real estate sale values in this zone. Zone 2 and Zone 6 also have a negative and statistically significant impact on home values with an increase in distance from wild. The negative impact on Zone 6 of San Bernardino County residential real estate values with an increase in distance from wild is less than Zone 2 in that same county and less than Zones 3 and 5 of Riverside County. The statistically significant negative impact on residential real estate values with a marginal increase in distance from wild indicates the RCIP influence in Riverside County for a complementary relationship between housing and open space in terms of increased residential real estate sale value through the RCIP policy’s attempt to balance housing and open space objectives. The interaction terms of distance from parks and zones (DP*Zone in Table 2) shows a marginal increase in distance from park has the highest negative impact on residential real estate values in Zone 6 (San Bernardino County) and the second highest negative impact on residential real estate values in Zone 1 (Riverside County). City parks have not been designated in San Bernardino County for over 20 years (Bluffstone et al. 2008). The average distance from residences to parks in San Bernardino County is high relative to Riverside County as shown in the county comparison of Fig. 2. The scarcity of parks in Zone 6 of San Bernardino County can contribute to the high amenity value generated from the regression involving the interaction terms. Perhaps, there is a spillover effect due to the RCIP policy in Riverside County. The third highest negative impact on residential real estate sale value with an increase in distance from parks is observed in Zone 3 of Riverside County. Figs. 2 and 3 depict the differences in wild and park open space distances between the counties before and after the RCIP policy aside from the summary statistics in Tables 1a–1c. Note, San Bernardino County has permanent national forest acreage in close proximity included in Fig. 3. With pooled observations across both counties, the interaction of distance wild with year (DW*Year in Table 2) has a higher marginal impact on residential property sale values in 2000-04 in the post-policy period. The interaction between distance from parks (DP) and years
Fig. 3. Average Distance of Residential Real Estate from Wild Habitat Area.
(DP*Year) also show a larger negative impact on residential real estate prices during 2000-04, post-policy period. Both of these results imply that the marginal value for open space increases after the RCIP policy is implemented. The real estate market upswing in residential real estate sale values during the time period for the analysis is recorded through the positive and statistically significant coefficients increasing for each consecutive year dummy variable. The zone fixed effects in our model also help to control for three factors including average date of establishment of municipality, population size and municipality size. The coefficients of this model are robust as verified by running heteroskedastic SEM regressions. The p value of a couple variables change by rounding digits and the statistical significance remains the same with the majority of coefficients significant at 1%. To explore the timing of the RCIP policy, the interaction between zones and years before and after the policy is announced yields results as follows. Zones 1, 3 and 5 have a positive and statistically significant effect at a higher magnitude for the later years of the time series versus the earlier time period whereas the opposite is the case for Zones 2 and 6 in San Bernardino County. This is consistent with Riddel (2001), which finds that the economic impact of an open space purchase takes some time to be fully realized. We conduct a three variable interaction of county, time (pre/post policy) and distance wild variable to identify the impact (if any) brought about by the RCIP policy on the marginal value of open space in Riverside and San Bernardino counties. A 1% increase in distance from wild leads to a 0.07% drop in residential real estate prices during 1996–99 and 0.20% drop in residential real estate prices in 2000–04. These coefficients indicate a higher marginal value for open space in the post-policy period in Riverside County. A 1% increase in distance from wild leads to a 0.04% drop in residential real estate prices during 1996–99 and 0.11% drop in real estate prices during 2000–04 in San Bernardino County. The increase in marginal value of wild habitat open space in San Bernardino County (where policy is not implemented) captures the spillover effect of the neighboring county (Riverside County) to some extent. Our results indicate that the residents in San Bernardino County also values open space more after an open space policy is implemented in the neighboring county (Riverside County). These variables are strong variables of interest in our study which helps to highlight that an open space policy can increase the marginal value of open space not only in the county it is implemented but also in the neighboring county where it's not implemented.
4. Discussion Of impact in monetary values 4.1. Impact on residential property Sale values by Zones Recall the cities that constitute each of these zones used in our analysis. Zone 1 consists of Murrieta and Temecula of Riverside County. Zone 3 consists of Beaumont, Calimesa and Banning of Riverside
Fig. 2. Average Distance of Residential Real Estate from Parks.
484
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
Table 1a Average of variables for Zone 1- Zone 6 from 1996 to 2004.
Zone1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6
AGE (years)
SQFT_STRU (sq ft)
LOT_SQFT (sq ft)
SALE_VALUE ($)
DIST_WILDS (meters)
DIST_PARKS (meters)
P_URBAN02 (%)
P_WILD02 (%)
7.36 25.08 30.00 31.85 27.25 40.28
2,121.88 1,649.64 1,398.13 1,741.74 1,613.39 1,407.77
21,370.95 9,591.70 22,251.16 15,046.65 12,442.78 10,838.33
248,191.55 195,576.27 133,734.88 196,097.76 166,253.85 128,010.24
2,649.09 1,702.13 2,064.97 900.65 2,319.80 1,544.46
820.14 1,976.01 709.47 1,614.69 1,796.12 4,274.88
58 70 49 63 73 61
1 1 2 1 4 4
Table 1b Average of variables for Zone 1- Zone 6 from 1996 to 99.
Zone Zone Zone Zone Zone Zone
1 2 3 4 5 6
AGE (years)
SQFT_STRU (sq ft)
LOT_SQFT (sq ft)
SALE_VALUE ($)
DIST_WILDS (meters)
DIST_PARKS (meters)
P_URBAN02 (%)
P_WILD02 (%)
10.07 28.80 33.00 35.10 30.62 40.85
2,075.61 1,600.38 1,366.13 1,659.65 1,584.93 1,363.17
20,540.27 9,448.78 18,416.22 15,646.15 12,271.54 10,749.32
309,762.48 241,554.98 156,133.30 229,866.74 209,232.93 156,253.69
2,599.94 1,739.76 2,085.78 957.36 2,299.27 1,577.19
821.01 1,987.24 729.18 2,008.97 1,747.57 4,496.68
57 73 50 63 74 61
1 1 2 1 4 4
Table 1c Average of variables for Zone 1 - Zone 6 from 2000 to 04.
Zone Zone Zone Zone Zone Zone
1 2 3 4 5 6
AGE (years)
SQFT_STRU (sq ft)
LOT_SQFT (sq ft)
SALE_VALUE ($)
DIST_WILDS (meters)
DIST_PARKS (meters)
P_URBAN02 (%)
P_WILD02 (%)
4.65 21.36 27.00 28.59 23.88 39.71
2,168.15 1,698.89 1,430.14 1,823.83 1,641.85 1,452.37
22,201.64 9,734.62 26,086.10 14,447.16 12,614.01 10,927.34
186,620.63 149,597.55 111,336.47 162,328.77 123,274.77 99,766.78
2,698.23 1,664.50 2,044.15 843.94 2,340.32 1,511.73
819.27 1,964.78 689.76 1,220.41 1,844.67 4,053.07
58 72 48 62 73 60
1 1 2 1 4 4
Zone 1, 3 and 5 respectively with a 1mile increase in distance from parks. Whereas there is a $15,166 and $7500 decrease in residential real estate sale prices in Zone 2 and Zone 6 with a 1mile increase in distance from parks in San Bernardino County. Recall, Fig. 2 depicts the shortage of parks in San Bernardino County. The results show the amenity value is capitalized into formal real estate value within the one mile radius range of standard real estate appraisal required by underwriters for financial transactions in real estate requiring comparable values for land appraisal occur within a mile of the property in question. Thus not only is the RCIP statistically significant for capitalizing permanent endangered species habitat into residential real estate values, but also the policy impact can be directly tied to the formal real estate market transactions guidelines within a mile radius for comparable values (Section B4-1.3-08: Comparable Sales (01/31/2017) Selling Guide and FreddieMac (www.freddiemac.com/hve/hve.html).
