Sunday, December 30, 2012

Post Peak Oil Economic and Population Scenarios: Part 1 Introduction and the Exporter Regions

As promised in my last post, I will spend some time writing about my analysis of the implications of the results from my recent ten part series, “Predicting Global and Regional Petroleum Consumption Trends” (starting here at Part 1). In that series, to remedy some of the deficiencies of my previous ELM analysis, I developed and presented my PIE analysis process, to predict future petroleum consumption trends for nine different regions, based upon the recent petroleum Production, Import and Export rate trends for each of the nine regions.

In this series of posts, I will consider some of the likely economic and population scenarios, based upon those predicted consumption and import/export trends for each of the nine regions, and, some mitigation measures that might help delay or soften the most negative implications.

Introducing some basic assumptions for the scenario

Presenting scenarios of economic and population changes, in view of peak oil, require some assumptions, and I want to articulate as clearly as possible what my assumptions are. Of course, these assumptions are somewhat speculative, because the world as a whole, has never faced an extended period of declining petroleum consumption. Optimists will simply write these scenarios off as being based on faulty assumption and continue merrily on their way to a bigger brighter future, where everyone in the world lives at the same standard of living as in the western developed countries.

Nevertheless, I believe that there are a few reasonable assumptions that can be made:

1) The relative economic decline in any one region, as quantified by a relative decline in per capita GDP, will be proportional to the relative decline in per capita petroleum consumption rate for that region.
It is inherent in my PIE analysis that the future import and export trends for each region have been accounted for, and, for the purpose of this analysis, I am going to assume that alternative fossil fuels could not or will not be ramped up in time to make much of a difference for the one area that petroleum has been critical for economic growth: transportation.

My hunch is that a decline in per capita petroleum consumption would cause a proportion decline in per capita GDP, but of course, a declining GDP could cause “demand destruction” leading to a decline in decline in per capita petroleum consumption. So, I am satisfied to just assume that petroleum consumption and GDP will change contemporaneously in the same direction.
Indeed, a study of 88 countries by Sinha found that per capita GDP growth and per capita energy consumption growth are “co-integrated,” which is to say, an increase in energy consumption will cause a increase in GDP, and, an increase in GDP will cause an increase in energy consumption. However, this does not prove the inverse, that a decrease in energy consumption will cause a decrease in GDP, or, vice versa.

Proving the inverse causal relationship might be more difficult when considering just per capita petroleum consumption and GDP. For instance, it might be hard to show that declining regional production can cause a decline in GDP for any one region because the domestic decline in production could be offset with increased import rates, or decreased export rates, to improve the region’s net import situation, or, perhaps other fossil fuels, gas and coal, could be used to partially offset declining petroleum use. There is also the possibility that there could be a significant lag time between a drop in petroleum consumption and drop in GDP.

For all of these reasons, there is not a lot of data out there to test a causal relationship between petroleum consumption decline and GDP decline. One would need a circumstance where, for a region, there was a large and enough sustained drop in production and/or imports to assess the effects on GDP. There are, however, a few examples of this happening.

For instance, there is evidence that declining oil production caused (“Granger caused,” to be more specific) a decline in GDP in the (see Reynolds and Kolodziej) in the Soviet Union/former Soviet Union in the late 1980s–early 1990s. Perhaps the relationship was clearer here because an external oil import source was not readily available to the Soviet Union during this period.

And, there is evidence of a positive relationship between the per capita petroleum consumption rate and per capita GDP for different countries, and, between downward changes in oil consumption and GDP, that occurred during the oils shocks in the USA in the 1970 (briefly reviewed and referenced in: An Export Land Model Analysis for the USA-Part 4; Relating Per Capita GDP to Petroleum Consumption and Exports for the MENA Countries).

2) Population decline in any one region would be contemporaneous with a decline in per capita petroleum consumption rate, after the per capita consumption rate drops below 1 barrel per persons per year (“bp/y”).

