Significance

Lakes are a crucial resource for the esthetic, recreation, drinking water, habitat, and other services they provide. Yet, many lakes are threatened by nutrient loading that compromises water quality. We use spatially explicit data on property values and lake water quality to provide the most extensive nationwide estimates for benefits of water quality capitalized through property values. We find property values appreciate by $6 to 9 billion if water quality is improved by 10% in the United States. These benefit estimates may support national, state, and local lake policies. The property value effects are particularly important because owners of properties around lakes are significant beneficiaries of lake policies, and they are the last line of protection from nutrient loadings.

Abstract

High-quality water resources provide a wide range of benefits, but the value of water quality is often not fully represented in environmental policy decisions, due in large part to an absence of water quality valuation estimates at large, policy relevant scales. Using data on property values with nationwide coverage across the contiguous United States, we estimate the benefits of lake water quality as measured through capitalization in housing markets. We find compelling evidence that homeowners place a premium on improved water quality. This premium is largest for lakefront property and decays with distance from the waterbody. In aggregate, we estimate that 10% improvement of water quality for the contiguous United States has a value of $6 to 9 billion to property owners. This study provides credible evidence for policymakers to incorporate lake water quality value estimates in environmental decision-making.
The United States has a rich tapestry of freshwater lakes that provide multiple benefits including water-based recreation, habitat that supports biodiversity, esthetic values, cultural values, and drinking water. Yet, the water quality in many lakes has been reduced by an influx of nutrients and other pollutants. Keiser and Shapiro (2019) (1) note that “over half of rivers and substantial shares of drinking water systems violate (water quality) standards.” Under the auspices of the Clean Water Act, the USEPA (U.S. Environmental Protection Agency), USDA (U.S. Department of Agriculture), and other federal and state agencies take actions to protect the quality of these waters through regulations, fines for violations, and incentives. These actions can be divisive, in part, because the monetary benefits from improvements have been inadequately quantified (2). This is illustrated by the recent debate over the Waters of the US rule (e.g., refs. 3 and 4). While this debate focused on connectivity between navigable water and upstream wetlands, the central issue is the incomplete quantification of the monetary benefits accruing from surface water quality improvements. In the absence of credible estimates of specific economic benefits that are salient to the public, policy costs receive more emphasis in public debates and decision-making than benefits.
While policy applications focused on air quality have benefited from influential national-scale studies (e.g., refs. 58), the empirical evidence on the value of water quality for agencies like the US EPA and USDA at the national scale is quite thin. Most studies estimating public benefits from improving surface water quality have been conducted at the local level [e.g., Wolf and Klaiber (2017) (9)]. As local studies are not broadly applicable across space, policy analysts must use “benefits transfer,” a technique whereby the results from existing local or regional studies are combined and “transferred” to regions where studies have not been undertaken to develop national benefits estimates (10, 11). However, the validity of benefits transfer can be compromised by the thinness of the empirical literature, its limited spatial representation, and differences in regional preferences and lake characteristics across the United States (see refs. 1215). An alternative to benefits transfer is to conduct a nationwide analysis of which we know of only four studies related to surface water quality: A stated preference study from 1983 by Carson and Mitchell (1993) (16); a recreation study by Keiser (2019) (17) that examines the USDA’s Conservation Reserve Program; a hedonic study on the downstream benefits of wastewater treatment grants by Keiser and Shapiro (2019) (1); and a recent hedonic study by Moore et al. (2020) (18), valuing lake water quality using property sales data. Each of these studies provides valuable information about different types of benefits of water quality and specific policies related to water quality. The study by Moore et al. (2020) (18) in particular provides a direct link between continuous measures of surface water quality and a market good, property sales. However, while this study features nationally a representative sample of lakes, the resulting water quality observations used by the authors were quite small (1,462 sales), limiting the authors’ ability to test key model assumptions and leaving an opening for future studies.
In this paper, we build on Moore et al.’s (2020) (18) study by leveraging spatially extensive data on water quality and property sales to estimate the price premium for homes near lakes with clean water. We combine water quality data from LAGOS-NE (LAke multi-scaled GeOSpatial and temporal database for Northeastern US) (19) and Water Quality Portal (WQP) (20) with property sales data from Zillow (21) to develop a sample of nearly 674,000 property transactions near 1,632 lakes across the United States. We estimate the impact of lake water quality on surrounding home prices using two different measures: Secchi depth, a measure of water clarity, and chlorophyll-a (chl-a), a measure used by limnologists to monitor the effects of nutrient pollution on lake water quality, which is one of the factors affecting water clarity. Our dataset allows us to apply a variety of rigorous modeling strategies and robustness checks. We further calculate the national-scale economic benefits that would accrue from improving water quality to two policy-relevant levels: restoring lakes to water quality levels typical of original natural conditions, as well as less ambitious, and likely more achievable, improvements of 10% from current baselines in all lakes across the United States. Together, these two scenarios demonstrate the scale of possible benefits from policy interventions to improve water quality in lakes and their surrounding watersheds.
Our results provide compelling nationwide evidence that homeowners substantially value lake water quality. On average, homeowners are willing to pay $3,681 more for a home near a lake with 0.1 m greater Secchi depth and $4,359 more for a 1.0-µg/L lower level of chl-a. When extrapolated nationally, our estimates indicate that a 10% improvement in Secchi depth generates $9.22 billion in economic benefits and restoring lakes to pristine ecological condition results in $26.67 billion in additional capitalized home values.

