Gasoline Prices and Driving Behavior

The effects of rising gasoline prices can be seen in changing highway traffic volumes and speeds and in shifting consumers’ choices about the kinds of vehicles to drive. In the short run, rising gasoline prices affect the number of vehicles on the highway and the speeds at which those vehicles are driven in free-flow conditions. It is a simple matter for motorists who make such behavioral adjustments to undo them if gasoline prices decline. Other adjustments motorists could make—including changing the kinds of vehicles to buy or where to live or work—are not as easily reversed. The greater price sensitivity of gasoline consumption in the long run reflects those other adjustments.

This Congressional Budget Office (CBO) study illustrates both kinds of effects that rising gasoline prices have had on consumers, and it suggests the kinds of consumer effects that could be expected from policies that would seek to discourage gasoline consumption and, by extension, limit the associated carbon dioxide emissions.

CBO’s analysis of the influence of gasoline prices on motorists’ behaviors is based on four years of data collected from metropolitan freeways in California between 2003 and 2006 and on statewide average gasoline prices and wages over that period. (Appendix A describes the data used, and Appendix B explains CBO’s analytical approach and presents the econometric results of the analysis.)

One way motorists can reduce transportation costs is to drive less, for example by using public transportation, alternative modes of transportation, or car pools; by consolidating trips; or by telecommuting to work. They also might make shorter trips, substituting nearby recreation or shopping locations for more-distant alternatives that they otherwise prefer or, in the long run, they might move closer to work or choose jobs closer to home.

The likelihood that a driver will make one of those changes depends on the price of gasoline and on other factors that determine how attractive driving is compared with the alternatives. For a motorist who routinely faces heavy traffic or high parking fees, the benefit of switching to public transportation can be quite large. But motorists who previously had been willing to accept those costs without switching must therefore place a relatively high value on driving. That said, work commuters are more likely to switch to public transportation—especially to rail, which is usually less affected by traffic congestion—if the available transit alternatives are convenient to workplaces and commuting routes.

Recent empirical research suggests that total driving, or vehicle miles traveled (VMT), is not currently very responsive to the price of gasoline. A 10 percent increase in gasoline prices is estimated to reduce VMT by as little as 0.2 percent to 0.3 percent in the short run and by 1.1 percent to 1.5 percent eventually.¹ A 2003 study of corporate average fuel economy (CAFE) standards, published by the National Research Council, cited slightly older estimates of the responsiveness of VMT to fuel costs that ranged from about 1 percent to 2 percent.²

Some of the VMT response comes from drivers who switch to commuter rail. An increase in gasoline prices raises the relative cost of driving compared with rail transit. As an illustration—in the opposite direction—of the price sensitivity of the demand for rail transit, a 10 percent increase in transit fares is estimated to reduce ridership by about 5 percent in the short run and by about 10 percent in the long run.³ Survey research has indicated that a change in the cost of driving is the most important factor motorists consider when deciding whether to continue to drive or to switch to some other mode of transportation.⁴ CBO’s findings suggest, however, that a large increase in the price of gasoline might cause only a small shift from automobiles to public transportation, at least in the short run.

CBO’s analysis is based on traffic volume (total vehicles per day) on metropolitan highways, rather than on total vehicle miles traveled. Those measures should be correlated, however: A general decline in VMT should reduce the volume of highway traffic, and most of that traffic occurs in metropolitan areas.⁵ However, because recent research indicates that VMT is relatively insensitive to gasoline prices, the higher prices of the past several years should not be expected to cause large changes in freeway traffic volume.

When faced with an increase in gasoline prices, motorists should most readily curtail their lowest value trips. If they consider weekend trips generally less important than weekday trips, then weekend traffic volumes should be more sensitive to the price of gasoline. (CBO has no information about how weekday driving is valued compared with weekend driving.) However, freeway traffic volume should be more responsive to changing gasoline prices in places where transit rail service is available, particularly on weekdays. That is because rail service is probably a better substitute for weekday driving to work than it is for weekend driving, when transit service often is less frequent, some destinations (such as sports fields or places of worship) may be less well served by public transportation, and trips are more likely to involve hauling purchased items or recreational gear. The relative sensitivity of weekend versus weekday traffic to the price of gasoline ultimately is an empirical question, which is addressed by CBO’s analysis.

