Analyzing Affordability: Supporting Households under New York’s Cap Trade and Invest Policy

This report, produced in partnership with the New York City Environmental Justice Alliance, investigate how different allowance-funded investment and dividend strategies can affect transportation and residential energy costs faced by New York households.

Date

Jan. 13, 2025

Authors

Molly Robertson, Alan Krupnick, Wesley Look, Eunice Ko, Conor Bambrick, Celeste Perez, and Eddie Bautista

Publication

Reading time

Overall, we find that the CTI program could result in net savings for many New Yorkers, especially when dividends are targeted by region and income. Furthermore, a high allowance price paired with a targeted dividend distribution strategy yields the greatest benefits for many New Yorkers, particularly those making less than $200,000 per year. The CTI revenue could be invested to reduce the transportation and energy cost burdens and put dollars back in New Yorkers' wallets while yielding greater emissions reductions and climate benefits.

This analysis leverages two energy models that estimate energy use, one for residential buildings and one for light-duty vehicles. The assumptions on allowance prices and revenues are derived from the NYSERDA and DEC preliminary analyses, which tested the effects of three price ceilings for allowances in the CTI program.

The residential model used in this study is an updated version Unlike previous iterations of the model, the model is not constrained by linearity assumptions and determines heterogeneous causal effects of different heating appliances on fuel expenditures depending on fuel prices, by taking advantage of state-of-the-art machine learning methods, particularly gradient boosting regressions and classifiers (Friedman 2001) and causal forests (Athey et al. 2019). of the model presented in Poblete-Cazenave and Pachauri (2021) and Poblete-Cazenave and Rao (2023). In this discrete-continuous choice model, households choose among various technologies, fuels, and consumption levels to satisfy their heating needs based on household, dwelling, and location characteristics. Different parts of the model are estimated separately: the adoption of heating technologies is estimated on a biannual panel of households from the American Housing Survey (AHS) for the years 2019-2023 and tested on the subset of households in the mid-Atlantic region. The heterogeneous causal effects of heating appliances on expenditures are estimated with data from the Residential Energy Consumption Survey (RECS) corresponding to the year 2020 and tested on the subset of households in New York State. Finally, the models are applied to the sample of New York households in the American Community Survey of 2020, which is projected to the year 2030 in line with the assumptions and scenarios from the Annual Energy Outlook 2023. The model simulates heating appliance and fuel consumption choices for each household in the sample (approximately 75,000 households), which are then aggregated in various ways for the presentation of results.

We use RFF's light-duty transportation model to predict new vehicle sales, miles traveled, consumer expenditures, and emissions through 2030. The model distinguishes consumer groups based on income, number of household-owned vehicles, population density by zip code, and region (Leard et al. 2023). Each simulation year, consumers first decide whether to scrap their older vehicles, depending on the vehicles' expected resale value net of scrappage subsidies. Subsequently, consumers may adjust their vehicle holdings by selling remaining vehicles or purchasing new or used vehicles. Consumers make these choices to maximize their subjective well-being, based on the vehicles' prices, expected fuel costs, horsepower, size, body type (sedan, sport utility vehicle, etc.), local electric vehicle (EV) charging availability, and other attributes. Having chosen which vehicles to own, consumers decide how much to drive those vehicles, depending on the vehicles' age, fuel costs, manufacturer and whether they are classified as cars or light trucks.

New-vehicle manufacturers choose vehicle prices and attributes to maximize their profits, given consumer preferences and policies such as federal GHG standards, EV subsidies in the Inflation Reduction Act (IRA), and California's Advanced Clean Cars Program (ACC2). New-vehicle prices and attributes, as well as compliance credit prices in the GHG and ACC2 programs, are determined to balance new-vehicle supply and demand. Across scenarios, the effects of EV subsidies on vehicle purchase prices account for price negotiations between vehicle buyers and sellers as well as changes in GHG and ACC2 credit prices. Nearly all model parameters are estimated using consumer and manufacturer choices observed in the InMoment New Vehicle Customers Study from 2010-2020 and the National Household Travel Survey from 2017-2022.