Materials and Methods

Property Data.

The property dataset used in this analysis is constructed by linking parcel boundary data with Zillow’s Transaction and Assessment Database (ZTRAX, version: Oct 09, 2019) (21). ZTRAX provides sales-related information (e.g., sale dates, prices, interfamily transfer), and parcel and building characteristics (e.g., house size, number of bedrooms, lot size, age). We link parcel boundaries (collected from open sources and Regrid’s data with purpose program) with ZTRAX using assessor’s parcel number and a pattern-matching algorithm (22). Observations where parcel subdivision and consolidation results were unsuccessful or partial linking of parcels with ZTRAX data occurred were excluded from the estimation. To address anomalies in the ZTRAX transaction values and coordinates, we followed best practices as described by Nolte et al. (2021) (23). Fair market values are ensured by 1) removing transactions involving interfamily transfers based on name similarity index; and 2) removing transactions involving public buyers and public sellers. In addition, we remove transactions if the sale price is less than $10,000 (24) and the top 1 percentile of sale price per hectare in order to address remaining outliers. We identify residential developed properties if there is any building area and Zillow reported a building code of “RR” (residential).

Water Quality Data.

Water quality data were obtained from two national-level data repositories: WQP (20) from the USEPA and the LAGOS-NE (19). We merge LAGOS-NE and WQP datasets to create a comprehensive dataset. We used the National Hydrography Dataset (NHD) (25) to combine the two sources of water quality data using a spatial join technique.
The two most commonly available water quality measurements in these data are Secchi depth and chl-a concentration. Secchi depth is measured by inserting a circular disk, colored white and black on alternating quadrants, into a water column and determining the distance from the surface at which the disk is no longer visible. Hence, it measures how deep sunlight can penetrate into water. As water clarity is the salient feature of water quality homeowners observe, we use Secchi depth as one measure of water quality in our hedonic pricing model.
Chl-a is a measure of the number of algae in a waterbody and is a common metric used by limnologists to measure lake water quality This provides a scientific measurement of lake water quality that is not directly observable by a lay person. However, chl-a concentrations have an indirect effect on what people observe as it is a factor affecting lake water clarity. Thus, chl-a and Secchi measurements are correlated but not perfectly, which presents a challenge for including both water quality measurements in a single hedonic model. We estimate the effect for these two water quality measures in two separate hedonic models.

Merging Property Data with Water Quality Data.

The property dataset contains transaction records of sales of properties within 2,000 m from the shoreline of lakes between 2000 and 2019. The properties within 2,000 m refer to the properties that have a distance between the parcel centroid and the nearest lake boundary of less than or equal to 2,000 m. The lake water quality dataset consists of water quality measures for lakes taken between 1900 and 2020. However, due to the high time and monetary costs of collecting these data (26), water quality measures are not taken every year, nor do they necessarily capture differences in water quality in different areas of a lake. This means there is not always a lake water quality measurement that matches in time with the occurrence of property sales. Thus, we used the water quality sample from the closest lake to the property that is closest in time to the property sale. If the difference between the water quality sample year and property sale year is more than 5 y, we omit the transaction. We also check robustness based on alternative cutoff values of this difference (i.e., 1 y and 3 y).
After merging the water quality and property transaction data, our sample consists of 746,424 transactions around 1,632 lakes greater than 4 ha (which is the minimum lake area in LAGOS-NE) (Fig. 1A). The lakes in our sample are clustered in lake-rich regions like the Midwest and East Coast. Adding restrictions to the sample based on the number of sale observations within 100 m of the lake greatly reduces the total number of lakes included in the sample and overall spatial coverage (Fig. 1B). Thus, we select the widest sample as the baseline for comparison and conduct several robustness checks. We observe wide variation in both Secchi depth and chl-a throughout the sample, with generally poor water quality conditions in the upper Midwest (Fig. 1 C and D).
Fig. 1.
(A) Distribution of study lakes with at least one water quality sample and one property sale within 100 m from lakefront (N = 1,632). (B) Distribution of study lakes with at least one water quality sample and 100 property sales within 100 m from lakefront (N = 109). (C) Most recent Secchi depth (m) samples for lakes averaged over census tracts. (D) Most recent chlorophyll-A (µg/L) samples for lakes averaged over census tracts.
The Table 1 reports the summary statistics of our main variables after merging property transactions with water quality values. It shows the wide range of the Secchi depths from a few centimeters to 10 m with a median value of 1.52 m. Lakes with Secchi depths less than 1 to 2 m are classified as having poor water quality (27, 28). Chl-a similarly exhibits wide variation, ranging from 0.01 to 200 with a median of 11.6 µg/L. A detailed description of data cleaning and combining is provided in SI Appendix. While our dataset has wide spatial coverage, it is important to note that our merged dataset does not have panel structure. The same lakes are not sampled frequently and the same properties are not sold multiple times. Only 29.03% properties are sold twice or more and 22.79% lakes are sampled 10 y or more over 20 y of sample. We discuss more about property price and water quality trend over the study period in SI Appendix.
Table 1.
Summary statistics for baseline model observations
 MinMaxMedianMeanSTD
Property location within:    
0 to 100 m buffer (0,1)0100.090.28
100 to 300 m buffer (0,1)0100.150.35
Secchi (m)0.029.701.522.011.59
Chl-a (ug/L)*0.01200.0011.6019.0121.54
Lot area (m2)1005,865,9051,0863,03519,898
Lake size (km2)0.041,650.281.1239.82122.67
Price (2019 $)10,00039,524,515276,544344,307348,070
Note: N = 746,424.
*
Fewer Chl-a observations (N = 637,548).