I have spent time analyzing and discussing this topic in previous posts, so I will simply refer you to Part 9: Estimating the critical levels of petroleum consumption necessary to sustain the modern food production system, and, to my previous series The relationship between hunger and petroleum consumption-Part 1. To summarize my thinking on this subject, it looks like the world’s modern petroleum dependent food production and distribution system needs about 1 bp/y to function at a level that avoids various proxy indicators of starvation. My assumption is that once a region’s per capita petroleum consumption drops below 1 bp/y, the food production system for that region would become limited, and further declines in petroleum consumption will mean proportional population decline. In other words, the population of region will contract in other to keep petroleum consumption at 1 bp/y.

To take an example for a hypothetical region: above a per capita petroleum consumption rate of 1 bp/y, there is adequate food production and the population will continue changing at whatever is the current trend for population change (up or down depending on demographica and other factors). However, below a per capita petroleum consumption rate of 1 bp/y, the food system is compromised, and with it, so to the population that can be sustained. Consequently, the population will decline to keep the petroleum per capita petroleum consumption rate equal to 1 bp/y. Assume, for example, that our hypothetical region has dropped to a petroleum consumption rate of 1 bp/y, has a population of 100 million people and has a population growth rate of 1 percent per year. So long as that region’s petroleum consumption is greater than 0.1 billion barrels per year (“bby”) the population will continue to grow at 1 percent year. However, if petroleum consumption were to drop to say 0.08 billion barrels per year, then I hypothesize that the population will drop to 0.08 billion or 80 million such that the per capita consumption rate adjusts back to 1 b/py.

Of course, the 1 b/py limit before population decline is only a rough approximation, and, I would not be surprised to see region-to-region variations from this assumption. For instance, other regions could send food aid to the region facing starvation. This would be the equivalent of exporting oil to the starving region for free, since the food would have been produced with the aid of the petroleum-driven food production system in the other region. Perhaps, some optimists will say that a starving region could produce the same or a growing amount of food (e.g., using sustainable food production practices) without any petroleum whatsoever. However, I am skeptical about the extent such measures, if even possible, could be implemented on a large enough scale fast enough to effectively mitigate starvation and death. Perhaps, sustainable food production practices could help somewhat. In this case, I see these measures as the equivalent of just assuming the critical per capita petroleum consumption rate number need to sustain the food production and distribution system is not 1 bp/y for the region, but some other lower number. That’s fine, pick whatever number between 1 and 0 b/py that you want to use; as you will see, for most regions, this would only shift my population decline scenarios back by a year or two.

3) The minimum that each region’s population can drop to, after reaching 1 b/py, will equal the population level that existed prior to the petroleum-driven food production system.

I will make a similar assumption to what I made in my earlier series (Part 10: Peak oil exports, peak oil and implications for population change), that the population of each of my nine regions in the year 1900 is a reasonable estimate of what populations are possible to sustain without significant petroleum inputs. If you want to go back to before the discovery of oil, in the 1850s, that’s fine, but it does not change the principle assumption, which is that, without petroleum, the present population of a region would stop growing and drop back down to a level that is much lower than the present number, or, the future number based on the present population growth trend.

Just how fast would the population drop to the pre-petroleum number? See assumption (2). The rate of change will be governed by the rate at which petroleum consumption declines, after the per capita rate drops below the 1 b/py limit.

The specific year 1900 population numbers for my regions are obtained or derived from The World at Six Billion, HYDE version 3.1, and Population Statistics, and are summarized below:

1) North America (NA): 82 million (from The World at Six Billion)
2) South America (SA): 74 million (from The World at Six Billion)
3) Europe (EU): 400 million (from The World at Six Billion)
4) The former Soviet Union (FS): 121 million (from Hyde, statistics for Commonwealth of Independent States)
5) Middle East (ME): 37 million (from Hyde)
6) Africa (AF): 141 million (from The World at Six Billion)
7) China (CH): 400 million (from Population Statistics)
8) Japan (JP): 43.8 million (from Population Statistics)
9) The remaining Asia-Pacific (rAP): 463 million (HYDE's population estimate of Asia in 1900 plus Oceania (5 million) minus the population estimates for FS, ME, CH and JP)

That tallies up to 1.66 billion, or, about 24% of the world’s present population of 7 billion.