With those two models, we analyze three cases to investigate the affordability of a CTI program:

• Business-as-usual (BAU) case, with no New York CTI;
• New York CTI using prices from Scenario A in the NYSERDA and DEC analysis (2024); and
• New York CTI using prices from Scenario C in the NYSERDA and DEC analysis (2024).

The BAU case has no CTI program in effect and no new emissions-reducing investments beyond current policies. The household costs in this case act as a baseline for comparing the scenarios. The BAU includes current New York and federal policies (like the IRA) that might influence emissions and household costs. We then model two CTI allowance prices using the highest and lowest prices assessed in the NYSERDA and DEC analysis (2024). In that analysis, Scenario A (our high-price case) starts at $23 in 2025 and goes up to $64 in 2030; the price in Scenario C (our low-price case) is $14 in 2025 and increases to $30 in 2030. Both cases reflect price ceiling options that are binding according to the state analysis, meaning the ceilings set the market allowance price. We assume the CTI revenue is used for investments in programs and policies such as heat pump and EV subsidies, detailed in Appendix A. We also test the effect of allowance prices alone (without investments) on households and the effect of removing IRA subsidies to reflect possible actions by the new presidential administration and provide a richer and possibly more relevant policy analysis.

For each scenario we assess effects by income group and region. We include eight income groups, which are defined consistently throughout the analysis. For regional comparison, we use Public Use Microdata Areas (PUMAs), which are designed to have a minimum of 100,000 residents. PUMAs are smaller than counties but larger than census tracts and vary in size depending on population density.

5.1. Electrified households have lower costs than households that run on fossil fuels.

Table 1. Annual Savings for Fossil Fuel vs. Fully Electrified Households, by Scenario, 2030

Table 1 demonstrates that households across income groups could save hundreds of dollars a year by electrifying their transportation and home heating. In the BAU, without any additional carbon cap restrictions, households operating EVs (the number of vehicles varies by income group) and a heat pump save on average $424 to $1,706 depending on income, compared with "fossil fuel households"-those still operating gasoline vehicles and a gas furnace. This table does not assess savings compared to households operating a propane or a fuel oil heating system. They are included in the rest of the analysis unless otherwise specified. Most of the savings are associated with the switch from gasoline vehicles to electric vehicles, with relatively smaller savings in home heating. Our residential model uses AHS and RECS data to estimate the adoption and energy consumption of different appliances. Observed data in RECS indicates households with heat pumps in New York State have lower operating costs than households with other equipment. This may be in part because dwellings that currently have heat pumps are better prepared to maximize heat pump efficiency than the ones who don't. While our operating cost data is informed by home size, age, a self-reported measure of insulation quality and other factors, the cost of operating heat pumps in the remaining stock is still estimated based on costs of current heat pump owners.
Indeed, in the lowest income group, heat pump operation costs are roughly equivalent to gas furnace operating costs. With a CTI program in place, savings from full electrification are higher. In the high allowance price case, savings in 2030 are larger than in the low-price case (ranging between $492 and $1,965 per year) because the relative difference between the cost of electricity and the cost of fossil fuels is greater. These cost differences are based solely on operating expenses for existing equipment, excluding investment incentives and dividends. Irrespective of the price ceiling level, savings from electrification are relatively lower among low-income households because they tend to spend less overall on fossil fuels. They own fewer cars, for instance, so they drive fewer miles and use less gasoline (even though their cars may be older and less fuel efficient). Across all income groups, these household savings highlight the importance of supporting electrification and energy efficiency by lowering initial purchase costs for new vehicles and appliances (through subsidies like those we model), building infrastructure like charging stations, and readying homes for electrification through weatherization, electric panel upgrades, and other building "shell" enhancements.

5.2. Existing subsidies drive electrification, and new investments lower costs for households.

In our modeling, we observe that heat pump subsidies funded by the IRA drive the majority of heat pump adoption in New York State. Additional subsidies funded by CTI proceeds make the most difference in adoption in the highest income categories ($150,000 to $200,000, and over $200,000) for heat pumps. EV adoption is largely driven by the zero-emissions vehicle (ZEV) mandate in the BAU. Generally, our results indicate that most of the state funding dedicated to heat pump and EV subsidies does not catalyze additional clean technology deployment, but it lowers the upfront purchase costs for households who would already have made EV or heat pump purchases because of IRA subsidies and the ZEV mandate. For example, average annualized household spending on heat pumps decreases from $10.32 in the BAU case to 40 cents in the high allowance price case. The average annualized spending includes homes that don't buy heat pumps, those are fully subsidized (zero cost), and those who pay up to full price for a heat pump.

