User Tools

Site Tools


Author: Bryan P. White

Original Publication: 10/28/2018


US automobile emissions have been both a contentious issue from the political spectrum as well as from the consumer spectrum due to car manufacturers need to maximize profits and automobile efficiency, but also adhere to state and federal environmental regulations. While the US Environmental Protection Agency (EPA) has laid out clear goals for auto manufacturers that sell cars in the US, its unclear what efforts those manufacturers have made to meet those standards. One way to gain insight into how auto manufacturers have been responding to recent regulations regarding CO2 emissions would be a shift from gasoline-only vehicles to more greenhouse-gas friendly vehicles, like electric vehicles or gasoline-electric hybrids. In this study, I found that there were overall trends towards an increasing use of gasoline-hybrid vehicles, but also an increase in gasoline-only vehicles, suggesting vehicle manufacturers are offering hybrid vehicles as more of an additional consumer option rather than a replacement for traditional gasoline-only engines. This study highlights the need for further research in this important US regulatory issue.


Automobile emissions are one of the greatest contributors to total global greenhouse gas (GHG) production per year [1] through the release of Carbon dioxide (CO2) and other greenhouse influencing gasses. While United States requirements for fuel efficiency shift from year to year and administration to administration, the consensus is that car manufacturers must meet some specific deadlines for the implementation of fuel-efficient car fleets to offset some of the late-stage, negative effects of global climate change [2]. Two leading efforts to implement scientifically informed climate regulations are California’s own state auto emission standards [3] that tend to be stricter than federal standard, and an international effort called the Paris Climate Agreement [4]. Recently, the US Environmental Protection Agency (EPA) has indicated that it might revoke California’s waiver to set its own auto emissions standards [5], and the US also announced that it would withdraw from the Paris Climate Agreement [6], but follow the 4 year draw down for an effective withdrawal by 2020, meaning that US states could begin implementing local regulations despite the impending federal withdrawal [7]. These events raise the question as to how much effort US auto manufacturers are being held accountable to for implementing cars that might help reduce GHG emissions.

A lack of fuel-efficient vehicle options might suggest that relaxed federal regulations have led to more relaxed fuel-efficient implementations at the fleet level. Understanding current trends in vehicle manufacturing related to fuel efficiency is an important step towards making informed decisions about present and future policy making as well making purchasing decisions from a consumer perspective. In this study, I utilize fuel economy data provided by the EPA to gain insight into how both policy makers and consumers might make decisions to influence that will influence the implementation of auto emission standards in the US.

Study Population and Data

To better understand what efforts car manufacturers have taken to reduce the overall GHG impact of their car fleets I analyzed a data set of fuel economy metrics made available from the EPA’s National Vehicle and Fuel Emissions Laboratory (NVFEL) in Ann Arbor, Michigan []. The NVFEL is the primary laboratory used by the EPA to study fuel emissions and supports the overall operations of the EPA in terms of vehicle emissions programs and works to develop technologies aimed at reducing vehicle emissions and increasing fuel efficiency. Data from the EPA Green Vehicle Guide were obtained from the EPA NVFEL website for years 2016, 2017, and 2018 [9]. Initially, this combined data set contains fuel economy data from 873 vehicle models and 50 vehicle manufacturers under both federal and California-specific emissions regulations. Since my initial research question is to better understand how car manufacturers might be implementing auto emission regulations across their fleets, I reduced the data set to vehicles from within the top 10 manufacturers by number of models produced. I also removed models that were specifically designed for California-only regulations leaving only vehicle models adherent to federal regulations behind. After the data reduction step, a total of 10 vehicle manufacturers and 472 vehicles remained. Data from the EPA NVFEL has been used in previous studies to assess the impacts of biodiesel on oil dilution in engines [10]. Ongoing uses for NVFEL data include the ranking of vehicles Greenhouse Gas Score and Air Pollution Scores, as well as determining official miles per gallon (MPG) estimates that are used to certify vehicles according to federal regulations.

Statistical Methods

Summary tables were produced in Microsoft Excel for the number of unique vehicle models produced by top 10 vehicle (by unique model abundance) manufacturers made available from the NVFEL website. The count and proportion of each manufacturer’s contribution of unique vehicle models to the total population of vehicle models for each fuel type were summarized for the years 2016-2018 in Table 1A. The proportion of each fuel type within a manufacturer’s unique number of vehicle models was summarized for the year 2018 only in Table 1B.

To assess trends in the number of each vehicles produced by fuel type for each vehicle manufacturer, a Pearson’s correlation coefficient was calculated for the number of unique vehicle model within each year for each manufacturer (Table 2). To understand overall trends in the change manufacturer fleet production over time, manufacturer’s that had a positive correlation (increasing production of a fleet type) over time were counted as “Increasing”, and those that had a negative correlation (decreasing production of a fleet type) were counted as “Decreasing”. To determine if the overall trends observed according to the counts of Increasing/Decreasing correlations, both a Chi-Square and Fisher’s exact tests was performed using R (R Core Team, 2013) in R Studio (R Studio Team, 2015).