Average weekday traffic volumes on some freeways have declined slightly in response to higher gasoline prices, CBO’s analysis shows. The routes on which that response was detected are adjacent to commuter rail systems. Weekly average gasoline prices appear to have had little effect on traffic volume at other freeway locations or on weekends.⁶ In the California data that CBO analyzed, higher gasoline prices also are associated with slightly greater ridership on transit rail systems.

The data consist of daily traffic counts for a dozen diverse freeway locations in metropolitan areas of California. The data cover the period from early 2003 through the end of 2006 and come from the state’s four primary metropolitan areas (Sacramento, the San Francisco Bay Area, Los Angeles and Orange County, and San Diego County). For each area, CBO collected data at representative freeway locations adjacent to the commuter rail system in that region and at other locations in the region where rail transit was not available. (The figures and tables in Appendix A give details on the data.)

On average, over all locations, the price of gasoline in a given week had a negligible effect on the volume of weekend traffic, but on weekdays, higher gasoline prices had a small but statistically significant effect (see Table 1-1). A 20 percent increase in price, or 50 cents if the base price is $2.50 per gallon, would reduce weekday freeway traffic by an average of 0.4 percent. The effect would occur entirely in the response at rail-accessible freeway locations (as shown in the last two rows of the table). At those places, a 20 percent price increase would reduce weekday traffic by an average of 0.69 percent. That result is strongly statistically significant, although it amounts only to about 730 fewer vehicles out of an average of more than 106,000 vehicles per weekday at those locations. Gasoline prices did not affect weekend traffic volume at any of the locations, nor did they affect weekday traffic counts where rail commuting was not an option.⁷

		Weekends	Weekdays


Average Effect,
	All Sampled Routes	0.12	-0.40
Statistical Significance		Not significant	1.4 percent*

No Rail Option		0	0
Statistical Significance		Not significant	Not significant

Parallel Rail		0.20	-0.69
Statistical Significance		Not significant	0.04 percent**

A 20 percent increase would be 50 cents per gallon if gasoline costs $2.50 per gallon or 60 cents if the price is $3.00.

The weekday response for rail-accessible freeway routes implies a traffic volume elasticity of -0.035 with respect to the price of gasoline. That result is roughly consistent with the short-run VMT elasticity estimate of -0.03 noted earlier, although that estimate represents all driving, not just driving on rail-accessible freeways. However, both elasticities indicate that gasoline prices have had a measurable, although small, effect on total driving.

Reductions in traffic volume correspond closely to increases in transit ridership. CBO analyzed monthly ridership totals for the municipal light-rail systems in Sacramento, Los Angeles, and San Diego and for the subway systems in Los Angeles and the San Francisco Bay Area. (In order to combine the data from different-sized transit systems into the same analysis, CBO expressed each system’s ridership as a percentage of its average in a baseline period—the same treatment it applied to freeway traffic volumes.) Adjusting for long-term ridership trends on each system, seasonal effects, and inertia (the tendency for ridership totals to persist from one month to the next), CBO estimates that the same increase of 20 percent in gasoline prices that affects freeway traffic volume is associated with an increase of 1.9 percent in average system ridership. That result is moderately statistically significant: It can be asserted with 95 percent confidence that higher gasoline prices are associated with increased ridership.

For an average-sized system, that result translates into about 1,870 additional rail trips per day in each direction, throughout the system.⁸ For a transit system that has several branches running alongside freeways—as the systems in CBO’s sample have—that implies an additional 625 to 935 riders per line for systems with, respectively, three or two such lines. Thus, as gasoline prices have increased, the average number of riders gained by the rail transit systems in CBO’s sample has been reasonably consistent with the reduction in the number of vehicles per weekday, about 730, on the adjacent freeways. In CBO’s analysis, all five transit systems exhibited positive relationships between ridership and gasoline prices, although for the two (interconnected) Los Angeles systems, the effect was small and not statistically different from zero.

Another way that motorists can reduce their fuel costs is to drive more slowly. The incentive to slow down will depend on how much gasoline prices have increased, how much fuel would be saved by slowing down, and how much motorists value their time while driving.The value of the potential fuel savings from slowing down is rather small compared with reasonable measures of many motorists’ value of time, so the likely effect of gasoline prices on highway speeds also should be rather small. For any given reduction in speed, however, the fuel savings are greater at faster speeds and for less-fuel-efficient vehicles.

