Measuring Water and Sewer Service Affordability

Section 4: Analysis

This section will propose implementing an alternative metric, demonstrating how the EPA’s current metric misrepresents unaffordability and providing empirical evidence of the actual extent of unaffordability, as well as the distributional impacts of the current rate structures in Southeastern Michigan on various groups. This empirical evidence is then used to inform discussion of policy alternatives to address utility revenue requirements, cost allocation, and capital improvements, as well as funding requirements for customer assistance programs or alternative rate designs in the Alternatives section.

Methodology

This analysis identifies and quantifies the gap in the measurement of affordability issues by first applying EPA’s metrics to GLWA, a large and complex system which serves a diverse population, and then applying a two alternative metrics. The ramifications of this gap are then analyzed in the context of local affordability programs and future capital expenses, noting how under-estimation of affordability by current metrics negatively impacts decisions around funding for assistance programs and distorts cost allocation and revenue recovery inputs for utility capital planning and rate design. GLWA wholesale customers (all communities besides Detroit) and DWSD retail (Detroit) customers are analyzed separately to acknowledge the administrative regime stemming from the GLWA/DWSD lease and operation agreements. This analysis shows meaningful outcomes that should inform GLWA and DWSD-R customer practices—most notably the pursuit of an effective, robust affordability plan or alternative rate structure. Policy alternatives and recommendations based on the identified gap and estimated needs for both low-income ratepayers and utility planning are proposed.

Choice of Alternative Metrics

A number of alternative metrics have been proposed by both academics and practitioners. Figure 2 from NAPA’s meta-analysis summarizes a selection of proposals:

Figure 10: Residential Indicator Metric Alternatives

Figure 10: Residential Indicator Metric Alternatives
Image source: Czerwinski, et al., 2017.

Most proposals attempt to gain a clearer picture of user affordability by using income quintile instead of MHI. Other proposals not mentioned in the NAPA table suggest using local community MHI rather than EPA’s nationwide small system MHI or utility-wide MHI to identify clusters of unaffordability missed by larger-jurisdiction MHI.60 Multiple proposals use the average ratepayer water bill rather than system costs to more directly compute user affordability before analyzing utility financial capability (acknowledging the interconnection between the two). Finally, most proposals include supplemental contextual information such as trends, poverty rates, housing cost burden, and analysis of sub-populations within a utility service area.

The alternative metrics used here will follow NAPA’s recommendations of simplicity, readily available public data sources, and applicability for comparative analysis among different utilities (see Appendix 2).61 This is consistent with the desire for this study’s metrics to be replicable by multiple stakeholders—from the government to community members. It will also follow recommendations to include drinking water and wastewater costs together, to focus on low-income users rather than median users, and to identify the size of the population of vulnerable users relative to the utility’s total ratepayer base.62

Implementation of Two Alternative Metrics

The alternative metrics in this analysis will be done in two rounds to illustrate the extent that EPA’s current metric misrepresents affordability. Both will focus on place-based analysis, which was consistently emphasized during NAPA’s stakeholder engagements63, as opposed to EPA’s current use of aggregated MHI—an approach that is not sensitive to local variation.

In this study, we first measure affordability using the EPA’s metrics as interpreted by the Affordability Assessment Tool for Federal Water Mandates published by the American Water Works Association, United States Conference of Mayors, and Water Environment Federation.64 They describe the EPA’s wastewater RI as “average per household cost of wastewater bills relative to median household income in the service area”65 rather than strictly adhering to the EPA’s 1997 guidance that describes costs as a function of annual operations and maintenance expenses, annual debt service, residential share of costs, and number of households.66 Similarly, the Affordability Assessment Tool interprets the EPA’s drinking water RI to include annual household bills as the cost indicator. This technique is also among the metric options recommended by the NAPA report.

