12.4: In Depth- Incentives in US Automobile Policy

Figure 12.4.1 United States emissions of CO₂ from transportation and electric power. Data from the US Energy Information Administration.

Transportation, mostly in the form of private cars, is the single largest source of carbon emissions in the United States. This is a sharp reversal from past decades, when electric power emissions were much larger. Figure 12.4.1 shows the history of these two sectors in the United States since the early 1970s, which is the last time that transportation emissions were larger than those from electric power. In this section, we take an in-depth look at the automobile sector to understand the different incentives for reductions in carbon emissions when using a command-based approach versus a market-based approach.

Gasoline use is also an especially important example for considering the equity impacts of carbon taxation: lower-income households in the United States (and especially those located in rural areas) often need to drive many miles for work or school. They could feel disproportionate financial impacts from a carbon tax: Box 12.4.1 gives an example of one system that could counteract this effect.

History

Figure 12.4.2 The history of fuel economy in the United States. Data from the National Highway Traffic Safety Administration. CAFE = Corporate Average Fuel Economy.

Figure 12.4.2 shows the time path of fuel economy in the US since 1955, measured in miles per gallon (MPG). The first interesting feature to note is that fuel economy declined steadily for almost 20 years. Technology was improving, so why were cars getting fewer and fewer MPG? Rising incomes meant that households could afford more power and weight, and they went for it. The low point was reached in 1973.

Not coincidentally, cars in the muscle car era of the late 1960s and early 1970s were often equipped with powerful V-8 engines, some producing well over 300 horsepower.

Next came two incredible runs upward in the 1970s and early 1980s, when fuel economy almost doubled. The rate of increase in that era was far more rapid than any command-based policy we have seen, and it was achieved with only small advances in technology. Why, and how, was this accomplished? It turns out both answers are straightforward: gasoline got much more expensive, and so people bought smaller and lower-horsepower vehicles.

Box 12.4.1 Income Distribution and Gasoline Use

It is often noted that overall gasoline use in the United States does not respond very rapidly to gasoline price, and so perhaps quite a large carbon tax would be needed to create much change. This is almost certainly true if we want to make a big change in only a few years, as everything from the location of people’s homes to the number and size of highways and the immense parking structures at many shopping malls seems designed to keep people driving. Change under a carbon tax will move slowly. However, it is important to note that all of the things a carbon tax would eventually affect (housing sprawl, parking garages, and so on) are important—perhaps even necessary—in terms of the long-run change needed to solve the climate problem.

What are the implications of having small short-term responses in transportation infrastructure? Carbon taxes, when they are passed through into gasoline, may be mostly unavoidable in the short term before housing options and infrastructure design start to change. Lower-income households will feel these (unavoidable in the short term) taxes most acutely, since gasoline is a much larger share of their budget, especially for those people living in places with little public transportation service. This has been one of the greatest political challenges of implementing cost-effective climate policy for automobiles. The ideal economic solution would be to, at the same time a carbon tax is imposed, offer low-income households other benefits to make up for the increased tax burden. The carbon dividend (a form of basic income) proposed in the economists’ Wall Street Journal letter discussed above is one potential way to mitigate the inequitable impacts a carbon tax could have.

Concerns surrounding the inequity of gasoline taxation (one aspect of carbon taxation) have been the subject of much research in economics. In the United States, low-income households drive a lot of miles and so are very exposed to a carbon tax via their gasoline consumption. For example, US households with $25,000 to $50,000 in annual income consume about 800 gallons of gasoline per year. Households with approximately double the income ($50,000 to $75,000 annually) consume about 1,100 gallons per year, much less than double the amount of gasoline. A carbon tax of $40 per ton works out to 35 cents per gallon of gasoline and so would cost the lower-income households 800 @ $0.35 = $280 per year. The higher-income households would pay $385—a higher amount, but a smaller fraction of their income.

One important economic finding is that if the total revenue from a gasoline tax were given back evenly (the same dividend check to everyone), lower-income households would, on average, be better off than they had been before the tax. In the example above, if we have the same number of households in each of the two income ranges, then the dividend check would be $332.50.* The lower-income households would pay $280 per year but get a $332.50 dividend check (a gain of $52.50), while the richer households would pay $385 in but get the same dividend of $332.50 back (they would lose $52.50 overall).

The key challenge for a dividend system like the one above is when there are differences between households that have the same level of income. Consider two households that both have $35,000 in annual income. One household uses public transport and consumes zero gallons of gasoline. The other drives a lot and uses 1,600 gallons per year. On average, this is exactly the 800 gallons above. But notice that a tax-and-dividend system would be very hard on the 1,600-gallon household (they would pay $560 in per year and get only $332.50 back) and very generous to the household with no cars (they would pay zero and still get $332.50 back). Avoiding this problem by targeting the dividends is difficult, if not impossible, and the subject of ongoing research. Notice that it is also important not to target the dividends too well, since part of the overall goal is to encourage a household like the one using 1,600 gallons yearly in the example above to find ways to reduce.

