An Optimistic Energy/GDP Forecast to 2050, Based on Data since 1820

We talk about the possibility of reducing fossil fuel use by 80% by 2050 and ramping up renewables at the same time, to help prevent climate change. If we did this, what would such a change mean for GDP, based on historical Energy and GDP relationships back to 1820?

Back in March, I showed you this graph in my post, World Energy Consumption since 1820 in Charts.

Figure 1. World Energy Consumption by Source, Based on Vaclav Smil estimates from Energy Transitions: History, Requirements and Prospects and together with BP Statistical Data on 1965 and subsequent. The biofuel category also includes wind, solar, and other new renewables.

Graphically, what an 80% reduction in fossil fuels would mean is shown in Figure 2, below. I have also assumed that non-fossil fuels (some combination of wind, solar, geothermal, biofuels, nuclear, and hydro) could be ramped up by 72%, so that total energy consumption “only” decreases by 50%.

Figure 2. Forecast of world energy consumption, assuming fossil fuel consumption decreases by 80% by 2050, and non fossil fuels increase so that total fuel consumption decreases by “only” 50%. Amounts before black line are actual; amounts after black lines are forecast in this scenario.

We can use actual historical population amounts plus the UN’s forecast of population growth to 2050 to convert these amounts to per capita energy equivalents, shown in Figure 3, below.

Figure 3. Forecast of per capita energy consumption, using the energy estimates in Figure 2 divided by world population estimates by the UN. Amounts before the black line are actual; after the black line are estimates.

In Figure 3, we see that per capita energy use has historically risen, or at least not declined. You may have heard about recent declines in energy consumption in Europe and the US, but these declines have been more than offset by increases in energy consumption in China, India, and the rest of the “developing” world.

With the assumptions chosen, the world per capita energy consumption in 2050 is about equal to the world per capita energy consumption in 1905.

I applied regression analysis to create what I would consider a best-case estimate of future GDP if a decrease in energy supply of the magnitude shown were to take place. The reason I consider it a best-case scenario is because it assumes that the patterns we saw on the up-slope will continue on the down-slope. For example, it assumes that financial systems will continue to operate as today, international trade will continue as in the past, and that there will not be major problems with overthrown governments or interruptions to electrical power. It also assumes that we will continue to transition to a service economy, and that there will be continued growth in energy efficiency.

Based on the regression analysis:

  • World economic growth would average a negative 0.59% per year between now and 2050, meaning that the world would be more or less in perpetual recession between now and 2050. Given past relationships, this would be especially the case for Europe and the United States.
  • Per capita GDP would drop by 42% for the world between 2010 and 2050, on average. The decrease would likely be greater in higher income countries, such as the United States and Europe, because a more equitable sharing of resources between rich and poor nations would be needed, if the poor nations are to have enough of the basics.

I personally think a voluntary worldwide reduction in fossil fuels is very unlikely, partly because voluntary changes of this sort are virtually impossible to achieve, and partly because I think we are headed toward a near-term financial crash, which is largely the result of high oil prices causing recession in oil importers (like the PIIGS).

The reason I am looking at this scenario is two-fold:

(1) Many people are talking about voluntary reduction of fossil fuels and ramping up renewables, so looking at a best case scenario (that is, major systems hold together and energy efficiency growth continues) for this plan is useful, and

(2) If we encounter a financial crash in the near term, I expect that one result will be at least a 50% reduction in energy consumption by 2050 because of financial and trade difficulties, so this scenario in some ways gives an “upper bound” regarding the outcome of such a financial crash.

Close Connection Between Energy Growth, Population Growth, and Economic Growth

Historical estimates of energy consumption, population, and GDP are available for many years. These estimates are not available for every year, but we have estimates for them for several dates going back through history. Here, I am relying primarily on population and GDP estimates of Angus Maddison, and energy estimates of Vaclav Smil, supplemented by more recent data (mostly for 2008 to 2010) by BP, the EIA, and USDA Economic Research Service.

If we compute average annual growth rates for various historical periods, we get the following indications:

Figure 4. Average annual growth rates during selected periods, selected based on data availability, for population growth, energy growth, and real GDP growth.