Methods.

In this study, we use a hedonic model to estimate the impact of lake water quality on housing prices. In this model, properties can be characterized as bundled commodities differentiated by their attributes. Consumers choose a house by selecting their preferred set of attributes, where one attribute is water quality, along with other goods, to maximize their utility given a budget constraint. The hedonic function defines the market price of a house as a function of its structural, neighborhood, environmental, and other attributes. An empirical estimate of the hedonic price function allows measurement of the implicit price of water quality; the model of behavior connects the implicit price to consumers’ marginal willingness to pay for water quality. We estimate the following hedonic model:
ln(Pit)=αln(Qit)+βsSit+βfFit+θslnQit×Sit+θflnQit×Fit+γXit+σYl+τct+it,
[1]
where Pit is the sale price of property i in transaction year t , Qit is the water quality assigned to property i sold in the year t , Sit and Fit are dummy variables indicating a property is within 100 m and 100 to 300 m of its nearest lake, respectively, Xit is the vector of housing characteristics, Yl denotes lake characteristics, τct denotes census tract by year fixed effects, and it is the random error term. Nominal home prices are deflated to real terms using the housing price index (29).
The existing hedonic literature finds that homeowners will pay a premium to buy a property on a lake with high water quality and lakefront owners will pay an even higher premium (9, 30, 31). Therefore, distance to the lake is an important variable affecting residential decisions. However, the cutoff distance that defines the lakefront area varies across studies in past literature. Here, we divided our properties into three distance bins according to the proximity of the property to its nearest lake: 0 to 100 m, 100 m to 300 m, and the area beyond 300 m.
We estimate separate models for the Secchi depth and chl-a measurements of water quality due to the correlation in these measurements noted above. In addition, we interact the distance dummy variables with water quality to investigate potential heterogeneity in the impact of water quality based on distance from lakes. Our primary interest focuses on the parameters α , θs, and θf – the direct and interaction effects of lake water quality on property values. Given the log-log functional form, α measures the percentage change in home prices from a 1% change in Secchi depth for properties within 2,000 m but beyond 300 m of the nearest lake. The parameters θs and θf capture the additional impact of water quality on the price premium for properties within 100 m and 100-300 m of lakes, respectively. Given that homeowners living within 300 m of lakes have easier access to lake amenities, we expect the values of θs and θf to increase the magnitude of the impact of water quality on property prices.
In addition to these water quality variables, we also include variables of housing characteristics that might potentially be correlated with water quality to avoid bias from omitted relevant variables. We use six variables to control for property attributes: lot size, average slope of the lot, elevation of the lot, proximity to a highway, age of the building, and the area of the building. As lakes with larger sizes usually provide more recreational amenities, we also included the size of the lake associated with the water quality measurement.
Census tract by year fixed effects are included to capture annually varying, spatial neighborhood attributes. In the absence of a quasiexperimental design, our identification relies on the fixed effects to control for local unmeasured variables that might be correlated with the water quality measures. In doing so, the set of fixed effects captures the influence of omitted amenities and disamenities that affect property values within a tract in a given year; these could include, for example, road access, natural shoreline, and the presence of parks. As a robustness check, we examine the sensitivity of our results to the inclusion of spatial fixed effects at different levels. We also cluster SEs at the census tract level to account for unobserved correlation within the census tract.
Some prior literature adds restrictions on the number of lakeshore observations and minimum number of water quality records per lake (32). However, we lose a large amount of data if we impose strict restrictions. It is important to include a wide coverage of lakes and property transactions that can represent our study area at a nationwide scale. As an alternative, we also estimated models with different restrictions on water quality and property transaction data as robustness checks.

