Equity, Tools & Resources Series 1: Household Vehicle Fuel Use: Share of Income at the Census Tract Level (Text Version)
This is a text version of the video for Equity, Tools & Resources Series 1: Household Vehicle Fuel Use: Share of Income at the Census Tract Level presented on June 28, 2021.
Sandra Loi: Okay. Well, welcome, everyone. Welcome to today's webinar. I'm Sandra Loi from the National Renewable Energy Laboratory. Today's webinar will feature a review of the outcomes of a study done by Argonne National Laboratory titled "Affordability of Household Transportation Fuel Costs by Region and Socioeconomic Factors." This webinar is the first in the Equity, Energy, and Environmental Justice Webinar Series supported by VTO's Technology Integration Program.
Household vehicle fuel consumption, one part of the transportation energy costs, varies geographically, and lower income households generally face higher energy cost burdens. This study on affordability of household fuel costs provides a more detailed understanding of the geographical variation and burden by connecting vehicle miles traveled, fuel economy, fuel costs, and income data at the census tract level. The study's baseline data and framework can help coordinators and others assess impacts of additional transportation costs or transportation policies on affordability for households at the census tract level. This information can help provide a foundation for equity-oriented transportation projects.
Before we get started, I would like to review a few items so you know how to participate in today's webinar. The audio connection options for today's webinar include listening in through your computer or by calling in through the telephone. For the best audio connection, we recommend calling in through a phone line. Be sure to mute your computer if you're calling in over the telephone to avoid possible feedback.
All attendees are muted and will remain so for the entirety of today's webinar. There will be a Q&A session at the conclusion of the presentation. We encourage you to submit questions as the presentation is taking place. You can do so by typing them into the Q&A pane. The Q&A pane is noted as a question mark icon, which should be located on the right-hand side of your control panel or screen. Once we arrive at the Q&A portion, we will address as many questions as time allows.
Should you have any technical difficulties, please reach out to me via the chat feature, which appears on your control panel as a comment bubble.
We are recording today's webinar, and it will be available on the Clean Cities website within the next week or so.
So just wanted to read off this quick disclaimer notice. So as I just mentioned, this webinar is being recorded and may be posted on DOE's website or used internally. If you speak during the call or use video, you are presumed to consent to recording and use of your voice and image.
Now I'd like to go ahead and formally introduce today's speakers. Today, we have two presenters. Our primary presenter will be Dr. Joann Zhou, with Dr. David Gohlke on the line as well.
Dr. Joann Zhou is a principal analyst and group leader of the Mobility and Deployment Group at the Energy Systems Division at Argonne National Laboratory. She leads several DOE and industry-sponsored research projects on energy and emissions impacts analysis of alternative fuel and vehicle technologies, electric vehicle charging infrastructure modeling, and transportation energy burden analysis. Dr. Zhou is a member of the Transportation Research Board's Alternative Transportation Fuel and Technologies Committee.
Dr. David Gohlke is an energy and environmental analyst in the Systems Assessment Center at Argonne National Laboratory. His work focuses on understanding the societal impacts of advanced vehicle technologies, including vehicle electrification and connected and automated vehicles. Previously, he held an AAAS Science and Technology Policy Fellowship at the Vehicle Technologies Office at the Department of Energy.
Now I'd like to pass things over to Joann. Joann, you may begin.
Joann Zhou: Okay. Let me see—well, do you see the next slides? You see the next slide? My slides have advanced?
Marcy Rood: Yes.
Sandra Loi: Yes. Mm-hmm.
Marcy Rood: We see them.
Joann Zhou: Okay. Great. Thanks, Sandra, for your introduction. So I'm Joann Zhou with Argonne National Lab. For people who don't know where the lab is, we're located just outside of the Chicago area, 30 minutes to downtown. So we are one of the 17 labs managed—that belong to the Department of Energy, but we are managed by University of Chicago, actually.
So today, I have my—one of my co-authors, Dave Gohlke, here. We are here to present our study, titled "Affordability of Household Transportation Fuel Costs by Region and Socioeconomic Factors." So this is the first webinar—Sandra mentioned, this is the first presentation of a series of webinars focused on transportation equity that is sponsored by—that is supported by the Vehicle Technologies Office. However, the study was founded by EERE Strategic Analysis Program. We like to acknowledge our sponsor for supporting the study.