The most important advantage of a tax, relative to a command-based rule, in terms of managing equity concerns is that the tax produces revenue. The revenue can be used to help households that are the most affected by the policy. A command-based system—for example, banning traditional gasoline cars and requiring hybrid or electric drive—would also be very burdensome for low-income households. However, a command-based rule like this wouldn’t raise any government revenue, and so dividends or other programs to help reduce the impact on disadvantaged groups would need to come from outside funds.

*To see why, imagine there are 10 households at each of the two income levels (20 households all together). Low-income households would pay 10 @ $280 = $2,800 into the system each year. High-income households would pay in 10 @ $385 = $3,850. This would be $2,800 + $3,850 = $6,650 in revenue per year. Then $6,650 / 20 = $332.50 in dividends available for each household.

The era of corporate average fuel economy

Partly in response to high gasoline prices and security concerns in the Middle East, the first significant command-based policy on gasoline use was put into place in 1978—the Corporate Average Fuel Economy (CAFE) standard. It is named a “corporate average” standard since it applies separately for each company, and within companies it applies separately for sedans and light trucks. Note that the “light trucks” designation includes SUVs, minivans, and also many of today’s crossover vehicles.

The level of the command-based standard was held quite flat for many years. For example, for every model year between 1990 and 2010, the standard for sedans was 27.5 miles per gallon. That is, all the different sedans that any one company made (calculated as a sales-weighted average) had to get at least 27.5 MPG. Since CAFE standards did not change over these two decades, manufacturers were able to use advances in technology exclusively for improvements in horsepower and weight, holding fuel economy flat. For light trucks (SUVs, minivans, etc.) the standard was weaker, although it did increase slightly from 20.7 MPG between 1996 and 2004 to 23 MPG by 2010.

From Figure 12.4.2 we can see that the true fuel economy of all vehicles on the road (the black line) actually declined somewhat through the 1990s, rather than remaining flat. If the CAFE rule was exactly flat, how is it that fuel economy slipped back downward? The answer lies in the composition of vehicle types. Many car buyers during this period were choosing to replace their sedans (which got an average of 27.5 MPG, following the rule) with SUVs (which got an average of 21 MPG, also following the rule). The compositional shift allowed manufacturers to stay within the law but still add power and weight back into the vehicle fleet.

In 2012 we saw the most ambitious and far-reaching reform to CAFE since its 1978 inception, with the following three important changes:

1. The target nearly doubled, to 54.5 miles per gallon (averaged across all private vehicles, and including some extra credits for electric vehicles) by 2025.
2. The original two categories (cars and light trucks) were divided into many more categories based on the width and length, or footprint, of the vehicles. Vehicles with very large footprints were assigned a much weaker standard to meet than vehicles with small footprints.
3. Trading of the standard among manufacturers was allowed. If a company chooses to exceed the standard, it can sell the credit to another company that can then fall below it.

Incentives with CAFE

What about the incentives, and our list of cost-effective actions, for climate? The first thing to notice is that the decision to apply the rule separately to every company means the strength of the incentive is different for different companies. Ford, General Motors, and Chrysler (the “big three” American automobile companies) were initially constrained by the standard and met the 27.5 MPG requirement almost exactly over the decades. There is evidence that, even though the outcomes were the same at 27.5 MPG, achieving them was much harder for GM and Chrysler (they were more well known for large and powerful cars) than it was for Ford (which sold more small cars to start with and so didn’t have as much to do to meet the rule).

To see an even sharper difference in incentives, consider Toyota, the largest foreign company selling cars in the US. Toyota’s fleet over this period had an average fuel economy of 30 MPG, reaching almost 35 MPG in the early 2000s. The CAFE rule had no effect on Toyota at all, even though it would still have had an ability to improve the efficiency of its cars if it had been asked. More troubling, some of the fuel-saving opportunities Toyota had left on the table would have been much cheaper per ton of carbon saved—that is, more cost-effective—than the changes that GM and Chrysler were forced to make. The CAFE rule did reduce carbon emissions (at least from some companies), but it didn’t find the most cost-effective ways to do so.

One key improvement to cost-effectiveness, and a change that was made to the rule in 2012, is item (3) in the list above: the ability for car companies to trade compliance. If in 2019 Toyota finds a cheap way to boost fuel economy by 1 MPG, for example, the trading system gives the company a reason to do that even though it is still more than complying with the rule. Toyota can sell its over-compliance to another company, like GM or BMW, that might find it cheaper to buy that 1 MPG than to implement it. The same amount of carbon would be saved, and it would cost less.