Results

Using transactions within 2 km of lakes, we found positive and statistically significant impacts of measures of water quality on home prices for properties, with larger impacts for properties closer to the shoreline (Table 2). Using Secchi depth as the measure of water quality, we find that properties within 100 m of lakes have the highest price premium for water quality and those located beyond 300 m are not affected by water quality (first column of Table 2). The coefficient of the interaction term shows that the premium for houses within 100 m of lakes increases with water quality improvement. Additionally, the joint F-test of both the Secchi variable and the interaction term with Secchi is significant (at 1% level), suggesting that a 1% increase in the value of Secchi on average increases home prices by 0.18% within 100 m of lakes. In the same sense, both the price premium for being close to the lake and the impact of water quality on the proximity premium are smaller when moving farther from the lakeshore: For properties in the 100 to 300 m of lakes, a 1% increase in the value of Secchi leads to a 0.05% increase in home prices on average. These results are consistent with other spatial fixed-effect levels (e.g., census block group and county). Further analysis, reported in SI Appendix, suggests that percent increases in home prices are higher around lakes with greater Secchi depth. For properties located beyond the 300 m threshold, water quality has a positive but statistically insignificant effect on home values. A similar insignificant overall impact has been found in other studies (9, 31).
Table 2.
Estimation results for Secchi and chl-a
 Secchichl-a
0 to 100 m buffer
0.4473***
(0.0165)
0.7600***
(0.0331)
100 to 300 m buffer
0.0496***
(0.0065)
0.1283***
(0.0151)
ln(Secchi)
0.0076
(0.0117)
NA
ln(Secchi) × 0 to 100 m buffer
0.1673***
(0.0172)
NA
ln(Secchi) × 100 to 300 m buffer
0.0405***
(0.0072)
NA
ln(chl-a)NA
-0.0115**
(0.0057)
ln(chl-a) × 0 to 100 m bufferNA
-0.1047***
(0.0127)
ln(chl-a) × 100 to 300 m bufferNA
-0.0276***
(0.0052)
N746,424637,548
adj. R-squared0.65670.6662
Number of lakes1,6321,425
States4342
Significance: ***P < 0.01; **P < 0.05; *P < 0.1. Clustered SEs at census tract level are in parenthesis. Fixed effect is census tract by year. ln() refers to natural log of the variable.
The average value of Secchi depth in our sample is 2.61 m and the average home price within 100 m of lakes is $549,245. Calculating the implicit price of water quality at the average values, a 0.1 m improvement in water clarity, which is a 4% increase in the average value of Secchi depth, increases the home price by $3,681 on average. Correspondingly, the implicit price for a 0.1 m increase in Secchi results in a $764 increase for properties within 100-300 m of lakes.
The results suggest that chl-a also affects home values for properties near lakes (Table 2). A 1% decrease in chl-a leads to a 0.1% increase in home prices for properties within 100 m of lakes and 0.0391% increase in home prices for properties within 100 to 300 m of lakes. For the average concentration of chl-a of 14.52 µg/L for properties within 100 m of lakes, households are willing to pay $4,395 for a 1 µg/L (7%) decrease in the level of chl-a. Similarly, residents living in the 100-300 m buffer area of lakes are willing to pay $783 for the same reduction in the level of chl-a. Water quality measurements aside, the dummy variables on proximity to lakes in the Secchi and Chl-a equations indicate that properties located closer to lakes are more highly valued relative to those located beyond 300 m.
Our estimates are robust to various modeling choices and sample inclusion criteria. In particular, we have estimated models with fixed effects at different geographic levels, with different minimum number of water quality observations required for each lake, and different minimum observed transactions required for each lake; we have also considered repeated sales models, though data limitations hindered accurate estimation. We also explored the nonlinearity of the estimates in the robustness section with an alternative log-linear specification (SI Appendix, Table S11). In Fig. 2, we summarize our robustness checks by plotting point estimates for key coefficients obtained from each alternative specification, with the baseline estimate highlighted in red. These analyses demonstrate that our baseline results are largely robust to a series of reasonable modeling choices, as most alternative estimates remain within close range from our baseline specification. We also note that there is considerable heterogeneity of our estimates. We further investigate whether there is heterogeneity based on spatial (ecoregions, political boundary) and temporal units (over time) and, baseline level of water quality. Details of robustness checks and heterogeneity analyses can be found in SI Appendix.
Fig. 2.
Violin plot of Secchi depth results across all model specifications and robustness tests. The red dots and labels indicate results from the baseline model. Colored areas are used to highlight relative concentration of coefficient estimates in certain ranges. In this figure, log() refers to natural log of variable.

Capitalization Effects.

To understand the full benefits of water quality reflected in lake housing markets, we consider the combined effects of water quality improvements for all properties throughout the study area. Given the broad spatial coverage of our dataset, we evaluate capitalization effects for lake water quality at a national scale. To do so, we consider all properties surrounding lakes larger than 4 ha within the continental United States. These properties are divided into two buffer groups: those within 100 m and those between 100 and 300 m from the lakeshore, indicated by subscripts s and f, respectively. The total combined capitalization effects are:
α+θsPsns+α+θfPfnf×%Change
[2]
where α+θs and α+θf are the combined elasticity estimates for the water quality variables in the estimated models, Ps¯ and Pf¯ represent the last sale price or assessed value averaged for each buffer group at the lake level, ns and nf are the numbers of properties within the buffer group for each lake, and % Change is a percent change in water quality. The numbers of properties around each lake are derived from parcel data to incorporate all properties that may expect increases in property values rather than properties where sale data are available.
For a marginal change in water quality, if the shape of the hedonic price function remains unchanged and preferences, income, and technology remain constant, we can interpret the capitalization effects as marginal willingness to pay (33). However, what constitutes a marginal change in water quality is not well defined. For example, both 1% and 10% changes in Secchi depth for a lake with 1 m of clarity (1 cm and 10 cm, respectively) are small. Considering that the median Secchi depth value for our sample is 1.52 m, we evaluate the effects of 10% improvements of both Secchi and chl-a using average assessed market values and average last sale prices. Our results indicate capitalization effects for marginal changes in Secchi range from $9.22 to $11.09 billion (Table 3). Most of these benefits are from properties within 100 m of the lake: $8.09 billion compared to $2.99 billion for properties 100-300 m from the lakeshore, based on last sale prices. For chl-a, we find smaller capitalization effects that range from $4.33 to $5.21 billion. The estimates for Secchi and Chl-a are not directly comparable because they measure different aspects of lake water quality and in different units of measurement. However, one might expect smaller capitalized impacts from Chl-a relative to Secchi from 1% changes, all other factors equal, because Secchi is a summative measure that reflects other components of the water column beyond Chl-a concentrations.
Table 3.
Estimation results for Secchi and chl-a
 