So I mentioned that I have Dave Gohlke with me here today. We're here to present the study together. We have another contributor to the study, Spencer, which right now is on vacation, so we wish him to have a good vacation, a good trip. But any questions, we have our contact in the end of the slides. Any questions, please feel free to ask during the webinar or send us an email later. Okay?
So we know that transportation energy is actually an important component of a household budget. So you can see there's a bar chart showing here that transportation actually accounts for about one-third of the total household expenditure categories. If we further break down the transportation categories, then we see that besides the vehicle purchase costs, that people pay for the gasoline and motor oil fuel, that is actually the second biggest transportation cost component at the household level.
So if you look at it from a different angle, that's looking at how much the household would pay for their energy to run the house, and vehicle fuel also count about over half of the total cost that a household would pay for energy. It's equal to the sum of what they would pay for electricity and natural gas together. So from those perspective, the transportation energy costs, or in other terms, is how much a household would pay for vehicle fuel, is one of the important cost components.
At a national level, the household energy cost is counted as 3 percent, 3.4 percent, of household income. However, that varies a lot geographically across the region and across income groups, and the lower income groups are always facing much higher transportation burden comparing to other households and communities.
So with that said here we define transportation—household transportation fuel cost and the affordability as a transportation energy burden as a percentage of the household income spent on the annual fuel cost. So how do we calculate household fuel costs on an annual basis? It depends on three important factors, besides the income.
So first, it depends on what we call efficiency of the household is running. Also, it depends on the fuel price, meaning how much they would pay for the gasoline or other fuel types they use. It also depends on the annual vehicle miles the household would have driven. So for the other information, like the household income, the vehicle efficiency, and the fuel price, and we can collect data on a zip code base or the census tract basis, given the data resource.
However, how do we estimate annual vehicle miles driven by each household or representative household at zip code or census tract level? That's the key research question, how that would vary by the socioeconomic and demographic factors across the U.S.
So the research framework actually can define by small steps. So here, back to the functions that we showed earlier, we need to calculate the percentage of the household income spent on annual vehicle fuel costs, and that depends on three important factors, besides income, which are vehicle efficiency, fuel price, and household vehicle miles driven per year.
So the first step is to project the household annual vehicle travel, based on the socioeconomic factors. We use the very high resolution demographic characteristics for each census tract to project household annual travel. And the table here is just showing a representative census tract. For this census tract ID, which is in the suburban Chicago area, that you see we further divided the census tract, the households in the census tract, to 220 groups, based on the important socioeconomic factors, such as number of workers, number of vehicles per household, number of workers per household, and household income groups. So for each census tract, we have over 220 groups defined.
We project the VMT, the annual vehicle miles traveled, for each of those 220 groups per census tract. After that, we estimate the stock-weighted vehicle efficiency—it's also known as on-road vehicle efficiency—by combining the vehicle registration data with the fuel economy by make and model. So this step is to understand what's the on-road vehicle efficiency driven by a representative household per census tract.
The next step is to collect the fuel price. Currently, most vehicles on the road are running on gasoline, so one important step is to collect the gasoline information at the zip code level. And we collect our gasoline information from a website called GasBuddy.com. They have a very high resolution gas price at zip code level, but we did compare the gasoline price with the state level and the national level numbers reported by EIA. Other fuel types, such as diesel, natural gas, electricity, we use the information from the Alternative Fuel Data Center that's hosted by National Renewable National Lab.
With all this information identifying the first three steps, annual VMT, vehicle efficiency, and fuel price they would pay, so we can calculate the cost that a household would pay for running their vehicle on an annual base, and we divide it by that household income. So we've got the household transportation energy burden.
So how do we estimate the household annual vehicle travel, also known as VMT? As I said, that's one of the key research questions of this study. So in order to project that, we actually applied machine learning techniques on National Household Travel Survey samples to develop models to project the household annual VMT, based on the socioeconomic factors. In National Household Travel Survey, there's a variable called self-reported annual miles. So the respondent of the survey actually reports annual miles driven by each vehicle within the household. So the sum of the miles driven by each vehicle from the same household, we got the annual miles driven by the household in total.
So this chart here showing the raw data we summarized on the National Household Travel Survey by census region and by urban, suburban, and rural areas. So you can see that there's a variation in total annual mileage driven by households across the census region, as well across the urban, suburban, and rural areas. The color here indicates the census region that is showing in the map here, and you see which states are covered in each census region. And so you can see the yellow dots representing the Northeast area, and you see Northeast urban area has the lowest—the household in Northeast urban area has the lowest annual miles. And the rural area on especially the South—I think it's the South Central—yes. The household in South Central rural area has the highest annual miles.
