Devils (and angels) in the details, Part 3
In early January the California Water Board released its draft proposal for a statewide low-income water bill assistance program. My last couple posts summarized the proposal and discussed the perverse incentives for rate design that the subsidy might create. Continuing the theme of unintended consequences, this one looks at what the statewide assistance program might mean for poor-performing small water systems.
All else equal, when it comes to water systems, bigger is better. Water and sewer systems require big infrastructure investments and lots of operating costs to work. When those costs can be spread across large numbers of people and businesses, the average price of water is lower. That makes utilities textbook examples of economies of scale.
Less obviously, the organizations that operate utilities also are subject to economies of scale. Larger organizations can hire and retain higher-quality employees and maintain specialized personnel. As my past research work with David Switzer has shown, these advantages help larger water systems take advantage of human capital in ways that smaller systems can’t.
Larger systems also enjoy significant financial advantages. Larger customer bases tend to be more economically diverse, providing a degree of financial stability to a utility. Financial markets favor large utilities, too, allowing them access to finance capital at lower interest rates than their smaller cousins.
Thing is, just as in the rest of the United States, most of California’s water utilities are really, really small. More than three quarters of California’s 2,800 water systems serve fewer than 3,300 people, but collectively serve just 2% of the state’s population. Meanwhile, the 183 largest systems serve 80% of the state’s population.
Small system struggles
Unfortunately, small systems tend to suffer disproportionately from poor water quality. Here’s the correlation* between annual Safe Drinking Water Act health violations and utility size in California, based on EPA data from 2010-2018:
Affordability also correlates with system size. So do poverty, income, and unemployment. My recent analysis of water and sewer rates nationwide shows that, all else equal, both AR20 and HM decline as utilities increase in size—that is, on average, water and sewer affordability gets better as utilities get bigger.
That small systems struggle is no secret, and California has in recent years sought to encourage system consolidation through a variety of carrots and sticks. The process is slow, however, and fraught with political conflict. Consolidation is not a panacea, and there are plenty of struggling large and medium-sized systems. But the data are clear on this point: on average, with water utilities, bigger is better.
Props and levers
What does all this have to do with low-income assistance? The proposed program would transfer significant resources from relatively affluent to relatively poor communities. In many cases, those communities are served by the small, perennially underperforming utilities that too often provide lousy water quality at very high prices. Financial pressures from ratepayers are among the strongest incentives for small utilities to consolidate. A low-income assistance program will disproportionately benefit customers of those systems (which is good), but could inadvertently prop up systems that would otherwise move toward consolidation due to financial pressures (which is bad). To some failing small systems, low-income assistance could be financial life support system.
But if structured correctly and implemented carefully, low-income water assistance could help drive small system consolidation. Financial assistance for low-income customers could make consolidation more attractive for the larger utilities and private firms that might otherwise be reluctant to take on the responsibility for troubled small systems with less affluent customers. More directly, the state could make assistance funds contingent on consolidation or compliance with SDWA management requirements. Statewide assistance might be a potent lever to push recalcitrant small systems toward consolidation and its blessings.
That’s why statewide low-income water bill assistance should not be considered in isolation, but rather as part of a comprehensive strategy to improve water systems. The same applies to any plan for a national water affordability assistance program.
Whether a statewide assistance program would prop up failing small systems or lever them into consolidation and sustainability will depend in large part on the administrative structures and processes used to implement the program. I’ll take up those administrative angels and devils in my next post.
*Fitted with a negative binomial regression.
†There are probably another dozen or so studies that I could link here.
A reasonable expectation
What low-income households pay for essential service in the United States
Over the past 18 months I’ve been working to develop a valid, generalizable depiction of water and sewer affordability in the United States. There’s currently no nationally-representative dataset on water and sewer rates, so my Texas A&M research assistant Jake Metzler and I painstakingly gathered water and sewer rates data from an original, nationally-representative sample of American water utilities.* We then calculated basic single-family residential water and sewer service prices and developed affordability estimates for each utility using the double-barreled methods I published earlier this year.
All that effort is now ready to bear fruit. A forthcoming AWWA Water Science article reports the full methodology and findings in gory academic detail. If there’s interest here, I’ll share more about our findings here over the next few weeks and months – without the academic gore.
Today I’ll start by sketching the basic picture: how we measured affordability, and how affordable water and sewer services are in the US.
The sample & data
We drew our sample from the EPA’s Safe Drinking Water Information System (SDWIS), which SDWIS contains basic system information for the country’s nearly 50,000 water systems. An overwhelming majority of systems serve fewer than 3,300 people, and so a simple random selection would result in a sample of mostly small systems that collectively serve a tiny minority of the total population. So we stratified the sample—that is, we intentionally “oversampled” different kinds of utilities in order to get a wide variety of systems. Our analysis then adjusted mathematically for the sampling procedure to make sure we get truly representative results.
Single-family residential rates data were collected in the summer of 2017. To call the process laborious is an understatement (I’ll probably blog about that some other time!). For most large and medium-sized utilities, rates data were readily available on websites. When rates were unavailable online—especially common among smaller utilities—we inquired by telephone, repeating calls as necessary. Efforts to collect rates data were abandoned after multiple refusals or non-responses. The final dataset included full water and sewer rates data for 329 utilities. That might not seem like much, but it’s a larger sample than most, it’s representative, and we’re highly confident in the validity of the data.
