EPA’s CEIP Should Directly Help Low-Income People

EPA should ensure that the CEIP's energy efficiency portion does not increase emissions

Between the time the Environmental Protection Agency proposed and finalized the Clean Power Plan—the first-ever limits on the nation’s largest source of carbon dioxide emissions, fossil fuel-fired power plants—the agency spoke to states, power companies, business groups, and advocates. Among the things they heard were that states wanted to be rewarded for early actions to reduce carbon pollution, and that EPA needed to use the Clean Power Plan (CPP) to help redress the lack of investment and high pollution burden inflicted on low-income and minority communities.

In response, EPA included the Clean Energy Incentive Program (CEIP) in the final Clean Power Plan. The CEIP provides early action emission allowances or emission rate credits to eligible renewable energy projects, and also solar and energy efficiency projects that benefit low-income communities. I say “early action” because the CEIP would give projects these credits before the CPP goes into effect in 2022. Allowances or credits can then be sold to power plants, generating revenue for supported projects. In June, EPA issued a proposed rule on CEIP design details; NRDC and other groups have until early November to respond.

This blog gives a preview of NRDC’s comments on one important issue: which energy efficiency projects should be eligible for the low-income portion of the CEIP. My NRDC colleague Kevin Steinberger recently posted a blog on the renewables portion of the CEIP.

The purpose of the low-income portion of the CEIP

EPA, in the preamble to the Final Rule, says that the low-income portion of the CEIP was developed to help alleviate “historic economic, logistical, and information barriers” that have kept energy efficiency projects from being implemented in low-income communities, ensure the benefits of the CPP are shared broadly across society, and avoid adverse impacts on low-income customers: i.e., bill increases.

This rationale makes sense. Doing energy efficiency projects—installing insulation, heating and cooling equipment, new, efficient refrigerators and light bulbs—in the housing where low-income people live is more expensive than energy efficiency projects elsewhere because these buildings are often in poor condition, and the project developer generally cannot expect for tenants or owners to make a substantial co-pay toward the cost of the project.

Should the CEIP’s low-income portion be opened up to all sorts of projects?

Some advocates, including groups we often agree with, have argued for an expanded scope for the low-income portion of the CEIP: to invest broadly in areas where lots of low-income people live, supporting things like energy efficiency projects in public buildings, schools, hospitals, and even industrial or small businesses. I don’t think that makes a lot of sense.

Projects at public institutions in low-income communities could allow local governments to reduce taxes or improve services, indirectly benefitting low-income people. Projects at for-profit businesses in low-income communities could allow a business to stay open, or the savings from projects could just increase the profits of the business owner, very indirectly benefitting low-income people. In contrast, projects in the apartments and houses where low-income people live reduce the energy bills landlords or tenants pay and improves housing quality, directly benefitting low-income people.

The CEIP’s primary purpose should therefore be to improve the efficiency of the housing where low-income people live. Projects that improve efficiency at public institutions should be next in line, and the program should not support projects at for-profit businesses.

But the pool of available credits is so big relative to the energy efficiency opportunity! Who cares if we give credits to projects that only indirectly benefit low-income people?

The pool of available early action credits from EPA is big: 150 million short tons. But for the reasons my colleague Kevin Steinberger lays out, giving away those credits to projects that don’t need help would actually increase emissions. Because of the “historic economic, logistical, and information barriers” that EPA cited, the size of the market for low-income residential energy efficiency is small, so even if EPA gave credit to business-as-usual low-income residential projects it would not increase emissions much. But opening the low-income portion of the CEIP to all energy efficiency projects that happen to take place in a defined low-income geographic area would increase emissions. The graphic below explains why. If EPA gives allowances from its matching fund (which exists outside state emission budgets) to projects that would have happened without the CEIP, this increases emissions because this allowance grant is not offset by emission reductions.

How should EPA define “low-income communities”

The CPP final rule states that eligible projects include “energy efficiency implemented in low-income communities,” but does not define “low-income communities.” The proposed rule on design details suggests four definitions that states could use in their state plans and likely get EPA’s approval:

  1. The definition of used in the New Markets Tax Credit (NMTC) program: a census tract is a “low-income community” if:

    1. its poverty rate is at least 20 percent, or

    2. for tracts outside of metro areas, if the tract’s median family income (lining up families by income, the median income is the income of the middle family) is less than 80 percent of the state’s median family income, or

    3. For tracts inside metro areas, if the tract’s median family income is less than 80 percent of the state’s or metro area’s median family income.