Figure 11: The EPA metric and this study’s conservative alternative (Click to enlarge)

Figure 11: The EPA metric and this study’s conservative alternative

The first alternative metric in this study will change the EPA metric’s numerator from “selected operations costs” and “planned upgrade costs” to actual costs faced by households by substituting total household bills (water, wastewater, and combined) based on average usage volume. This shifts the focus from theoretical cost allocation to actual average costs faced by households. The first alternative metric will also change the denominator from national small-system MHI for water and system-wide MHI for wastewater to local MHI for sub-populations within the service area for both water and wastewater. This shifts the focus from aggregated measures to place-based analysis that more accurately represents utility users and gives a sense of the size of vulnerable users relative to total ratepayer base. This amounts to a conservative estimate of affordability that does not radically challenge the EPA’s chosen threshold values and is not responsive to the much-agreed-to need to focus on lower-income households. This setup biases the specification of the alternative metric in favor of the status quo tendency to mask unaffordability, meaning that if unaffordability is found, then the extent of those issues are severe enough to show through a model prone not to reveal it. Accordingly, findings from applying the second alternative metric are more believable because the results do not disappear if preferred modeling assumptions are removed.67

Figure 12: The EPA metric and this study’s realistic alternative. (Click to enlarge)

FIGURE 12 shows a side by side of the EPA metric and this study’s realistic alternative

The second alternative metric will again change the numerator in the same way as the first round. The denominator, however, will change from MHI for sub-populations to income of the upper bound sub-populations to income of the upper bound of the first quintile (equivalent to the twentieth percentile) for those sub-populations. This refocuses the income analysis away from the unrepresentative median and instead to what is often considered the lower boundary of the middle class68 to better reflect the set of individuals who are most sensitive to rate increases. This sensitivity is defined as those subjected to greater harm from rate increases but also more likely to adjust consumption behavior, the first being critical for equity analysis and the second being critical for efficiency analysis. (See Feature Box at the end of Section  3)

Data Sources

Demographic and economic data for primary affordability calculations come from the United States Census Bureau’s American Community Survey (ACS) 2016 five-year estimates. Spatial data comes from the United States Census Bureau Topologically Integrated Geographic Encoding and Referencing (TIGER) products. Demographic and economic data for time series and trend analysis come from ACS one-year, three-year, and five-year estimates, with decennial census as needed. Retail water and sewer rate information was obtained from GLWA, DWSD, and individual cities, towns, townships, and charter townships. When rate information was not available online, it was obtained via telephone from local officials. System usage and capital financial data were obtained from GLWA and DWSD.

Great Lakes Water Authority Service Area

The GLWA service area (including the City of Detroit and leased DWSD infrastructure) is large and complex, spanning 1,069 square miles, 2,700 miles of transmission lines, five water treatment plants, and 19 booster pump stations for water delivery, as well as three interceptors, eight combined sewer overflow facilities, and five pump stations for wastewater management.69 The water system includes 127 member partners that purchase water from GLWA representing 119 individual communities.70 The wastewater system serves 87 member partners representing 76 individual communities.71

Figure 13: GLWA Water Service Area (Click to enlarge)

Map shows GLWA Water Service AreaImage source: GLWA Year in Review, 2017

To apply place-based affordability metrics, the complex GLWA service area was created according to census geographical units. This analysis uses county subdivisions (census code 060) for community-to-community analysis, since Michigan is one of 12 states where county subdivisions are legal entities—as opposed to statistical areas derived by census bureau staff—that serve as local governments (called Townships or Charter Townships) similar to small cities or towns.72 This means that census cities, towns, townships, and charter townships directly correspond to legal cities, towns, townships, and charter townships that enter into wholesale utility agreements.

Analysis of economic and demographic factors takes place at a smaller resolution, using individual census tracts (census code 140) to identify more nuanced differences among groups within individual communities, akin to neighborhoods. Census tracts within the GLWA service area were computed using the MABLE/GeoCorr14 geographic correspondence engine.