Unlike item (3) in the list of changes, which improves cost-effectiveness, the change in item (2) reduces cost-effectiveness. To see why, observe that the footprint basis of the rule now allows compositional effects of the type we saw in the 1990s (switching from cars to SUVs) but throughout the whole fleet instead of just across two categories. Switches to wider and longer vehicles (for example, from compact sedans to midsize sedans) undermine the overall average, since that increase in square footage reduces the fuel economy target that a manufacturer has to meet. This unintended incentive within the regulation has meant that even though the nominal standard has risen quickly in recent years (the colored lines in Figure 12.4.2), the actual fuel economy of new vehicles has remained quite flat between 2014 and 2018. This problem is not unique to the US. Indeed, the same unintended incentives worked to reduce the effectiveness of the fuel economy standard in Japan.

Incentives with a carbon tax

Returning to our study of cost-effectiveness, let’s revisit the list of all possible ways to conserve carbon. New technologies to improve fuel economy are definitely on the list, and those are incentivized by the command-based CAFE standards. How about some of the other items on the cost-effectiveness list? Examples include:

• Combining trips to work and the store
• Reducing the number of cars in a household from three to two
• Choosing to live closer to work or school
• Walking or bicycling to work instead of driving

All of these (and many more) can be excellent, cost-effective ways to save gasoline. These options will make more sense for some people than others, because individual circumstances determine how much time or convenience is given up when taking these actions.

As an example, think about the reduction from three to two cars for a particular suburban household. Suppose this household observes that the cost of their third car plus the gasoline they put in it is $5,000 per year. The third car sometimes lets the children avoid having to wait as long to be picked up, and it can make getting to work for one of the parents more convenient. Suppose this time and convenience is worth $5,100 to the household. They will keep the car and will get to enjoy $100 of “consumer surplus”—the difference between what something actually costs and the value the person gets from it.* CAFE policy will make that third car a little bit more efficient, reducing the amount that has to be spent on gas. CAFE might also raise the price of the third car a little bit (by putting new technologies into it). Because these two effects tend to offset each other, CAFE does not usually change the decision of the household: people still buy almost as many cars as they always did.

On the other hand, a price incentive on carbon could make a big difference for this household. Even a small carbon tax (which would show up in the price of gasoline) might be enough to raise the overall cost to $5,200 dollars, for example, and so the car would no longer be worth it. This is a very cost-effective action for saving carbon—the household only loses $100 of surplus, while society gains an entire car’s worth of carbon reductions.

The advantage of a carbon tax is that it incentivizes every one of the actions on the list above, and in fact every other action that we could write down, to save gasoline. The complete list of actions in the automobile sector is very long (we could probably fill this entire book with it!), while the list of beneficial actions implemented as a result of the CAFE standard (mainly, better technology on new cars and reductions in horsepower and weight) is very short. Estimates in the literature vary, but my own research finds that a carbon tax could achieve the same amount of carbon savings as CAFE for one-fifth to one-third the cost. The carbon tax can do this because it utilizes much cheaper actions, much higher up on the cost-effectiveness list.

Double dividends and consumer choices

Separate from our goal of reducing carbon emissions, many advocates of command-based policies like CAFE argue that the policy will also help consumers avoid making mistakes in their car purchase decisions. This is the idea of a “double dividend”—getting two things out of one policy—in this case a cleaner environment and better decisions by consumers.

In this setting, the mistake would be that the typical car buyer isn’t thinking very far ahead about gasoline purchases. The hypothetical buyer picks a big and very powerful model and then regrets that choice after learning how much it costs to have the tank filled. A policy like CAFE could mean that the carmaker doesn’t even offer a vehicle with such a low MPG for sale anymore, or that it sets the purchase price high enough that most people can’t afford it.

Either way, the hypothetical buyer is forced into a smaller, more modestly powered car that was actually a better choice to start with. That extra surplus from the improved decision-making means the environmental goals can be reached more cheaply. Much has been written about this effect in the economics literature, and no consensus has yet emerged. Some authors contend that statistical evidence shows that car buyers (particularly used-car buyers, who account for the majority of the car market) are very careful about fuel use. Small increases in the price of gasoline push used-car buyers very rapidly toward smaller and less powerful models, and so a policy like CAFE might not have the extra benefit of correcting consumer choice.

Other authors find instead that as much as 20% to 30% of future gasoline cost is ignored by new-car buyers. This opens up an opportunity for a policy like CAFE to improve decision-making and therefore improve somewhat on cost-effectiveness. There is so much ground to make up (if the carbon tax starts at only one-third or one-fifth the cost of CAFE), however, that it could be difficult for this consumer-choice effect to reverse the overall ranking of policies. Questions surrounding the ability of consumers to make good decisions, and the ability of carbon policy to improve on those decisions, are quite important and an active area of research.

*We can calculate the consumer surplus here as $5,100 (the value to the household) minus $5,000 (the cost).