We can see from Figure 4 that energy growth and GDP growth seem to move in the same direction at the same time. Regression analysis (Figure 5, below) shows that they are highly correlated, with an r squared of 0.74.

Figure 5. Regression analysis of average annual percent change in world energy vs world GDP, with world energy percent change the independent variable.

Energy in some form is needed if movement is to take place, or if substances are to be heated. Since actions of these types are prerequisites for the kinds of activities that give rise to economic growth, it would seem as though the direction of causation would primarily be:

Energy growth gives rise to economic growth.

Rather than the reverse.

I used the regression equation in Figure 5 to compute how much yearly economic growth can be expected between 2010 and 2050, if energy consumption drops by 50%. (Calculation: On average, the decline is expected to be (50% ^(1/40)-1) = -1.72%. Plugging this value into the regression formula shown gives -0.59% per year, which is in the range of recession.) In the period 1820 to 2010, there has never been a data point this low, so it is not clear whether the regression line really makes sense applied to decreases in this manner.

In some sense, the difference between -1.72% and -0.59% per year (equal to 1.13%) is the amount of gain in GDP that can be expected from increased energy efficiency and a continued switch to a service economy. While arguments can be made that we will redouble our efforts toward greater efficiency if we have less fuel, any transition to more fuel-efficient vehicles, or more efficient electricity generation, has a cost involved, and uses fuel, so may be less common, rather than more common in the future.

The issue of whether we can really continue transitioning to a service economy when much less fuel in total is available is also debatable. If people are poorer, they will cut back on discretionary items. Many goods are necessities: food, clothing, basic transportation. Services tend to be more optional–getting one’s hair cut more frequently, attending additional years at a university, or sending grandma to an Assisted Living Center. So the direction for the future may be toward a mix that includes fewer, rather than more, services, so will be more energy intensive. Thus, the 1.13% “gain” in GDP due to greater efficiency and greater use of “services” rather than “goods” may shrink or disappear altogether.

The time periods in the Figure 5 regression analysis are of different lengths, with the early periods much longer than the later ones. The effect of this is to give much greater weight to recent periods than to older periods. Also, the big savings in energy change relative to GDP change seems to come in the 1980 to 1990 and 1990 to 2000 periods, when we were aggressively moving into a service economy and were working hard to reduce oil consumption. If we exclude those time periods (Figure 6, below), the regression analysis shows a better fit (r squared = .82).

Figure 6. Regression analysis of average annual percent change in world energy vs world GDP excluding the periods 1980 to 1990 and 1990 to 2000, with world energy percent change the independent variable.

If we use the regression line in Figure 6 to estimate what the average annual growth rate would be with energy consumption contracting by -1.72% per year (on average) between 2010 and 2050, the corresponding average GDP change (on an inflation adjusted basis) would be contraction of -1.07% per year, rather than contraction of -0.59% per year, figured based on the regression analysis shown in Figure 5. Thus, the world economy would even to a greater extent be in “recession territory” between now and 2050.

Population Growth Estimates

In my calculation in the introduction, I used the UN’s projection of population of 9.3 billion people by 2050 worldwide, or an increase of 36.2% between 2010 and 2050, in reaching the estimated 42% decline in world per capita GDP by 2050. (Calculation: Forty years of GDP “growth” averaging minus 0.59% per year would produce total world GDP in 2050 of 79.0% of that in 2010. Per capita GDP is then (.790/ 1.362=.580) times 2010′s per capita income. I described this above as a 42% decline in per capita GDP, since (.580 – 1.000 = 42%).)

Population growth doesn’t look to be very great in Figure 4, since it shows annual averages, but we can see from Figure 7 (below) what a huge difference it really makes. Population now is almost seven times as large as in 1820.

Figure 7. World Population, based on Angus Maddison estimates, interpolated where necessary.

Since we have historical data, it is possible to calculate an estimate based on regression analysis of the expected population change between 2010 and 2050. If we look at population increases compared to energy growth by period (Figure 8), population growth is moderately correlated with energy growth, with an r squared of 0.55.

Figure 8. Regression analysis of population growth compared to energy growth, based on annual averages, with energy growth the independent variable.