I can imagine both optimists and pessimists objecting to this assumption. On one hand, some optimists will argue that sustainable food production practices could support much higher populations than these. On the other hand, some pessimist will argue that there is no way that the world could go back a non-petroleum food production system and still be able to support these population levels. That is, other resource limitations will limit food production, such as: scarcity of fresh water, farm land being encroached by urban sprawl, overused soils being depleted of nutrients, declining fish stock and climate change.

I don’t see how to resolve this kind of debate; it is just too speculative to say what kind of offsetting adjustments that 21st century knowledge, applied in a post-petroleum, 19th century like world, can make to mitigate these other effects of resource depletion. I do think however, that absent an energy replacement for petroleum, an energy source that can do the same things and be used in the same way as petroleum can, wewill see the economy and food production go down, and with it, population.

My main goal here is to assess which regions will most vulnerable to these changes over the next 50 years.

As I am sure you can appreciated by now, predicting the per capita consumption rate for each region is central to my analysis. I already have predictions of future petroleum consumption from my PIE analysis, done in my previous series “Predicting Global and Regional Petroleum Consumption Trends” (starting here at Part 1).

So, now I need predictions of future population trends for each of the nine regions. Fortunately, the US Census Bureau International Database provides population data from 1950, and projected population these trends forwards to 2050, for nearly every country in the world, and from theses the regional population data for each or my nine regions can be calculated. Additionally, since the census bureau only extrapolated population trends out to 2050, but, I want to extend my prediction scenario out to 2065, I need to extend the census bureau’s population trend prediction out another 15 years. To extend this, I did a linear regression analysis on the last ten years of the census bureau’s predicted growth rate trend, (i.e., the yearly change in population from 2040 to 2050), and, used the best fit slope and intercept to interpolate the growth rate trend out to 2065.

I think that covers the assumptions that underlie my analysis, so let’s start with the net petroleum exporters.

Economic and Population Trends for the Net Petroleum Exporting Regions

Middle East (ME)

As this is the first region discussed, I will spend a little bit more time explaining the data presentation, whichwill be the same for the subsequent regions.

For each of the nine regions, I show the population growth trend, that is, the yearly percent change in population. Actually this is better states as being the population change trend, as there are some regions that presently, or soon will, have negative population growth. In each case, the open circles are the data reported by the US census bureau, and the solid line is the bureau’s predicted trend, plus my interpolation of that trend out to 2065.

Figure 1 shows the population change trend for ME.

As you can see, the present growth rate is about +1.7 %/y, which is down from a ridiculously high growth rate of +4.1 %/y in the early 1980s. As you can also see, although the population growth rate is predicted to continue declining, it will still be positive in 2065. That means that the ME’s population is predicted by the census bureau to continue to increase beyond 2065, probably hitting a zero growth rate in the 2070s.

The right hand side axis in Figure 2 shows the actual population (blue circles) and predicted population (dark blue line), based on the population trend shown in Figure 1.

The solid blue line indicates that, if the present population growth rate trend continues, then ME’s population will expand from 212 million (0.212 billion) in 2011 to 341 million (0.341 billion) by 2065.

The left hand side axis in Figure 2 shows the calculated per capita petroleum consumption rate (red circles), and, my predicted per capita consumption rate, which is based on the petroleum consumption rate predicted for this region from my PIE analysis, divided by the predicted population.

The ME’s per cap petroleum consumption rate was 13.7 b/py in 2011, and, the per cap rate is predicted to peak at almost 14 b/py in 2014. Thereafter, per cap petroleum consumption stops growing and goes into decline. It is worth noting that this predicted peak is 6 years earlier than the ME’s predicted absolute peak in petroleum consumption of 2020 (see e.g., Figure 4 in Part 2 of my earlier series). This is because of the on-going population growth rate in the ME of +1.7 to 1.5 %/y from present to 2020.
Overall then, from 2011 to 2020, the predicted drop is about -2.8% or about -0.3 %/y. 
The per cap petroleum consumption rate drops even faster from 2020 to 2040, from 13.3 b/py to 7.2 b/py, a 46% decrease, or, -2.3 %/y for 20 years. This steep and steady decline reflects the combination of predicted declining petroleum consumption and increasing population. From 2040 to 2060, the per cap petroleum consumption rate decline is still steeply declining on an annual percentage basis (-3.1 %/y), although the absolute change (7.2 to 2.7 b/py) is not quite as large. Towards the end of the study period, the decline is less steep, as the population grow rate approaches zero.