Computation of Average Bills

Average per capita water consumption in the United States is estimated to be between 2,400 and 3,000 gallons per month.73 Average usage for the GLWA service area was computed using monthly service volume within Detroit (the jurisdiction with the clearest reported monthly service volume) divided by the number of households and the average number of individuals per household.74 The resulting per capita estimate was 2,640 gallons per month, squarely within the national average. Total average household monthly usage given the average family size of 2.675 is 6,864 gallons. This is roughly confirmed across GLWA, which reports average monthly usage for a family of three as 7,000 gallons.76

That monthly average was combined with individual community water rates per unit, wastewater rates per unit if applicable, fixed charges (water and wastewater), special charges (water and wastewater), and number of billing periods per year to compute average annual bills broken out by water and wastewater (see Appendix 1 for details). When billed by meter size, 5/8 connector (or smallest available) was chosen, as most residential meters are this size.

The image below shows average annual household water and wastewater bills for each community served by GLWA. 77

Figure 14: Average Annual Household Water and Wastewater Bills (6,864 Gallons per Month per household - Click to enlarge)

Figure 14: Average Annual Household Water and Wastewater Bills  (6,864 Gallons per Month per household)

Average total annual bills for communities within GLWA range from $492.80 to $1,762.86, with wastewater costs typically $32.23 higher than water costs, although a portion of that for some communities could be attributable to wastewater infrastructure outside of GLWA management (see Appendix 3). The average annual total bill for Detroit is $1,246.51, with wastewater comprising 63 percent of that total bill. Aside from a cluster of relatively high costs in and around Flint, as well as a cluster of relatively lower costs south of Detroit, there are no clear spatial patterns to costs. Figure 15 below is a scatter plot showing annual water costs on the y-axis, median household income on the x-axis, and with each point representing a community scaled by the number of households.

Figure 15: Scatter Plot of Annual Water Costs, Median Household Income, and Number of Households (Click to enlarge)

Figure 15: Scatterplot of Annual Water Costs, Median Household Income,  and Number of HouseholdsImage source: US Census, Individual local governments, author calculations

Aside from a few outliers within MHI, annual water costs, and number of households, there are no clear relationships or trends between any of the dimensions. If there were relationships between MHI, annual water costs, and/or number of households, a more linear pattern would emerge in the scatter plot.

Community Economic Characteristics

Median Household Income across the GLWA service area is represented below in Figure 16:

Figure 16: Median Household Income by Census Tract, 2016 (Click to enlarge)

Figure 16: Median Household Income by Census Tract, 2016Image source: US Census Bureau

These data show patterns of low-income households in urban areas such as Detroit, Flint, and Pontiac, with higher-income areas encompassing the suburban areas between the urban centers, such as Bloomfield Township, Pittsfield Township, and Northville Township. Part of the reason for such patterns include the well-reported economic decline of Detroit and similar urban centers, as mirrored in unemployment rates seen in Figure 17 below.

Figure 17: Unemployment Rate by Census Tract, 2016 (Click to enlarge)

Figure 17: Unemployment Rate by Census Tract, 2016Image source: US Census Bureau

The data represented on these maps, and the map of annual water costs, are combined as inputs into the metric calculations shown in the results section below.

 

 

 

Results

Affordability Results of EPA Metric

Recall that the EPA’s RI for drinking water uses the median household income of nationwide small systems and an unaffordability threshold of 2.5 percent, while the RI for wastewater uses MHI for the system and an unaffordability threshold of 2 percent.

Individual community MHI within GLWA has a wide range, with a minimum of $21,196 and a maximum of $169,265. The EPA’s 1997 guidance instructs system-wide MHI be calculated from a weighted average that scales each community’s MHI by the proportion of total households it represents. GLWA’s weighted average MHI is $60,570.17, while Detroit’s MHI is $26,249.

Combining these inputs into the RIs yields an average combined water and wastewater result of 2.20 percent, a water result of 1.25 percent, and a wastewater result of 0.96 percent. The highest result was 3.52 percent, with the lowest being 0.97 percent (See Appendix 4 for full table), as seen in Figure 18.

Figure 18: EPA Metric Results (Click to enlarge)

Figure 18: EPA Metric ResultsImage source: US Census Bureau, individual local governments, author calculations

Detroit, although found to have a higher combined water/wastewater percentage, is still deemed affordable at 2.35 percent, with water and wastewater individually at 1.05 percent and 1.29 percent, respectively.