Across the policy cases, the IRA policies invest approximately $18 billion in New York electrification between 2026 and 2030. That would amount to about 80 percent of CTI revenues in the low-price case and 40 percent of revenues in the high-price case over the same period. Due to uncertainty about whether federal subsidies will continue under the new administration, we modeled a case in which New York adopts CTI but the IRA incentives are eliminated. We found that energy costs and spending on vehicles and appliances were affected, particularly for middle- and high- income households that were taking advantage of IRA subsidies to electrify (Table 2).

Table 2. Annual Household Cost Effects of Removing Federal IRA Subsidies, by Scenario, 2030

Even without the IRA, our modeling indicates that CTI encourages significant household electrification: CTI allowance prices and subsidies prompt heat pump adoption in about 3 million households in the high-price case and 2.9 million households in the low-price case. These adoption rates are significantly higher than those estimated by NYSERDA and DEC (2024). We include a sensitivity that tests affordability impacts under lower heat pump adoption in Appendix B.
This indicates that even the $5.55 billion in heat pump subsidies between 2026 and 2030 in the low allowance price case could support significant electrification. Additional funds available in a high allowance price scenario could be directed toward complementary electrification readiness policies, such as lead, asbestos, and mold removal, insulation, and electric panel upgrades, though we cannot represent those in our modeling. Appendix A offers more detail on the subsidies modeled and the amount of spending in each CTI policy case.

A limitation of our modeling is that we cannot accurately account for total construction or installation costs associated with heat pumps. Demand for heat pumps in the model is implicitly informed by these additional costs but shows up in our modeling as additional "friction" that makes certain households very unlikely to adopt heat pumps even with generous subsidies. This friction is based on observed data and represents barriers in the real world that prevent people from switching to heat pumps. Policymakers may need to address these overlooked challenges to accelerate home electrification across the state. Our model estimates that low-income households are particularly unlikely to purchase heat pumps if they have any of the following: elderly residents, a larger home with more rooms, a newer home with newer appliances, or water-heating and cooking appliances that rely on natural gas. These sources of friction offer supporting evidence that policies and additional investments beyond direct heat pump subsidies could help foster heat pump adoption.

5.3. Expenditure changes are proportional to household spending on energy.

We calculated household costs under a CTI program across income groups before using all the CTI revenue for dividends. In general, we find that costs are not regressive, meaning that on a percentage basis, costs in 2030 relative to BAU increase more for high-income households, irrespective of the allowance price. For instance, with a high allowance price, costs for the lowest income group increase 3 percent over the BAU, whereas costs for households with income between $50,000 and $150,000 increase 5 to 7 percent (Table 3). Overall, gross household cost increases are slightly higher with the higher allowance price, as expected. However, we find that we can largely offset these higher costs with the higher revenue generated for the CCAA dividend payments, discussed below.

Table 3. Average Annual Home Expenditure Change from BAU, Including Capital Costs, 2030

We also consider how these cost changes relate to household income. Energy and transportation burdens on households are typically calculated by dividing the amount of spending on home energy and transportation fuels by a household's income. A household is considered energy burdened when it spends 6 percent or more of its income on home energy and is considered transportation burdened when it spends more than 4 percent of its income on transportation fuels (NREL 2024). In the BAU, the average household in the bottom two income groups spends a significant amount of its income on such costs and is considered burdened in both categories. Higher costs under a CTI program could increase those burdens for LMI households, absent the dividend payments. Table 4 shows how energy and transportation burdens could increase with a CTI program before dividends are distributed to households. Even though cost increases for low-income households are smaller in absolute terms, they make up a greater share of their household income, compared with the average cost increases for high-income households. When we explore options for distributing dividends in this analysis, maximizing savings for these burdened low-income groups is a top priority.