The development of freeway congestion-pricing projects—which charge tolls that rise with the amount of traffic congestion—has enabled researchers to estimate motorists’ value of time during congested commuting hours. Estimates for California’s high-occupancy toll lanes along State Route 91, west of Riverside, and Interstate 15, north of San Diego, indicate, on the basis of tolls and travel time savings in toll lanes versus free lanes, that motorists value their time between $20 and $45 per hour of reduced travel time.⁹

In other contexts, where toll-based estimates are not available, economists typically use average hourly after-tax wage rates as a proxy for motorists’ value of time.¹⁰ Consistent with that, motorists’ preferred driving speeds have been found to be positively associated with income.¹¹

There have been studies that link driving speeds and motorists’ value of time, but CBO’s analysis is among the few published studies on the relationship between driving speeds and gasoline prices. Previous work, now decades old, also found a link between higher gasoline prices and slower driving.¹² And new research has identified a relationship between higher gasoline prices and lower motor vehicle fatality rates, although the researchers attributed it to a reduction in vehicle miles traveled and did not consider whether slower driving also could have contributed to the decline in fatalities.¹³

In response to higher gasoline prices, drivers optimally would slow just to the speed at which the value of the fuel saved equaled the value of time lost to slower driving. By that logic, motorists who valued their time more would slow down less, or not at all, than drivers who valued their time at a lower rate per hour. At any given gasoline price, a motorist’s preferred speed also depends on factors that are unrelated to gasoline prices or the value of time, such as the local speed limit and its enforcement, the time of day, the time of year, the physical characteristics of the road at that location, and traffic density.¹⁴ However, in keeping with the evidence cited earlier, speed should be correlated with the value that motorists place on their time.

CBO’s analytical results are consistent with that observation. In all likelihood, few if any motorists know what their optimal response should be (with respect to driving speed) when gasoline prices change. But CBO’s analysis suggests that drivers’ responses may be proportionate to their value of time. That is, if motorists can be said to have particular speed preferences, then those who tended to drive more slowly than average before gasoline prices increased appear to have slowed slightly more than other drivers did, and faster drivers have not reduced their speeds at all.¹⁵ Overall, as described with the rest of the findings, the amount of fuel saved as a result is consistent with recent estimates of the price elasticity of the demand for gasoline. (The variation in motorists’ driving speeds as a response to higher gasoline prices may have some implication for highway safety; see Box 1-1.)

If the speeds at which motorists drive are positively correlated with how much they value their time, then a disproportionate number of slower-driving motorists would have lower-than-average values of time and faster-driving motorists would have higher values.¹ If that is the case, then the findings of the Congressional Budget Office (CBO) about the effects of the price of gasoline on highway speeds are consistent with the prediction that motorists with lower values of time will be more responsive to an increase in gasoline prices than will drivers with higher values of time. (CBO has no information about whether driving speeds and values of time are correlated.) The results also could be explained if all motorists have a similar distribution of driving speeds and a similar response to gasoline prices, or if drivers of less-fuelefficient vehicles tend to drive more slowly than do drivers whose vehicles get better mileage. However, those explanations would require that all motorists value their time about the same. Given the wide variation in motorists’ effective wage rates, that premise seems unlikely.

If different motorists do have different propensities to drive faster or slower, the findings also imply that higher gasoline prices increase the variance in highway driving speeds at a given time. That would mean more interactions, such as passing. In that case, motorists who wish to maintain a given level of safety would need to devote slightly more attention to tasks such as monitoring other vehicles and maintaining their desired following distance.² However, because higher gasoline prices also would cause drivers to reduce vehicle speeds slightly, the effect on safety is indeterminate but probably negligible because the changes in speed are small.

1. One useful way of thinking about the concept of motorists’ value of time is the amount they would be willing to pay to reduce their travel time by one hour.

2. For a discussion of the possibility of a negative link between variance of speed and highway safety, see Charles A. Lave, "Speeding, Coordination, and the 55-MPH Limit," American Economic Review, vol. 75, no. 5 (1985), pp. 1159–1164; and Theodore E. Keeler, "Highway Safety, Economic Behavior, and Driving Environment," American Economic Review, vol. 84, no. 3 (1994), pp. 684–693.

