Modelling Long Term Oil Demand

Trying to predict long term oil demand is important for society. Governments and businesses need to anticipate future energy trends in order to make decisions on investments. There are several factors involved in making these models.

1.      Population – how many potential consumers will there be in the future? Other demographic factors such as ageing populations.

2.      GDP – how much money will these consumers have and what portion of this wealth are they able or willing to spend on energy?

3.      Technology – how efficient will be the machines used to generate and spend the wealth that the consumers (people, governments and businesses) have? What materials will people use and in what quantities?

4.      Environment – the need to maintain a reasonably healthy environment and sustain resources for future generations.

5.      Society – how consumers view energy and environment? Trends in housing, urban vs rural vs suburban living, personal vs public transport, conspicuous consumption vs virtue signalling frugality, effect of other demands on income such as taxes or debt payments?

I have built a simple model combining population and oil per capita for the most important oil consuming countries to produce a low mid and high forecast. This was then compared to public domain forecasts.

·        The low case forecast for 2040 has humanity consuming 90% of 2015 oil consumption

·        The mid case forecast for 2040 has humanity consuming 122% of 2015 oil consumption ( I think however, that producing this much oil would be challenging geologically)

·        The high case forecast for 2040 has humanity consuming 156 % of 2015 oil consumption ( I think that this level of consumption is not realistic from a geological reserves and capacity perspective, let alone for any environmental consequences, but this case illustrates what the demand could be in an unconstrained world)

The low and mid cases are within the general range of other public domain forecasts. 

Figure 1 – comparison of forecasts for oil consumption, relative to 2015

Building a Simple Model

I have built a relatively simple model that looks at how future demand may unfold. This model has two variables:

a.      Population projections until 2040 based on the United Nations world populations prospects – ( data is available here = https://population.un.org/wpp/Download/Standard/Population/)

b.     Per capita oil consumption – projected using data from the BP Statistical Review of World Energy – ( data is available here = https://www.bp.com/en/global/corporate/energy-economics/statistical-review-of-world-energy.html )

A matrix was created for each of the top 15 countries which account for 68% of present day oil consumption. This matrix consisted of high, mid and low case forecasts for both population and per capita oil consumption as illustrated below

Figure 2. Prediction matrix.

Three cases were then taken for each country a Low-Low case, a High-High Case and an average (mean) of all cases.

Population Predictions

The United Nations population division publishes an annual prognosis for the world’s population. The projection is illustrated below.

Figure 3. UN population projections

The UN produces mid, low and high estimates as well as a constant fertility rate estimate, and a momentum estimate. By 2040 the world is forecast to have between 9 and 9.6 billion people compared to 7.4 Billion in 2015.

The Top 15 Countries.

The top 15 Countries account for 68 % of the world’s oil consumption and therefore modelling their future consumption would be an effective proxy for the whole world.

Figure 4. Oil consumption share by country (based on BP statistical review of world energy)

Past Consumption Patterns

Between 1990 and 2017 the world has increased oil consumption by 42% from 3154 million tonnes/year (66.5 MMBO/day) to 4470 million tonnes /year (93 MMBO/day). The OECD’s consumption has gone up by only 9% during this time span while the non OECD countries have increased consumption by 95%.  

Figure 5. Oil consumption by region, 1990 to 2017 (BP statistical review of world energy)

China and India have seen the greatest growth while Russia, Japan and to a lesser extent Europe have seen declines in oil consumption.

Looking at consumption per capita we actually see a steady state for oil, with declines in OECD and former communist countries offsetting an increase in growing non OECD countries such as China and India. Coal has seen a rise since 2000 in per capita consumption (almost entirely due to China) and gas has seen a steady rise from 1990 to 2016. Note the small dip in consumption of all three fuels due to the 2008 worldwide recession.

Figure 6. Per Capita fossil fuel consumption, 1990 to 2015 (based on BP statistical review of world energy and UN population database)

Per capita oil consumption is of course not evenly distributed. The chart below shows the per capita consumption of the top 15 countries.

Figure 7. Per capita oil consumption, top 15 countries

Almost all of the OECD countries show a decline in per capita consumption between 1990 and 2015. The Non OECD countries (other than Russia) show a growth in per capita consumption. Note the large increase in Saudi Arabia.

The Link between GDP and Oil Consumption

Oil consumption is related to GDP but this is not a clear one to one relationship. There appear to be four phases to the GDP and oil consumption equation, illustrated below.

Figure 8. The Oil consumption and GDP relationship model

·        Phase 1. This is when per capita oil consumption grows with per capita GDP. The society is developing and people are using more energy as they buy their first motorbike or car. This is where China, India and Brazil are today

·        Phase 2. Oil consumption is growing more slowly. People still consume more oil with GDP growth but fuel efficiency leads to a slower rate of growth. This is where the USA was during the 1990’s and early 2000’s and where Russia has been since the post Yeltsin recovery started around 2000.

·        Phase 3. This is when oil consumption decouples from GDP growth. Energy efficiency means that people get a higher income from a similar oil consumption. This is where the USA is now.

·        Phase 4. This represents a more post-industrial world. GDP continues to grow, albeit slowly. While per capita oil consumption declines, due to even more energy efficiency and deindustrialisation. This is where most of the EU and Japan currently are.