County. Zone 5 consists of Moreno Valley, Riverside and Mira Loma of Riverside County. Zone 2 consists of the cities of Montclair, Rancho Cucamonga, Ontario, Fontana and Upland, which falls within San Bernardino County. Zone 4 consists of Yucaipa, Redlands and Loma Linda of San Bernardino County. Zone 6 consists of San Bernardino, Highland, Grand Terrace, Colton and Bloomington of San Bernardino County. In order to avoid multicollinearity, we drop Zone 4 from the regression, considering it as the default zone. Zone 1 contains the key initial areas of focus for open space preservation and Zones 5 and 6 contain the largest cities and are the county seats for Riverside and San Bernardino, respectively. Table 3 shows the impact for each of the independent variables on residential real estate sale prices in dollar values.2,3 We find that with a 1mile increase in distance from park there is a $23,100 decrease in residential real estate sale values whereas with a 1mile increase from distance from wild habitat areas there is a $43,333 decrease in residential real estate sale values. In Riverside County, a 1mile increase in distance from wilds yields an approximately $35,000 decrease in residential real estate sale values in Zone 1 and $36,666 decrease in Zone 3 and $33,333 decrease in Zone 5. On the other hand, in San Bernardino County a 1mile increase in distance from wilds leads to a $61,666 decrease in residential real estate sale values in Zone 2 and a $36,666 decrease in real estate sale values in Zone 6. There is a $70,000, $41,666 and $14,333 decrease in residential real estate sale prices in
4.2. Impact on residential property values by years During the years 1996-99, there is a $626 decrease in real estate sale values in Riverside whereas a $36,666 decrease in 2000–04 in real estate values in Riverside with a one-mile increase in distance from wild. The real estate values decreases by $330 during 1996–99 and $20,000 in 2000–04 with a one mile increase in distance from wilds in San Bernardino. Although we do not estimate the impact of distance from parks in monetary values separately for Riverside and San Bernardino counties, there is $2313 decrease during 1996–99 and $8333 decrease in the real estate values during 2000–04 with a one mile increase in distance from parks, overall.
2 The impact on residential real estate sale values in dollars is estimated based on the coefficients (elasticities) in Table 1 and the average real estate sale price of residential properties in each zones. 3 As the units for distance used in our analysis are in meters, we use this unit for estimating the marginal impact. However, we use these estimates in further calculation of the impact on home values due to a 1 mile change in distance, as this is a more conventional unit used for real estate transactions. Our discussion of monetary values is also based on the 1mile change in distance. We report the estimates both for a 1 mile change in distance in Table 3.
5. Summary and conclusions Our results indicate a high value for proximity to wild habitat areas to residential real estate in the zones of Riverside County in the post485
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
Table 2 SEM results with all variables and interaction term.4 Variables
Coefficients
Variables
Coefficients
Variables
Coefficients
Structure Lot Distpark Distwild Age Pwild02 Purban02 DWZone1 DWZone2 DWZone3 DWZone5 DWZone6 DPZone1 DPZone2 DPZone3 DPZone5 DPZone6 DP(96-99) DP(2000-04)
0.985*** (0.000) 0.167*** (0.000) −0.118*** (0.000) −0.261*** (0.000) −0.001*** (0.000) 0.013*** (0.000) −0.101*** (0.000) −0.227*** (0.000) −0.321*** (0.000) −0.342*** (0.000) −0.276*** (0.000) −0.258*** (0.000) −0.139*** (0.000) −0.093*** (0.000) −0.132*** (0.000) −0.093*** (0.000) −0.149*** (0.000) −0.016 (0.169) −0.025*** (0.000)
DW(96–99) DW(2000–04) Year1997 Year1998 Year1999 Year2000 Year2001 Year2002 Year2003 Year2004 Riv*(96–99) Riv*(2000–04) SB*(96–99) SB*(2000–04) Riv*(96–99)*DW Riv*(2000–04)*DW SB*(96–99)*DW SB*(2000–04)*DW Zone 1
−0.046*** (0.000) −0.076*** (0.023) 0.217*** (0.000) 0.286***$ (0.000) 0.366*** (0.000) 0.718*** (0.000) 0.845*** (0.000) 0.985*** (0.000) 1.172*** (0.000) 1.441*** (0.000) 0.062*** (0.000) 0.191*** (0.000) 0.037*** (0.000) 0.038*** (0.000) −0.071*** (0.000) −0.202*** (0.000) −0.041*** (0.000) −0.106*** (0.000) 2.445 (0.000)
Zone 2 Zone 3 Zone 5 Zone 6 Zone1*(96-99) Zone1*(2000-04) Zone2*(96-99) Zone2*(2000-04) Zone3*(96-99) Zone3 *(2000-04) Zone5*(96-99) Zone5*(2000-04) Zone6*(96-99) Zone6*(2000-04) Lambda Rsquare Adjusted R square
2.938 (0.000) 2.935 (0.000) 2.443 (0.000) 2.634 (0.000) 0.134*** (0.006) 0.147*** (0.002) 0.209*** (0.000) 0.023** (0.000) 0.063* (0.105) 0.177*** (0.000) 0.016*** (0.014) 0.033** (0.030) 0.123*** (0.000) 0.023*** (0.002) 0.495*** (0.000) 0.750 0.749
value due to a marginal increase in distance to wilds is approximately half the value of Riverside County for each of the time periods (before and after the RCIP policy is implemented). The significant variation of the values of open space across different zones within a county provides evidence of the existence of a segmented market even at the county level. With more accurate and efficient estimates of value of open space and other variables, the spatial error hedonic model can be used in the decision-making process associated with habitat conservation policy and urban land-use planning. While Mahmoudi et al. (2012) note a reconfiguration of a park matters for amenity values associated with urban property and Tyravainen (1997) and Poudyal et al (2009) note larger, nearby parks increase housing values, our analysis econometrically estimates the impact on residential property value of a county’s policy to preserve endangered species habitat. The policy context of our study is that the habitat is for more than one endangered species formally listed in federal policy. Since the initial focus on the Stephens’ Kangaroo Rat, now 142 species exist in the habitat preserved through the RCIP policy. The improvement in residential real estate values is a complementary benefit to the endangered species goal of the county policy. We find that amenity values vary spatially similar to Cho et al (2008) for other types of open space. Overall, the results from this study highlight the importance of conservation policy in influencing value of habitat areas for multiple endangered species. Furthermore, our pre-post policy analysis in both counties for distance to wild habitat shows how a conservation policy in a particular county can influence the residential real estate sale values not only within that county but also residential real estate sale values in the neighboring county. The amenity values generated in this study for different zones within counties can help in estimating the benefit of conservation of habitat for endangered species, which can be used as a tool by policy makers to set the conservation fees, e.g., development impact fees that help finance conservation. Additionally, this study can prove significant for land-use planning and conservation decisions, not only in the Inland Empire region but also for any other region with similar geographical characteristics and residential markets.