Based on these predictions, it is hard to imagine how the ME region as a whole, will have economic growth past 2014, after the per capita petroleum consumption rate stops growing. Rather, after a brief plateau, from 2020 and on, we are likely looking at economic decline, perhaps GDP declining on the order of -2.3 %/y for the next 20 years, assuming direct proportionality between GDP and petroleum consumption.

Going from a per capita petroleum consumption rate of 14 to 1.9 b/py corresponds to an 86% decline. Assuming directly proportional behavior, this would mean a huge relative drop in GDP per capita.

Still, even by the end of my study range, in 2065, the cap petroleum consumption rate is equal to 1.9 b/py, which is well above my starvation threshold of 1 b/py.

This probably means a very crowded ME with a still growing population of 340 million, about 60% larger than today, but with a much lower standard of living.

Former Soviet Union (FS)

Figure 3 shows the population change trend for FS.

As you can see the FS’s population growth stopped and then plummeted in the late 1980s to early 90s, going from positive to negative in 1994, and, has never recovered into positive territory since then. According to the US census bureau, the present population “growth” rate is about -0.12 %/y, i.e., a population de-growth rate. The census bureau predicts that this negative population change trend will continue, reaching about -0.43 %/y by 2050. If that trend continues, the population change will be -0.58 %/y by the end of my study period in 2065.

The right hand side axis in Figure 4 shows the actual population (blue circles) and predicted population (dark blue line) based on the population trend shown in Figure 3.

As you can see, FS's population has been gradually declining since 1994, and if the trend continues, the population decline will continue on a somewhat similar pace. The solid blue line indicates that, if the present population change rate trend continues, as predicted by the US census bureau, then FS’s population will decline from its present 0.283 billion in 2011 to 0.235 billion by 2065.

The left hand side axis in Figure 4 shows the calculated per capita petroleum consumption rate (red circles), and, my predicted per capita consumption rate (solid red line).

The FS’s per cap petroleum consumption rate peaked at 11.4 b/py way back in the 1980s, then plummeted down to 4.3 b/py in 2001 and then has recovered to 5.0 b/py in 2011. Notice that although the population growth rate went negative by 1994, there was no dramatic drop in population from 1980 to 2001 as the per capita consumption rate was plummeting. This is consistent with my hypothesis that population decline due to decline petroleum consumption would not occur until a region drops below the 1 b/py threshold.

In my earlier series, predicting petroleum consumption rates for FS (Figure 11, Part 3), I showed how FS’s predicted petroleum production and export rate trends would result in FS’s domestic consumption rate steeply declining and hitting zero by 2027.

The consequence of this prediction is reflected in Figure 4: a steeply declining per capita petroleum consumption rate that hits zero by 2027. The predicted decline in petroleum consumption is so steep that the predicted more mildly declining population trend doesn’t do much to mitigate the per capita petroleum consumption decline. For instance, by 2015 the per capita petroleum consumption decline rate is about -9%/y, and by 2020 the decline rate is -17 %/yr.

But, according to this scenario, FS would have bigger problems than the economic declines associated with declining per capita petroleum consumption: by 2022 the per capita petroleum consumption rate drops below my threshold of 1 b/py. It only takes another 5 years before hitting 0 b/py, implying a sharp decline in population from its predicted level of 278 million in 2022 to the year 1900 pre-petroleum level of 121 million (light blue line in Figure 4 labeled “population predicted w peak oil”). That is a 56% population decline in 5 years—a disaster in other words.

In summary then, if the predictions of FS’s domestic petroleum production (peaking now and about to decline as sharply as it went up), and, increasing export rate trend are correct, and continue on their present path, then this region would suffer a very steep economic contraction and population collapse within the next 10-15 years.