A total of 2,108 data points representing fuel type for unique vehicle models from 10 vehicle manufacturers were analyzed in this data set. On average, gasoline-only vehicle models represented 89.6% of vehicle manufacturers fleet, with diesel, gasoline/electric hybrids, electric-only, and ethanol representing less than 5% each, and hydrogen vehicles accounting only for 0.4%. Only two vehicle manufacturers included hydrogen vehicles in their 2018 fleets, which suggests adoption of hydrogen technology has been extremely limited.

According to a correlation coefficient analysis, vehicle manufacturers had trended towards increasing the number of unique vehicle models for gasoline-only, gasoline-electric hybrids, and electric-only vehicles. Manufacturers tended to decrease the inclusion of diesel and ethanol-gas hybrid vehicles in their fleets. Hydrogen vehicle inclusion was on average unchanged, but only represented in two out of 10 manufacturers. Tests of significance on the number of increasing and decreasing inclusions of vehicles by fuel type were negative according to both Chi-Square test (X2 = 13.5, df = 12, p = 0.33) and Fisher’s exact test (p = 0.138) using a 5% level of significance.

Limitations and Strengths

Some of the limitations of this data set are that it is unknown what types of decisions car manufacturers go into when deciding how many different types of car model to produce. In other words, it could be that Manufacturer A decides that cars that differ in something simple like a stylistic change between bumper plates are labeled as different models, whereas Manufacturer B might lump stylistic changes together. This could mean that the data reduction step where data were reduced to the top 10 manufacturers based on fleet size could be biased towards including manufacturers that tend to split similar vehicles into different models rather than combine them. However, if a manufacturer followed that practice consistently within a company, the results of the correlation analysis should still give a relative estimate of the change in frequency of fuel-type from 2016-2018.

Some other limitations of this data are that even if fleet size by fuel-type itself were an accurate reflection of a manufacturer’s distribution of car models, its not entirely certain that the distribution itself accurately relates to fuel efficiency effort. In other words, car manufacturers might to tend to invest heavily in 1 or 2 fuel-efficient hybrid models (eg. Toyota Prius) while maintaining a more expansive gasoline-only fleet. Another way to look at this problem could be total fuel-efficient vehicle production as a percentage of total fleet production, but that data is not available in this data set.

Despite these limitations, these data do offer some insight into manufacturers efforts relating to fuel-efficient vehicle manufacturing. In general, gasoline/electric hybrids and electric only vehicles were increasing in number of models produced, with ethanol and diesel productions decreasing. However, gasoline only vehicles had one of the largest increases (8/10 manufacturers) suggesting total fleet diversification is increasing in general as opposed to just a fuel-efficient diversification push. The sparse data used in this data set also limited the usefulness of statistical tests like Chi-Square or other contingency table-type tests. It could be that as gasoline-only vehicle technologies approach hybrid-electric technologies vehicle manufacturers are investing more towards the gasoline-only vehicles while still able to meet the minimum requirements for fuel efficiency increases.

Future analysis should inspect the fuel efficiency of the gasoline-only vehicle models, which is a data element available in the NVFEL data set. Expanding the data set to include additional car manufacturers, say from the top 10 to the top 25, might increase the statistical power of tests to detect true differences in the distributions of changes in fleet inclusion over time. Overall, this study provides results that show some insight into how car manufacturers have been responding to environmental regulations and highlights the need for further studies to expand on this important US regulatory issue.

Figures and Legends

Table 1A. Distribution of fleet size (number of unique models produced) by fuel-type for 10 car manufacturers selected from the 2016, 2017, and 2018 EPA National Vehicle and Fuel Emissions Laboratory.

Table 1B. Most recent (2018 only) frequency and average of unique models by fuel-type for 10 car manufacturers selected from 2018 EPA National Vehicle and Fuel Emissions Laboratory.

Table 2. Correlation from 2016-2018 in number of unique vehicle models produced. Correlations were categorized as either “Increasing” (positive correlation coefficient) or “Decreasing” (negative correlation coefficient) and summed on the bottom row.


1. US EPA, Sources of Greenhouse Gas Emissions. (10/28/2018) Retrieved from:

2. US EPA, Regulations for Greenhouse Gas Emissions from Passenger Cars and Trucks. (10/28/2018). Retrieved from:

3. US EPA, California Greenhouse Gas Waiver Request. (10/28/2018). Retrieved from:

4. United Nations, Climate Change. The Paris Agreement. (10/28/2018). Retrieved from:

5. DiChristopher, Tom. (Updated 7/24/2018). EPA reportedly will block increase in fuel efficiency standards for autos, revoke California's authority over tailpipe emissions. Retrieved from:

6. BBC. Paris climate deal: Trump pulls US out of 2015 accord. (7/1/2017). Retrieved from:

7. Tabuchi, Hiroko and Fountain, Henry. (6/1/2017). Bucking Trump, These Cities, States and Companies Commit to Paris Accord. Retrieved from:

8. US EPA, About the National Vehicle and Fuel Emissions Laboratory (NVFEL). (10/28/2018). Retrieved from:

9. US EPA, Fuel economy data source. (10/28/2018). Retrieved from:

10. Thornton, M. J., Alleman, T. L., Luecke, J., & McCormick, R. L. (2009). Impacts of biodiesel fuel blends oil dilution on light-duty diesel engine operation. SAE International Journal of Fuels and Lubricants, 2(1), 781-788.

bpwhite/us_fuel_economy.txt · Last modified: 2019/09/09 04:53 by bpwhite