Figure 9. GDP vs Oil Consumption

The graph above shows trends for selected countries. China and Brazil show phase 1 growth. South Korea showed an almost instant transition from phase 1 to phase 3, spurred by the 1998 financial crisis. Russia shows a phase 1 in reverse during the Yeltsin depression of the 1990’s and a phase 2 growth since then. Japan, Germany and Britain show phase 4. Saudi Arabia shows off scale growth.

What is oil used for? What about substitution ?

Oil has several uses of which only some may be substituted by other energy sources. Oil has many advantages;

·        It is energy dense, the energy value of a barrel of oil (in Kw) is equivalent to 4.5 person years of manual labour.

·        Portability, oil is easy to store and transport.

·        Consistency, refinery quality control gave predictability in fuel quality. Industrial feedstock, oil is non – substitutable in plastics / polymers, solvents lubricants and petrochemicals.

But there are also significant drawbacks; Oil is a hazardous product and tragedies have happened in extraction, transport and refining despite a focus on safety. There are also significant environmental costs in terms of both immediate pollution, NOx, SOx, Carbon Monoxide and particulates as well as CO2. Oil production does have social and environmental impacts and there are concerns about long term supply as oil is a finite resource. There are also concerns in some countries about oil supply security and import dependency.

Figure 10. Usage of oil by sector (BP statistical review of world energy)

About 56 % of oil is used in transport but only about half of this might be substitutable by electrification of light vehicles (cars, light vans and taxis) and some heavy vehicles such as urban buses and trains. The portions used in industry and petrochemicals are non-substitutable. Note that only 5% of oil use is for generating electricity, this varies considerably between counters but is only about 1% in the USA. Humanity’s need for petroleum is therefore going to be around for a very long time.

Forecasting Future per Capita Oil Use.

All forecasts are by definition uncertain. I will show models for the biggest two countries USA and China.

Figure 11. USA per capita oil consumption – historical and forecast.

The US has shown a fall in per capita usage since the oil price rises of the early 2000’s, which slowed somewhat post 2010. The mid case (blue) continues on the present day trajectory to show a slight decline. The low case shows an acceleration in per capita usage decline due to an emergent energy transition and greater emphasis on energy efficiency. The high case shows a reversal of current trends with greater per capita oil use followed by a plateau. 

Figure 12. China per capita oil consumption- historical and forecast

China has shown a steady rise in per capita oil usage as Chinese people have become richer. The mid case (blue) shows a continuation of present day trends (the graph shows a 97% R2 correlation). The high case (red) shows a slight acceleration, followed by steady growth. The low case is a more revolutionary scenario where there is a rapid transition to low oil / high efficiency hybrid and electric vehicles spurred by government orders to combat pollution. Per capita oil consumption declines slightly, but economic growth continues at a slower pace than currently. I feel that this scenario is less likely but it is not that far removed from what happened in South Korea during the 2000’s.

Similar forecasts were produced for the other top 15 countries.

Figure 13 – USA Oil consumption Forecast. Min with a 15% decline represents a low case population model and a low case consumption model. Max with a 34% growth represents a high case population and a high case consumption model. The mean case is the average of all cases with a 9% growth in oil consumption (due to population growth).

Figure 14. – China Oil consumption Forecast. Min with a 14% decline represents a low case population model and a low case consumption model. Max with a 102% growth represents a high case population and a high case consumption model. The mean case is the average of all cases with a 53% growth in oil consumption.

Aggregated Results

Figure 15. Low Case Forecast

This is the sum of low cases for the top 15 countries. (I know that statistically this is not really the right thing to do, adding all the low cases, but this gives an indication of a possible floor, apologies to my high school maths teacher)

Figure 16. Mid Case Forecast

This is the sum of all the mean cases. I feel that existing reserves and future resources would be seriously stretched to meet this demand.

Figure 17 High Case

This is the sum of all the high cases (again apologies to my high school maths teacher). I feel that current oil reserves are highly unlikely to meet this demand but this case gives an idea of what an unconstrained scenario looks like.

Comparison with Other Forecasts

Figure 18. Summary of demand forecasts. All have been indexed to 100 in 2015 for ease of comparison.

Comparing to other forecasts the mid case is very similar to the forecasts of the Massachusetts Institute of Technology (MIT), BP and the Energy Information Administration (EIA). The Shell Sky Scenario, released earlier in 2018 shows a similarity to the mid case initially followed by a decline after 2030. The information behind the scenario with the forecast numbers in a spreadsheet is available here (https://www.shell.com/energy-and-innovation/the-energy-future/scenarios/shell-scenario-sky.html ). However Shell forecast a similar overall consumption in 2040 to what is currently being consumed. Equinor (formerly Statoil) have produced three scenarios (available at https://www.equinor.com/en/how-and-why/sustainability/energy-perspectives.html). Their renewal scenario is lower that the modelled low case in 2040 but still has a forecast demand of 75% of today’s.

Some of these demand forecasts would be difficult to meet with the hydrocarbon resources that we have. We need to explore and develop, running like Mo Farah just to stand still, let alone needing to run the 1500 m at Usain Bolt pace which is what is needed to meet some of the larger demand projections. This is illustrated by a diagram from the Equinor sustainability scenarios shown below

Figure 19, Need for new hydrocarbon investment to meet demand even under low case scenarios