The implication of this scenario, to me, is that it would be rational for the countries in the FS region to intervene in the present trends at some point, at least to mitigate economic and population collapse. It would most rational for these countries to start reducing the export rate right now if they appreciated how rapidly their remaining oil will deplete at the present increasing consumption rate and export rate trends.

However, there is still the hope of new sources of petroleum (e.g., Arctic oil or fracking), although, I don’t see evidence in the production data reported in the BP review to support this. And, of course, the present petroleum exports represent a significant source of income that helps support the economy. Therefore, my hunch is that it is most likely that intervention to reduce exports will occur later than sooner, and therefore, there would be a sharp decline in exports, as domestic per capita consumption reaches 1 b/py, and the threat of starvation loomed nearer.

How long could the FS keep itself at 1 b/py in order to stave off a sharp population decline, while still exporting some oil, and, what implications would this have on its major importers (e.g., Europe and Asia) of FS’s petroleum?

For now, I am going to set aside the answers to these questions, until I finish my survey of all nine regions.

Africa (AF)

Figure 5 shows the population change trend for AF.

Similar to the ME, AF population growth rate peak in 1981 at just over +3 %/yr and has been declining since then. But the rate of decline is quite slow, and so in 2011, the population growth rate was still +2.25 %/yr. The US census bureau projection and my extrapolation puts AF’s population growth rate still at +1.2 %/yr in 2065. At this rate of decline in the growth rate, zero population growth would not be reached until well into the next century.

The right hand side axis in Figure 6 shows the actual population (blue circles) and predicted population (dark blue line) based on the population growth trend shown in Figure 5.

A growth rate of 2.2 to 1.2 percent per year may not seem like much, but as illustrated by the dark blue line, it adds up to AF’s present population doubling from about 1 billion to 2 billion by 2047, and reaching 2.6 billion by the end of my study range in 2065.

The left hand side axis in Figure 6 shows the calculated per capita petroleum consumption rate (red circles), and, my predicted per capita consumption rate for AF.

As you can see, AF’s per capita consumption rate increased above 1 b/py in 1980, and has been ever-slightly increasing for the last 30 years, to about 1.2 in 2011.

Unfortunately, AF’s petroleum production rate appears to be peaking now, and, I predicted it to decline quite steeply in the immediate years to come (see e.g., Figure 16 in Part 4 of my earlier series). And assuming that the present export rate trend were to continue, AF’s domestic consumption rate would also decline quite steeply in the immediate years to come. This means that AF’s per capita petroleum consumption rate drops even more steeply—due to the combination of petroleum consumption declining and population increasing. Since AF is just barely presently over my 1 b/py threshold for starvation and population decline, this scenario predicts that a population decline starting right now in order to keep the per capita rate at 1 b/py (light blue line in Figure 6).

It looks like my predicted declines in petroleum consumption rate, per capita consumption rate and population are all earlier than what the most recent data are suggesting, however. I think this could be due to a couple of factors that my PIE analysis can not easily account for. For instancem from Figure 16 in Part 4 the most recent net export trend is down to a greater degree than predicted from the previous 10-trend used in the PIE analysis. Also, I suspect that there is substantial food aid going to some countries in AF is the equivalent to oil imports (since the food is produced by a petroleum-driven food production system), that are not accounted in the BP statistical review data, and which help sustain population growth. Additionally there is growing pipeline oil thefts and refining and black market selling for domestic and foreign use (see e.g., Nigeria’s booming illegal oil refineries; Africa's oil theft crisis), which are also not accounted for by BP’s statistics, and therefore are not part of my PIE analysis.

If the predicted population decline scenario for FS looks bad, then the population decline scenario for AF is an unmitigated disaster—a population drop from 1 billion to the 1900’s level of 141 million by 2026—that is, an 86 % drop.

This, of course, raises the question, similarly to what I raised for FS, to what extent could this devastating decline in population be mitigated or delayed by the oil producing countries in AF cutting exports to other regions? And what effect does this have on the major importer’s of AF’s oil (e.g., North America, Europe, Asia)?