Table 3 Monetary Values. Variables
Impact ($) with 1 mile increase
Variables
Impact ($) with 1 mile increase
Distpark Distwild DWZone1 DWZone2 DWZone3 DWZone5 DWZone6 DPZone1 DPZone2 DPZone3 DPZone5 DPZone6 DW*(96-99) DW*(2000-04) DP*(96-99) DP*(2000-04)
−23,100 −43,333 −35,000 −61,666 −36,666 −33,333 −36,666 −70,000 −15,166 −41,666 −14,333 −7,500 −5150 −20,000 −2,313 −8,333
Riv*DW*(96-99) Riv*DW*(2000-04) SB*DW*(96-99) SB*DW*(2000-04)
−626 −36,666 −330 −20,000
policy period. The results indicate a complimentary impact on appreciated residential real estate of a policy initiated for conservation of wild habitat areas in Riverside County during 1999. We also observe a higher value for proximity to wild habitat areas to residential real estate in San Bernardino County in the post-policy period which can be a potential spillover effect from the neighboring county. We also observe a high amenity value for parks in Zone 6 of San Bernardino County and Zone 1 of Riverside County. Evidence from previous literature (Bluffstone et al. 2008), shows a scarcity of parks in San Bernardino County that can influence its high amenity value in certain zones of this county. Also, it can be a spillover effect due to the RCIP policy in the neighboring county We observe a higher value for proximity to parks in Riverside County during 2000–04. Again, we observe that the post policy years of (2000–04) lead to a higher increase in residential real estate sale values in the zones of Riverside County compared to pre-policy years of (1996–99). This is as expected, as the announcement of the conservation policy could have played a major role in driving the residential real estate sale prices higher in these regions of Riverside County. For zones of San Bernardino County, the resulting change in residential real estate
References Acharya, G., Bennett, L., 2001. Valuing open space and land-use patterns in urban watersheds. J. Real Estate Finance Econ. 22, 221–237. Anderson, S., West, S., 2006. Open space, residential property values and spatial context. Reg. Sci. Urban Econ. 36, 773–789. Anselin, L., 1988. Spatial Econometrics: Methods and Models. Kluwer Academic Publishers, Dordrecht; Boston.
4 DW: Distance from wild; DP: Distance from park, PW: Percentage wild habitat area; PU: percentage urban area.
486
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
nonmarket benefits. Resour. Future. Mukherjee, M., Fernandez, L., 2011. Analysis of the Influence of Open Space on Residential Values, Giannini Foundation Grant Report for Agricultural and Resource Economics Update, 15-1, September 2011. Nordman, E., Wagner, J., 2009. Public Purchases and Private Preferences: A Hedonic Model of Open Space Acquisitions, Working Paper. 27 p. Available at:. Grand Valley State University, Allendale, MI accessed 19 May 2009. http://works.bepress.com/ erik_nordman/1/. Poudyal, N., Hodges, D., Ronn, B., Cho, S., 2009. Valuing diversity and spatial pattern of open space plots in urban neighborhoods. For. Policy Econ. 11, 194–201 Riverside County Integrated Project http://www.rcip.org/. Riddel, M., 2001. A dynamic approach to estimating hedonic prices for environmental goods: an application to open space purchase. Land Econ. 77 (4), 494–512. Riverside County, 1999. 2003, Riverside County Integrated Plan. http://www.rcip.org. Thorsnes, P., 2002. The value of a suburban forest preserve: estimates from sales of vacant residential building lots. Land Econ. 78 (3), 426–441. Tyravainen, L., 1997. The amenity value of the urban forest: an application of the hedonic pricing method. Landscape Urban Plann. 37 (3), 211–222. USDA, 2006. Society’s Choices: Land Use Changes, Forest Fragmentation and Conservation, Pacific Northwest Research Station Science Findings Newsletter, Issue 88, November. Wooldridge, J., 2006. Introductory Econometrics: A Modern Approach, third edition. Thomson South-Western. Wu, J., Adams, R., Plantinga, A., 2004. Amenities in an urban equilibrium model: residential development in Portland, OR. Land Econ. 80, 19–32. Yoo, J., Browning, J., Minesing, K., 2017. Open space premium near commercial zones- a case study in the City of corona, CA. Int. J. Urban Sci online access 11/26/17.
Bluffstone, R., Braman, M., Fernandez, L., Scott, T., Lee, P., 2008. Housing, sprawl, and the use of development impact fees: the case Of Inland Empire. Contemp. Econ. Policy 26 (3), 433–447. Brander, D., Koetse, M., 2011. The value of open space: meta analysis of contingent valuation and hedonic pricing results. J. Environ. Manage. 92 (10), 2763–2773. Brasington, D., Hite, D., 2008. A mixed index approach to identifying hedonic price models. Reg. Sci. Urban Econ. 38 (3), 271–284. Cho, S.C., Clark, W., Park, S.Kim, 2009. Spatial and temporal variation in the housing market values of lot size and open space. Land Econ. 85 (1), 51–73. Cho, S.-H., Poudyal, N.C., Roberts, R.K., 2008. Spatial analysis of the amenity value of green open space. Ecol. Econ. 66, 403–416. Doss, C., Taff, S., 1996. The influence of wetland type and wetland proximity on residential property values. J. Agric. Resour. Econ. 21 (1), 120–129. Geoghegan, J., Wainger, L., Bockstael, N., 1997. Spatial landscape indices in a hedonic framework: an ecological economics analysis using GIS. Ecol. Econ. 23, 251–264. Hao, L., Naiman, D., 2007. Quantile Regression. Quantitative Applications in the Social Sciences, vol. 149 Sage Publications. Irwin, E., 2002. The effects of open space on residential property values. Land. Economics 78 (4), 465–480. Kroeger, Tim, 2008. Open space property value premium analysis. National Council for Science and the Environment 2006 Wildlife Habitat Policy Research Program. Kuminoff, N.V., Smith, K.V., Timmins, C., 2013. The new economics of equilibrium sorting and policy evaluation using housing markets. J. Econ. Lit. 51 (4), 1007–1062. Mahmoudi, P., MacDonald, D., Crossman, N., Summers, D., van der Hoek, J., 2012. Space matters: the importance of amenity in planning metropolitan growth. Aust. J. Agric. Econ. 57, 38–59. McConnell, V., Walls, M., 2005. The value of open space: evidence from studies Of
487
Contents lists available at ScienceDirect
Land Use Policy journal homepage: www.elsevier.com/locate/landusepol
The effect of conservation policy and varied open space on residential property values: A dynamic hedonic analysis
T
Linda Fernandeza,⁎, Monobina Mukherjeeb, Thomas Scottc a
Virginia Commonwealth University, United States Moulton Niguel Water District, United States c University of California, Riverside, United States b
A R T I C L E I N F O
A B S T R A C T
Keywords: Open space Wild habitat Conservation Policy Hedonics Valuation Spatial
We investigate open space value in an ideal setting for a natural experiment between Riverside County, with an open space conservation policy, and neighboring San Bernardino County without the policy. With spatial econometrics, time series and spatial data, this study accounts for both spatial and temporal variation of open space values. The novelty of our paper is that we combine an investigation of the effect of open space proximity on residential property value with an analysis of the effect of endangered species habitat preservation policy distinguishing between types of open space (wild habitat for endangered species versus developed parks) in a two county study. We find that proximity to open space has a positive and statistically significant influence on increased value of residential real estate, with some distinction among type of open space between the counties. Conservation policy for open space with wild habitats contributes to increased value of this amenity in Riverside County.