So using the sample data from the National Household Travel Survey, and we first identified five important socioeconomic factors that are contributing to the household annual travel based on literature review and data availability. So here is the five variables in order of feature importance, how important they are correlating to the household annual travel.
So the first one is the number of vehicles owned by the household. The second one is number of workers per household. The third is household income. And the number four is the house and unit density. And this is the variable we used to define whether the census tract belonged to urban or suburban or rural areas. Number fifth factor is called a life cycle factor. This one indicates number of children or number of senior people per household.
So with all these five variables, in order to consider the heterogeneity among the census regions and rural/urban areas, which is showing in the chart—in the figure here, we actually developed 18 different VMT projection models. So that's considering six census regions, and for each census region, we considered three urbanicity groups. So in total, we have 18 different VMT projections modeled. And we run the model 220 times for each census tract. Remember, for each census tract, we have 220 groups because of the number of vehicles, number of workers, and household income groups they have.
So we run the model for each census tract 220 times to get a VMT projection for each socioeconomic group, based on the factors showing above. So we got this fine resolution of household annual VMT projected, and that will serve as the base for us to further calculate household annual fuel costs, and also quantify the burden as a percentage of the fuel cost—as a percentage of their income.
So here, I'm showing the results. So back to the function, so this is the annual household VMT. We used the machine learning model to project the household annual VMT based on the socioeconomic factors, and we are showing the result here by county. And the results are actually available by each census tract. However, in order to show the result in a small map, we have to aggregate to the county level.
So as you can see from this map, there's a wide variation in average household VMT across the U.S. The national average annual household VMT is about 18,500 miles. However, at a county level, that number could vary from 2,500 miles to over 40,000 miles by county. And if you look at this map, you can see that households in Pacific and Mountain region have the lower annual VMT, and households in Middle West, some of Middle West regions, has the highest VMT. Suburban and rural households have a higher annual VMT than urban households. And you almost can see the urban area lighting up on the map, and those are the areas that has the lower household annual VMT.
We also found out that annual household VMT increase as the household average income increase. And the chart here showing the annual household VMT by income groups. So the X axis showing the income groups, and the Y axis showing the annual household VMT. And the half-violin plots just showing you the distribution of the household VMT for each of the income groups, and each dot here represents one census tract. And you can see as the income group increase, the average, which is showing in the central line of that box, increase over the income, with the household income increased, so meaning that wealthier households actually driven more miles than the lower income groups.
We also found out there's a much wider variation in the higher income groups compared to the lower income groups, possibly due to the variation in the number of vehicles owned by those households. Imagine a household, rich household in a suburb area or rural area could have multiple vehicles per household, and a household in—living in downtown Chicago or New York City, they could have zero or just one vehicle. So the variation in the number of vehicles owned by the wealthy household create possibly contribute to the variation in annual miles driven by them.
However, our focus is the lower income groups, and you can see there's lower—not only lower annual incomes, but lower annual VMT driven by those households, and the variation is much less than higher income groups.
So back to our function, so another important—besides the VMT, which I talked about in the last slides, another important variable is the vehicle efficiency. So what kind of vehicle you are driving does contribute to how much you would pay per mile in a year for fuel. So most of the vehicles on the road right now is a gasoline vehicle. For other fuel types, we convert the MPG to GGE, gasoline gallon equivalent.
So we combined MPG per make—by make and model, that data from FuelEconomy.gov, combined that vehicle efficiency, MPG, with the vehicle registration data from IHS Markit registration database. So we developed this database showing average on road vehicle MPG by census tract across the U.S. Again, result is available for each census tract. The map here is showing aggregated efficiency of on-road vehicle by county based on 2018 registration, which was the latest data we had when we conducted the study.
So you can see from the map that on road vehicle efficiency actually varies from 15 miles per gallon to over 23 miles per gallon. Again, on this map you can see the light area, lighting up on the map, indicating metropolitan areas, which indicates that vehicles in urban areas are much more efficient than a vehicle owned by a household in a suburb or even rural areas. And you can see that in the Middle West region, again, you have one of the lowest on-road vehicle efficiency, where in California and the Northeast region have much higher vehicle efficiency, and most of them are because of the high adoption of alternative fuel vehicles, such as electric vehicles, and also adoption of newer vehicles, as we know that newer vehicles are more efficient than older vehicles, because the vehicle technology constantly increase, improving over time.
