In the last Musings we wrote about the challenge of understanding why gasoline demand has collapsed in recent months at the same time the economy is recovering and new car sales are ramping up. In that article we focused on the acknowledgment by the Energy Information Administration (EIA) that it had misestimated gasoline export volumes and as a result may have overstated the magnitude of the year-to-year decline in domestic gasoline consumption.

Unfortunately, based on the chart on the EIA's website showing the impact on export volumes from its prior estimation methodology and its current one, we still have a ways to go before this change will correct the data reporting problem -- assuming the EIA doesn't make further adjustments in its estimating procedures.

As can be seen from Exhibit 2, there has been a significant increase in monthly gasoline export volumes since the switch to the new methodology in August of last year, and there is a noticeable difference between the monthly export figures from the Petroleum Supply Monthly, which gets its volume figures from the Census Bureau, and the 4-week average volume estimate derived from an historical analysis of five years' worth of data from the Weekly Petroleum Supply Report. By using the latter analysis, export volumes were estimated to be lower than the monthly data report projected, which had the effect of leaving the domestic volume too high. With higher estimates for export volumes, domestic demand is reduced making the year-to-year comparison less stark.

While we suspect this flaw in the statistical methodology has impacted the outcome, we also remain convinced there are structural changes underway in the gasoline market that have impacted, and likely will continue to impact, consumption. As shown in Exhibit 3, there has been a decline in total vehicle miles driven in the United States since the financial crisis in 2008. But the growth rate in vehicle miles driven had been slowing before the peak. The two factors impacting vehicle miles and gasoline consumption that we have focused on are the age mix of drivers along with their driving habits and the impact of the Internet and social media on driving habits. A number of surveys have shed light on these issues.

If one examines Exhibit 4 showing the historical growth of the U.S. population, the number of vehicle registrations and licensed drivers, it becomes clear that some of the past trends are changing. For example, in recent years there has been a decline in the number of vehicles in America. At the same time, there has been a flattening in the growth rate for licensed drivers at a time when our population was growing at its historical rate. It is interesting to note that it required just 17 years for the number of licensed drivers to increase from 100 million to 150 million (1965-1982), yet 23 years were needed to add the next 50 million drivers (1982-2005).

In Exhibit 5, we see similar changes in growth trends for the number of registered vehicles, licensed drivers and driving-age population. In that chart, during the last half of the 1990s, there was a noticeable increase in the driving-age population, yet no comparable rise in the number of licensed drivers. This suggests that a portion of the driving-age population has not secured driver's licenses, which would likely be the youngest age group that is just entering that populations segment.

Several studies of Federal Highway Transportation Department data for licensed drivers by age and sex substantiate that these teens are the drivers who are not procuring licenses. We will explore some of the reasons later, but first we want to address the gasoline consumption question. As shown by Exhibit 1, high gasoline prices have impacted gasoline consumption as people are driving less. The most recent weekly data from the MasterCard SpendingPulse survey that measures gasoline sales showed the latest weekly demand being down 1.5% from the prior week, while it fell 7% from the same period a year ago. This marked the 53rd straight weekly decline in gasoline consumption. A survey showed that Americans under the age of 40 were driving significantly less than 10 years ago, but the unemployed people in this age group had reduced their driving by between 19% and 24%. That trend prompted the surveying firm to question whether the American car culture was waning as younger Americans have become more environmentally aware.

The decline in driving by younger drivers is shown clearly in Exhibit 6 where the blue line shows miles driven per year per capita for 1995, the red line for 2001 and the green line for 2009. What it shows is substantially fewer miles driven by 18-39 year olds in 2009 compared to prior years. The magnitude of the declines between 1995 and 2009 (purple) and 2001 to 2009 (blue) confirm this trend.

One explanation is the sharp decline in the percentage of younger Americans who are obtaining their drivers' licenses during the 25-year period 1983-2008. While the percentage is down for Americans from 18 years old to 40, the corresponding licensure trend has increased for drivers 50 years and older. Some of this increase probably reflects middle-aged drivers maintaining their licenses as they grow older.

To further understand the impact these licensed driver trends might have on gasoline consumption, we looked to see how miles driven varied by age. Unfortunately, there is not the granularity in miles driven by age as there is for licenses (Exhibit 8).

What this data shows is that the most miles driven are racked up by those in the 20-54 year age bracket. Individuals in the 16-19 and 65+ age groups drive about the same number of miles annually, but there is a marked difference in the number of miles driven by sex, especially for the older group. The data suggests that there shouldn't be much difference in gasoline consumption, based on miles driven per year, between teenagers and American drivers 65+ years old. While that may be what the statistical averages show, we believe that as people age their energy consumption declines, which is consistent with other energy surveys, and especially for gasoline. The more important question becomes what may happen in future years as the American population ages?

Based on the demographics from the U.S. Census Bureau, there has been dramatic growth in the number of Americans who are 70 years old or older. In 1983, the total number of teenage and 70+ drivers were essentially equal, but by 2010 (the latest data available) there had been more than a doubling of oldsters with no appreciable growth in the number of teenagers, meaning there has been a significant increase in the relative importance of this older group of drivers for gasoline demand.