Under the EPA’s current metric, no community or neighborhood is found to be unaffordable either in the total bill or water/wastewater component part.

Affordability Results of Two Alternative Metrics

Specification 1: Conservative Estimate (Local MHI)

This conservative specification of the alternative metric changes the EPA metric’s denominator from national small-system MHI for water and system-wide MHI for wastewater to local MHI for sub-populations within the service area for both. This amounts to a conservative estimate of affordability that does not radically challenge the EPA’s chosen threshold values, or even use the much-agreed-on need to focus on lower-income households. Since this model is conservative, a finding of unaffordability here means that the true extent of unaffordability is severe enough to show through a model that is still biased against revealing it. Figure 19 shows the results of the conservative alternative metric and highlights census tracts that breach the 4.5 percent combined water and wastewater indicator in yellow.

Figure 19: Conservative Alternative Metric Results (4.5 percent Combined Water & Wastewater Threshold - Click to enlarge)

Figure 19: Conservative Alternative Metric Results (4.5% Combined Water & Wastewater Threshold)

Under this specification of the metric, unaffordability begins to appear at MHI below $31,300, as seen in Table 1 below.

 

 

 

 

Table 1: Communities Below $31,300 Median Household Income

Name Local MHI Annual Water Costs78 EPA Metric Alternative Metric
Imlay City $31,287 $1,418.40 2.87 4.53
Inkster79 $31,285 $1,230.58 2.29 3.93
Pontiac $30,152 $1,412.52 2.81 4.68
Royal Oak Township $27,837 $1,258.80 2.36 4.52
River Rouge $27,463 $1,098.79 2.09 4.00
Ecorse $26,496 $738.53 1.44 2.79
Detroit $26,249 $1,246.51 2.35 4.75
Flint $25,650 $1,213.53 2.45 4.73
Hamtramck $23,609 $1,333.36 2.5 5.65
Mount Morris $21,196 $1,702.71 3.52 8.03

It is notable that not only low-income urban areas face unaffordability issues. While the core urban centers and their neighboring communities of Mount Morris, Hamtramck, Ecorse, and River Rouge face unaffordability driven by a mixture of low incomes and high costs, the suburban communities of Inkster and Imlay City face very similar issues. Although they have the highest incomes of the locations that rise above the 4.5 percent threshold, they are only barely above 150 percent of the federal poverty level ($30,240 for a family of three), which is a common threshold for some public assistance benefits.80 Additionally, the communities of Lapeer, Montrose, Melvindale, Eastpointe, and Almont each face partial affordability for water or wastewater (rising above the threshold value for one of the two), even though their MHI rises above $31,300 (See Appendix 4 for full table).

In Detroit, 272,774 households reside within census tracts whose RI breaches the 4.5 percent threshold, representing households at risk for unaffordability. Of those households, 72,185 had annual household incomes under $29,999,81 making them extremely likely to face unaffordability. Within the rest of GLWA, 59,381 households are at risk, and 35,864 are extremely likely to face unaffordability. 

When the metric is tailored to actual on-the-ground economic characteristics, unaffordability issues become clear, even under modeling assumptions biased against finding unaffordability.

Specification 2: Realistic Estimate (Local MHI and Focus on twentieth Income Percentile)

This realistic specification of the model makes the same changes to introduce local data for place-based analysis that the conservative specification made, but also shifts the focus of the denominator to the upper bound of the first quintile of the income distribution (twentieth percentile) as recommended by NAPA. This amounts to a more realistic representation of affordability issues that reflect struggles faced by households and also provides more accurate inputs for utility financial planning based on actual customer ability to pay. Figure 20 shows the results of the realistic alternative metric and again highlights census tracts that breach the 4.5 percent combined water and wastewater indicator in yellow.