So we want to use one region as example to show how the MPG is correlated with vehicle efficiency. Here, the map is showing Washington, D.C. vehicle MPG, the on-road vehicle MPG, and the vehicle age by census tract. As you can see that the on-road vehicle efficiency are correlated with vehicle age. And in the census tract, the lower MPG, which is showing the darker color on the left chart, also has a darker color on the right chart, which is showing the vehicle age, meaning the households in that census tract has a much—are driving a much older vehicle compared to the households that has a higher income, which are in the northeast or northwest part of the city. For people who know the city well, you know those parts of the city are the households that has a relatively higher income than the households on the other side of the city.
So vehicle age, vehicle efficiency, is one big contributing factor to the on-road vehicle efficiency driven by the household.
So again, back to our function, another factor important to our calculation is the gasoline—is the fuel price. And most vehicles on the road right now are still running on gasoline, so one important fuel price is the gasoline price. We collect zip code level gasoline price from GasBuddy.com. They have a self-reported gasoline price by zip code. And we further aggregated that zip code level information to state level and region level, even to the national level, in order to compare with the number reported by Energy Information Administration on numbers reported on a weekly base. And those numbers, those two numbers agree well, so that gives us confidence to use the zip code level, high resolution zip code level fuel price from GasBuddy.com.
And in order to pick representative gasoline price across the region, we chose the February 26, 2020 fuel price on that day as a snapshot across the U.S. by census tract and zip code in our calculation, and then we compared that gasoline price, which is $2.46 per gallon, with the last five years' gasoline price, and we found out it's about—it's close to now the average across the five-year period. So we think that gasoline price is more or less representative in our calculation. It's not too high nor too low to—for example, during COVID times, the gasoline price was dropped to a very low level, and in some other times, the gasoline price was pretty high.
So the gasoline price actually varies across temporally, and also across the region, which is showing on the other chart that you can see even for that given day, February 26th, there's a significant variation across regions, across states. And you see the spike in California, and also again in New York City, and in Chicago, Illinois. And we chose the different times, just get snapshot of different months, February, April, June, September, in 2020, just to give you the impression that gasoline price varies both temporally and also across regions, and those regions, such as California, Illinois, New York City, constantly have a higher gasoline high price comparing to other regions, other states.
So for other fuel types, other than gasoline, for example, electricity, diesel, natural gas, we gathered the information from the Alternative Fuel Data Center. These are at the county level or state level. Those fuel prices are—those type of vehicles are small, small portion of total registration, so they don't really affect our result for developing this baseline.
For other fuel types, we multiplied on-road MPG by the cost of fuel, each fuel type, to find a cost per mile of the operating—operating each of that vehicle model, that we first aggregated to census tract level for our first calculation.
So with that all said, so the gas—now we have the household annual VMT projected per household. We also collect the gasoline fuel price and other fuel price from different data sources, and we calculate how much people—a household would pay by rounding that vehicle per mile. So you say it's $1.00 per gallon, and then we times with the efficiency of the vehicle owned by the household by each census tract. So with all those three variables, we could further quantify the annual fuel costs that a household would pay for running that vehicle, paying that vehicle price, and rounding that even mileage, divided by the household income, so we can quantify the household transportation energy burden at the census tract level.
So the charts here are showing you the aggregated result at a state level, and the bubble size indicate population size for each state. And the color indicates census region. And you can see that relatively speaking, you can see from the chart that the household income—sorry, the household transportation energy burden actually improves as the median household income increases, so as the Y axis increase in terms of income, we see the percentage they spend on fuel costs would decrease, meaning the burden is less. You can see that D.C. is a kind of outlier on the far right, because the residents and households in D.C. are dependent on public transportation, and they would pay less fuel for running their own vehicles. Some of the households there doesn't even own a vehicle.
And from state level, you can see that some of the states has a higher transportation energy burden. For example, California, that could be because of different reasons. They pay a higher gasoline price, but even though their vehicle efficiency is higher than some of the other states, they pay a higher gasoline price, and the annual household vehicle miles are higher than other states.
For some other—the highest state average is in Missouri. It's 4 percent. So the average in the state household is spending 4 percent of their income on fuel costs, which is higher than the national average.
Each of the bubbles here represent one state, and each state, the energy burden are contributing by different factors. As I mentioned, in California, even though they have a good vehicle efficiency, they could pay high gasoline price, and they could run more annual miles. Some other states, they could have a lower efficiency of the vehicle running on the road, even though their fuel price is reasonably low compared to other regions.
