**3. Results**

A weighted tabulation and corresponding percentage of too high/other responses for select variables are presented in Table 2 which provides descriptive information about respondents. Tabulations and percentages were calculated for these variables because they are discrete and non-ordinal, and therefore, will be included as dummy variables in the regression analysis that follows. Most households (63.3%) reported that their water bills were about right or should be higher. Approximately 36.7% reported their water bills were too high. Several demographic and socio-economic factors impacted the perceptions of water bills. Females were more likely to indicate their water bills were too high, as were racial/ethnic minorities. Blacks, Hispanics, and respondents identifying as some other race (e.g., Native American, Native Hawaiian or Pacific Islander, or Middle Eastern) were more likely to indicate their water bills were too high. People with lower levels of educational attainment were also more likely to report that their water bills were too high. In particular, people without a high school education were the most likely to report that their water bills were too high. Relatedly, people with incomes under USD 50,000 reported feeling water bills were too high.


**Table 2.** Water Cost Perceptions and Household Demographic and Socio-Economic Characteristics.

Table 3 is similar in layout to Table 2 and presents additional information about other household characteristics including geographic location, employment status, and health insurance coverage according to water cost perception responses. The table also presents policy related information such as water bill frequency and respondents' enrollment in water payment programs. The table suggests there are regional differences in the percentage of respondents who felt their water bills were too high. The Pacific Northwest (40.9%), Southeast Michigan (51.9%), and Southern California (40.8%) were regions where the largest proportion of individuals reported their water bills were too high. Regions where most respondents said their water bills were about right or too low are in the Piedmont Atlantic (69.2%), the Mid-Atlantic (69.9%), and the Sun Corridor (67.1%). Some respondent and household characteristics had a much higher rate of reporting their water bills were too high. Some individuals reported their water bills were too high at a greater rate than the overall survey rate of 36.7% including those on Medicaid (49.0%), with no health insurance (46.5%), the unemployed (47.9%), and living in a mobile home/trailer (47.3%).

Table 4 presents the logistic regression results that help us understand which of the variables presented in Table 1 are explanatory variables of water bill perceptions, even after controlling for these factors simultaneously. Overall, income, geographic location, and race explained whether individuals considered their water bills to be too high. Compared to Whites, Black, Asian, and Hispanic individuals were more likely to perceive their water bill charges as too high: Hispanic respondents were 27.4% more likely to report water bills were too high, Black respondents were 43.8% more likely, and Asians were 32.1% more likely.

Income was also a strong indicator of whether respondents felt water bills were too high. Respondents in the lowest income bracket, making less than USD 50,000 per year were approximately 2.3 times more likely to report their water bills were too high compared to those making over USD 100,000 per year. Individuals in households making between USD 50,000 and USD 100,000 were approximately 75% more likely to report their water bills were too high compared to those making over USD 100,000.

There were also statistically significant geographic trends in water bill perceptions. Compared to the Piedmont Atlantic region, four regions were statistically more likely to have respondents that perceived their water bills to be too high. In Eastern Massachusetts, respondents were 45.2% more likely to report water bills were too high. In Southern California, respondents were 63.8% more likely to indicate that water bills were too high. In Southeast Michigan respondents were 2.59 times more likely to indicate they were billed

too much for water, while in the Pacific Northwest, respondents were 70% more likely to indicate their water bills were too high. From a water provider perspective, two significant variables are particularly interesting. Billing frequency and enrollment in a water payment assistance program were significant explanatory factors behind perceptions of water bills. Households on a quarterly schedule for water bill payments were 18% more likely to consider their water bills to be too high. However, respondents indicating they paid their water bills annually or had their water included in their homeowners' association (HOA) fees were approximately half as likely to indicate they perceived their water bills as too high. Households enrolled in a payment assistance program were about 26% less likely to perceive their water bills to be too high.


**Table 3.** Water Cost Perceptions and Other Household Characteristics.


**Table 4.** Logistic Regression Results: Odds Ratios for Factors affecting Perceptions of Water Bills.

Note: \* *p* < 0.1; \*\* *p* < 0.05; \*\*\* *p* < 0.01. N = 6198 F-statistic = 9.305 \*\*\*.