1. Introduction Open space may be protected or unprotected, public or private land (USDA, 2006). Public and private open space is highly valued (Irwin, 2002; Thorsnes, 2002). Open space includes parks, stream and river corridors, forests, and other natural lands within urban and suburban areas. The rate of open space conversion to urban development doubled in the late 1990s from rates of earlier decades for some areas (USDA, 2006). From such conversion, we are losing 6,000 acres of open space each day across the United States, at a rate of 4 acres per minute (USDA, 2006). This means losing public goods and services such as food, fiber, recreation, natural hazard mitigation, and habitat for endangered species (USDA, 2006). In order to address the loss of open space, it is important to understand the value of open space. This study aims to estimate the value of open space in the context of a formal conservation policy to preserve open space for habitat. This study’s estimated values help to evaluate the conservation policies that have been implemented in Riverside County, California for addressing endangered species habitat loss and open space protection. In this study, open space consists of the lot size aside from the structure size on a residential land parcel, and alternatively parks, forests and wild land for endangered species habitat. The objective of our analysis is to investigate how residential real estate
⁎
value is affected by proximity to a habitat and open space policy. A comparison of pre and post policy residential real estate price variation in neighboring counties of our study with and without open space conservation policy leads to a rigorous analysis of the impacts of the conservation policies on residential property values, controlling for similar characteristics in the two counties. Our results are particularly policy relevant because they apply to open space preservation on the wild land-urban frontier. This frontier is often where ecological value of open space is high because there is room to maintain contiguous habitat. Contiguous refers to uninterrupted habitat that is not broken up by development amidst habitat acreage. Our research suggests that there are statistically significant benefits to residential real estate values in addition to these ecological benefits. There is a history of effort for conservation and preservation of open space areas by Riverside County that follows a trend elsewhere of establishing habitat conservation plans for open space. The Riverside County Integrated Plan (RCIP) is a comprehensive, three-part, integrated program initiated by the Riverside County Board of Supervisors in 1999 to determine future conservation, transportation, and housing and economic needs in Riverside County (Riverside County, 1999). Protection of the natural environment by conserving endangered species habitat on open space through a Multi-Species Habitat Conservation Plan (MSHCP) is central to the RCIP. The MSHCP
Corresponding author. E-mail address: [email protected] (L. Fernandez).
https://doi.org/10.1016/j.landusepol.2017.12.058 Received 5 December 2016; Received in revised form 21 December 2017; Accepted 22 December 2017 0264-8377/ © 2017 Elsevier Ltd. All rights reserved.
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
(2011). For a time horizon of three years in one city, Corona, CA, Yoo et al (2017) estimate value from developed urban park open space amidst commercial and industrial areas and residential real estate. Prior studies reviewed by McConnell and Walls (2005) did not have a spatial weight matrix to take into account spatial correlation in the dependent variable (i.e. house price in this case) and error terms. Past studies in the large literature review by McConnell and Walls (2005) include cross sectional hedonic pricing and contingent choice methods instead of the dynamic and spatial approach we use. Open space is clearly not a homogeneous good (Kroeger, 2008). We are able to distinguish between different types of open space where the distance change occurs to test the policy influence on the heterogeneous open space. With data of different residential real estate from 1996 to 2004, we conduct a spatial econometric analysis in this paper. The spatial weights in this analysis offer a useful framework for studying cross-sectional dependence with geographic distance changes (Anselin, 1988). The data used in our study include real estate sales information for residential properties in Riverside and San Bernardino Counties over a significant time horizon. These are two neighboring counties with similar characteristics except for the RCIP policy in Riverside County and not in San Bernardino County. The data in our analysis provides information useful for conducting a study over time and space, both of which together are not accounted for to a considerable extent in the past hedonic studies. For example, the data used in Brasington and Hite (2008) is only for one year in Ohio. Although their data covers the spatial dimension, it does not provide any information on the temporal dimension. Furthermore, in the study by Nordman and Wagner (2009), although they use a time series data for 1999–2005, they limit their study only to the town of Brookhaven in Suffolk County, New York. Therefore, the authors focus only on the temporal dimension but not on the spatial aspect of the hedonic study. Our study includes different zones across neighboring Riverside and San Bernardino counties sharing comparable demographic characteristics except for the RCIP policy, hence accounting for segmented markets even within counties. In this manner, we overcome the limitation that McConnell and Walls (2005) note about county level analyses in their hedonic pricing literature review. Estimates of open space values at a county level may lead to biased estimates as open space values differ by location and distance of open space from residential real estate (McConnell and Walls, 2005). Acharya and Bennett (2001) indicate the importance of the spatial scale to determine the value of open space variables. Our early view of small scale zones separately (Mukherjee and Fernandez, 2011) did not lead to a proper comparison across the entire two county dataset. The incomplete view lacks an evaluation with interaction terms to help gauge the change over time and space of residential real estate values between similar counties in all but habitat conservation policy occurring only in one. We analyze the influence of open space variables on residential property values before and after the RCIP policy is implemented in Riverside County versus San Bernardino County, without any conservation policy. In addition to the quasi-experimental setting, our analysis divides the two county study region into several zones. These zones are paired across counties based on similar population size and date of establishment of municipalities and census information. This pairing helps control for factors from the last sentence in residential real estate aside from open space policy in one county. As observed by Kuminoff et al., (2013), households "sort” across neighborhoods according to their wealth and their preferences for public goods, social characteristics, and commuting opportunities. The aggregation of these individual choices in markets and other institutions influences the supply of amenities and local public goods. Our policy analysis is an improvement over previous studies to address what Kuminoff et al. (2013) suggest in terms of providing a pseudo experiment of a control and treatment context with the conservation policy. Our analysis is at a finer scale compared to previous studies that have looked at open space
will provide habitat open space and protect watersheds and the environmental needs of the County. “Its (RCIP) objective is to retain and enhance the integrity of existing residential, employment, agricultural, and open space areas by protecting them from encroachment of land uses that would result in impacts from noise, noxious fumes, glare, shadowing, and traffic.″ (Riverside County, 2003). According to the Riverside County General Plan (2003), "The population of Riverside County and its cities is expected to double between the years 2000 and 2020, growing by approximately 1.4 million people. Efficient land use may have growth strategically located into existing developed areas, thus minimizing development pressures on rural, agricultural, and open space areas.″ The MSHCP addresses permanent opportunities for habitat through land use policy. The RCIP does strive for the balance of its three parts: open space conservation, transportation and housing with some limits to insure open space is not less of a priority that the other two parts (Riverside County, 1999). The RCIP designates permanent public land habitat through county acquisition leading to no threat of land conversion in the future. Our primary objective of this study is to analyze the RCIP policy and its impact on property values. This study’s analysis assigns Riverside County as the treatment group where the RCIP has been implemented, and San Bernardino County as the control group where the RCIP policy has not been implemented. This study also compares residential property values for Riverside County before and after the county’s involvement in the RCIP program. We use a hedonic model with spatial econometric techniques for the valuation of open space. These econometric models will be discussed further in the methodology section. 1.1. Literature review The novelty of our paper is that we combine an investigation of the effect of open space proximity on residential property value with an investigation of the effect of open space preservation policy with a distinction between types of open space (wild habitat for endangered species versus developed parks). We investigate the effect of a county level endangered species habitat preservation policy with a basic premise supported by the literature of open space that is nearer to residential property is worth more than further away. We are able to show the marginal negative effect of greater distance from residential property to wild open space is absolutely larger in both a county with a preservation policy and one without. Amidst the large literature on economic value of open space, few papers (for example, Cho et al., 2009) account for simultaneous changes in time and space with respect to open space and residential real estate values and others have not included a policy effect with their hedonic pricing investigation of value of open space. Cho et al. (2009) find the marginal effects of different measures of land amenities (lower housing density, greenways, parks and water bodies) on property value were statistically positive and significant. There is a statistically significant increase in the time periods they include (1989–1991 and 1999–2001). In their study, while the marginal effects of lot size and proximity to golf course were also positive and significant, they decreased from the first to the second time period. Brander and Koetse (2011) estimate the value of urban open space using meta-analyses of contingent valuation and hedonic pricing methods of existing literature including Anderson and West (2006), Geoghegan et al. (1997), Irwin (2002), Acharya and Bennett (2001), Poudyal et al (2009) who include a percentage of urban open space within a given buffer distance and Wu et al. (2004) who include explicit distance to open space. Brander and Koetse (2011) focus mainly on forests and urban parks for their open space variables. Increase in distance from open space has a negative and statistically significant impact on house prices. The time series data used in our paper help capture changes in value of open space over time for a longer continuous time horizon than Cho et al. (2009) for open space that supports endangered species habitat unlike the open space in Cho et al. (2009) and in Brander and Koetse 481
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
2. Material and methods
The sales information on residential real estate of Riverside and San Bernardino counties were obtained from Dataquick data. The Euclidean distance measurements were computed using GIS.