Energy and Transportation Burdens in the BAU and Percentage Point Changes under CTI, without Dividends, 2030

5.4. Average costs for LMI households vary across the state.

Costs also vary by geography. Suburban and rural households tend to spend more on oil, natural gas, and gasoline than their New York City counterparts. Figure 1 shows the gross increases in expenditures for households making less than $125,000 a year, relative to the BAU, for high and low allowance prices before dividends are distributed. Expenditures include average spending on heating oil, natural gas, gasoline, electricity, home heating appliances (i.e., furnaces and heat pumps), and vehicles. Increased spending in downstate New York and Long Island is partly driven by the concentration of higher-income households, which tend to spend more on energy and transportation.

Average Annual Expenditure Changes for LMI Households, by Scenario, 2030

5.5. Dividends from the CCAA can cover costs for many households and even offer a net financial benefit.

New York State has pledged to set aside 30 percent of CTI program revenues for the Consumer Climate Action Account to "mitigate any consumer cost increases that could result from the NYCI program," (NYS n.d.). The high allowance price could generate more than $1.5 billion to distribute in 2027 and more than $3.5 billion in 2030 (NYSERDA & DEC 2024). The low allowance price generates less revenue to distribute: more than $900 million in 2027 and more than $1.6 billion in 2030. Table 5 shows the annual estimated budget for the CCAA, based on 30 percent of estimated program revenues.

Table 5. Annual Revenue for Consumer Climate Action Account (Billions), by Scenario

5.6. Equal dividend payments can cover average costs for the lowest-income households.

To explore the effects of dividend distribution options, we first look at equal distributions to all households-a simple (if unlikely to be implemented) approach that gives us a baseline for comparison with other options. If funds are distributed equally among all households, the 2030 revenues are high enough to fully cover average increased costs for a New York household making less than $125,000 per year. Not only are average costs fully covered, but many LMI households even see financial gains compared with the BAU. Figure 2 shows net costs by income group and policy case in 2027 and 2030. Expenditure changes associated with energy use (red bars), procurement of home heating equipment and vehicles (orange bars), and the receipt of dividend payments (green bars) are represented. Increased costs are shown as positive numbers (above 0), and savings (household financial gains) are shown as negative numbers. The black dashes represent net expenditure. If the black dash is below zero, the program provides a net financial benefit based on the average household costs for that income group, and if the dash is above zero, the program results in a net cost increase.

As illustrated, the ability to cover the costs scales with the amount of revenue available for distribution in the form of dividend payments. For low- to middle-income groups, a higher allowance price and a larger dividend make them better off relative to the BAU (no CTI program), with the black dash for net costs well below zero, and relative to the low allowance price case. For higher-income households, the dividend payments cover a majority but not all the average price increases. By 2030, on average, LMI households are noticeably better off under the high allowance price case, with higher dividend payments outpacing cost increases. We also see evidence that investing the CTI revenue in the form of subsidies effectively lowers household spending on vehicles and home heating equipment. For low-income groups, negative orange bars show the average financial benefit of the scrappage subsidy. For high-income groups, negative orange bars show the use of heat pump and EV subsidies leveraged by those households.

Figure 2. Annual Costs and Savings with Equal Dividends, by Income Group, 2027 and 2030

The equal distribution approach could make average LMI households better off across many communities, but LMI households in certain regions could face average expenditure increases. Figure 3 shows that average households making less than $125,000 per year face different net expenditures depending on where they live. Energy spending varies geographically, so equal distribution of the dividend payments would benefit communities with lower average spending, like many areas upstate and in and around New York City. Differences in regional expenditures provide a compelling reason to distribute dividends in a targeted way that takes into account higher costs in specific regions.

Figure 3. Average Annual Net Expenditures for LMI Households after Equal Distribution, by Scenario, 2030

5.7. Targeted distribution of dividend payments can cover increased expenditures across households more evenly, especially LMI households

Since CTI-driven expenditure changes vary by region and income group, we also evaluate household costs and savings when we target the distributions of dividend payments by region. We apply the same proportion of total CTI program revenue for dividend payments (30 percent of the overall Climate Action Fund) and distribute the funds proportionally to the 10 economic development regions in New York State (ESD n.d.) based on the share of overall costs borne by each region for households making less than $200,000 per year. This approach weighs regional spending by the number of households and different energy consumption patterns (which reflect different household income levels).