A study conducted by Oak Ridge National Laboratory (ORNL) showed that slowing from 70 miles per hour (mph) to 65 mph—a 7.1 percent reduction—would reduce a typical vehicle’s fuel consumption from 3.7 to 3.4 gallons per 100 miles, an 8.2 percent reduction. At $3 per gallon, the fuel savings would be worth 0.9 cents per mile. Travel time would increase by about 4 seconds per mile.¹⁶

Figure 1-1 shows the average relationship between speed and fuel consumption for speeds from 15 mph to 75 mph.

Source: Congressional Budget Office based on data from the Bureau of Transportation Statistics. See Stacy C. Davis, Transportation Energy Data Book: Edition 21-2001, ORNL-6966 (prepared by Oak Ridge National Laboratory for the Department of Energy Office of Transportation Technologies, October 2001), www.ornl.gov/~webworks/cppr/y2001/rpt/111858.pdf, Tables 7.21 and 7.22; B.H. West and others, Development and Validation of Light-Duty Modal Emissions and Fuel Consumption Values for Traffic Models, FHWA-RD-99-068 (Federal Highway Administration, 1999).

Note: The results are based on nine representative vehicles selected and tested by Oak Ridge National Laboratory.

Table 1-2 is based on the data underlying Figure 1-1. It shows average fuel savings per hour of additional travel time caused by slower driving, compared with the fuel and time consumed when driving at 70 mph. For example, when the price of gasoline is $2.50 per gallon, slowing from 70 mph to 69 mph would generate fuel savings of $7.47 for every hour "lost" to slower driving. Thus, for motorists who value their time at less than $7.47 per hour—and who would prefer to drive 70 mph or faster if gasoline were less expensive—the value of fuel saved by slowing to 69 mph would exceed the cost in terms of additional travel time. For drivers who value their time at more than $7.47 per hour, the financial benefit of slowing would be less than the time cost.

Table1-2.2.1.htm


Price per	Fuel Cost	Target Speed	Value of Fuel Saved	Time Lost (Seconds)		Savings per
Gallon	per Mile^a	(Miles per hour)	per Mile^b	Per Mile	Per Penny Saved	Hour^c


2.50	0.093	69	0.0015	0.75	4.9	7.47

		69	0.0019	0.75	4.0	8.96
3.00	0.112	65	0.0092	3.96	4.3	8.37
		60	0.0164	8.57	5.2	6.89

3.50	0.131	69	0.0022	0.75	3.4	10.45

Source: Congressional Budget Office based on Oak Ridge National Laboratory (ORNL) estimates of average fuel consumption versus speed.

More generally, Table 1-2 suggests that an increase in the price of gasoline will cause some motorists, but not all, to slow down when they are driving on uncongested freeways. For example, no motorist with a value of time above $10.45 per hour and a preference for driving 70 mph would slow down as long as the price of gasoline remained below $3.50 per gallon. The table also suggests that a motorist’s optimal response should vary inversely with that motorist’s value of time. For example, if the price of a gallon of gasoline rose from $2.50 to $3.00, a motorist who prefers to drive 70 mph would slow by anywhere from 1 mph to 10 mph for values of time ranging from around $9 per hour to below $7 per hour, respectively. In September 2005, when the price of gasoline first reached $3 per gallon in California, the median after-tax wage rate in that state was $11.27 per hour—close enough to the values in Table 1-2 that the price should have caused motorists to drive more slowly if their value of time was several dollars (or more) below the statewide median wage.¹⁷

It would be difficult to detect the effect of gasoline prices on speeds in heavy traffic, because motorists already must drive more slowly than they prefer in such conditions. For that reason, CBO analyzed driving speeds only for weekends, when freeways are demonstrably less congested than they are on weekdays.¹⁸ If motorists value their time less on the weekends, their weekend driving speeds also should be more sensitive to gasoline prices. Thus, if no slowdown could be detected in weekend driving, probably none could be detected in weekday travel either.

Higher gasoline prices from 2003 through the end of 2006 caused many motorists to drive a little more slowly on uncongested highways. Median speeds in free-flow conditions declined slightly as gasoline prices increased. The slowdown was more pronounced for vehicles moving at the somewhat lower 5th percentile speeds; there was no discernible effect on 95th percentile speeds. The median effect is consistent with recent estimates of gasoline price elasticity, which indicate that short-run demand declines by around 0.6 percent when the price rises by 10 percent, all else being equal.¹⁹ The diverse effects of gasoline prices on vehicles traveling at different speeds are consistent with the notion that motorists who set a lower value on their time may be more willing to trade (slightly) longer travel times for (slightly) lower fuel costs.