  2. The definition used for the Department of Housing and Urban Development’s (HUD) Qualifying Census Tracts (QCTs): a tract is a QCT if the median household income is less than 60% of the Area Median Gross Income,

  3. The Federal Poverty Guidelines

  4. The definitions used by states in the Weatherization Assistance Program.

The first two definitions are geographic: they identify census tracts where low-income people live. Census tracts are contiguous areas of around 4,000 people. The second two definitions are household: they identify low-income individuals or groups of people who live together in the same unit.

Until recently I lived in Census Tract 263.02, in the Crocker-Amazon neighborhood of the far-southern part of San Francisco, with (according to the 2014 American Community Survey 5-year estimates) around 5,295 neighbors, plus or minus seven percent. You can find your census tract by entering your address here.

Median household income (again from those 2014 5-year estimates) was $68,281, median family income was $72,780, and around 53 of 1,125 families live on below-poverty incomes. In comparison, the San Francisco-Redwood City-South San Francisco Median Family Income for 2016 was $104,700, and HUD’s 2016 median income for San Francisco County was $107,700.

This means my former census tract would not qualify for the NMTC based on poverty criteria, but it would qualify based on median family income, because in 2016 its median family income was only 69 percent of the metro area’s. My former census tract would not be considered a HUD Qualified Census Tract, because its median household income was 63 percent of HUD’s median income for San Francisco County, while the threshold is less than 60 percent.

I considered my old neighborhood a middle-class one—it is one of the last places in San Francisco where people who work normal, non-tech jobs can own a home. But yet it meets the New Markets Tax Credit definition of “low-income community.” My hunch, explored below, is that the NMTC definition is too broad.

Why have geographic definitions at all?

If the CEIP’s low-income portion is primarily focused on housing—where we can identify the income status of households and focus investment on houses and apartments where low-income people live—what is the point of a geographic definition? The answer is lower transaction costs: a geographic definition would allow program providers to sweep through the community, touching every house. The alternative of checking the income of every eligible household sounds daunting in comparison.

But there are holes in this justification for using a geographic definition of low-income community. People with different incomes live in the same geographic area: high-income people live in geographies we would define as low-income, and low-income people live outside of these low-income geographies too. Under any geographic definition of low-income community, if we allow program providers to touch every house, a lot of CEIP credits would flow toward higher-income households. Moreover, EPA would need to establish a household-based definition of low-income to allow the CEIP to benefit low-income people (and it definitely should: we are learning more and more about the benefits of de-concentrating poverty).

Just get rid of geographic definitions of low-income community

EPA should just get rid of geographic definitions of low-income community. They do not make sense given the purpose of the CEIP; under the geographic definitions offered by EPA, they would allow too much program benefits to flow to higher-income people.

Why do I go this far? My colleague Emily Barkdoll and I gathered data on population, race, and income for each census tract from the U.S. Census’ 2010-2014 American Community Survey 5-Year Summary File, identifying HUD Qualified Census Tracts (QCTs) and NMTC-eligible tracts.

We found:

  1. Many more people live in NMTC census tracts compared to HUD QCTs: 113 million people instead of 50 million people

  2. Taking into account the larger population covered by the NMTC definition, the HUD QCT definition does a better job of identifying geographies where non-white people live and geographies that have a high portion of at-or-near poverty families, and a better job of excluding households with higher incomes. Though the total population in QCTs is only 44% of the total population in NMTC tracts, QCT’s contain 56% of the non-white population of NMTCs, around 55% of the below-poverty population, and only 30% of the population who have incomes 5 times the poverty guideline or more.

  3. But even QCTs include 2.6 million households who had annual incomes higher than $75 thousand. The number of greater than $75 thousand annual income households in NMTC tracts is about 37% of the number of sub-$40 thousand annual income households.

It makes no sense, given additionality concerns outlined above, to allow states to choose either/or definitions that would have such varying effects in communities. Given that even the more conservative of the two geographic definitions—HUD Qualifying Census Tracts—would allow 2.6 million middle- and higher-income households to get projects that receive credit, EPA would be justified in getting rid of geographic definitions of low-income community entirely and instead go with household-based definitions.

But which household-based definitions? That’s the subject of my next blog.