So we can first quantify the result by census tract, and we do have the information available by each census tract. But here, again, we presented the result by the county level. Remember, I mentioned on the national level the transportation energy burden is 3.4 percent, and that percentage could vary by region, by community, and by rural and urban area greatly. So what's the number here, so what is the real number? So we can see at a county level that variation could vary from almost 0 percent to over 23 percent, so meaning in some of the counties also could pay over 23 percent of their income, out of their income, for the vehicle fuel.
And I have a blowout of Illinois just as an example. So for Illinois, that percentage varies from 0.6 percent to almost 13 percent by census tract, and you can see from Illinois that Chicago region, which is on the north corner that has a lighter color, which meaning the household in this urban area has a less transportation energy burden. However, in other part of the state, especially in some of the southern portions of the states, they are facing—the households here are facing much higher transportation energy burden.
And back to the national lab, again, you can see that across the U.S., the urban area has a lower energy burden compared to the household in the suburban region and the rural regions. And in California, again, even though you see the San Francisco area, and maybe Los Angeles area, has a lower transportation energy burden, however, in some other parts of that state, the households face much higher energy burden compared to the state average and the national average.
So we first run the statistical analysis to see, out of the three variables, the vehicle efficiency, gasoline price, and household annual VMT, what is the major contributing factor to the energy burden? This is—the motivation is what's the action we can take, or learning from the result here, that what we can do to relieve the transportation energy burden at the household level, and to further help the community and neighborhood in facing higher energy burden.
So besides the income, as you can see from this table, the numbers are less relevant, more important to see the correlation, the magnitude of correlation. So you can see from the table, we compared the major contributing factors, household income, fuel consumption, which is the conversion - the fuel efficiency of the vehicle, and the vehicle—household VMT, and the fuel price they will pay. And out of these variables, besides the income, fuel costs, fuel consumption, actually highly correlated—sorry, the energy burden actually highly correlated with vehicle efficiency.
So back to our function, out of the three variables besides the income, vehicle efficiency is contributing the most to the household transportation energy burden. So what that means is that adoption of more fuel efficient vehicles, especially among the lower income households that have the highest impact on improving the household transportation energy burden, because that's the most important contributing factors to the variation in transportation energy burden across the U.S.
We did a quick—we did a calculation comparing to the vehicle to—comparing the vehicle on-road vehicle efficiency between 2018, which is the number we used in the study, with 2016 registration data. So we have the registration data for 2016 and 2018, and consider the gasoline price and household VMT do not change from 2016 to 2018, we want to quantify how much energy cost that household could save just because the vehicle efficiency improvement. So comparing the 2016 with 2018 vehicle registration and the fuel vehicle efficiency of those vehicles registered at the household level, they have a 3 percent improvement across the country, which is shown in the map here. Some of the states has a negative change, but most of the states has an improvement in on-road vehicle efficiency, especially in the West Coast, California, Washington, and Oregon, and some in the Northeast regions. So most of the regions—most of the census tracts in the United States has improvement in on-road vehicle efficiency from 2016 to 2018.
And that 3 percent improvement in the stock MPG actually saved American households over $8 billion, and this improvement is largely due to improvement of conventional vehicle fuel economy. So people start to adopt new vehicles, and a newer vehicle has a newer technology, which is more efficient than older vehicles. It's also due to the increasing electrical vehicle adoption in those regions as well. So you can see in California, in the Northeast, where the electric vehicle and the newer vehicle registration are higher, their improvements are higher than the rest of the states.
So other benefits, such as the GHG emission reductions, could also be quantified using this framework. We only now quantified the fuel cost savings as an example here.
So the study developed a framework that you identify the baseline of the regional affordability level, and quantify the overburdened fraction of the households in each census tract, counties, states, or for the entire United States. So the baseline showed percentage of households that are spending above a given affordability threshold or burden threshold on household vehicle fuel, which is also called the transportation energy burden.
So we have a chart here to show an example. For example, the dotted black line is showing the national average of transportation energy burden on the household level, and you can see how many percent of the households in the U.S. by each of the selected states are over that burden.