Table 5 presents information from the U.S. Census Bureau and the Environmental Protection Agency's Environmental Quality Index (EQI) [38] for each of the regions to provide context to the regression results. These data correspond to the counties containing the city pairs of interest in each region, as listed earlier in Table 1. Social and demographic information comes from the U.S. Census Bureau's American Community Survey 2015– 2019 [39]. The EQI index presents a county-level ranking of overall environmental quality according to five categories: air, water, land, built, and sociodemographic environments across the U.S [38]. Table 5 includes a measure of the total overall ranking, as well as the water subset of the EQI. Low rankings represent lower levels of degradation. The rankings are based on percentiles across U.S. counties as follows: lowest (0–5th percentile); very low (5th–20th percentile); low (20th–40th percentile); moderate (40th–60th percentile); high (60th–80th percentile); very high (80th–95th percentile); highest (95th–100 percentile). In Table 5, the regions are divided into two groups according to the previous regression results: regions where respondents were less likely to perceive their water bills to be too high and regions that were more likely to perceive their water bills to be too high.

The regions more likely to say their water bills are too high have on average, a higher percentage of individuals with a high school education or below (38.24% compared to 31.3%), a higher non-White population (43% compared to 30%), and higher population densities. Median household income and poverty levels were similar for both. Regions with a higher percentage of households more likely to say their water bills are too high are located in counties with a ranking of water quality problems ranging from high to highest (75% for regions more likely to report bills too high compared to 50% for those less likely). These regions also have higher levels of environmental degradation (37.5% compared to 30%). Therefore, respondents from regions that perceived their water bills to be too high are more likely to live in areas of lower water and lower environmental quality.

Table 6 presents tabulations of survey questions for respondents who felt their water bills were too high, which provide important contextual information about respondents' experiences with water and utilities (e.g., water and electricity). Based on the information presented in this table, the majority of households who perceive their water bills to be too high worry about the cost of water and are less likely to feel they can easily afford their water bills. Only 44.7% of these households reported they could easily afford their water bills and 81.2% say they worry about the cost of water. However, a lower percentage of these same respondents have had prior experience with utility affordability issues. Of the respondents who indicated their water bills were too high, just over a third had experienced prior restrictions on water use (32.8%) or had received a water (32.2%) or electric shutoff notification (36.3%);23.3 percent and 20.5 percent had experienced a water or electric shutoff respectively.

Interestingly, these views and experiences do not appear to have impacted respondents' trust in public institutions at the time of the survey. Table 7 presents tabulations of survey questions about trust in public institutions, which may be a driver of water bill perceptions; households with low trust in institutions may be more likely to perceive water bills to be too high. The table indicates however, that the majority of respondents felt confident in institutions such as their local water utility (62%), flood control district (54.6%) and public health agencies (58.5%). A somewhat lower percentage of respondents felt confident about their city/town government (50.6%).





**Table 5.**

*Cont*.


Note: Data sources: U.S. Census Bureau 2015–2019 County Level Estimates [39] and the U.S.

Environmental

 Protection Agency's

Environmental

 Quality Index [38].


**Table 6.** Contextual Survey Questions Related to Experiences with Water Services.

Note: Affirmative responses (yes or agree) are presented in bold in this table.

**Table 7.** Survey Questions Related to Trust in Institutions.


#### **4. Discussion**

The United States is in an era of infrastructure replacement, which will require massive investments totaling an estimated USD 600 billion towards water infrastructure over the next two decades [3]. These investments, along with shutoffs in water service in several cities across the United States and the Flint water crisis, suggest that trust in water service and also the perceptions of water services are important to analyze at this juncture in history. Aside from consumer reactions to water costs, the perceptions of these costs are also important for water utilities to bear in mind since a sizable customer base that considers water bills to be too high may lead to the inability or unwillingness to pay for water services. It may also cause consumers to switch to alternate water sources, such as private wells or bottled water, which could erode the revenue streams of utilities [40]. Combined, these coping strategies may erode the long-term customer base of utilities and public engagement in local water policy decisions [41]. To this point in time, however, studies of water perceptions in the developed world have assessed dimensions of water services (e.g., quality and willingness to pay) other than perceptions of water costs. To address this research gap, the goal of this paper was to analyze the perceptions of households regarding the cost of water services and to assess the characteristics of households who felt their water bills were too high.