Figure 20: Realistic Alternative Metric Results (4.5 percent Combined Water & Wastewater Threshold - Click to enlarge)

Figure 20: Realistic Alternative Metric Results  (4.5% Combined Water & Wastewater Threshold)Image source: US Census Bureau, individual local governments, author calculations

It is worthwhile to note that a spatial component to the analysis is important since it reveals additional relationships that most tables alone do not. When communities that tip affordability thresholds are mapped, it becomes clear that “pockets of poverty”82 extend beyond impoverished urban areas, as well as smaller neighborhood-level pockets of unaffordability within communities.

Under the realistic alternative metric unaffordability issues again begin to arise around roughly $30,000 annual household income (except for Lenox Township, which has the highest annual costs of all communities). However, many more communities have twentieth percentile incomes below this level. Whereas the conservative metric found six communities to have RIs above the threshold, the realistic metric found 45 to be above the threshold. Table 2 below shows the upper five and lower five communities ranked by twentieth percentile income from this group (see Appendix 4 for full table):

Table 2: Selected Communities Above 4.5 percent RI Threshold Under Realistic Metric Specification

Name 20th Percentile Income Annual Water Costs83 EPA Metric Alternative Metric
Lenox Township $33,727 $1,762.86 3.27 5.23
Royal Oak $30,520 $1,400.01 2.57 4.59
Clawson $28,345 $1,465.39 2.73 5.17
Gibraltar $27,458 $1,658.89 3.15 6.04
Montrose Township $26,232 $1,288.17 2.64 4.91
... ... ... ... ...
Inkster $10,351 $1,230.58 2.29 11.89
Flint $9,897 $1,213.53 2.45 12.26
Hamtramck $9,818 $1,333.36 2.5 13.58
Detriot $9,574 $1,246.51 2.35 13.02
River Rouge $9,435 $1,098.79 2.09 11.65

Under this specification, the previous trends around urban centers and their immediate neighbors remain, however, many more suburban communities—including communities far from urban centers—are found to face potential affordability issues since the metric is now calibrated to focus on the segment of the customer base where costs are more acutely felt.

In Detroit, 104,181 households are located within census tracts whose RI breaches the 4.5 percent threshold and have annual household incomes under $29,999, while the rest of GLWA has 204,308 households that fall under the same criteria. It is also important to note that although the number of communities found to face unaffordability pressures jumps by 650 percent, the number of households within each community represented in this metric specification only increases by 470 percent for GLWA and 44 percent for Detroit.84 This is because the conservative specification uses median household income, under which half of all households fall, whereas the realistic specification uses the upper bound of a single quintile for household income, which represents one-fifth of all households. Thus, although the geographic spread of unaffordability grows with this specification of the metric, the absolute number of households does not grow at the same rate.

This is precisely what this specification of the metric is meant to do: reveal the actual areas of unaffordability that are masked by aggregate median measures, but not to inflate the scale of the problem. The metric should reveal the extent of unaffordability pressures on households susceptible to such pressures but remain useful as a tool to utilities and regulators.

The EPA metric attempts to simplify review by using aggregate measures, but has been shown to mask unaffordability. At the same time, more accurate measures are shown not to be unnecessarily complex. Further, the accurate measures can be simplified in a number of ways as a final step to get a single system-wide number as an input for planning purposes. This feature was the original intent with the EPA’s original specification of the metric.85 For example, the new final stage of the RI could be the proportion of census tracts or communities system-wide that face unaffordability issues above the 4.5 percent threshold. This would essentially turn the current RI (after the adjustments to local MHI and twentieth percentile income) for each community into a first stage matrix that is then condensed into a new ratio of unaffordable census tracts to total census tracts as a final single-number output. This achieves the dual goal of making sure the metric accurately reflects the real-world information it is trying to convey, while still being simple enough to use as an input into complex planning models without being a burden on limited utility, community, or regulator staff resources.

When the metric is tailored to on-the-ground economic characteristics and the segment of the population most sensitive to unaffordability, the true scope of the problem is revealed and utilities are better able to plan accordingly in revenue requirement, cost allocation, and capital improvement planning calculations.