So for example, if you defined—look at the dot there, we give an example. That's 33 percent of the households in the U.S. actually spending more than 3 percent of their household income on fuel costs. So if you define the threshold as 3 percent, you want to improve on the transportation—household transportation energy burden, you want first an understanding, what fraction of the households in your community are spending higher than that threshold. So that dot gives you that 33 percent of the households in the United States are spending higher than that. And that number could vary by state. So you can see in California and in Oregon, a certain percentage of households are spending more than that threshold defined, and—however, in Illinois and New Jersey, maybe the percentage of households—there's no percentage of households spending that higher than 3 percent of their income on vehicle fuel.
If you move that threshold to a lower number, then you see more households that would be over that—would be overburdened than that threshold. For example, if you move your threshold to 2 percent, you'll get more households spending more than 2 percent.
So this framework identified a baseline with the current situation, so it helped the community—helped the stakeholders to identify underserved communities, where they are, how much is their energy burden, and what's the population that are included in those communities, so we can further invest in the clean energy technologies, and to benefit those communities.
So we identify a community that could benefit the most from energy efficiency technology, so we can direct future investment to those communities, help them to improve the vehicle technology or other technology that which further will reduce their energy burden.
We found out that a community with higher transportation energy burden are actually also facing higher environmental burdens. So here, we used Chicago as an example. So the left chart is showing you the percentage of income spent on vehicle fuel, so the darker color, again, showing they spend more of their income on vehicle fuel, which means higher energy burden. And the Chicago area which is circled in the bluer rectangle, that is actually corresponding to the chart that we got from NRDC that's showing the energy—I'm sorry, the environmental burden by each neighborhood in the same area.
So again, in the chart on the right, the darker color is showing higher environmental burden. So as you can see, some of the households has a higher transportation energy burden, which showed on the left chart—for example, the southern Chicago area—that are also facing a higher environmental burden, which is showing on the right chart.
So again, so this is showing that our framework to help indicate the communities that can benefit from energy efficiency technology, that not only reduce their transportation energy burden, but also could improve the environment, to further reduce their environmental burden faced by those communities.
And our data is actually available on a website created by our DOE sponsors. Thank them for making the website available. So the link is listed here. Let me see whether I can switch to the website. I've got it open here. Yeah, if you click the link I showed on the slides, which we will share after the presentation, you will go this OpenEI website, and this is the beta version. We are still putting some missing numbers for Alaska and Hawaii, and there's one missing county in Nevada. But most of the data are available here, that you can see the household energy burden across the country by each county, and you can navigate through—you can navigate through here to show, if you want to put a burden as a different threshold, that how many percentage of households still are over that burden, the fraction of households would pay over that threshold we defined.
You can switch to different states, and for each state, you can further see the countries. Ok. Yeah. You can choose the states and the counties to see the percentage of households that are spending more than the threshold you define here.
And moving down, you can see other data, such as energy expenditure, by income groups, and also there's other information available, such as energy burden. So this is the energy expenditure, that's the cost, and then we divide it by income, that's energy burden, and also, we have miles traveled for each of those counties.
We also have the result available at the census tract level. If you need that information, please feel free to contact us. And this website is live now, so you can click the link and explore and see how that can also help you to identify the communities that are facing high energy burden.
So in summary, so we estimate household transportation energy burden based on the household annual vehicle travel, the on-road vehicle efficiency owned by those households, and the fuel price they would pay in their zip code. The conclusion is the variation in transportation energy burden is—largely varies by region, by census tract, by the socioeconomic factors. However, it's largely explained by the vehicle fuel efficiency that—of the vehicles adopted by each household.
So currently, wealthier census tract has vehicles with much better fuel economy, so the household—wealthy households, they either have much newer vehicles or they have electric vehicles, which has a much better fuel efficiency on average, comparing to the lower income communities.
So the application of the study can help the stakeholders to identify the community that could benefit most from energy efficient technology, and how much households you could help. And future work includes we will plan to include other vehicle ownership costs into the same framework. For example, the vehicle purchase cost, the vehicle registration, insurance, maintenance costs. As we talked in the first slide, that vehicle purchase cost is the biggest component, and there's other cost components, such as insurance, maintenance, are also contributing to the total ownership cost. So we want to integrate all the cost components into the same framework, so we can further quantify the transportation burden at the household level per census tract.
With that, I finish, and any questions, please let us know now or later by email. And again, we'd like to thank our sponsor for funding this study. Thank you. Thank you for your time and interest.
Sandra Loi: Thank you, Joann. That was an excellent presentation. Thank you for walking us through the study and the results there. For those who may not have seen it, definitely check your chat box. We have a link to the tool that Joann just called out, as well as a link—direct link to the report, so you can access those there. And they'll also be linked in the presentation.