Market value
(billion $)*
Last sale price
(billion $)
Marginal improvements
 10% increase in Secchi9.2211.09
 10% decrease in chl-a6.467.81
Pristine ecological condition
 2012 NLA Secchi25.18-
 2017 NLA Secchi26.67-
*
Overall impacts for all properties within 300 m of lakefront for all NHD lakes (n = 76,131) larger than 4 ha identified using PLACES parcel data.
Most recent market values and last sales prices from ZTRAX averaged at the lake level within a county for each buffer distance (0 to 100 m and 100 to 300 m).
A limitation of analyzing marginal improvements in water quality is that the resulting capitalization effects may underestimate the benefits of water quality policies, with intended outcomes that are nonmarginal in nature. Thus, to provide an upper bound, we consider the capitalization effects if all lakes were restored to pristine ecological conditions. We define pristine as the water quality conditions in “reference lakes” from the National Lakes Assessment (NLA), which provides water quality samples for an ecologically representative national sample of lakes. These reference lakes refer to the least disturbed lakes for each ecoregion and serve as the baseline by which the NLA gauges lake health within an ecoregion (34, 35). Using the 2012 and 2017 NLAs, we calculate the percent changes in Secchi depth required to shift the average water clarity in each ecoregion to its respective reference lake level by:
%Change=(Q¯eco,ref-Q¯eco)Q¯eco
[3]
where Q¯eco is the average Secchi depth for lakes in each ecoregion and Q¯eco,ref is the average Secchi depth for the reference lakes in each ecoregion. Thus, the percent change required to reach pristine conditions varies across ecoregions from 20 to 75%. One region required a −5% change, which we removed from the analysis. We obtain the capitalization effect by multiplying the percentage change of Secchi depth with the elasticity estimated from the baseline model reported in Table 2. In this scenario, we find capitalization effects ranging from $25.18 to $26.67 billion in assessed market values using the 2012 and 2017 NLAs, respectively. Because the changes in water quality in Eq. 3 can no longer be considered marginal, we cannot interpret these results in marginal willingness to pay terms. However, they do provide an upper-bound estimate of the value of water quality as capitalized in housing markets, which can be used to consider important economic insights. Furthermore, property taxes are a large source of government revenue that support local school systems, public safety, and other government programs. Thus, the capitalized benefits of improving water quality at the national level extend beyond gains to individual property owners and are relevant to policymakers, and this upper bound suggests the magnitude of the potential community benefits.

Discussion

The results of this study are qualitatively consistent with the findings from prior lake water quality hedonic studies that relied on evidence from relatively small spatial scales (e.g., refs. 30, 3640). Like those studies, we find clear evidence that households living near lakes have a positive willingness to pay for improvements and that the willingness to pay is greatest within 100 m of a lake. Secchi measures of water clarity, which are generally acknowledged to capture eutrophication and other harms from long-term nutrient loading, are strongly predictive of house prices. We find that a 0.1-m increase in Secchi depth leads to a $3,681 price premium for an average lakefront home located within 100 m of lakes. This estimate falls between the bounds of the regional literature, which estimates implicit prices as low as $850 in Northern Maine (22) and as high as $5,540 in Orange County, Florida (30). While Moore et al. (2020) (18) report a nationwide implicit price estimate of $4,354, their sample included only 113 lakes that may be more representative of popular lakes with more frequent and abundant lakefront sales. In contrast, we consider 1,632 lakes, and our results are robust to multiple sample requirements and model specifications. We construct different samples of lakes by restricting the number of water quality records and the number of housing transactions per lake and find that the results across our broadest sample of lakes are the most robust. Only in our most restrictive models that drastically reduce the number of lakes below 100 do we observe differences in the estimated effects, providing evidence of potential sample selection bias in broad-scale hedonic analysis with limited numbers of observations.
Given the strength of our results and broad spatial coverage, which includes lakes in 43 states, we conclude that water quality, as measured both by water clarity and algal concentration, is capitalized in lakefront property values. We find the largest effects on properties within 100 m from the lakeshore and no significant effects beyond 300 m. Prior studies have found similar decay in the effects of water quality moving away from the waterfront (31). The results allow us to provide national willingness-to-pay estimates for lake water quality improvements that are captured via capitalization into housing prices. These results provide previously missing estimates for benefit–cost assessment for regulatory and policy decision-making at the national scale. They can also be used to understand distributional impacts by revealing census tracts that experience higher-than-average damages from degraded water quality. Higher valued lakefront properties may suggest possible distributional impacts of an increase in house prices and socioeconomic characteristics (such as individual income and race). However, this individual information is not available and further research is needed to merge with mortgage data to explore distributional impacts deeper. It is important to note that the effects estimated in this study do not represent a comprehensive value of water quality. A comprehensive estimate would require considering a range of different use and nonuse values; benefits accruing to recreational users who live more than 300 m from lakes, who may value water quality indirectly for wildlife and habitat benefits; and preservation values. However, the benefits associated with esthetic and local recreational enjoyment captured in home prices are an important piece.
Our results are drawn from available national-scale ZTRAX dataset. Zillow’s ZTRAX program is shutting down in June 2023. In the absence of such ZTRAX, we foresee at least two options that can allow a researcher to accumulate national-level analysis. First, other companies provide similar housing datasets. However, while the price of such dataset varies, the cost can be substantial for a national-level dataset. Alternatively, a group of researchers could coordinate to collect and harmonize publicly available country-level housing data. This combined open-access dataset could be used to foster future research on housing markets. While our research might not be reproducible in the absence of ZTRAX dataset access, we believe our results are replicable with a similar dataset: Key coefficients will have the same sign and similar magnitude when estimated with a combined national-level dataset.
In building out the national benefits of protecting and improving surface water quality, the national hedonic model provides important insights that can be useful for informing homeowners and local community. The model shows that people pay for higher water quality through price premiums on properties located on and near lakes. Understanding the effects of water quality on properties located in close proximity to lakes is important for accurate regulatory and policy decisions and public support for lake protection and enhancement policies. These homeowners will be major beneficiaries and also make decisions that directly impact lake water quality. Owners of these properties make decisions on landscaping and water management that directly affect local water quality. The explicit monetization of the benefits of water quality can be used to educate and influence homeowners to take actions such as reducing lawn fertilizer or maintaining riparian buffers. The empirical results can be used to illustrate to property owners that it is in their own financial interest to take actions to protect lake water quality and thereby their property values. Additionally, property taxes provide an important source of revenue, particularly to support primary and secondary education, in communities, and this link to hedonic model results can be used to provide a financial incentive at the community level to support water quality protection upstream in watersheds. Collectively, these insights demonstrate that hedonic models of lake water quality can be a powerful tool in policy analyses.
In aggregate, the benefits calculated in this study can lay the groundwork for national policies designed to protect and restore lakes. Our capitalization results show that even the modest policies aimed at 10% improvement in lake water quality have substantial impact on housing prices, ranging from over $6 billion for chl-a to over $9 billion for Secchi depth. The potential benefits increase in relation to large, policy-relevant, changes in water quality. We find that capitalized increases in property values at the national level could be in the range of $25 to 27 billion to return all lakes to pristine ecological conditions.

