Oil demand growth – of graphs and gasoline

A closer look at gasoline consumption suggests that central predictions for future global oil demand may still be too high

Supply is more interesting than Demand – it shouldn’t be

On Google trends, citations for “oil supply” are three times more common than “oil demand”. Why is this?

Perhaps when the scale of something has grown steadily over decades, to become large and pervasive, the original aim of it gets gradually disregarded, and the thing itself becomes the purpose. Just managing it, maintaining it, analyzing it and communicating about it can require a large amount of effort and energy, even ahead of ensuring its primary function still works properly.

Clear examples are large bureaucracies, such as government departments or sprawling financial entities such as major banks.

One more case in point may be the crude oil industry. Its vast infrastructure of pipelines, projects, products, companies, traders, employees and major newsworthy events such as Macondo and Nigerian oil attacks, ensure the supply side of the business is constantly gaining attention.

Internally, just keeping the business moving with all its engineering, political and financial complexity means producers and suppliers may rarely pay attention to external shifts and leanings that could seriously disrupt their business.

For the supply side IS the oil and gas business. When we discuss the oil industry, we are talking about it as a supplying body. It’s the same with almost all commodities – supply is tangible and stimulating, and intrinsically full of short-term narratives about people, safety, deals and changes.

Demand is different. It is diffuse, moves in longer-term cycles, and narratives are often subtle, hidden within quiet trends.

So, most theories of oil price reflect ideas and narratives about supply.

The oil price slump of over 60% since 2014 is the latest example. Most hypotheses centre on OPEC policy to maximize production, and the rise of oil shale in the US.

But a minority point to demand shocks from China’s sudden move away from heavy industry to services.

Whatever the case, and its likely both are involved, we are going to unashamedly review demand here on a regular basis, to keep the ratio of supply / demand analysis closer to one to one.

Lets start with gasoline.

Gasoline Demand

Whilst its maybe not surprising that oil supply often trends larger than oil demand, it’s a bit more unexpected that “oil supply” trends 10 times as large as “gasoline supply” or “gasoline demand”.

That’s because you can argue that gasoline demand is oil demand, as it’s the largest component of the barrel (45%) and the highest growth (global consumers via cars).

Whatever the reason, lets focus on a major oil industry fact – oil’s primary function today is to produce transport fuels (gasoline and diesel), and demand growth in this sector is almost exclusively in gasoline from non-OECD sources, mainly China and India.

Take away growth in gasoline, and crude oil demand flattens – this is because all other oil barrel products are in an ex-growth phase, serving mature industry sectors, where energy conservation and lower market growth have reduced the opportunities to expand fuel consumption.

This is why most major international and national oil companies take considerable effort to show convincing predictions of major gasoline demand growth over the next 10-20 years, and equally convincing predictions of the limited uptake or impact of alternative options such as bio-fuel and electric vehicles.

Lets use the latest BP Global Energy Outlook analysis to get at the numbers.

BP Fuel demand 2035

About 65% of a barrel of crude oil goes into making gasoline and diesel fuels – hence the BP data shows that of the global consumption of crude oil liquids of about 92 million barrels per day in 2015, around 53 million barrels went in to transport via gasoline and diesel.

Let’s be precise here for future reference, and use an American Petroleum Institute table on oil products from a barrel of oil.

Crude oil Fuels

The other major component from crude oil is used for “industry” – from the table above you can see that is heavier oil for use in heating and power. Note some of this is used in petrochemicals, but the majority of petrochemicals are derived from “NGLs”, non-gaseous liquids, which are different in kind to typical crudes.

Note also that from the BP chart the industry liquids consumption profile has been flat since 2005, so the large uptick projected looks more unrealistic than the transport profile. This is a point we will come back to later.

For these reasons, the focus of attention here is transport gasoline. As BP note, “The growth in the global consumption of liquid fuels is driven by transport and industry, with transport accounting for almost two-thirds of the increase. “

Boiling all this down, as it were, predicts that the largest reason for crude oil demand growth over the next 20 years will be the demand growth of mainly gasoline, with diesel usage likely to be flatter (consumer car growth far more robust than heavy industrial development).

How does that demand picture look, and how is the growth derived – BP again: “The growth in transport demand reflects rapid increases in vehicle ownership in emerging economies, partially offset by sustained gains in vehicle efficiency, which slow the sector’s growth post-2025”

Fueld demand growth BP

Lets take stock of this: gasoline demand growth drives crude oil demand growth: and gasoline demand growth is driven by the vehicle ownership uptake in emerging economies, especially China over the next two decades.