So we do have a few questions here, and feel free, audience attendees, to go ahead and post or submit additional questions. We'll take as many as we can. So Joann, my first question here, this was a little bit at the beginning of your presentation. You were talking about testing or running the data through about—I think you said 200 times. Is that to train the machine?
Joann Zhou: Two hundred groups for each census tract, based on the variables we selected. We have to further divide the census tract to over 200 groups, and then we project vehicle miles traveled for each of those groups within the census tract. It's just because of the variables we chose, we have to further develop the census tract to different groups.
Sandra Loi: Great. Thank you. The next question here, were households' VMTs adjusted for employment related to VMT?
Joann Zhou: Sorry, say that question again.
Sandra Loi: Sure. Were households' VMTs adjusted for employment-related VMT?
Joann Zhou: This is the total household annual VMT. It could—it could include employment-based VMT, but we didn't separate a fraction of that VMT is commute, or work-related trips. But as you can see, one of the key variables that we identified contributing to the annual household VMT is the number of workers per household. So I would imagine that most of the—that a big portion of the household annual VMT are contributed by the commute travel, employment-related travel.
Sandra Loi: Okay. Great. So a little bit of a long question. So on the slide showing variation of annual VMT by income, you suggest a number—just a second—the number of vehicles per household may play a factor. It seems to me that the number of people is a bigger driver than number of cars, because cars have to have a driver to be out on the road. Did you check number of people per household? More people driving more cars definitely drives up VMT, but does the data show that the VMT increases even if you control the number of people? Like maybe wealthier people are joyriding in cars they drive for fun.
Joann Zhou: So that's a great question. So we—eventually, we identified those five variables that contributed to VMT, but we actually in the beginning of the study, we explored dozens of variables, based on the literature review and data availability. So household size, which is the number of people per household, is one of the factors we considered. But actually compared to the variables we listed here, especially the first four variables, the household size is—has a lower correlation compared to the first—the four variables here. So we decided to not include the household size eventually in our model, and instead, we used the five variables here. And that was considered—explored in the beginning of the study.
Sandra Loi: Okay. Great. And David, we want to invite you to jump in if you'd like as well. I know you've been listening in as well.
We have another question that came in here. When you look at other vehicle ownership costs, will you be including insurance and registration? My guess is those will also vary greatly by location.
Joann Zhou: Yeah. I'd like to have Dave Gohlke answer that question, since he is our expert in total cost ownership. Dave, you want to jump in?
David Gohlke: Oh, fun. Yes. So we are indeed looking towards that going forward. We focused on the energy costs, the fuel costs, to begin with, partially because focusing on the Department of Energy needs and looking at the cost of fuel obviously is directly tied to energy, but also secondarily because of the high volatility. So gas prices can change very rapidly by year, by state, even by day or by week, whereas there's less volatility, especially temporally, in insurance and maintenance and repair costs. But that is something that we are looking into, along with local differences in taxes and fees, which are something that could impact it, and also kind of what those vehicles are, because that changes the depreciation of any of these. The differences in depreciation change the overall cost of ownership as well.
So yes, looking forward, more holistic, beyond just fuel, is indeed where we'd like to take this project.
Sandra Loi: Great. Thank you.
Joann Zhou: Thank you, David.
Sandra Loi: Thank you both. And Joann, I think you covered this, but I just want to make sure that it was brought up. There's just a general question. Will you also share the mean annual household VMT?
Joann Zhou: Yes. So the county level information is available on the website I showed earlier. And you can download—I believe. I didn't try to download, but it should be available for you to download the given county or given state you chose. It will be here, that VMT is part of that. But if you want census tract level information, please contact us, and yeah, for any region that you'd like to see, we can share that projection with you.
Sandra Loi: Perfect. Thank you. I'm not seeing any new questions. Feel free to go ahead. We have a couple of minutes, if anyone else has any questions. Go ahead and submit them. Joann, did you have something else to add?
Joann Zhou: No. I don't. I welcome any questions you might have, and here is our contact if you have further questions later on.
Sandra Loi: Great. Wonderful. Yeah. We'll with Joann's and David's permission, we'll go ahead and post those slides as well, along with the report and you can have access to that.
Joann Zhou: The question says show the variables on transportation energy burden again? Okay, I can go back to that slide. Is this the one? Why it's not showing the first? Here is the one. Showing the variables? Okay. Slides where we show the fuel efficiency has a strong correlation. I think that that was—yeah, that was this one.