2.1. Data
2.2. Methodology
The applied econometric estimation utilizes approximately 148,000 residential real estate sales observations for Riverside and San Bernardino counties, the two largest counties in southeastern California. The Inland Empire has a population of about 3.8 million (Bluffstone et al., 2008). This region is useful to study due to the neighboring counties similar in many aspects except the RCIP policy in Riverside County. We number zones in the data in pairs in order to arrange a test zone and control zone between the two counties within each pair of zones that share three similar characteristics including population size, date of establishment of municipalities, size of the municipalities. This arrangement helps to compare the value of open space in a zone that has a conservation policy (RCIP) with its value in a zone without the policy, controlling for other relevant factors for statistical robustness. We collected those three characteristics from each municipality’s city planning department in the two counties. The characteristics and the zones help include fixed effects in the analysis. The history of urban development in both counties and the similarities they share in terms of demographics and timing of urban development of cities in both counties helps the process of establishing zones in our study region along with the data we collect from the cities and counties through an academic consultant for Riverside County’s acquisition of habitat preservation properties (Thomas Scott, personal communication). Zone 1 consists of Murrieta and Temecula of Riverside County. Zone 3 consists of Beaumont, Calimesa and Banning of Riverside County. Zone 5 consists of Moreno Valley, Riverside and Mira Loma of Riverside County. Zone 2 consists of the cities of Montclair, Rancho Cucamonga, Ontario, Fontana and Upland, which falls within San Bernardino County. Zone 4 consists of Yucaipa, Redlands and Loma Linda of San Bernardino County. Zone 6 consists of San Bernardino, Highland, Grand Terrace, Colton and Bloomington of San Bernardino County. The zones are depicted in Fig. 1. The pairing of zones (for example 1 and 2) is based on similarity between them across the two counties in terms of the average time of establishment of the municipalities, population size and municipality size. We are able to control for these factors through the zone fixed effects in our model. We investigate the temporal variation around the RCIP policy through regressions that compare 1996–99 and 2000–2004, prior and post RCIP implementation. The data contains residential property sales information for cities of Riverside and San Bernardino counties from 1996 to 2004, useful for conducting a study over time and space. The data provide information on all variables in the econometric regressions including: distance of the houses from forest and parks, residential lot size, structural area of house, age of house, percentage of wild habitat areas and percentage of urban density within 2 kilometers (km) of the residential real estate parcel. The data are also geo-referenced with latitude and longitude coordinates of each residential property that has been useful for our spatial analysis with digitized measure of distance from each residential property to each type of open space. The following sentences indicate variable names and definitions of these variables for the analysis. “Lot size” is the area of the residential lot aside from the residential structure itself measured in square feet. “Distance from park” is the Euclidean distance of the residential property to the nearest park in meters. “Distance from wild habitat” is the Euclidean distance of the residential property to the nearest wild habitat in meters. “Percentage wild habitat” is the percentage of density of wild habitat within 2km of a residential property. “Percentage urban” is the percentage of urban density within 2km of a residential property. “Age of the house” was calculated based on the years between the time it was built and sold.
With pooled data across two counties, our spatial econometric hedonic model helps estimate both policy and different open space value results. We analyzed three kinds of models - log-log Ordinary Least Square (OLS) regression model, log-log spatial autoregressive model (SAR) and log-log spatial error model (SEM). This paper presents results of the SEM model as it is the best fit to account for the spatial distribution of data observations and omitted variables within the study area depicted in Fig. 1. Since the SEM allows for uncertainty regarding an exact model to include independent variables such as the RCIP policy and there are no spatial lags of the dependent variable that corresponds with a SAR model, SEM is more appropriate for our analysis. Neglecting to account for error caused by spatial dependency could result in biased and inconsistent estimators. We do detect a significant spatial correlation in the error terms and hence, the fit of the model improves significantly with SEM model. The SEM provides more robust estimates compared to OLS that assumes data observations are independent of space or SAR.1 The R square of the spatial autoregressive model turned out to be much lower (∼0.4) compared to the R square (∼0.7) of the Spatial Error Model. The estimates in the SEM model had a higher statistical significance than the SAR model and standard error of estimates in the SEM model were lower than the standard error of estimates in the SAR model. Based on these goodness of fit statistics we chose SEM model over the SAR and the OLS regression models for our analysis. Our regression equation in logarithmic form accommodates different units of measure of variables, enabling us to estimate the coefficients as elasticities. The logarithmic form enables comparison in terms of relative size of estimated coefficients. In general, log transformation has improved fit over the raw scale (Hao and Naiman, 2007). The log-log coefficients are typically interpreted as elasticities with a 1% increase in the independent variables (Wooldridge, 2006).
values at a county level such as Doss and Taff (1996).
2.3. Spatial econometric model specifications In the spatial analysis performed in this paper, the weight matrix consists of the inverse of the Euclidean distance calculated based on the geo-reference latitude and longitude coordinates of the residential property and open space variables [in Eq. (i)]. Thus, the weight matrix assigns higher weights to the property that are situated closer to each other compared to those that are situated at a farther distance from each other. A Matlab program with the parallel processing toolbox has been used in this analysis to construct the spatial weight matrix faster by dividing the task into different processors of the computer. A spatial weights matrix provides the “structure” of assumed spatial relationships.
⎡ 0 1/ d12 1/ d13 ⎤ W = ⎢1/ d21 0 1/ d23 ⎥ ⎢1/ d 1/ d 0 ⎥ 32 ⎦ ⎣ 31
(i)
In Eq. (i) dijindicates the distance between residential property in location i and location j. The diagonal terms in the matrix is zero as there is no weight for location i on location i. The weight matrix assigns higher weights to residential property situated closer to each other compared to residential property, which are situated farther away from each other. The spatial weights matrix is row standardized. The specification of the Spatial Error Model (SEM) model is 1 The results produced by the log-log OLS model and log-log SAR are not reported in this paper but can be made available upon request.
482
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
Fig. 1. Study Region Map.
wild*year)it+ β10 ln(distance from park*year)it + β11 (year dummies) + β12 (zone*year) + β13 (countydummy*year) + ε.