An awful lot is riding on this demand chart then.

If the slowdown in US / OECD gasoline growth is quicker than predicted or the upturn in China does not follow the predicted growth, the impact on oil supply requirement could be large and in turn the industry could be caught wrong-footed, as it was with shale oil and Chinese industrial demand in 2014.

Taking each major component in turn:

US Fuel Demand

Assessing US fuel usage is important as the US consumes over about 25% of the world’s liquid transportation fuels.

In 2003, the US advisory body, the EIA predicted US oil demand to 2025 (shown below). The dotted line is their prediction, the solid line actual up to 2015. By last year, the actual demand out-turn was over 6 million barrels per day (mbpd) less than the prediction of 12 years ago. Almost all of this was due to a reduction in gasoline demand against the prediction (as shown). By 2025, the difference will be 10 mbpd.

WEF Projection 2003

The main reason for the large gap in demand was two factors: driving behavior (75%) and vehicle fuel efficiency (25%). Driving behavior is measured by Vehicle Miles Travelled (VMT), and as the chart below shows, over the period, gasoline prices and behavior conspired to make 2004 a peak gasoline demand year, with consumption retreating since. Baby boomers grew old, drove less, and their children became less in love with motor cars.


Thus, when demographics, aging of the population, and some basic economic factors are added to a simple model the following profile occurs (note differing timeframes).
Models for VMT v3

There is still an error – some factor is missing as the VMT does actually increase above the model, but the use of a simple predictive model avoids the major gap of 6 million bpd that the linear analysis allowed.

There are some major implications here however from this tale of expected gasoline demand growth:

• Simple linear trends can create major errors – the use of in this example of an innocuous 1.8%pa expected growth rate may be fine for a year or two, but relying on it for projections out 10-20 years has potentially serious consequences. Although projections were updated, and became more accurate, any car or oil firm or government agency using these projections for several years may have invested in plant, people or expansions that in the end were not necessary, or far less efficient than expected.
• The impact of demand shifts can be profound. Its worth restating that expected future oil consumption in 2015 was over 6 million bpd higher than reality. Its unclear whether any supply investment could have actually created such a rapid increase in fuels, although oil company plans would have attempted it given demand projections (leading to overhangs experienced now perhaps).
• Simple models may be better than scenarios. A stress test growth rate of say 1% rather than 1.8% would have “only” been in error by 2 mbpd by 2015. Not good, but unlikely to have caused major strategic mis-steps or investment cases. The addition of just one or two other variables also increased accuracy by about 80%.
• House preferences – finally, its always worth assessing where the analysis is coming from – independent observers, or industry participants. The EIA is an independent government agency, but it has a tendency to be at the top end of industry projections. Likewise, projections by firms such as BP may tend to favour certain scenarios or tools – as a general rule oil firms will tend to prefer high growth models as they indicate higher future revenues, and justify continued investments.

In summary, the assumption of flat US gasoline forward demand looks robust as projections are now more settled than several years ago. However, we need to be aware of changes that can cause demand shocks either way. These include deeper penetration of SUVs leading to more per capita fuel consumption, or the sudden rise of EVs.

Absent these, there appears to be no growth in demand from the US in the transport fuel and hence for crude oil, as predicted.

This points to growth in demand from elsewhere.

Chinese Fuel Demand

If growth is to occur, China, and then India will be the main protagonists. China commands over 40% of total Asian demand for gasoline, and accounted for over 70% of global growth in consumption of the product in 2015.

Today China consumes about 2.5 mbpd of gasoline, and major growth out to 2035 is predicted. Its diesel consumption has risen quickly this century, but has plateaued at around 3.0 mpbd since 2014 as heavy industry demand has declined. Diesel growth is expected to remain low moving forwards, mirroring China’s GDP softening.

In sum – heavy industry growth has reached plateau (diesel), consumer-led growth is to follow (gasoline).

Chinese cars fuel

Central scenarios of Chinese transport fuel demand growth, such as BP’s, are based on the trebling of car ownership to reach peer per capita models of about 300 vehicles per 1000 people. This leads to expected gasoline demand growth of around 2.5 – 3 mbpd by 2035, an equivalent estimated 5 mbpd increase in crude oil.

