# RECORD 8fb41eefcd89b6d6746d9ab5af9f6f61e2da05b2
# TITLE: Spatial Heterogeneity in Hedonic Price Effects for Lake Water Quality
# YEAR: 2023.0  VENUE: Land Economics  DOI: 10.3368/le.100.1.102122-0086r
# AUTHORS: Kristen Swedberg; Diego S. Cardoso; Adriana Castillo-Castillo; Saleh Mamun; K. Boyle; C. Nolte; M. Papenfus; S. Polasky
# PAGES: 42  CANDIDATE TABLE PAGES (1-based): [33, 34]
# ABSTRACT: This study uses Zillow’s Transaction and Assessment Database to investigate variation in hedonic price effects of water clarity on single-family houses throughout the United States. We consider five spatial scales and estimate models using different sample selection criteria and model specifications. The results indicate considerable spatial heterogeneity within and across the four U.S. census regions. However, we also find heterogeneity resulting from different types of investigator decisions, including sample selection and modeling choices. Thus, it is necessary to use practical knowledge to consider the limits of market areas and to investigate the robustness of estimation results to investigator choices.


===== PAGE 1 (score 6, numeric-paren 0, table-words 0, var-words 6) =====

1 
 
Spatial Heterogeneity in Hedonic Price Effects for Lake Water Quality 
Kristen Swedberg1,2, Diego S. Cardoso3, Adriana Castillo-Castillo4, Saleh Mamun4,5,6,  
Kevin J. Boyle1,7, Christoph Nolte8,9, Michael Papenfus10, Stephen Polasky4,5,11 
 
1 Department of Agricultural and Applied Economics, Virginia Tech, Blacksburg, VA 
2 ORISE Fellow in Office of Water, Environmental Protection Agency, Washington, DC  
3 Department of Agricultural Economics, Purdue University, West Lafayette, IN  
4 Department of Applied Economics, University of Minnesota, St. Paul, MN 
5 The Natural Capital Project, University of Minnesota, St. Paul, MN 
6 Natural Resources Research Institute, University of Minnesota – Duluth, Duluth, MN 
7 Blackwood Department of Real Estate, Virginia Tech, Blacksburg, VA 
8 Department of Earth & Environment, Boston University, Boston, MA 
9 Faculty of Computing & Data Sciences, Boston University, Boston, MA 
10 Environmental Protection Agency, Corvallis, OR 
11 Department of Ecology, Evolution & Behavior, University of Minnesota, St. Paul, MN 
 
Abstract 
This study uses Zillow’s ZTRAX property transaction database to investigate variation in hedonic 
price effects of water clarity on single-family houses throughout the United States. We consider 
five spatial scales and estimate models using different sample selection criteria and model 
specifications. Our results indicate considerable spatial heterogeneity both within and across the 
four U.S. Census regions. However, we also find heterogeneity resulting from different types of 
investigator decisions, including sample selection and modelling choices. Thus, it is necessary to 
use practical knowledge to consider the limits of market areas and to investigate the robustness of 
estimation results to investigator choices. (JEL Q51) 
 
 
doi:10.3368/le.100.1.102122-0086R
by guest on October 20, 2026. Copyright 2023
Downloaded from 