The data CBO examined consist of distributions (percentile values) of weekend vehicle speeds over a month, with a separate distribution for each hour of day and each month.²⁰ CBO collected data for three locations, recording each month’s 5th, 50th (median), and 95th percentile weekend speeds at each location for each hour of the day. Table 1-3 reports sample statistics for those data and median traffic volumes for each location. Interstate 405 in Orange County is the busiest of the three sampled locations, with a median volume of 5,530 vehicles per hour over the entire sample.²¹ That location also has the slowest 5th percentile speeds, averaging 59 mph over all hours of the day, compared with a little less than 63 mph on I-680 in San Ramon and nearly 67 mph on I-8 in San Diego. Median speeds are more similar across the three locations, ranging from 66.4 mph to 69.5 mph, on average. The averages in Table 1-3 include some congested traffic; as described in the appendix, data that appear to indicate congested travel were excluded from the analysis.

Excluding 1 a.m to 6 a.m. In the bottom three rows, each figure is the average of the given percentile speed in each month for each hour on Saturdays and Sundays. With 19 hours between 6 a.m. and 1 a.m., each figure is an average of 48 months × 19 hours = 912 values.

For this study, CBO developed a statistical model of driving speed as a function of the price of gasoline and other factors, including seasonal and freeway-specific effects. Holding those other factors constant, the model indicates that a 50 cent increase in the price of gasoline would cause median freeway speeds in the sample to decline by a little more than three-quarters of a mile per hour. (For example, at the study’s mean gasoline price of $2.35 per gallon and the median freeway speed of 67.8 mph on uncongested freeways, when the price of gasoline reached $2.85 per gallon, the median speed would have declined, on average, to 67.0 mph.) The effect on slower vehicles is 50 percent greater: Fifth percentile speeds would decline by about 1.2 mph. By contrast, higher gasoline prices do not appear to have affected 95th percentile speeds: Faster-moving traffic appears not to have slowed down as gasoline prices increased, at least over the period observed in the data (see Table 1-4).

Note: mph = miles per hour; ** = significant at 1 percent. The differences between effects also are significant at 1 percent.

The median result translates into an elasticity of speed of about -0.05 with respect to the real price of gasoline—which is to say that a 10 percent increase in the price of gasoline would cause the median speed to decline by about 0.5 percent. That result, applied to the findings of the ORNL study, implies that for an average vehicle driven at highway speed, fuel consumption would decline by about 1.2 percent, or slightly less than one-twentieth of a gallon per 100 miles. Those savings—a teaspoon of gasoline every 2.6 miles—imply a gasoline demand elasticity in urban highway driving of about -0.06 with respect to the price of gasoline.

Thus, along with conventionally understood sources of elasticity in the demand for gasoline—changes in the length and frequency of automobile trips and in the types of vehicles people drive—the way vehicles are operated could be a meaningful source of short-run elasticity in the demand for gasoline. In particular, although the implied elasticity of -0.06 is objectively small, it is consistent with current estimates of the overall short-run elasticity, which range from about -0.03 to -0.09.²³

The analysis of driving speeds shows that if gasoline costs $2.85 per gallon, motorists with average fuel economy vehicles who slow from the median speed by 0.75 mph would cut their fuel expenditures by about 0.13 cents per mile. Those savings would accumulate at a rate of about $8 per hour of additional travel time—about 30 percent less than the median hourly after-tax wage rate for California in 2005.²⁴ At the 5th percentile speed, slowing by 1.2 mph would cut fuel expenditures by about 0.16 cents per mile, or about $5.24 per additional hour of travel time compared with the baseline speed of 62.8 mph. Motorists who travel at the 95th percentile speed, and who do not slow in response to higher gasoline prices, would be implicitly valuing their driving time at $8.60 per hour or more.²⁵ (Another way motorists can reduce the per-mile cost of driving is to switch to a lower grade of gasoline; see Box 1-2).

Higher gasoline prices have induced many drivers to make small changes in the way they operate their vehicles. The higher prices also may have caused some consumers to switch to lower octane formulations, which generally are sold at slightly lower prices. There currently tends to be a difference of about 20 cents per gallon between the prices of regular and premium gasoline; the price for midgrade gasoline tends to fall in the middle.¹

Anecdotal evidence suggests that some consumers are satisfied with their vehicles’ performance using a lower octane gasoline than that recommended by the manufacturer. As gasoline prices rise, consumers may become more willing to ignore manufacturers’ recommendations and switch to a less expensive grade of gasoline. It is also the case, however, that if the price of each grade increases by the same amount, the relative price of the higher grade falls in comparison to the price for each lower grade. That shift can cause some consumers to substitute toward a higher grade, although the empirical evidence for that phenomenon is mixed and rather sparse.² In either case, given the low elasticity of overall demand for gasoline, grade switching has little effect on total U.S. gasoline consumption.