One of the issues in forecasting population using regression analysis is that in the period since 1820, we don’t have any examples of negative energy growth for long enough periods that they actually appear in the averages used in this analysis. Even if this model fit very well (which it doesn’t), it still wouldn’t necessarily be predictive during periods of energy contraction. Using the regression equation shown in Figure 8, population growth would still be positive with an annual contraction of energy of 1.72% per year, but just barely. The indicated population growth rate would slow to 0.09% per year, or total growth of 3.8% over the 40 year period, bringing world population to 7.1 billion in 2050.

Energy per Capita

While I did not use Energy per Capita in this forecast, we can look at historical growth rates in Energy per Capita, compared to growth rates in total energy consumed by society. Here, we get a surprisingly stable relationship:

Figure 9. Comparison of average growth in total world energy consumed with the average amount consumed per person, for periods since 1820.

Figure 10 shows the corresponding regression analysis, with the highest correlation we have seen, an r squared equal to .87.

Figure 10. Regression analysis comparing total average increase in world energy with average increase in energy per capita, with average increase in world energy the independent variable.

It is interesting to note that this regression line seems to indicate that with flat (0.0% growth) in total energy, energy per capita would decrease by -0.59% per year. This seems to occur because population growth more than offsets efficiency growth, as women continue to give birth to more babies than required to survive to adulthood.

Can We Really Hold On to the Industrial Age, with Virtually No Fossil Fuel Use?

This is one of the big questions. “Renewable energy” was given the name it was, partly as a marketing tool. Nearly all of it is very dependent on the fossil fuel system. For example, wind turbines and solar PV panels require fossil fuels for their manufacture, transport, and maintenance. Even nuclear energy requires fossil fuels for its maintenance, and for decommissioning old power plants, as well as for mining, transporting, and processing uranium. Electric cars require fossil fuel inputs as well.

The renewable energy that is not fossil fuel dependent (mostly wood and other biomass that can be burned), is in danger of being used at faster than a sustainable rate, if fossil fuels are not available. There are few energy possibilities that are less fossil fuel dependent, such as solar thermal (hot water bottles left in the sun to warm) and biofuels made in small quantities for local use. Better insulation is also a possibility. But it is doubtful these solutions can make up for the huge loss of fossil fuels.

We can talk about rationing fuel, but in practice, rationing is extremely difficult, once the amount of fuel becomes very low. How does one ration lubricating oil? Inputs for making medicines? To keep business processes working together, each part of every supply chain must have the fuel it needs. Even repairmen must have the fuel needed to get to work, for example. Trying to set up a rationing system that handles all of these issues would be nearly impossible.

GDP and Population History Back to 1 AD

Angus Maddison, in the same data set that I used back to 1820, also gives an estimate of population and GDP back to 1 AD. If we look at a history of average annual growth rates in world GDP (inflation adjusted) and in population growth, this is the pattern we see:

Figure 11. Average annual growth in GDP in energy and in population, for selected periods back to the year 1 AD.

Figure 11 shows that the use of fossil fuels since 1820 has allowed GDP to rise faster than population, for pretty much the first time. Prior to 1820, the vast majority of world GDP growth was absorbed by population growth.

If we compare the later time periods to the earlier ones, Figure 11 shows a pattern of increasing growth rates for both population and GDP. We know that in the 1000 to 1500 and 1500 to 1820 time periods, early energy sources (peat moss, water power, wind power, animal labor) became more widespread. These changes no doubt contributed to the rising growth rates. The biggest change, however, came with the addition of fossil fuels, in the period after 1820.

Looking back, the question seems to become: How many people can the world support, at what standard of living, with a given quantity of fuel? If our per capita energy consumption drops to the level it was in 1905, can we realistically expect to have robust international trade, and will other systems hold together? While it is easy to make estimates that make the transition sound easy, when a person looks at the historical data, making the transition to using less fuel looks quite difficult, even in a best-case scenario. One thing is clear: It is very difficult to keep up with rising world population.

Source: Our Finite World

About the Author

Actuary
GailTverberg [at] comcast [dot] net ()
randomness