Data, Materials, and Software Availability

We used publicly available water quality data and proprietary housing data. We provided code in the SI Appendix that can be used to download water quality data and perform analyses. But we cannot provide property data that we obtained from Zillow Inc. as the data sharing has restriction (21). More information on accessing the housing data can be found at http://www.zillow.com/ztrax.

Acknowledgments

We thank David Keiser, Tihitina Andarge, PNAS editor and two anonymous reviewers, and USEPA reviewers for their valuable comments and suggestions in preparing the manuscript. We acknowledge the Minnesota Supercomputing Institute at the University of Minnesota for providing resources that contributed to the research results reported within this paper. URL: http://www.msi.umn.edu. We thank Zillow, Inc. for having made Zillow Transaction and Assessment Dataset (ZTRAX) available free of charge for US academic, nonprofit, and governmental researchers. More information on accessing the data can be found at http://www.zillow.com/ztrax. Christoph Nolte acknowledges support from the Department of Earth & Environment at Boston University, the Junior Faculty Fellows program of Boston University's Hariri Institute for Computing and Computational Science, and The Nature Conservancy. Parts of an earlier version of this paper written by this author team were used as part of coauthor Jiarui Zhang’s dissertation. The results and interpretations are those of the authors and do not reflect the position or policies of Zillow Group, U.S. Environmental Protection Agency, and the participating universities.

Author contributions

S.M., A.C.-C., K.S., J.Z., K.J.B., D.C., C.L.K., C.N., M.P., D.P., and S.P. designed research; S.M., A.C.-C., K.S., J.Z., K.J.B., D.C., C.L.K., C.N., M.P., D.P., and S.P. performed research; S.M., A.C.-C., K.S., J.Z., and C.N. analyzed data; and S.M., A.C.-C., K.S., J.Z., K.J.B., C.L.K., D.P., and S.P. wrote the paper.

Competing interests

The authors declare no competing interest.

Supporting Information

Appendix 01 (PDF)