Consumption of midgrade and premium gasoline has been declining in absolute terms since 2000 (see the accompanying figure on the facing page); consumption of regular fuel has increased. Some of the decline in the use of higher octane fuels might reflect a change in vehicle designs and in sales of vehicles that require higher grades of gasoline. However, as with grade switching, the Congressional Budget Office did not analyze the extent to which changing engine designs and consumer preferences have contributed to the decline in sales of higher octane fuels.

Retail Prices and Consumption of Gasoline

(Millions of gallons per day) (Dollars per gallon)

Source: Congressional Budget Office based on data from the Department of Energy, Energy Information Administration.
Notes: Prices are nominal at-the-pump prices and include all taxes, ending in May 2007. The smoothed price series is a six-month moving average, computed by CBO.
Consumption totals were converted to logarithms (left scale) by CBO.

1. In 2006, the U.S. average prices for regular and premium grades were $2.59 and $2.80 per gallon, respectively. See Department of Energy, Energy Information Administration, Monthly Energy Review (October 2007), p. 132, Table 9.4, Motor Gasoline Retail Prices, U.S. City Average, www.eia.doe.gov/emeu/mer/pdf/pages/sec9_6.pdf.

2. See Robert Lawson and Lauren Raymer, "Testing the Alchian�Allen Theorem: A Study of Consumer Behavior in the Gasoline Market," Economics Bulletin, vol. 4, issue 35 (2006), pp. 1–6, http://economicsbulletin.vanderbilt.edu/ 2006/volume4/EB-06D00021A.pdf; and Todd M. Nesbit, Excise Taxation and Product Quality: The Gasoline Market, Working Paper 05-11 (Morgantown: West Virginia University, Department of Economics, 2005), www.be.wvu.edu/ div/econ//work/pdf_files/05-11.pdf.

The theorem was originally explained by Armen Alchian and William Allen in Exchange and Production: Competition, Coordination, and Control (Belmont, Calif.: Wadsworth, 1983; originally published as University Economics: Elements of Inquiry, 1964) and further developed by Yoram Barzel in "An Alternative Approach to Analysis and Taxation," Journal of Political Economy, vol. 84, no. 6 (1976), pp. 1177–1197.

Such small responses are unlikely to result from conscious calculations. Few motorists would have the information required to gauge their responses so acutely, nor the time or inclination to do so. However, higher prices make drivers pay more attention to speed. The modest reductions in speed suggest that drivers may have responded by easing off slightly on the gasoline pedal or dialing back cruise-control settings a notch. If only a minority of drivers have that response, their reduced speeds could cause nearby drivers to slow down as well, even if gasoline prices alone would not have that effect. Both kinds of response contribute to elasticity in the demand for gasoline.

Although they are based on California data, the findings of this study are more or less applicable to other, similar metropolitan areas in the United States. Motorist populations, highway and mass transit infrastructure, and vehicle stocks differ somewhat from one part of the country to another, and gasoline prices vary from state to state because of differences in state taxes and regional supply.²⁶ Retail gasoline prices tend to be higher in California not only because gasoline pipelines that serve many other parts of the country do not extend into California but also because the gasoline sold in that state’s metropolitan areas is reformulated as required by the Clean Air Act, adding about three cents to the retail price.²⁷ Despite the differences, changes in gasoline prices tend to be highly correlated, so consumers throughout the country have had similar financial incentives to reduce their consump-ion of gasoline.²⁸ This study’s findings on highway speeds may be more generally applicable than are its findings on the volume of traffic, which apply to areas served by rail transit systems.