Distributional Impacts of Affordability Challenges

In addition to economic, cost, and population characteristics, it is important to know the demographic characteristics of populations facing unaffordability, especially if those populations are vulnerable in certain ways.86 A nationwide study of affordability found that census tracts with median incomes below $32,000 have higher percentages of disabled people, higher unemployment rates, higher levels of public assistance income, a higher incidence of sole female heads of household, larger percentages of minority residents, and lower rates of health insurance.87 Local stakeholders who have addressed affordability in southeastern Michigan have pointed to the lack of research on these particular issues, which this study attempts to rectify.88

As can be seen in the figures below, the nationwide trends are reflected in the Michigan data. Figure 21 shows the percentage of African Americans by census tract with tracts from the conservative specification over the 4.5 percent RI threshold (i.e., the extreme cases of unaffordability) highlighted in yellow. The correlation between the two is striking.

Figure 21: Census Tracts by Percent African American and Above 4.5 percent RI Threshold

Figure 21: Census Tracts by Percent African American and Above 4.5 percent RI ThresholdImage source: US Census, author calculations

Figures 22 and 23 below show the extent of public assistance89 used within census tracts and the census tracts that rise above the 4.5 percent affordability threshold evaluated at the twentieth percentile.

Figure 22: Census Tracts by Percent Receiving Public Assistance

Figure 22: Census Tracts by Percent Receiving Public AssistanceImage source: US Census, author calculations

Figure 23: Census Tracts by Percent Receiving Public Assistance and Above 4.5 percent RI Threshold

Figure 23: Census Tracts by Percent Receiving Public Assistance and Above 4.5% RI ThresholdImage source: US Census, author calculations

Again, the connection between the two is visible, with the highest rates of public assistance occurring in and around the metro centers of Detroit, Pontiac, and Flint, but also spreading to the surrounding suburbs, especially towards the southwest, alongside the tracts that fall above the 4.5 percent threshold.

Lastly, similar comparisons can be drawn between the percentage of the population with a disability90 and the tracts that rise above the 4.5 percent threshold, as seen in Figures 24 and 25.

Figure 24: Census Tracts by Percent with a Disability

Figure 24: Census Tracts by Percent with a DisabilityImage source: US Census, author calculations

Figure 25: Census Tracts by Percent with Disability and Above 4.5 percent RI Threshold 

Figure 25: Census Tracts by Percent with Disability and Above 4.5 percent RI Threshold Image source: US Census, author calculations

Here, the connection between the two is less obvious, since there are more widespread lower percentages of disability across the region that is not limited to the tracts that breach the affordability threshold, but it is still clear that the areas suffering the most intense affordability issues are generally the places that have disability rates above 30 percent.

It is clear that water unaffordability disproportionately impacts already-vulnerable communities.

Trends Over Time

One of the common critiques of existing metrics is that they are static snapshots in time that do not take into account temporal trends. It is important to analyze how demographic and economic patterns shift over time to see what trends the near future may hold, especially when comparing different localities.

Figure 26, below, shows the population shifts in Detroit (a low-income urban center), Bloomfield Township (a high-income Pontiac suburb), Van Buren Township (a median-income Detroit suburb), and Harrison Township (a median-income township not located near an urban center). These locations represent a sample of different income levels and geographic identities, but still show clear trends.

Figure 26: Population Estimates 1990-2016

Figure 26: Population Estimates 1990-2016Image source: US Census American Community Survey 5-year estimates for 2012-2016, 3-year estimates for 2009-2011, 1-year estimates for 2007-2008, and decennial census for 1990-2000.

Detroit shows a steady and significant decline in population of over 300,000 people, while the individual suburbs grow slightly (Van Buren Township) or stay relatively stable (Bloomfield Township, Harrison Township).91

Figure 27 displays poverty rates among the locations over time.92 Although poverty rates climb for all locations in the aftermath of the Great Recession, only Detroit’s was climbing significantly beforehand, and again only Detroit’s shows minimal recovery after the nation’s economic turnaround.