References

1
D. A. Keiser, J. S. Shapiro, Consequences of the clean water act and the demand for water quality. Q. J. Econ. 134, 349–396 (2019).
2
D. A. Keiser, C. L. Kling, J. S. Shapiro, The low but uncertain measured benefits of US water quality policy. Proc. Natl. Acad. Sci. U.S.A. 116, 5262–5269 (2019).
3
K. J. Boyle, M. J. Kotchen, V. K. Smith, Deciphering dueling analyses of clean water regulations. Science 358, 49–50 (2017).
4
D. A. Keiser et al., A water rule that turns a blind eye to transboundary pollution. Science 372, 241–243 (2021).
5
United States Environmental Protection Agency (USEPA), The benefits and costs of the clean air act from 1990. https://www.epa.gov/sites/default/files/2015-07/documents/fullreport_rev_a.pdf. Accessed 15 June 2022.
6
United States Environmental Protection Agency (USEPA), The benefits and costs of the clean air act, 1970 to 1990–Retrospective study. Accessed 15 June 2022.
7
K. Y. Chay, M. Greenstone, Does air quality matter? Evidence from the housing market. J. Polit. Econ. 113, 376–424 (2015).
8
J. Currie, R. Walker, What do economists have to say about the clean air act 50 years after the establishment of the environmental protection agency? J. Econ. Perspect. 33, 3–26 (2019).
9
D. Wolf, H. A. Klaiber, Bloom and bust: Toxic algae’s impact on nearby property values. Ecol. Econ. 135, 209–221 (2017).
10
S. Newbold et al., Benefit transfer challenges: Perspectives from U.S. practitioners. Environ. Resour. Econ. 69, 467–481 (2018).
11
J. Corona et al., An integrated assessment model for valuing water quality changes in the United States. Land Econ. 96, 478–492 (2020).
12
J. Rolfe, J. Windle, R. J. Johnston, “Applying benefit transfer with limited data: Unit value transfers in practice” in The Economics of Non-Market Goods and Resources (Springer, Dordrecht, 2015), pp. 141–162.
13
D. Guignet, M. T. Heberling, M. Papenfus, O. Griot, Property values, water quality, and benefit transfer: A nationwide meta-analysis. Land Econ. 98, 191–218 (2021).
14
K. Moeltner, J. A. Balukas, E. Besedin, B. Holland, Waters of the United States: Upgrading wetland valuation via benefit transfer. Ecol. Econ. 164, 106336 (2019).
15
R. J. Johnston, E. Y. Besedin, R. Stapler, Enhanced geospatial validity for meta-analysis and environmental benefit transfer: An application to water quality improvements. Environ. Resour. Econ. 68, 343–375 (2017).
16
R. T. Carson, R. C. Mitchell, The Value of clean water: The public’s willingness to pay for boatable, fishable, and swimmable quality water. Water Resour. Res. 29, 2445–2454 (1993).
17
D. A. Keiser, The missing benefits of clean water and the role of mismeasured pollution. J. Assoc. Environ. Resour. Econ. 6, 669–707 (2019).
18
M. R. Moore, J. P. Doubek, H. Xu, B. J. Cardinale, Hedonic price estimates of lake water quality: Valued attribute, instrumental variables, and ecological-economic benefits. Ecol. Econ. 176, 106692 (2020).
19
P. A. Soranno et al., LAGOS-NE: A multi-scaled geospatial and temporal database of lake ecological context and water quality for thousands of US lakes. Gigascience 6, 1–22 (2017).
20
National Water Quality Monitoring Council, Water quality portal. https://www.waterqualitydata.us. Accessed 2 March 2021.
21
Zillow, Zillow transaction assessor dataset 2019 Q4. http://www.zillow.com/ztrax. Accessed 19 May 2021.
22
C. Nolte, High-resolution land value maps reveal underestimation of conservation costs in the United States. Proc. Natl. Acad. Sci. U.S.A. 117, 29577–29583 (2020).
23
C. Nolte et al., Studying the impacts of environmental amenities and hazards with nationwide property data: Best data practices for interpretable and reproducible analyses. SSRN (2021), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3900806.
24
M. Gindelsky et al., “Valuing housing services in the era of big data: A user cost approach leveraging Zillow microdata" in "Big data for 21st Century Economic Statistics,” K. G. Abraham, R. S. Jarmin, B. C. Moyer, M. D. Shapiro, Eds. (University of Chicago Press, Chicago, IL, 2021).
25
United States Geological Survey, USGS National Hydrography Dataset (NHD). https://www.sciencebase.gov/catalog/item/4f5545cce4b018de15819ca9. Accessed 2 March 2021.
26
L. A. Sprague, G. P. Oelsner, D. M. Argue, Challenges with secondary use of multi-source water-quality data in the United States. Water Res. 110, 252–261 (2017).
27
B. Moss et al., The determination of ecological status in shallow lakes—a tested system (ECOFRAME) for implementation of the European Water Framework Directive. Aquat. Conserv. Mar. Freshw. Ecosyst. 13, 507–549 (2003).
28
M. Søndergaard, E. Jeppesen, J. P. Jensen, S. L. Amsinck, Water framework directive: Ecological classification of Danish lakes. J. Appl. Ecol. 42, 616–629 (2005).
29
Federal Housing Finance Agency, House price index datasets. Available at https://www.fhfa.gov/DataTools/Downloads/Documents/HPI/HPI_PO_monthly_hist.xls. Accessed 24 Jun 2021.
30
P. J. Walsh, J. W. Milon, D. O. Scrogin, The spatial extent of water quality benefits in urban housing markets. Land Econ. 87, 628–644 (2011).
31
P. Walsh, C. Griffiths, D. Guignet, H. Klemick, Modeling the property price impact of water quality in 14 chesapeake bay counties. Ecol. Econ. 135, 103–113 (2017).
32
J. Zhang, D. Phaneuf, B. Schaeffer, Property values and cyanobacterial algal blooms: Evidence from satellite monitoring of inland lakes. Ecol. Econ. 199, 107481 (2022).
33
N. V. Kuminoff, J. C. Pope, Do “capitalization effect” for public goods reveal the public’s willingness to pay? Int. Econ. Rev. (Philadelphia) 55, 1227–1250 (2014).
34
United States Environmental Protection Agency (USEPA), National Lakes Assessment 2012: A Collaborative Survey of Lakes in the United States. https://www.epa.gov/sites/default/files/2016-12/documents/nla_report_dec_2016.pdf. Accessed 30 September 2021.
35
United States Environmental Protection Agency (USEPA), National Lakes Assessment 2017: A Collaborative Survey of the Lakes in the United States. Available at https://www.epa.gov/national-aquatic-resource-surveys/data-national-aquatic-resource-surveys. Accessed 30 September 2021.
36
H. J. Michael, K. J. Boyle, R. Bouchard, Does the measurement of environmental quality affect implicit prices estimated from hedonic models? Land Econ. 76, 283–298 (2000).
37
J. P. Gibbs, J. M. Halstead, K. J. Boyle, J.-C. Huang, An hedonic analysis of the effects of lake water clarity on new hampshire lakefront properties. Agric. Resour. Econ. Rev. 31, 39–46 (2002).
38
W. Weng et al., Coupling natural and human models in the context of a lake ecosystem: Lake Mendota, Wisconsin, USA. Ecol. Econ. 169, 106556 (2020).
39
T. Liu, J. J. Opaluch, E. Uchida, The impact of water quality in Narragansett Bay on housing prices. Water Resour. Res. 53, 6454–6471 (2017).
40
P. J. Walsh, J. W. Milon, Nutrient standards, water quality indicators, and economic benefits from water quality regulations. Environ. Resour. Econ. 64, 643–661 (2016).