Not surprisingly, income was one of the more important factors in explaining water bill perceptions. Households making less than USD 50,000 were more likely to feel that their water bills were too high. Even after controlling for income, race was also a significant factor behind households' perceptions of water bills. Non-white, minority households were more likely to perceive that their water bills were too high. This finding is in line with recent research, which finds high water costs disproportionately affect communities of color [42,43]. Studies suggest that these high costs are a result of population decline in urban areas and postindustrial divestment [42]. It may also reflect the fact that Black and Hispanic neighborhoods are at higher risk for water shutoffs due to non-payment than predominantly White neighborhoods [29].

Another important finding of this study was variations in household perceptions across particular regions of the country. Households in four regions of the country (e.g., Eastern Massachusetts, Southern California, Southeast Michigan, and Pacific Northwest) were more likely to perceive water bills as being too high. This may reflect the higher cost of living in three of these areas of the country (Eastern Massachusetts, Southern California, Pacific Northwest). In Southeast Michigan, which includes the cities of Flint and Detroit, these results may reflect consumer awareness of shutoffs in Detroit and also rising water rates in these cities [34,35].

Model results also indicated that the frequency of billing affects perceptions of water bills. Respondents billed quarterly were more likely to consider their water bills to be too high compared to customers billed monthly or annually. Therefore, one recommendation based on these findings is for water companies to bill monthly, which prior work indicates helps household budget their money better [44]. Alternatively, companies could also bill households annually, allowing for customers to easily anticipate this one-time annual payment without focusing on water costs for the rest of the year. Another important result was that water payment programs reduced the likelihood that households perceived their water bills to be too high. This finding suggests that water providers should work to establish water assistance programs for customers in need. At present there is no federal framework guiding the implementation of customer assistance programs (CAPs) [45], which provides utilities with a good deal of flexibility in structuring these programs. Types of CAPs that may be offered range from water efficiency programs to bill discounts to lifeline rates [46]. Important considerations in CAP design that influence program cost include the program size and the type of assistance offered [46]. State laws governing utility regulation and the wording and interpretation of state statutes are also important considerations to keep in mind when designing programs because the legal barriers to CAPs do vary across states and utility type [8]. If a utility already has a CAP in place, providers may want to develop outreach programs to communicate with customers and enhance their awareness of CAPS.

That said, it is important to acknowledge some limitations of the present study. One, the SWISSH survey contains several questions that ask about pollution in nearby water bodies and experiences with water pollution. These questions were not incorporated into the logit model because pollution is not related to the primary topic of this paper. In addition, as noted previously in the introduction to this paper, research on public perceptions of the quality of local water resources finds that perceptions of the quality of local water resources are based predominantly on organoleptic properties such as taste and turbidity, which are not based on measurable safety or water quality metrics [23,47]. Studies also show the perceived risk of local water resources is strongly associated with perceived (not necessarily measurable levels) of chemicals in water, external information, past health problems, and trust in water suppliers [47]. Two, the SWISSH survey does ask respondents to estimate the amount of their last water bill. We elected not to include this information in the models because prior research has indicted that biases in responses are likely to arise related to recall problems [48]. Studies have also found that consumers do not have an accurate understanding about how their water bills are calculated or how much they pay for water services [30]. There is also no nationwide data available in the United States about water rates to use in place of survey data. The American Water Works Association (AWWA) has a survey, but it is only for AWWA member utilities and is not representative of all utilities across the nation. The University of North Carolina also provides some rate data [49], but the coverage of these data is not national. It is also prohibitive from a time and financial perspective to collect rate data for the nation as a whole. Collecting this information would require collaboration with thousands of water providers. Harmonizing these data would also be quite complex because utilities use different pricing strategies for water in the United States, which contributes further to the infeasibility of creating a nationwide water rate database. A third limitation of this study is that it does not control for household water use, which could impact the amount of water bills and also perceptions of water bills. To control for this, water usage data would need to be acquired from individual utilities which may be infeasible because of privacy concerns for customers. It is also not feasible to acquire usage data with the same coverage as the SWISSH survey.