===== PAGE 33 (score 57, numeric-paren 36, table-words 1, var-words 12) =====

33 
 
Table 2 Descriptive Statistics by Region  
 
Full Sample 
Northeast 
Midwest 
South 
West 
Sale Price (2020 $) 
450,060 
344,023 
356,748 
642,456 
(426,614) 
(250,921) 
(282,424) 
(377,061) 
Secchi Depth (m) 
2.49 
1.67 
1.49 
2.93 
(1.61) 
(1.22) 
(0.87) 
(1.85) 
Lake Distance (m) 
1,016 
924 
1,016 
1,325 
(734) 
(687) 
(708) 
(695) 
Lake Distance <150 m (1/0) 
0.13 
0.14 
0.11 
0.04 
(0.34) 
(0.34) 
(0.32) 
(0.19) 
Lake Distance 150-300 m (1/0) 
0.09 
0.10 
0.09 
0.05 
(0.29) 
(0.30) 
(0.28) 
(0.21) 
Lake Area (ha) 
8,944 
967 
5,334 
2,451 
(28,427) 
(5,083) 
(17,441) 
(7,227) 
Urbanized Area (1/0) 
0.60 
0.67 
0.67 
0.76 
(0.49) 
(0.47) 
(0.47) 
(0.43) 
Median Income (2016 $) 
83,769 
78,565 
66,228 
82,623 
(35,013) 
(32,392) 
(30,839) 
(37,863) 
Slope (deg) 
3.89 
2.69 
3.41 
3.93 
(3.27) 
(2.22) 
(3.41) 
(3.88) 
Elevation (m) 
148.95 
291.20 
136.82 
885.74 
(130.50) 
(59.09) 
(142.42) 
(791.58) 
N 
396,988 
1,241,194 
1,388,818 
344,394 
Lakes 
1,112 
4,012 
830 
308 
States 
9 
12 
16 
10 
Ecoregions 
5 
10 
7 
8 
 
Samples Restricted by Property Attribute Availability 
Northeast 
Midwest 
South 
West 
Lot Size (sqft) 
43,190 
30,916 
31,008 
17,659 
(78,291) 
(64,055) 
(55,912) 
(40,489) 
n = 380,289 n = 1,057,113 n = 1,291,411 n = 323,733 
Building Age (years) 
48.60 
43.19 
27.23 
32.70 
(29.12) 
(30.09) 
(21.85) 
(27.29) 
n = 318,287 
n = 870,459 
n = 1,069,065 n = 286,611 
Building Size (sqft) 
1,924 
1,820 
2,057 
1,969 
(871) 
(972) 
(1,024) 
(846) 
n = 345,396 
n = 860,124 
n = 684,586 
n = 271,974 
Number of Bathrooms 
2.07 
2.39 
2.71 
2.24 
(0.92) 
(1.33) 
(1.30) 
(0.84) 
n = 258,968 
n = 725,467 
n = 473,398 
n = 170,154 
Number of Bedrooms 
3.20 
3.59 
3.61 
3.27 
 
(0.98) 
(1.46) 
(1.36) 
(0.97) 
n = 258,968 
n = 725,467 
n = 473,398 
n = 170,154 
Means reported with standard deviations in parentheses.  
 