Higher gasoline prices could have a smaller effect on VMT in rural areas, in California or elsewhere, because there are fewer alternatives to driving and because trip distances may be greater. However, the largest share of total VMT occurs on highways in metropolitan areas.²⁹ This study’s findings on vehicle speeds in urban highway driving may, if anything, understate the effect of higher gasoline prices on driving speeds in rural areas because median household income—and presumably motorists’ valuation of their time—tends to be considerably higher in urban areas.³⁰

The VMT effect is slightly smaller, therefore, than the more general effect of gasoline prices on the demand for gasoline, estimated in recent research to be about -0.6. See Kenneth A. Small and Kurt Van Dender, "Fuel Efficiency and Motor Vehicle Travel: The Declining Rebound Effect," Energy Journal, vol. 28, no. 1 (2007), pp. 25–51. Other estimates of VMT elasticity (in the literature reviewed in that study) were higher, ranging from -0.10 to -0.16 in the short run and -0.26 to -0.31 in the long run. Small and Van Dender’s estimates are based on recent data, and they attribute their estimates’ being lower than those in the reviewed literature to growth of real (inflation-adjusted) income and lower real fuel prices, which have combined to make the cost of driving a smaller share of personal disposable income.

Technically, what is estimated is the VMT elasticity with respect to fuel costs, or the percentage change in VMT that results from a 1 percent change in per-mile fuel costs. See National Research Council, Effectiveness and Impact of Corporate Average Fuel Economy (CAFE) Standards (Washington, D.C.: National Academy Press, 2002), available from www.nap.edu/catalog.php?record_id=10172. For VMT elasticity estimates, that report cites David L. Greene, James Kahn, and R. Gibson, "Fuel Economy Rebound Effect for U.S. Household Vehicles,"Energy Journal, vol. 20, no. 3 (1999), pp. 1–31; J. Haughton and S. Sarker, "Gasoline Tax as a Corrective Tax: Estimates for the United States. 1970–1991," Energy Journal, vol. 17, no. 2 (1996), pp. 103–126; and C.T. Jones, "Another Look at U.S. Passenger Vehicle Use and the "Rebound’ Effect from Improved Fuel Efficiency," Energy Journal, vol. 14, no. 4 (1993), pp. 99–110.

Richard Voith, "Fares, Service Levels, and Demographics: What Determines Commuter Rail Ridership in the Long Run?" Journal of Urban Economics, vol. 41 (1997), pp. 176–197. In earlier research, Voith reported a somewhat lower elasticity of ridership; see "The Long-Run Elasticity of Commuter Rail Demand," Journal of Urban Economics, vol. 30 (1991), pp. 360–372.

The primary costs considered in the survey were tolls and parking charges, but gasoline prices also affect driving costs. Lesser factors included increases in the time to drive a given distance and improvements in the quality of other modes of transportation. See Kevin Washbrook, Wolfgang Haider, and Mark Jaccard, "Estimating Commuter Mode Choice: A Discrete Choice Analysis of the Impact of Road Pricing and Parking Charges," Transportation, vol. 33, no. 6 (2006), pp. 621–639.

The correlation would be lower to the extent that the reduction in VMT attributable to higher gasoline prices occurred less on metropolitan highways and more on surface streets or rural highways. (That would be the case, for example, if urban-area residents responded to higher gasoline prices by cutting back on their out-of-town highway travel.)

Some results in Table 1-1 are sensitive to the inclusion of a location on eastbound Interstate 80 that often carries significant weekend recreational traffic toward Lake Tahoe. With I-80 in the sample, the "no rail option" weekend response is positive (indicating more driving in response to higher gasoline prices) and statistically significant, suggesting that the model is misspecified (in particular, it lacks indicator variables for weekends in summer and in ski season).

The systems in CBO’s sample averaged about 4.1 million trips per month in 2006, ranging from 8.8 million on Bay Area Rapid Transit to 1.2 million on Sacramento’s light-rail system. Thus, 1.9 percent translates to about 78,000 additional trips per month, or 1,870 trips per (nonholiday) weekday, under an assumption that each added rider takes two trips (one round-trip) per day— 21 days a month, on average.

The range of values is attributable to differences in the commuter populations on the two freeways. The estimates may slightly overstate true values of time, as they may also reflect motorists’ willingness to pay for greater reliability and perceived safer conditions on toll roads. See David Brownstone and others, "Drivers’ Willingness-to-Pay to Reduce Travel Time: Evidence from the San Diego I-15 Congestion Pricing Project," Transportation Research, Part A: Policy and Practice, vol. 37 (2003), pp. 372–387. For results from SR-91 in Riverside, see Kenneth A. Small, Clifford Winston, and Jia Yan, "Uncovering the Distribution of Motorists’ Preferences for Travel Time and Reliability," Econometrica, vol. 73, no. 4 (2005), pp. 1367–1382.