provided in Eq. (ii). In Eq. (ii) W is the spatial weight matrix from Eq. (i) and the parameter λ ρ is the coefficient for error correlation. The parameters βi indicate the statistical significance of the independent variable’s influence on the dependent variable ln(salevalue ) . The real estate sale value on left hand side of the Eq. (ii) is the real estate sale price of the residential property sold at time period t. The right hand side variables in Eq. (ii) constitutes of lot size (lotsize) of residential property, structural area (structure)of a house, Euclidean distance of a residential property from the nearest park (distance from park), Euclidean distance of a residential property from the nearest wild habitat area (distance from wild), percentage of urban density (percentage urban) and density of wild habitat (percentage wild) within 2km of a house, age of the house (age) when it was sold, year fixed effects (year dummies) and interaction of year and distance from park (distance from park*year) and interaction of year and distance from wild (distance from wild*year) and interaction of county with year dummies (countydummy*year) and interaction of zone with year dummies(zone*year) ε is a vector of error terms. The Spatial Error Model (SEM) used in this study is:
ε = λWε + μ
μ ∼ (0, σ 2In )
(ii) (iii)
The data are pooled for the regression of the SEM with relevant dummy variables for each zone assigned and also time dummy variables help in our comparative investigation. 3. Results The results in Table 2 from all observations pooled across two counties indicate the positive and statistically significant influence from onsite characteristics of house structure size and lot size on residential real estate sale price and a negative and statistically significant influence of house age on residential property sale price. The abbreviation of variables in Table 2 appears in parentheses in this section next to their complete name. The negative and statistically significant coefficient on distance from wild (DW) and distance from parks (DP) indicates a reduction in residential real estate sale price overall as distance to both types of open space increases. We find that distance to wild habitat area dominates in magnitude as the decrease in residential real estate sale price is higher due to a marginal increase in distance from wild habitat
ln (salevalue)it = β0 ln(structure)it + β1 ln(lotsize)it + β2 ln(distance from park)it+ β3 ln(distance from wild)it+ β4 ln(age)it+ β5 ln(percentage wild)it+ β6 ln(percentage urban)it+ β7 ln(distance from wild*zone)it+ β8 ln(distance from park*zone)it+ β9 ln(distance from 483
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
area compared to a marginal increase in distance from park. The positive and statistically significant coefficient on percentage wild (Pwild02) indicates that a homebuyer’s marginal value increases from more open space within a one mile circumference of the residential property. One mile is a standard distance for real estate market appraisers generating comparable values for residential properties. The negative and statistically significant coefficient on percentage urban (Purban02) indicates a reduction in residential real estate sale value from an increase in urban density within the one mile circumference. The interaction between distance from wild and each zone (DW*Zone in Table 2) shows a negative impact on residential real estate values in Zones 1, 3 and 5 with a marginal increase in distance to wild in Riverside County. The highest marginal value for proximity to wild habitat areas are observed in Zone 3 as a marginal increase in distance from wilds has the highest negative impact on residential real estate sale values in this zone. Zone 2 and Zone 6 also have a negative and statistically significant impact on home values with an increase in distance from wild. The negative impact on Zone 6 of San Bernardino County residential real estate values with an increase in distance from wild is less than Zone 2 in that same county and less than Zones 3 and 5 of Riverside County. The statistically significant negative impact on residential real estate values with a marginal increase in distance from wild indicates the RCIP influence in Riverside County for a complementary relationship between housing and open space in terms of increased residential real estate sale value through the RCIP policy’s attempt to balance housing and open space objectives. The interaction terms of distance from parks and zones (DP*Zone in Table 2) shows a marginal increase in distance from park has the highest negative impact on residential real estate values in Zone 6 (San Bernardino County) and the second highest negative impact on residential real estate values in Zone 1 (Riverside County). City parks have not been designated in San Bernardino County for over 20 years (Bluffstone et al. 2008). The average distance from residences to parks in San Bernardino County is high relative to Riverside County as shown in the county comparison of Fig. 2. The scarcity of parks in Zone 6 of San Bernardino County can contribute to the high amenity value generated from the regression involving the interaction terms. Perhaps, there is a spillover effect due to the RCIP policy in Riverside County. The third highest negative impact on residential real estate sale value with an increase in distance from parks is observed in Zone 3 of Riverside County. Figs. 2 and 3 depict the differences in wild and park open space distances between the counties before and after the RCIP policy aside from the summary statistics in Tables 1a–1c. Note, San Bernardino County has permanent national forest acreage in close proximity included in Fig. 3. With pooled observations across both counties, the interaction of distance wild with year (DW*Year in Table 2) has a higher marginal impact on residential property sale values in 2000-04 in the post-policy period. The interaction between distance from parks (DP) and years
Fig. 3. Average Distance of Residential Real Estate from Wild Habitat Area.
(DP*Year) also show a larger negative impact on residential real estate prices during 2000-04, post-policy period. Both of these results imply that the marginal value for open space increases after the RCIP policy is implemented. The real estate market upswing in residential real estate sale values during the time period for the analysis is recorded through the positive and statistically significant coefficients increasing for each consecutive year dummy variable. The zone fixed effects in our model also help to control for three factors including average date of establishment of municipality, population size and municipality size. The coefficients of this model are robust as verified by running heteroskedastic SEM regressions. The p value of a couple variables change by rounding digits and the statistical significance remains the same with the majority of coefficients significant at 1%. To explore the timing of the RCIP policy, the interaction between zones and years before and after the policy is announced yields results as follows. Zones 1, 3 and 5 have a positive and statistically significant effect at a higher magnitude for the later years of the time series versus the earlier time period whereas the opposite is the case for Zones 2 and 6 in San Bernardino County. This is consistent with Riddel (2001), which finds that the economic impact of an open space purchase takes some time to be fully realized. We conduct a three variable interaction of county, time (pre/post policy) and distance wild variable to identify the impact (if any) brought about by the RCIP policy on the marginal value of open space in Riverside and San Bernardino counties. A 1% increase in distance from wild leads to a 0.07% drop in residential real estate prices during 1996–99 and 0.20% drop in residential real estate prices in 2000–04. These coefficients indicate a higher marginal value for open space in the post-policy period in Riverside County. A 1% increase in distance from wild leads to a 0.04% drop in residential real estate prices during 1996–99 and 0.11% drop in real estate prices during 2000–04 in San Bernardino County. The increase in marginal value of wild habitat open space in San Bernardino County (where policy is not implemented) captures the spillover effect of the neighboring county (Riverside County) to some extent. Our results indicate that the residents in San Bernardino County also values open space more after an open space policy is implemented in the neighboring county (Riverside County). These variables are strong variables of interest in our study which helps to highlight that an open space policy can increase the marginal value of open space not only in the county it is implemented but also in the neighboring county where it's not implemented.
4. Discussion Of impact in monetary values 4.1. Impact on residential property Sale values by Zones Recall the cities that constitute each of these zones used in our analysis. Zone 1 consists of Murrieta and Temecula of Riverside County. Zone 3 consists of Beaumont, Calimesa and Banning of Riverside
Fig. 2. Average Distance of Residential Real Estate from Parks.
484
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
Table 1a Average of variables for Zone 1- Zone 6 from 1996 to 2004.
Zone1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6
AGE (years)
SQFT_STRU (sq ft)
LOT_SQFT (sq ft)
SALE_VALUE ($)
DIST_WILDS (meters)
DIST_PARKS (meters)
P_URBAN02 (%)
P_WILD02 (%)
7.36 25.08 30.00 31.85 27.25 40.28
2,121.88 1,649.64 1,398.13 1,741.74 1,613.39 1,407.77
21,370.95 9,591.70 22,251.16 15,046.65 12,442.78 10,838.33
248,191.55 195,576.27 133,734.88 196,097.76 166,253.85 128,010.24
2,649.09 1,702.13 2,064.97 900.65 2,319.80 1,544.46
820.14 1,976.01 709.47 1,614.69 1,796.12 4,274.88
58 70 49 63 73 61
1 1 2 1 4 4
Table 1b Average of variables for Zone 1- Zone 6 from 1996 to 99.
Zone Zone Zone Zone Zone Zone
1 2 3 4 5 6
AGE (years)
SQFT_STRU (sq ft)
LOT_SQFT (sq ft)
SALE_VALUE ($)
DIST_WILDS (meters)
DIST_PARKS (meters)
P_URBAN02 (%)
P_WILD02 (%)
10.07 28.80 33.00 35.10 30.62 40.85
2,075.61 1,600.38 1,366.13 1,659.65 1,584.93 1,363.17
20,540.27 9,448.78 18,416.22 15,646.15 12,271.54 10,749.32
309,762.48 241,554.98 156,133.30 229,866.74 209,232.93 156,253.69
2,599.94 1,739.76 2,085.78 957.36 2,299.27 1,577.19
821.01 1,987.24 729.18 2,008.97 1,747.57 4,496.68
57 73 50 63 74 61
1 1 2 1 4 4
Table 1c Average of variables for Zone 1 - Zone 6 from 2000 to 04.