Its worth noting that the projection in the BP chart looks like a linear trend model of about 4-5% growth pa across the period. The main predictive variable is the ownership ratio, followed by consumption, and allowing for usage efficiency gains. This is applied across the OECD growth in general, but China (and India) clearly the largest contributors.

Some commentators suggest that these robust growth projections still may underestimate Chinese consumer growth of gasoline and vehicle ownership and travel patterns.

However, as the growth is predicated on consumer behavior, and central government policies, its worth being cautious, again, about linear trends.

A number of factors will almost certainly impact how the Chinese car market grows – to be sure some may increase consumption, but others will cause demand to flatten more quickly. Some of these are highlighted below:

City ownership statistics: an overall ownership number target may overlook that in several large cities, ownership ratios are already quite high by peer standards eg Bejing and Shanghai. This has led to severe congestion issues, and governmental policies to limit new registrations and impose stricter quotas on sales
Health and Safety: city smog and congestion are already at serious levels in several cities, leading to curtailment of travel and engine sizes. In addition, more critically, the rapid rise in ownership has lead to an increase of fatalities to over 220,000 in 2015, at about 20 times the rate per vehicle in the UK.
Transport Policies: Chinese policies on car ownership are moving in several directions – including incentives for smaller engine vehicles, and toward hybrid cars, especially home-grown ones. Targets for car fuel economy have been set to increase from 35mpg in 2015 to 47mpg by 2020.
New energy vehicles: private and public moves to introduce hybrid and EVs are increasing rapidly with China’s latest Five Year Plan aiming to introduce 500,000 EVs in 2016, and over 1 million by 2020. This envisages investment in EV subsidies and infrastructure of over $15bn over the next ten years.
Demographic and technology shifts: as in the US, younger city dwellers are trending more toward public or internet-based transport services such as Didi, which may significantly alter the demand for gasoline. China also enters the main vehicle market when efficiency and electric technologies are far more mature and diverse.

It is therefore difficult to predict where demand will transition over the next few years. However, it is likely that a simple linear trend projection could mis-read the outcome significantly. This will have serious consequences for policies and investments in either direction.

Its surprising that models of demand for such critical areas seem quite simple and often weighted toward preferences.

In such as case, the actual oil demand for China could be significantly different from current projections. As most of the current policies and directions are aimed at energy efficiency, and curtailment of excess use, the likeliest outcome is that the linear trend over-estimates ultimate demand, as we saw in the US case. In addition, the global context of weaker GDP expectations adds more downside uncertainty.

The overall transport growth predicted in the BP model of about 12 mbpd seems optimistic given these factors – potentially by 50%, using China as a proxy for the overall projection.

Industrial Crude Demand

A final point to note is that the growth projected in the BP chart for industry growth of an increase of 6 mbpd over the 2015-35 period. This is a compound growth of 1.5% for a sector that has shown no real growth for a decade. Without a justification or model for the uptick, it seems hard to see why something close to flat growth should very rapidly increase in growth at several times the current trend.


The study of oil and oil product demand seems to be a different discipline from the study of supply.

Supply often assumes a certain rate of demand in the background and then emphasizes current myriad issues such as OPEC negotiations, field disruptions, new project updates and investment issues.

Demand is more a domain of gradual trends and tendencies, and the subtleties (and dangers) of compound growth.

We have quickly reviewed longer-term (to 2035) gasoline demand as a proxy for crude demand, using a base case (BP Global Outlook 2016) industry projection.

Three key findings:

1. Generally, we need to ensure we are very cautious about accepting demand projections based on linear trends, especially greater than 5 years – life is too intricate for that. The linear trend analysis seems to be a “best case” growth model, which is at the very high end of expectations, and is susceptible to significant impacts, especially when analyzing consumer behavior.
2. Gasoline, and hence crude demand growth is heavily predicated on Chinese (and Indian) gasoline demand growth. At the moment, the demand growth models for this expansion are based on linear trends from current data – this risks missing actual outcomes significantly, with serious consequences for long-term investors such as oil firms and governments.
3. On balance, given the direction of Chinese policies on vehicle efficiency, technology substitution and urban policies, its likely that actual gasoline and crude growth will be less than currently estimated in these models – realistic lower growth scenarios should be more clearly indicated.

Given this, the alternative estimate for 2035 demand using this analysis is closer to 100 mbpd, or some 12 mbpd lower than the BP industry estimate examined.

Other industry analysts, eg McKinsey, have proposed similar outcomes.

We’ll return to demand regularly.