Basic water & sewer as share of disposable income
We measured affordability in two ways; the first is with the Affordability Ratio (AR).
AR estimates basic water and sewer costs as a share of disposable income, which for present purposes is total income minus essential living costs (taxes, housing, food, health care, and home energy), estimated using Consumer Expenditure Survey data. To focus on low-income households, we measure AR at the 20th-income percentile (AR20). The 20th percentile household typically identifies the lower boundary of the middle class. These working poor households have very limited financial resources, but may not qualify for much public assistance.
Here’s what we found:
AR20 averaged 9.7. In substantive terms, that means that in the average utility, a household at the 20th income percentile must spend about 9.7 percent of its disposable income on basic water and sewer service. But AR20 varies widely, from a low of 0.6 to a high of 35.5. As the figure above indicates, AR20 is less than ten for about two-thirds of systems. A handful are very high—but these values reflect very low 20th percentile incomes as much as they do water/sewer costs.
Basic water & sewer in hours at minimum wage
My other way of measuring affordability is to express basic water/sewer costs in terms of Hours’ Labor at Minimum Wage (HM). This approach may not express affordability quite as accurately, but it’s precise and wonderfully intuitive: it represents the cost of water/sewer service in terms of labor. For those of us who have worked at minimum wage at some point in our lives, it is a powerful way to understand the cost of something.
Here’s affordability in the United States measured as HM:
As you’d expect, the distribution is very similar to AR20, ranging from 1.1 to 25.6. The overall average is 9.5, meaning that in the average utility, paying for basic water and sewer service requires about 9.5 hours of labor at minimum wage.
So is water affordable in America?
When confronting affordability, utility leaders are really asking: how much is it reasonable to expect households of limited means to pay? What economic sacrifices are reasonable?
These are ultimately questions about community values. There is no scientific answer to philosophical questions. No metric can define what is “affordable,” but good measurement can help us think about our values.
In the past I’ve suggested AR20 less than 10% and HM less than 8.0 as rules-of-thumb or point of departure to guide discussion. By these guidelines, 51 percent of the sampled utilities are affordable as measured by AR20 and just 39 percent are affordable according to HM.
We’re now in the process of analyzing the policy, organizational, economic, and demographic correlates of affordability. I’ll be publishing and blogging about what we find over the next several months. For now, I hope these figures provide some food for thought and help frame an agenda for tackling a tough issue.
*Collecting rates data is surprisingly difficult labor-intensive. Jake and I wrote a fun article on that topic–look for it next month.
Another way in which it’s tough to be poor
Drinking water utilities are great, but they aren’t perfect. Sometimes there are problems. Do those problems occur randomly? Or are there observable patterns in the water service problems?
Recently I’ve been posting about some findings from a Texas A&M Institute for Science, Technology & Public Policy (ISTPP) national public opinion survey. The survey’s carefully-designed sample of nearly 2,000 individuals is representative of the US population, and so offers an extraordinary look at public perceptions about water service. Earlier posts reported on attitudinal differences between water professionals and the general public, and on the ways that gender predicts opinion on water issues. I’m continuing to write up interesting findings from the ISTPP survey as time allows.
Today I’m looking at income.
Water service problems
The ISTPP survey asked respondents to say whether they had experienced each of the following problems with their drinking water with a simple yes/no answer:
- The water does not taste good (31.5% yes)
- The water is cloudy or dirty (19.5%)
- Water pressure is low (29.2%)
- The water causes sickness (3.8%)
- Water billing or payment problems (10.2%)
Importantly, this survey captures perceived water service problems, not actual problems—we don’t know that any given respondent actually experienced low water pressure, for example. We only know whether a respondent thinks (s)he experienced a problem. Likewise, we don’t know whether water actually caused sickness, only whether the respondent believes that it did. Fortunately, the large majority of respondents said “no” to all of these.
But the “yes” responses didn’t happen by chance. I fitted logistic regression models to identify correlates of water service experiences using the demographic variables in the ISTPP survey, such as race, ethnicity, age, urban/rural location, region, and income. These models estimate the likelihood of experiencing each of the five service problems.
A troubling pattern
The demographic correlates of water service problems vary, but across all five items, household income was the single strongest and most consistent predictor of water service problems. The graph below shows the likelihood of reporting that water billing problems at various income levels, with all else held equal (vertical spikes represent 95% confidence intervals):
At a $20,000 household income, there is a 13% chance of reporting billing problems. At $50,000, the likelihood is to about 9%; at $100,000 the likelihood drops to about 8%. That all makes some sense; we’d generally expect billing problems to correlate with income.
But the same pattern emerges for other kinds of water service problems, too. Here is the likelihood of reporting that water tastes bad at various income levels, again with other variables held constant:
At a $20,000 household income, there is a 37% chance of reporting bad-tasting tap water. At $50,000, the likelihood is to about 30%; at $100,000 the likelihood drops to about 25%.
Here’s the likelihood of experiencing cloudy or dirty water by household income:
Here’s the likelihood of reporting low water pressure by income:
And finally, here’s the likelihood of reporting that water caused illness by income:
Taken together, this is a sobering picture.* There is a clear relationship between income and the way that Americans experience their drinking water utility service. These results resonate with recent research finding a positive relationship between tap water consumption and income, with attendant implications for public health.
*In a future post I’ll look at race and drinking water experience; the picture won’t be much prettier.