Zone Zone Zone Zone Zone Zone
1 2 3 4 5 6
AGE (years)
SQFT_STRU (sq ft)
LOT_SQFT (sq ft)
SALE_VALUE ($)
DIST_WILDS (meters)
DIST_PARKS (meters)
P_URBAN02 (%)
P_WILD02 (%)
4.65 21.36 27.00 28.59 23.88 39.71
2,168.15 1,698.89 1,430.14 1,823.83 1,641.85 1,452.37
22,201.64 9,734.62 26,086.10 14,447.16 12,614.01 10,927.34
186,620.63 149,597.55 111,336.47 162,328.77 123,274.77 99,766.78
2,698.23 1,664.50 2,044.15 843.94 2,340.32 1,511.73
819.27 1,964.78 689.76 1,220.41 1,844.67 4,053.07
58 72 48 62 73 60
1 1 2 1 4 4
Zone 1, 3 and 5 respectively with a 1mile increase in distance from parks. Whereas there is a $15,166 and $7500 decrease in residential real estate sale prices in Zone 2 and Zone 6 with a 1mile increase in distance from parks in San Bernardino County. Recall, Fig. 2 depicts the shortage of parks in San Bernardino County. The results show the amenity value is capitalized into formal real estate value within the one mile radius range of standard real estate appraisal required by underwriters for financial transactions in real estate requiring comparable values for land appraisal occur within a mile of the property in question. Thus not only is the RCIP statistically significant for capitalizing permanent endangered species habitat into residential real estate values, but also the policy impact can be directly tied to the formal real estate market transactions guidelines within a mile radius for comparable values (Section B4-1.3-08: Comparable Sales (01/31/2017) Selling Guide and FreddieMac (www.freddiemac.com/hve/hve.html).
County. Zone 5 consists of Moreno Valley, Riverside and Mira Loma of Riverside County. Zone 2 consists of the cities of Montclair, Rancho Cucamonga, Ontario, Fontana and Upland, which falls within San Bernardino County. Zone 4 consists of Yucaipa, Redlands and Loma Linda of San Bernardino County. Zone 6 consists of San Bernardino, Highland, Grand Terrace, Colton and Bloomington of San Bernardino County. In order to avoid multicollinearity, we drop Zone 4 from the regression, considering it as the default zone. Zone 1 contains the key initial areas of focus for open space preservation and Zones 5 and 6 contain the largest cities and are the county seats for Riverside and San Bernardino, respectively. Table 3 shows the impact for each of the independent variables on residential real estate sale prices in dollar values.2,3 We find that with a 1mile increase in distance from park there is a $23,100 decrease in residential real estate sale values whereas with a 1mile increase from distance from wild habitat areas there is a $43,333 decrease in residential real estate sale values. In Riverside County, a 1mile increase in distance from wilds yields an approximately $35,000 decrease in residential real estate sale values in Zone 1 and $36,666 decrease in Zone 3 and $33,333 decrease in Zone 5. On the other hand, in San Bernardino County a 1mile increase in distance from wilds leads to a $61,666 decrease in residential real estate sale values in Zone 2 and a $36,666 decrease in real estate sale values in Zone 6. There is a $70,000, $41,666 and $14,333 decrease in residential real estate sale prices in
4.2. Impact on residential property values by years During the years 1996-99, there is a $626 decrease in real estate sale values in Riverside whereas a $36,666 decrease in 2000–04 in real estate values in Riverside with a one-mile increase in distance from wild. The real estate values decreases by $330 during 1996–99 and $20,000 in 2000–04 with a one mile increase in distance from wilds in San Bernardino. Although we do not estimate the impact of distance from parks in monetary values separately for Riverside and San Bernardino counties, there is $2313 decrease during 1996–99 and $8333 decrease in the real estate values during 2000–04 with a one mile increase in distance from parks, overall.
2 The impact on residential real estate sale values in dollars is estimated based on the coefficients (elasticities) in Table 1 and the average real estate sale price of residential properties in each zones. 3 As the units for distance used in our analysis are in meters, we use this unit for estimating the marginal impact. However, we use these estimates in further calculation of the impact on home values due to a 1 mile change in distance, as this is a more conventional unit used for real estate transactions. Our discussion of monetary values is also based on the 1mile change in distance. We report the estimates both for a 1 mile change in distance in Table 3.
5. Summary and conclusions Our results indicate a high value for proximity to wild habitat areas to residential real estate in the zones of Riverside County in the post485
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
Table 2 SEM results with all variables and interaction term.4 Variables
Coefficients
Variables
Coefficients
Variables
Coefficients
Structure Lot Distpark Distwild Age Pwild02 Purban02 DWZone1 DWZone2 DWZone3 DWZone5 DWZone6 DPZone1 DPZone2 DPZone3 DPZone5 DPZone6 DP(96-99) DP(2000-04)
0.985*** (0.000) 0.167*** (0.000) −0.118*** (0.000) −0.261*** (0.000) −0.001*** (0.000) 0.013*** (0.000) −0.101*** (0.000) −0.227*** (0.000) −0.321*** (0.000) −0.342*** (0.000) −0.276*** (0.000) −0.258*** (0.000) −0.139*** (0.000) −0.093*** (0.000) −0.132*** (0.000) −0.093*** (0.000) −0.149*** (0.000) −0.016 (0.169) −0.025*** (0.000)
DW(96–99) DW(2000–04) Year1997 Year1998 Year1999 Year2000 Year2001 Year2002 Year2003 Year2004 Riv*(96–99) Riv*(2000–04) SB*(96–99) SB*(2000–04) Riv*(96–99)*DW Riv*(2000–04)*DW SB*(96–99)*DW SB*(2000–04)*DW Zone 1
−0.046*** (0.000) −0.076*** (0.023) 0.217*** (0.000) 0.286***$ (0.000) 0.366*** (0.000) 0.718*** (0.000) 0.845*** (0.000) 0.985*** (0.000) 1.172*** (0.000) 1.441*** (0.000) 0.062*** (0.000) 0.191*** (0.000) 0.037*** (0.000) 0.038*** (0.000) −0.071*** (0.000) −0.202*** (0.000) −0.041*** (0.000) −0.106*** (0.000) 2.445 (0.000)
Zone 2 Zone 3 Zone 5 Zone 6 Zone1*(96-99) Zone1*(2000-04) Zone2*(96-99) Zone2*(2000-04) Zone3*(96-99) Zone3 *(2000-04) Zone5*(96-99) Zone5*(2000-04) Zone6*(96-99) Zone6*(2000-04) Lambda Rsquare Adjusted R square
2.938 (0.000) 2.935 (0.000) 2.443 (0.000) 2.634 (0.000) 0.134*** (0.006) 0.147*** (0.002) 0.209*** (0.000) 0.023** (0.000) 0.063* (0.105) 0.177*** (0.000) 0.016*** (0.014) 0.033** (0.030) 0.123*** (0.000) 0.023*** (0.002) 0.495*** (0.000) 0.750 0.749
value due to a marginal increase in distance to wilds is approximately half the value of Riverside County for each of the time periods (before and after the RCIP policy is implemented). The significant variation of the values of open space across different zones within a county provides evidence of the existence of a segmented market even at the county level. With more accurate and efficient estimates of value of open space and other variables, the spatial error hedonic model can be used in the decision-making process associated with habitat conservation policy and urban land-use planning. While Mahmoudi et al. (2012) note a reconfiguration of a park matters for amenity values associated with urban property and Tyravainen (1997) and Poudyal et al (2009) note larger, nearby parks increase housing values, our analysis econometrically estimates the impact on residential property value of a county’s policy to preserve endangered species habitat. The policy context of our study is that the habitat is for more than one endangered species formally listed in federal policy. Since the initial focus on the Stephens’ Kangaroo Rat, now 142 species exist in the habitat preserved through the RCIP policy. The improvement in residential real estate values is a complementary benefit to the endangered species goal of the county policy. We find that amenity values vary spatially similar to Cho et al (2008) for other types of open space. Overall, the results from this study highlight the importance of conservation policy in influencing value of habitat areas for multiple endangered species. Furthermore, our pre-post policy analysis in both counties for distance to wild habitat shows how a conservation policy in a particular county can influence the residential real estate sale values not only within that county but also residential real estate sale values in the neighboring county. The amenity values generated in this study for different zones within counties can help in estimating the benefit of conservation of habitat for endangered species, which can be used as a tool by policy makers to set the conservation fees, e.g., development impact fees that help finance conservation. Additionally, this study can prove significant for land-use planning and conservation decisions, not only in the Inland Empire region but also for any other region with similar geographical characteristics and residential markets.