by guest on October 20, 2026. Copyright 2023
Downloaded from 


===== PAGE 34 (score 123, numeric-paren 97, table-words 3, var-words 6) =====

34 
 
Table 3 National Regression Results  
Parameter 
Estimate 
Std. Error 
Parameter 
Estimate 
Std. Error 
const 
7.123*** 
(0.207) 
Secchi * MD 
-0.259** 
(0.085) 
Secchii,ii,iii 
0.231*** 
(0.034) 
Secchi* ME 
2.428*** 
(0.077) 
Secchi * Lake Area 
Demeaned3 
0.009* 
(0.004) 
Secchi * MI 
-0.0680 
(0.048) 
Secchi * <150 m 
0.025** 
(0.009) 
Secchi * MO 
0.6730 
(0.739) 
Secchi * Lake Distance 
-0.034*** 
(0.005) 
Secchi * MT 
-2.021*** 
(0.465) 
Secchi * Urbaniv 
-0.0150 
(0.014) 
Secchi * NC 
-0.1830 
(0.123) 
Secchi * Ecoregion 5.2iv 
0.075* 
(0.036) 
Secchi * ND 
-0.2230 
(0.292) 
Secchi * Ecoregion 5.3 
0.0430 
(0.033) 
Secchi * NE 
-0.1790 
(0.101) 
Secchi * Ecoregion 6.2 
0.5940 
(0.462) 
Secchi * NH 
-0.0650 
(0.047) 
Secchi * Ecoregion 7.1 
0.9910 
(0.596) 
Secchi * NJ 
-0.0030 
(0.052) 
Secchi * Ecoregion 8.2 
0.0570 
(0.077) 
Secchi * NY 
-0.0740 
(0.04) 
Secchi * Ecoregion 8.3 
0.0550 
(0.048) 
Secchi * OH 
-0.1980 
(0.16) 
Secchi * Ecoregion 8.4 
-0.0170 
(0.069) 
Secchi * OK 
-0.1240 
(0.266) 
Secchi * Ecoregion 8.5 
0.0260 
(0.031) 
Secchi * OR 
-0.8670 
(0.525) 
Secchi * Ecoregion 9.2 
0.150 
(0.1) 
Secchi * PA 
0.154*** 
(0.036) 
Secchi * Ecoregion 9.3 
0.2230 
(0.201) 
Secchi * RI 
0.102* 
(0.047) 
Secchi * Ecoregion 9.4 
0.2030 
(0.149) 
Secchi * SC 
-0.0530 
(0.084) 
Secchi * Ecoregion 10.2 
0.3060 
(0.462) 
Secchi * SD 
-0.1270 
(0.287) 
Secchi * Ecoregion 11.1 
0.6620 
(0.462) 
Secchi * TN 
-0.1130 
(0.078) 
Secchi * Ecoregion 13.1 
0.3660 
(0.475) 
Secchi * TX 
-0.3930 
(0.233) 
Secchi * Ecoregion 15.4 
0.0320 
(0.048) 
Secchi * VA 
-0.1340 
(0.083) 
Secchi * ALiv 
0.0030 
(0.202) 
Secchi * VT 
-0.0780 
(0.129) 
Secchi * AR 
0.1210 
(0.15) 
Secchi * WA 
-0.9890 
(0.597) 
Secchi * AZ 
-0.3960 
(0.475) 
Secchi * WI 
-0.0330 
(0.08) 
Secchi * CA 
-0.7280 
(0.465) 
<150 m 
0.124*** 
(0.009) 
Secchi * CO 
-0.4980 
(0.461) 
Lake 
Distance 
-0.095*** 
(0.005) 
Secchi * CT 
-0.266* 
(0.12) 
Lake Areav 
0.013* 
(0.005) 
Secchi * DE 
-0.0860 
(0.402) 
Median 
Income 
0.131*** 
(0.014) 
Secchi * FL 
-0.0190 
(0.037) 
Slope 
0.019*** 
(0.002) 
Secchi * GA 
0.0330 
(0.128) 
Elevation 
-0.0230 
(0.023) 
Secchi * IA 
-0.221* 
(0.11) 
Lot Size 
0.101*** 
(0.004) 
Secchi * IL 
-0.1070 
(0.08) 
Building Age 
-0.125*** 
(0.003) 
Secchi * IN 
0.20 
(0.515) 
Building Size 
0.547*** 
(0.01) 
Secchi * KY 
-1.956*** 
(0.058) 
Quarter 2 
0.044*** 
(0.002) 
Secchi * LA 
0.0840 
(0.211) 
Quarter 3 
0.05*** 
(0.002) 
Secchi * MA 
-0.0270 
(0.024) 
Quarter 4 
0.015*** 
(0.002) 
Notes: N = 485,860 Standard Errors Clustered at Tract Level. *p<.05 **p<.01 ***p<.001 
iInteraction terms for MN and Ecoregion 8.1 were pre-selected as the excluded categories, 
because the area intersection contains the most observations in the sample. 
iiDuring estimation, interaction terms for KS, NM, WV, Ecoregion 10.1, Ecoregion 12.1 were 
fully absorbed by fixed effects and dropped from model. Observations for Ecoregion 12.1 all 
by guest on October 20, 2026. Copyright 2023
Downloaded from 


===== PAGE 35 (score 7, numeric-paren 0, table-words 2, var-words 3) =====

35 
 
fall within AZ. Variation in estimates for KS, NM, and WV can be explained by the 
corresponding ecoregion interactions.  
iiiThe log of lake area is demeaned in the interaction term with Secchi. Thus, the Secchi 
coefficient alone can be interpreted as the effect of water quality for a lake with the mean log-
lake area. Secchi  * Lake Area Demeaned can be interpreted as the effect of an increase in lake 
size over the sample mean. 
ivBecause this model is estimated with census tract by year fixed effects, it is not necessary to 
include Urban, Ecoregion, and State constituents in the model as control variables.   
vLake area when not interacted with Secchi is not demeaned, as it only serves as a control 
variable in the model rather than a coefficient of interest.   
  
 
 
 
 
 
 
 
 
 
 
 
by guest on October 20, 2026. Copyright 2023
Downloaded from