Figure 27: Poverty Rate Estimates 1990-2016

Figure 27: Poverty Rate Estimates 1990-2016Image source: US Census American Community Survey 5-year estimates for 2012-2016, three-year estimates for 2009-2011, 1-year estimates for 2007-2008, and decennial census for 1990-2000.

It is clear from both trends that the economic and social pressures on utilities in struggling urban centers like Detroit (and by extension Flint and Pontiac) are not on a path to rectify themselves. Increasing poverty rates mean more households will face unaffordability concerns, while shrinking ratepayer numbers due to population decline will also put further strain on utilities that need to spread high fixed costs among a broad customer base.

  • 60. Teodoro, Measuring Household Affordability; Stratus Consulting, Affordability Assessment Tool.
  • 61. “Panel Recommendation #1:… [m]etrics used should meet the following criteria: 1) readily available from publicly available data sources; 2) clearly defined and understood; 3) simple, direct, and consistent; 4) valid and reliable measures; 5) applicable for comparative analyses among permittees.” Czerwinski, et al., Developing a Framework, 61.
  • 62. “Panel Recommendation #2:… elements defining the current component should be revised to: 1) include all water costs, not just selected clean water costs, to include all drinking water and clean water costs… 2) focus on the income of low-income users most vulnerable to rate increases rather than Median Household Income; 3) identify the size of the vulnerable users relative to the utility’s total rate payer base.” Czerwinski, et al., 62.
  • 63. Czerwinski, et al.,177.
  • 64. Stratus Consulting, Affordability Assessment Tool.
  • 65. Id., 4.
  • 66. This analytical choice was made because GLWA and DWSD report costs, volumes, and revenues according to retail grouping and wholesale grouping, rather than residential grouping and commercial grouping in financial, operations, and capital planning documents. Without access to internal computer systems, it is not possible to identify the residential share of costs, nor is it possible to review the RI screen done by EPA for DWSD/GLWA’s National Pollutant Discharge Elimination System permit. This does not impact the validity of this analysis, however, because use of average wastewater utility bill is guaranteed to include all costs the utility faces since Michigan law and the GLWA lease agreement set wholesale water rates at full cost pricing (many retail water bills even include line items for fixed costs and debt service). If anything, this improves the EPA metrics’ ability to detect unaffordability, making the fact that it still fails to do so all the more robust a finding.
  • 67. In layman’s terms, this is akin to saying “I grant your framework these certain points that I would argue against, yet even with those concessions and still using your own framework, my conclusion holds.” A fortiori arguments are common with economists to demonstrate robustness of findings before presenting preferred models that more accurately reflect reality.
  • 68. Teodoro, Measuring Household Affordability, 15.
  • 69. Great Lakes Water Authority, 2017 Year in Review (Great Lakes Water Authority, 2017), https://www.glwater.org/wp-content/uploads/2018/08/GLWA_YIR_2017_final_d... “About DWSD,” City of Detroit, accessed March 16, 2019, http://www.detroitmi.gov/Government/Departments-and-Agencies/Water-and-S....
  • 70. Great Lakes Water Authority, 2017 Year in Review; City of Detroit, “About DWSD.”
  • 71. It is important to note that GLWA wastewater coverage does not extend to all of the communities served by drinking water infrastructure. Some of the calculations in this study will be forced to use wastewater costs provided by utilities outside of GLWA, such as the Genesee County Drain Commissioner. This study is intended to analyze affordability at the neighborhood and household level, so the inclusion of some costs outside of those imposed by GLWA is necessary, since it is still their customers whose budgets are impacted by these outside wastewater costs, which in turn impacts GLWA’s cost recovery efforts on these same households.
  • 72. “Geographic Terms and Concepts – County Subdivision,” United States Census Bureau, accessed March 16, 2019, https://www.census.gov/geo/reference/gtc/gtc_cousub.html; “Glossary – County Subdivision,” Census Reporter, accessed March 16, 2019, https://censusreporter.org/glossary/.
  • 73. “Water Questions and Answers,” United States Geological Survey, accessed on March 16, 2019, https://water.usgs.gov/edu/qa-home-percapita.html.