Table 3 Monetary Values. Variables
Impact ($) with 1 mile increase
Variables
Impact ($) with 1 mile increase
Distpark Distwild DWZone1 DWZone2 DWZone3 DWZone5 DWZone6 DPZone1 DPZone2 DPZone3 DPZone5 DPZone6 DW*(96-99) DW*(2000-04) DP*(96-99) DP*(2000-04)
−23,100 −43,333 −35,000 −61,666 −36,666 −33,333 −36,666 −70,000 −15,166 −41,666 −14,333 −7,500 −5150 −20,000 −2,313 −8,333
Riv*DW*(96-99) Riv*DW*(2000-04) SB*DW*(96-99) SB*DW*(2000-04)
−626 −36,666 −330 −20,000
policy period. The results indicate a complimentary impact on appreciated residential real estate of a policy initiated for conservation of wild habitat areas in Riverside County during 1999. We also observe a higher value for proximity to wild habitat areas to residential real estate in San Bernardino County in the post-policy period which can be a potential spillover effect from the neighboring county. We also observe a high amenity value for parks in Zone 6 of San Bernardino County and Zone 1 of Riverside County. Evidence from previous literature (Bluffstone et al. 2008), shows a scarcity of parks in San Bernardino County that can influence its high amenity value in certain zones of this county. Also, it can be a spillover effect due to the RCIP policy in the neighboring county We observe a higher value for proximity to parks in Riverside County during 2000–04. Again, we observe that the post policy years of (2000–04) lead to a higher increase in residential real estate sale values in the zones of Riverside County compared to pre-policy years of (1996–99). This is as expected, as the announcement of the conservation policy could have played a major role in driving the residential real estate sale prices higher in these regions of Riverside County. For zones of San Bernardino County, the resulting change in residential real estate
References Acharya, G., Bennett, L., 2001. Valuing open space and land-use patterns in urban watersheds. J. Real Estate Finance Econ. 22, 221–237. Anderson, S., West, S., 2006. Open space, residential property values and spatial context. Reg. Sci. Urban Econ. 36, 773–789. Anselin, L., 1988. Spatial Econometrics: Methods and Models. Kluwer Academic Publishers, Dordrecht; Boston.
4 DW: Distance from wild; DP: Distance from park, PW: Percentage wild habitat area; PU: percentage urban area.
486
Land Use Policy 73 (2018) 480–487
L. Fernandez et al.
nonmarket benefits. Resour. Future. Mukherjee, M., Fernandez, L., 2011. Analysis of the Influence of Open Space on Residential Values, Giannini Foundation Grant Report for Agricultural and Resource Economics Update, 15-1, September 2011. Nordman, E., Wagner, J., 2009. Public Purchases and Private Preferences: A Hedonic Model of Open Space Acquisitions, Working Paper. 27 p. Available at:. Grand Valley State University, Allendale, MI accessed 19 May 2009. http://works.bepress.com/ erik_nordman/1/. Poudyal, N., Hodges, D., Ronn, B., Cho, S., 2009. Valuing diversity and spatial pattern of open space plots in urban neighborhoods. For. Policy Econ. 11, 194–201 Riverside County Integrated Project http://www.rcip.org/. Riddel, M., 2001. A dynamic approach to estimating hedonic prices for environmental goods: an application to open space purchase. Land Econ. 77 (4), 494–512. Riverside County, 1999. 2003, Riverside County Integrated Plan. http://www.rcip.org. Thorsnes, P., 2002. The value of a suburban forest preserve: estimates from sales of vacant residential building lots. Land Econ. 78 (3), 426–441. Tyravainen, L., 1997. The amenity value of the urban forest: an application of the hedonic pricing method. Landscape Urban Plann. 37 (3), 211–222. USDA, 2006. Society’s Choices: Land Use Changes, Forest Fragmentation and Conservation, Pacific Northwest Research Station Science Findings Newsletter, Issue 88, November. Wooldridge, J., 2006. Introductory Econometrics: A Modern Approach, third edition. Thomson South-Western. Wu, J., Adams, R., Plantinga, A., 2004. Amenities in an urban equilibrium model: residential development in Portland, OR. Land Econ. 80, 19–32. Yoo, J., Browning, J., Minesing, K., 2017. Open space premium near commercial zones- a case study in the City of corona, CA. Int. J. Urban Sci online access 11/26/17.
Bluffstone, R., Braman, M., Fernandez, L., Scott, T., Lee, P., 2008. Housing, sprawl, and the use of development impact fees: the case Of Inland Empire. Contemp. Econ. Policy 26 (3), 433–447. Brander, D., Koetse, M., 2011. The value of open space: meta analysis of contingent valuation and hedonic pricing results. J. Environ. Manage. 92 (10), 2763–2773. Brasington, D., Hite, D., 2008. A mixed index approach to identifying hedonic price models. Reg. Sci. Urban Econ. 38 (3), 271–284. Cho, S.C., Clark, W., Park, S.Kim, 2009. Spatial and temporal variation in the housing market values of lot size and open space. Land Econ. 85 (1), 51–73. Cho, S.-H., Poudyal, N.C., Roberts, R.K., 2008. Spatial analysis of the amenity value of green open space. Ecol. Econ. 66, 403–416. Doss, C., Taff, S., 1996. The influence of wetland type and wetland proximity on residential property values. J. Agric. Resour. Econ. 21 (1), 120–129. Geoghegan, J., Wainger, L., Bockstael, N., 1997. Spatial landscape indices in a hedonic framework: an ecological economics analysis using GIS. Ecol. Econ. 23, 251–264. Hao, L., Naiman, D., 2007. Quantile Regression. Quantitative Applications in the Social Sciences, vol. 149 Sage Publications. Irwin, E., 2002. The effects of open space on residential property values. Land. Economics 78 (4), 465–480. Kroeger, Tim, 2008. Open space property value premium analysis. National Council for Science and the Environment 2006 Wildlife Habitat Policy Research Program. Kuminoff, N.V., Smith, K.V., Timmins, C., 2013. The new economics of equilibrium sorting and policy evaluation using housing markets. J. Econ. Lit. 51 (4), 1007–1062. Mahmoudi, P., MacDonald, D., Crossman, N., Summers, D., van der Hoek, J., 2012. Space matters: the importance of amenity in planning metropolitan growth. Aust. J. Agric. Econ. 57, 38–59. McConnell, V., Walls, M., 2005. The value of open space: evidence from studies Of
487