The impact of these older drivers is shown in Exhibit 10. The oldest driver group has increased its share of total drivers by nearly four percentage points while teenagers declined by two percentage points. That amounts to a nearly six percentage point swing. If we assume that 70+ aged drivers don't drive the average number of annual miles suggested by the data for 65+ drivers in Exhibit 8 (page 8), we then conclude that there will be a permanent reduction in miles driven by this group even though there will be more adults in this age category. On the other hand, as this older group accounts for a larger percentage of all drivers, and not just at the expense of teenagers, the younger peak age drivers who account for the greatest number of miles driven will also be declining. All of these demographic trends on driving come before we consider the impact of more fuel-efficient vehicles in the fleet and other economic and social changes underway.

If we consider the population and recent projections from the Census Bureau, we can develop a feel for what may happen to future gasoline demand as a result of the demographic trends. In 2010, there were 21.8 million people in the 15-19 age range compared to 28 million who were 70+. By 2015, there is projected to be about 500,000 fewer teenagers but three million additional older adults. The spread between these two age groups becomes even wider by 2020 when there will be 22.6 million teenagers but 36.9 million who are 70+. As a result, the count of 6.2 million more 70+ adults in this country expands to more than 14 million by the end of 2020. To our way of thinking, there has to be a negative impact on miles driven and gasoline consumption from this trend.

In a new study published in the journal Traffic Injury Prevention, two professors associated with the University of Michigan Transportation Research Institute found that having a higher proportion of Internet users was associated with lower licensure rates among young people. The researchers compared data for 14 other countries to that of the United States and found that half of the other countries showed a similar age-related pattern of change in U.S. driver licensing -- a decrease in young drivers and an increase in older drivers. Canada, Great Britain, Germany, Japan, Norway and South Korea all have seen similar licensing declines. On the other hand, Israel, Finland, Poland, Latvia, Spain, Switzerland and the Netherlands have experienced an increase in both young and older drivers over time, although the increase in the younger age group was smaller.

According to the research, in 1983, a third of all licensed drivers in the United States were under age 30. Today, that is only about 22%. Also, about 94% of Americans in their 20s had a driver's license in 1983, compared to about 84% in 2008. The conclusion of the study was that "…countries with higher proportions of Internet users were associated with lower licensure rates among young persons, which is consistent with the hypothesis that access to virtual contact through electronic means reduces the need for actual contact among, young people." This conclusion is borne out by results from other surveys. A 2011 survey by Zipcar found that 68% of licensed drivers under age 35 sometimes choose to spend time with friends online rather than driving to see them. J.D. Power and Associates found that young adults care more about their cell phones than their cars. And the Gartner research firm found that 46% of drivers aged 18-24 would choose Internet access over owning a car.

These conclusions are consistent with the changes in social media and their impact on the mobility needs of younger people. But there is also the impact of the Internet on jobs and how that might impact driving patterns. A new study on the employment effects of Internet and wireless technology improvements found some interesting data.

According to the McKinsey Global Institute, the Internet contributed about 3% to global GDP growth in 2009 and was responsible for 21% of the U.S. GDP growth over the last five years. Furthermore, a recent study by the World Bank found that among high-income countries such as the United States, every 10 percentage-point increase in broadband penetration is associated with an additional 1.21 percentage-points of economic growth.

While we haven't seen research specifically showing that the increase in Internet use or greater access to cell phones has cut down on driving and thus gasoline consumption, intuitively we have to believe that is the case. We base our conclusion on changes we personally have made in recent year to our shopping habits. It is often easier to shop for things on Internet web sites and have them shipped to our home rather than for us to drive to the store. That along with bunching errands together into one single trip rather than making multiple trips reduces miles driven, albeit not materially. But if millions of people adjust their driving habits just as we have, there will collectively be a meaningful impact.

The last noteworthy social trend is the desire of young adults to live in dense, urban neighborhoods. According to the results of a study by RCLCO Consumer Research, 77% of Generation Y plans to live in an urban area. More than half of them would trade the size of their homes for proximity to shopping or work. A rule of thumb is that doubling the density of a neighborhood will reduce miles driven by 20%, although much of the reduction is a result of dense neighborhoods being more likely to have public transportation as well as nearby shops, entertainment and employment. The Midtown area of Houston, which has been redeveloped over the past thirty years, is a good example of an urban area that attracts younger adults by its proximity to work, shopping and entertainment.

How much of the reduction in gasoline consumption that we have experienced recently is due to each of these many factors? It is impossible to sort that out definitively. What we believe, though, is that these trends are firmly established and will contribute to a steady decline in miles driven and gasoline consumption by individuals. That said we are still talking about the United States using nearly 8.5 million barrels per day of gasoline, a substantial volume. Gasoline use is not going away, but its influence in the domestic energy market will diminish over time.