After regional allocations are identified, the funds are distributed evenly among each region's households making less than $200,000 per year. This approach can scale with the amount of revenue raised by the program because the total dividend budget can be allocated accordingly. The resulting targeted distributions in 2030 are shown in Table 6.

Table 6. Annual Targeted Distributions to Households Earning Less than $200,000, by Region, 2030

Targeting the amount of the dividend payments by region and income leads to higher payments for households making less than $200,000 per year, compared with the equal distribution case. A comparison of Figure 4 and Figure 2 makes it clear that LMI New Yorkers experience a greater net benefit from the CTI revenues with a targeted dividend distribution strategy. Furthermore, although net expenditures for households making more than $200,000 per year increase (modestly, Costs are higher for the income group making more than $200,000, but it is important to put those costs in context, relative to annual income. For average households in that group, the increased expenditure on energy represents only 0.2 percent of their income in the high price case, and just 0.1 percent of their income in the low price case. relative to their income) because they receive no dividends, the average costs across the lower-income groups are fully compensated or more than fully compensated (i.e., households save money) for their increased spending (black dashes below zero). This pattern holds for renters, who may have limited ability to electrify their heating systems. Figure 4 also shows that if dividend distributions are targeted, average LMI households are even better off in the high-price scenario than in the low-price scenario because the former generates more revenue to distribute. When the black bars representing net expenditures are lower in the high price bars than the low price bars, that represents savings for households under a high allowance price case.

Figure 4. Annual Costs and Savings with Targeted Dividends, by Income Group, 2030

Targeting dividends based on regional costs can help households cover higher expenditures under CTI in all regions-including areas with higher home heating and transportation costs. Figure 5 shows that for households making less than $125,000 per year in each PUMA, average costs could be fully covered by a targeted rebate strategy. In contrast, Figure 3 shows that in some regions, an untargeted dividend approach leaves the same households with average net expenditure increases. Not only does targeting dividends cover costs, but in most PUMAs, the households making less than $125,000 per year experience on average a significant financial benefit from the program: dividend payments far outpace average cost increases, putting dollars in New Yorkers' wallets and generating savings for households. This is particularly true for the high allowance price case, which shows the greatest financial benefit for LMI households across the state. The pattern holds for regions with disadvantaged communities. Appendix B shows versions of these maps with disadvantaged communities highlighted.

Figure 5. Average Annual Net Expenditures for LMI Households after Targeted Dividends, by Scenario, 2030

5.8. Targeted dividends can cover costs for many households that don't electrify right away.

Figures 4 and 5 include both electrified households and those that have not electrified. Electrified households spend little to nothing on fossil fuels, so they are more insulated from fossil fuel price increases associated with a CTI program. These households may receive the full dividend but experience little to no increase in costs. Figure 6 shows average net expenditures for households that still operate gasoline vehicles and a natural gas furnace in each income group, after the CCAA dividends are applied. As illustrated by the black dashes that often sit below zero (i.e., with net savings), for households making less than $200,000 a year, targeted dividend payments are greater than average costs even if they have not electrified by 2030.

Figure 6. Annual Costs and Savings with Dividends for Households Dependent on Fossil Fuels, by Income Group, 2030

5.9. Targeted dividends make up a greater share of income for low-income households than for high-income households.

Earlier in the report, we discussed the effect of increased costs on household energy and transportation burdens. Our modeling indicates that the dividend to households can offset much of the increase. In Table 7, we calculate the targeted dividend as a share of average income in each group. Compare the numbers in Tables 4 and 7 to see how costs compare to the targeted dividends. For the lowest income group, the dividend makes up a greater share of their income, equivalent to a 3.5 percent pay raise in the high price case and a 1.7 percent pay raise in low price case. Dividends that appear on energy bills as credits, as in California or as explored in the NYSERDA affordability analysis (2023), could also reduce household energy burdens.