Now here, this is just a simplified statistical analysis result. We actually ran the statistical result at census tract level, state level. Here is aggregated results. So the number is less relevant. It is more the correlation that is important. We want to show that—I mean, the VMT is certainly contributing to the fuel cost and energy burden. As you can imagine, the more you travel, the more you drive your vehicle, the more you will pay.
However, that variation is not a big explanatory variable to the variation in the fuel consumption, and—sorry, in the transportation energy burden of the final results. And the statistical analysis actually showed the vehicle efficiency, which converts to the fuel consumption per mile here, is the biggest contributing factor that can explain the variation in the transportation burden across the United States by census tract, which shows that if we improve the vehicle efficiency on road, that would have the biggest impact on reducing the energy burden.
Sandra Loi: Lori, does that answer your question? I'm not sure if you had additional—oh, bingo. Perfect. Okay. You explained it all. Thanks for confirming.
And I just got another question asking if the link will be shared with all registered attendees. Yes, we'll send a follow-up email with that link. And then I also put the link in the chat to where you can find where it will be housed as well. So you can look on there in the next week or so, and I also plan to send an email to everyone that did attend today as a follow-up. Great. Any other questions? I'm not seeing anything in here. I don't know if David or Joann, if you had anything else you wanted to add?
Joann Zhou: Dave, do you have anything to add?
David Gohlke: Nothing in particular, other than thanks to everybody for attending.
Marcy Rood: Dave, this is Marcy. I did add your study, the total cost of ownership study, since you were talking about that earlier. It's in the chat.
David Gohlke: Sure. I don't see it in the chat on my end, but I can send a link to that as well. So for those who don't know, there was a multi-lab study funded by the Department of Energy's Vehicle Technologies Office looking into total cost of ownership of both passenger and fleet, say commercial vehicles, as well. That's a very comprehensive study. It is at a—more of a national average level, as opposed to a individual vehicle owner level. There could very well be fluctuations that any individual owner has that aren't explicitly accounted for in that report. But it is a pretty comprehensive assessment of these different costs that make up total cost of ownership, including the vehicle and fuel costs, of course, but also as were asked earlier, insurance, registration, maintenance and repair, and costs of that nature. So I'll drop the link to that into the chat window as well.
Sandra Loi: I also had another question here about—it was interested—they'd be interested to know how people are using the data, and how to—and maybe ask them to report back. So I don't know if you have folks already using this information and different types of analysis and things, and just curious. I think maybe they're just more curious how people are using it and how they're—and maybe to give others ideas on how they could use it.
Joann Zhou: Right. So since our study published, we got several requests from local agencies, utility companies, and nonprofit organizations to get access to our regional data, like our projected VMT, the on-road vehicle efficiency. My understanding is they could use this to of course identify the communities, census tracts, that are facing higher burden, how much higher than the threshold they define, or just national average, state average, and help them to direct the future investments, and also help them to apply funding.
For example, even Volkswagen funding some other state or federal funding for deploying clean energy technologies, so they know which community that could be relieved, benefit from adoption of those advanced technologies, and how much households we are talking about, how much burden we can reduce, which transferred to how much cost you can save later on on the household level.
Another benefit like emissions that's also popular actually we didn't show a result here, but it's something that's relatively straightforward, based on the data we have to quantify the environmental benefits.
Sandra Loi: Yeah, that's great. And maybe those on the line, if you do end up using that data, maybe you could report back to Joann and David just to say this is how we're using it, and then they'll kind of collect that and be able to share that with others. I think it'd be interesting to kind of see how it, you know, is used in the future.
Yeah. Okay. Well, I'm not seeing any other questions coming in. And I think one of our Clean Cities coordinators had an idea on how to maybe use that data, so we can certainly share that with Joann and David and even others if they're willing, afterwards. So we can go ahead and wrap up today.
So thank you all for participating on today's webinar. Thank you, Joann and David, and Marcy as well, in the wings, for presenting today some really great information. If you do have any questions, feel free to reach out to myself, Sandra Loi, or anyone else here on today's webinar. And as I mentioned, it's being recorded, and we will post it within the next week or so, and we'll have the slides available as well. And I will send a follow-up email once that is available.
So thank you again, and have a wonderful day. Thank you.
Joann Zhou: Yeah, thank you, Sandra. Thank you, everyone, for your time and interest.