On the basis of his review of the economic literature, Kenneth Small concluded that "a reasonable average value of time for the journey to work is 50 percent of the [motorist’s] gross wage rate." See Kenneth A. Small, Urban Transportation Economics, vol. 51 of Fundamentals of Pure and Applied Economics (Newark, N.J.: Harwood Academic Publishers, 1992), p. 44. For value-of-time estimates not based on road tolls, see Orley Ashenfelter and Michael Greenstone, "Using Mandated Speed Limits to Measure the Value of a Statistical Life," Journal of Political Economy, vol. 112, no. 1 (2004), pp. S226–S267. See also Robert T. Deacon and Jon Sonstelie, "Rationing by Waiting and the Value of Time: Results from a Natural Experiment," Journal of Political Economy, vol. 93, no. 1 (1985), pp. 627–647. Deacon and Sonstelie report that drivers who voluntarily queued for lower-priced gasoline implicitly valued their waiting time at amounts similar to their after-tax wages. Finally, a long-ago survey of empirical studies of choices of commuting modes showed that, on average, the estimated value of time spent commuting was only about 20 percent to 30 percent of wages, although that analysis examined commuting by all modes of transit, not just private automobiles. See Nils Bruzelius, The Value of Travel Time: Theory and Measurement (London: Croom Helm, 1979).

Young-Jun Kweon and Kara Kockelman, "Driver Attitudes and Choices: Speed Limits, Seat Belt Use, and Drinking-and-Driving," Journal of Transportation Research Forum, vol. 45, no. 3 (2006), pp. 39–56.

A 1979 report (based on 1972 data) found that a 10 percent increase in gasoline prices would have induced a 3.5 percent decrease in annual statewide average highway speeds. See Carol A. Dahl, "Consumer Adjustment to a Gasoline Tax," Review of Economics and Statistics, vol. 61, no. 3 (1979), pp. 427–432. More-recent work finds a weak (not statistically significant) link between higher gasoline prices and slower speeds (Nicholas E. Burger and Daniel T. Kaffine, "Gas Prices, Traffic, and Freeway Speeds in Los Angeles," University of California at Santa Barbara, Department of Economics, unpublished working paper). The lack of significance may stem from a failure to account for correlations in highway speeds along different routes and at different times and in considering speeds only within a pair of two-hour nighttime periods.

See David Grabowski and Michael Morrisey, "Do Higher Gasoline Taxes Save Lives?" Economics Letters, vol. 90 (2006), pp. 51–55. A more in-depth exposition of that research—albeit with a focus on declining real gasoline prices—appears in Grabowski and Morrisey, "Gasoline Prices and Motor Vehicle Fatalities," Journal of Policy Analysis and Management, vol. 23, no. 3 (2004), pp. 575–593.

CBO does not have data on individual motorists, so it is not possible to determine the extent to which individual drivers tend to drive faster or more slowly nor to observe differences in how such motorists respond to gasoline prices. However, CBO’s findings are consistent with that suggestion.

See B.H. West and others, Development and Validation of Light-Duty Modal Emissions and Fuel Consumption Values for Traffic Models,FHWA-RD-99-068 (Federal Highway Administration, 1999).

The 0.6 percent represents the midpoint of a range of effects implied by recent elasticity. Jonathan E. Hughes, Christopher R. Knittel, and Daniel Sperling, Evidence of a Shift in the Short-Run Price Elasticity of Gasoline Demand, Research Report UCD-ITS-RR-06-16 (University of California, Davis: Institute of Transportation Studies, 2006), estimate that the short-run demand elasticity for

			I-405
		I-680	(South)	I-8
		(North)	Orange	(West)
		San Ramon	County	San Diego


Median Number of
	Vehicles per Hour	3,460	5,530	3,620
5th Percentile Speed		62.7	59.0	66.8
Median Speed		66.4	67.6	69.5
95th Percentile Speed		70.3	70.9	71.3


		5th Percentile		Median		95th Percentile


Baseline Speed (mph)		62.8		67.8		70.8
Change in Speed (mph)		- 1.2	**	- 0.8	**	No change

Elasticity of Speed with
	Respect to the
	Price of Gasoline	-0.09		-0.05		0

Implied Elasticity of
	Fuel Consumption^a	-0.08		-0.06		0