Information & Authors

Information

Published in

Go to Proceedings of the National Academy of Sciences
Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 120 | No. 15
April 11, 2023
PubMed: 37011190

Classifications

Data, Materials, and Software Availability

We used publicly available water quality data and proprietary housing data. We provided code in the SI Appendix that can be used to download water quality data and perform analyses. But we cannot provide property data that we obtained from Zillow Inc. as the data sharing has restriction (21). More information on accessing the housing data can be found at http://www.zillow.com/ztrax.

Submission history

Received: June 24, 2022
Accepted: February 17, 2023
Published online: April 3, 2023
Published in issue: April 11, 2023

Keywords

  1. Value of water
  2. Property values
  3. Hedonic pricing model
  4. Lake water quality
  5. National study

Acknowledgments

We thank David Keiser, Tihitina Andarge, PNAS editor and two anonymous reviewers, and USEPA reviewers for their valuable comments and suggestions in preparing the manuscript. We acknowledge the Minnesota Supercomputing Institute at the University of Minnesota for providing resources that contributed to the research results reported within this paper. URL: http://www.msi.umn.edu. We thank Zillow, Inc. for having made Zillow Transaction and Assessment Dataset (ZTRAX) available free of charge for US academic, nonprofit, and governmental researchers. More information on accessing the data can be found at http://www.zillow.com/ztrax. Christoph Nolte acknowledges support from the Department of Earth & Environment at Boston University, the Junior Faculty Fellows program of Boston University's Hariri Institute for Computing and Computational Science, and The Nature Conservancy. Parts of an earlier version of this paper written by this author team were used as part of coauthor Jiarui Zhang’s dissertation. The results and interpretations are those of the authors and do not reflect the position or policies of Zillow Group, U.S. Environmental Protection Agency, and the participating universities.
Author Contributions
S.M., A.C.-C., K.S., J.Z., K.J.B., D.C., C.L.K., C.N., M.P., D.P., and S.P. designed research; S.M., A.C.-C., K.S., J.Z., K.J.B., D.C., C.L.K., C.N., M.P., D.P., and S.P. performed research; S.M., A.C.-C., K.S., J.Z., and C.N. analyzed data; and S.M., A.C.-C., K.S., J.Z., K.J.B., C.L.K., D.P., and S.P. wrote the paper.
Competing Interests
The authors declare no competing interest.

Notes

This article is a PNAS Direct Submission.

Authors

Affiliations

Department of Applied Economics, University of Minnesota, St. Paul, MN 55108
The Natural Capital Project, University of Minnesota, St. Paul, MN 55108
Natural Resources Research Institute, University of Minnesota–Duluth, Duluth, MN 55811
Adriana Castillo-Castillo https://orcid.org/0000-0002-5353-0668
Department of Applied Economics, University of Minnesota, St. Paul, MN 55108
Department of Agricultural and Applied Economics, Virginia Tech, Blacksburg, VA 24061
Jiarui Zhang
Department of Agricultural and Applied Economics, University of Wisconsin Madison, Madison, WI 53706
Kevin J. Boyle
Blackwood Department of Real Estate, Virginia Tech, Blacksburg, VA 24061
Department of Agricultural Economics, Purdue University, West Lafayette, IN 47907
Dyson School of Applied Economics and Management, Cornell University, Ithaca, NY 14853
Atkinson Center for a Sustainable Future, Cornell University, Ithaca, NY 14853
Department of Earth & Environment, Boston University, Boston, MA 02215
Faculty of Computing & Data Sciences, Boston University, Boston, MA 02215
Environmental Protection Agency, Corvallis, OR 97333
Daniel Phaneuf
Department of Agricultural and Applied Economics, University of Wisconsin Madison, Madison, WI 53706
Department of Applied Economics, University of Minnesota, St. Paul, MN 55108
The Natural Capital Project, University of Minnesota, St. Paul, MN 55108
Department of Ecology, Evolution & Behavior, University of Minnesota, St. Paul, MN 55108

Notes

1
To whom correspondence may be addressed. Email: [email protected].

Metrics & Citations

Metrics

Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.


Citation statements




Altmetrics

Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

View Options

View options

PDF format

Download this article as a PDF file

DOWNLOAD PDF

Get Access

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Personal login Institutional Login

Recommend to a librarian

Recommend PNAS to a Librarian

Purchase options

Purchase this article to get full access to it.

Single Article Purchase

Valuing water quality in the United States using a national dataset on property values
Proceedings of the National Academy of Sciences
  • Vol. 120
  • No. 15

Media

Figures

Tables

Other

Share

Share

Share article link

Share on social media