  • 74. It is possible, even likely, that average usage varies between urban areas such as Detroit and rural areas in the rest of GLWA. However, the point of this analysis is to identify differences in structural affordability of water according to local characteristics, not to compute actual water consumption across jurisdictions. The goal is comparability between locations, so a single usage estimate for all communities is required. Where the usage estimate comes from is less important than the variance in cost according to that usage estimate.
  • 75. Average household size calculated for GLWA service area based on American Community Survey 5-year estimates for 2016.
  • 76. “Household Eligibility for WRAP Assistance,” Great Lakes Water Authority, accessed March 16, 2019, https://www.glwater.org/wp-content/uploads/2018/07/household-eligibility...
  • 77. Ten communities were not responsive to rate information requests after multiple telephone calls. They are represented as blank community boundaries in Figure 14 (see Appendix 1 for more details).
  • 78. Combined water and wastewater costs.
  • 79. Although Inkster does not break the combined 4.5% water and wastewater threshold community-wide, its sewer RI of 2.6% does breach the 2% threshold and a significant portion of its neighborhoods, represented in census tracts highlighted in the map, do break the combined 4.5% threshold.
  • 80. Kimberly Amadeo, “Federal Poverty Level Guidelines and Chart,” The Balance, last modified March 12, 2019, ://www.thebalance.com/federal-poverty-level-definition-guidelines-chart-330....
  • 81. This number was selected because of the way income groups are reported in ACS table B19001, which breaks out income groups in $5,000 increments. $29,999 and below best corresponds to roughly $30,000 which, as pointed out above is roughly 150% of the federal poverty level for a family of 3 (a common threshold for public assistance benefits) and tracks roughly along with the MHI at which community-wide water and wastewater costs become unaffordable.
  • 82. Elizabeth A. Mack and Sarah Wrase, A Burgeoning Crisis.
  • 83. Combined water and wastewater costs.
  • 84. Detroit’s percent change is small compared to the others because so much of Detroit’s income distribution is situated towards the bottom end, so the shift from median to 20th percentile was not as radical a shift as for other communities.
  • 85. United States Environmental Protection Agency, Drinking Water Requirements for States and Public Water Systems: Variances and Exemptions, accessed March 16, 2019, https://www.epa.gov/dwreginfo/variances-and-exemptions; United States Environmental Protection Agency, Variances and Exemptions: A Quick Reference Guide, accessed March 16, 2019, https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=901U0I00.txt.
  • 86. United States Water Alliance, An Equitable Water Future, 8.
  • 87. Elizabeth A. Mack and Sarah Wrase, A Burgeoning Crisis.
  • 88. Blue Ribbon Panel, Final Report, 25;; WRAP Advisory Group, Water Residential Assistance Program: Program Design Report (Detroit, MI: Great Lakes Water Authority, August 2015), p. 6, http://www.detroitmindsdying.org/wp-content/uploads/2015/11/2015-8-WRAP-....
  • 89. Here, “public assistance” refers to all non-Social Security public assistance tracked by the Census Bureau in Table S19058, including: General Assistance (sometimes called “GA” or “welfare”), Temporary Assistance to Needy Families (or “TANF”), and Supplemental Nutrition Assistance (sometimes called “food stamps” or simply “SNAP.”)
  • 90. Here, “disability” refers to all disabilities tracked by the Census Bureau in Table S1810, including: hearing difficulty, vision difficulty, cognitive difficulty, ambulatory difficulty, self-care difficulty, and independent living difficulty.
  • 91. Some of the variability in these charts is due to differences in methodology, both between the decennial vs. ACS surveys and also between the 1/3/5-year estimates. However, each year in the chart follows the same methodology between locations, so that any variability is applied to each jurisdiction alike and overall trends still show through. The differences in census methodology mean that the actual scale of the year-to-year differences may be an artifact of methodological changes rather than actual population shifts, however, the spread between localities reflects actual proportional differences between them.
  • 92. The note of caution on interpreting year-to-year differences in the preceding chart applies to this chart as well.