Table 7. Annual Dividends to Households as Share of Income, by Scenario, 2030

5.10. A CTI program delivers important climate and health benefits.

This analysis shows how CTI allowance prices and dividend programs affect household costs. Many previous analyses, including several from this research team, have provided evidence that a CTI program delivers other critical economic and health benefits, including savings from reduced health care costs and mitigated climate damages (Krupnick et al. 2023, 2024). Here, we compare the GHG and copollutant emissions reductions in a CTI program under different allowance prices. We find that a high allowance price delivers significantly higher GHG and copollutant emissions reductions (in CO2, SO2, NOX, and direct PM2.5) than a low allowance price.

Table 8 shows the emissions levels for GHGs and major copollutants attributable to the residential and transportation sectors and the percentage reductions associated with each CTI allowance price. Copollutant reductions are greater than GHG reductions because of the substantial emissions reductions in the residential sector, which uses a significant amount of natural gas. Reductions in the transportation sector, largely driven by the ZEV mandate in the BAU, are smaller. For the building and transportation sectors more broadly (including commercial buildings and transportation besides light-duty vehicles), the state estimates a 7 to 9 percent GHG emissions reduction from the reference case, depending on the allowance price. Our modeled emissions reductions do not align with those estimated in the NYSERDA and DEC analysis, which anticipates more reductions in the transportation sector and fewer reductions in the residential sector.

Table 8. GHG and Copollutant Emissions Effects in Residential and Light-Duty Transport Sectors, by Scenario, 2030

Those reductions offer significant climate and copollutant-related health benefits. In the NYSERDA preproposal analysis, the state estimated that pollution reductions associated with CTI could deliver $2.5 billion to $6.5 billion in statewide health benefits in 2030 (NYSERDA and DEC 2024). Benefits from GHG reductions (CO2 and methane) can also be quantified using the social cost of carbon. Based on the cumulative emissions reductions through 2030 associated with CTI (Table 8) and using RFF's most recent social cost of carbon estimate ($180/ton CO2e from Rennert et al., 2022), CTI could deliver $6.6 billion to $9 billion in climate benefits (e.g., reduced peak temperatures, mitigated sea-level rise) by 2030. These benefits ought to be considered alongside the household economic effects we analyze in this brief.

A CTI program can financially benefit households across most income groups and regions in New York State. Specifically, a CTI program with a high allowance price (NYSERDA Scenario A) generates more proceeds for investment and dividends, makes many New Yorkers financially better off, and drives more emissions reductions and associated public health improvements than a lower allowance price (Scenario C). Furthermore, households that electrify by 2030 spend much less on energy than those that remain on fossil fuels across all cases, including the business-as-usual case. We find that the additional revenue generated under the high allowance price case is sufficient to cover average household cost increases for those making less than $200,000 per year through a combination of investments (Climate Investment Account) and dividends (Consumer Climate Action Account) from the Climate Action Fund. Targeting dividends to regions and income groups with the greatest energy and transportation cost burdens enables low- and middle-income households to receive higher payments that in many cases outweigh the cost increases associated with CTI.

Our model estimated significant adoption of heat pumps under the IRA subsidies by 2030, with limited additional adoption if subsidies are increased under CTI. Although CTI investments work to reduce costs, many factors (e.g., installation costs and associated house retrofits) make it difficult for households to adopt heat pumps even if the cost of the equipment is fully covered. LMI households in particular may need more interventions to ready their homes for electrification (Levin et al. 2022). CTI revenue could be invested to lower these electrification barriers for households and increase heat pump adoption statewide. Similarly, EV adoption is largely driven by the requirements of the ZEV mandates, but CTI investments can be used to lower the capital costs of switching to EVs or to compensate lower-income households for scrapping older vehicles.

More detailed work could determine the optimal investment strategy for CTI revenues, but this paper illustrates that a CTI program could financially help many New Yorkers-addressing questions about the affordability of such a program. Our research shows that CTI could not only produce health and climate benefits but also generate dividends sufficient to cover or even exceed the costs for many New Yorkers.

Analyzing Affordability: Supporting Households under New York's Cap-Trade-and-Invest Policy

Read the full report with citations and appendices.

PDF - 9.1 MB

Topics

• Environmental Justice
• Carbon Pricing
• Electric Vehicles
• Energy Efficiency and Demand-Side Management

Authors

Related Content