The limitations of this paper present several opportunities for future research that expand on the present study. One, future work could collect information about water costs, water use, and survey data about perceptions of water costs, based on those provided by the SWISSH survey, to understand the linkages between water cost, water use, and perceptions of water costs. Acquiring these data would also require the cooperation of a utility and would require them to solicit information from customers. There may be privacy risks to customers in acquiring these data, however. Thus, the feasibility of this research path is questionable. Two, future work could collect information about actual water costs from customer bills, and pair these data with survey data from customers about their estimated costs of water. This would be useful in understanding the extent customers are aware of the actual cost of their water use and their recall accuracy. Again, the privacy risk to customers and the time burden this may place on utilities may render this research path unfeasible. Third, the results of our study suggest that news coverage about water issues may explain geographic differences in household perceptions of water costs, particularly in Southeast Michigan, that includes the cities of Flint and Detroit, which has received a lot of national news coverage related to water shutoffs and water rate increases [34,35]. Future work could test the extent that news coverage creates bias in household perceptions of water costs by collecting times series information about water rates from individual providers, survey data about customer perceptions of water trends and news reports from the media about water issues. This type of survey design has fewer data privacy risks for individual consumers, but is risky because the data collection would be time intensive and require a knowledgeable team of personnel which could also be quite costly.

#### **5. Conclusions**

This study provided the first examination of household perceptions of water costs across nine geographically, demographically, and socioeconomically diverse regions of the United States. In doing so, our study advances the water and public policy literature in three ways. One, it collected one-of-a-kind survey data to address the need for household resolution information about water issues given the absence of data at this scale in the United States. Two, it incorporated these one-of-a-kind survey data into logistic regression models to understand the drivers of household perceptions of water costs. Three, we assessed the impact of proposed solutions to improve water affordability on household perceptions of water costs. Model results indicated low-income and households in underrepresented groups, such as racial and ethnic minorities, were more likely to perceive their water bills to be too high. The perception of water costs also varied geographically. From a policy perspective, model results indicate utilities can positively affect perceptions of water bills via the frequency of water billing and provision of payment assistance programs. Utilities could also use the information from the survey and model results to focus outreach and communication activities to customers who feel their water bills are too high. As water utilities and city governments navigate the conflicting objectives of maintaining and upgrading water systems at prices that are affordable for a majority of water users, communication with customers will be key to maintaining good relationships during this period of change and adaptation.

**Author Contributions:** Conceptualization, L.M. and E.A.M.; methodology, L.M.; formal analysis, L.M.; resources, E.A.M.; data curation, L.M.; writing—original draft preparation, L.M. and E.A.M.; writing—review and editing, L.M. and E.A.M.; supervision, E.A.M.; project administration, E.A.M.; funding acquisition, E.A.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by National Science Foundation Grant Number 1444758 and Supplement Number 1444758. The APC was funded by Michigan State University's Department of Geography, the Environment and Spatial Sciences and Laura Medwid.

**Institutional Review Board Statement:** The study was approved and deemed exempt in accordance with federal regulations by the Institutional Review Board (IRB) of Michigan State University (IRB# x16-579e 16 April 2016) for studies involving human subjects.

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Data for this study may be accessed via the Harvard Dataverse https: //doi.org/10.7910/DVN/0ETQ74 (accessed on 24 June 2021).

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


**Jie Zhang 1, Chong Meng 2, Shugang Hu 1,\* and Wei Li 2,\***

+86-139-1012-3723(W.L.)


**Abstract:** The surface water ecosystem has important ecological value and plays an important supporting and guarantee role in the sustainable development of human society. In this study, an inexact two-stage stochastic programming (ITSP) model was developed for supporting water resource allocation for the four main water sectors (industry, municipal, agriculture, and ecological environment). Several scenarios corresponding to different flow patterns, which reflect different probabilities of water resource availability and environmental carrying capacity, were examined. On the basis of traditional water resource allocation, this model adds consideration of ecological value factors, which is conducive to the synergistic efficiency of socio-economic and ecological water consumption. Results revealed that the water resource carrying capacity, ecological value factors, and water environmental capacity are the main factors affecting the optimal allocation of water resources. Furthermore, the optimal allocation scheme for water resources coupled with ecological value factors were determined to realize the coordinated development of social economic benefits and ecological benefits. The current study findings are of great significance for establishing a rational water resource management system for water resource exploitation and utilization. This model can be used to guide various departments in Dalian to formulate an optimal water resources allocation scheme by considering ecological value factors, and provide a basis for realizing the coordinated development of Dalian's socio-economic development goals, water resource utilization, and environmental quality improvement.

**Keywords:** inexact two-stage stochastic programming; water management; ecological value factor; water resource allocation