Again, I am setting this aside until I finish my nine-region survey. However, I will say this—I do not see anyway that AF’s present and projected population growth rate can continue on as shown in Figures 5 and 6, and, I am very skeptical that even the present population of 1 billion could be sustained in a future where domestic petroleum production is in decline and the western developed nations are in economic decline resulting less domestic food production, less food imports and less food aid to AF in the near future. To the contrary, countries in the ME and Asia having been buying up farm land in AF with the intent of producing food for export to these regions (see e.g. articles cited in, Export land model analysis of food production and consumption—Background).

This is all too depressing, let’s move on to South America.

South America (SA)

If you have managed to get this far, then you should be used to the pattern: Figure 7 shows the population change trend for SA.

SA’s population growth trend has been declining since 1965 when it was at about +2.6 %/y. The rate in 2011 was +1.15 %/y and, if I extrapolate the US census bureau’s predicted trend, zero population growth is reached in about 2060, and the population declines thereafter.

The right hand side axis in Figure 8 shows the actual population (blue circles) and predicted population (dark blue line) based on the population growth trend shown in Figure 7.

The solid line blue line illustrates the continued population increase from present levels of 483 million to the predicted peak in 2060 of 632 million.

The left hand side axis in Figure 8 shows the calculated per capita petroleum consumption rate (red circles), and, my predicted per capita consumption rate for SA.

SA’s per capita consumption has been on a bumpy upwards trend since its low level of 2.9 b/py in 1965, peaking at 4.1 b/py in 1980, dropping back down to 3.6 b/py in 1985, finding a new peak of 4.4 b/py in 1999, dropping back down to 4 b/py in 2003 and then finding yet another high of 4.7 b/py in 2011. My PIE analysis, however, suggested that production is about to peak, but the net export rate trend was flat to slowly declining to zero by 2037 (see Figure 21 from Part 5 of my previous series). Consequently, the domestic consumption rate for SA was predicted to gradually decline going forwards.

Because the consumption rate is predicted to decline and the population is predicted to continue increasing until 2060, SA’s per capita petroleum consumption rate is predicted to be on a downwards trend dropping from 2 b/py in 2011 to 1.6 b/py in 2065.

In fact, like AF, over the last few years, SA’s per capita consumption rate has exceeded the predicted rate—mainly this is due to net exports being less than that predicted by the 10-year trend (again, see Figure 21 from Part 5).

In summary, SA is predicted to have a n economic decline because per capita consumption rate is predicted to decline. Still, a decline from 2 b/py in 2011 to 1.6 b/py, is only 20 % over a period of 54 year, or -0.37 %/yr, which is a much milder decline compared to the other three regions examined in this post. Plus the per capita consumption rate never drop to my threshold rate, of 1 b/py, for starvation and population decline, at least within my study range out to 2065.

Overall then, of the four present net petroleum exporters, SA looks best off for the next 50+ years with relatively mild economic decline. ME is next with relatively larger economic declines, but, at least the per capita petroleum consumption rate doesn’t decline to the point where there is starvation and population decline. FS, on the other hand, would experience a significant economic and population decline starting around the early-2020s. The situation for AF looks worst off of four regions, with a predicted steep economic and population decline starting now and continuing through the mid-2020s. This makes the FS and AF regions primary candidates for an intervention in their existing increasing export rate trends—a topic that I will explore later on in this series.

Although I try to keep this blog on a fairly neutral non-emotional level, it is hard to predict starvation and population declines of 56 and 86 percent for some regions of the world, and not be depressed by the shear magnitude of human suffering implied by these predictions.

My irrational hope for the new year is that my predictions are completely wrong, and, that humanity will find a way out of the predicament that it appears to be stumbling towards.
Next time, I will present my economic and population scenarios for the remaining five regions—the net importers.


  1. Hi CW

    Not sure if my comment went through. Anyway keep up the good work. Love to see you circle back and take a critcal look at your assumptions, and see if refinments are in oreder


  2. I think your article is missing the coal liquefaction as solution. There is maximum long term price for oil. After that, coal liquefaction becomes profitable and investments are made. China has already one successful demonstration plant using MTG (Methanol-to-gasoline) process.
    It's run by Shanxi Jincheng Anthracite Mining Group.

    This will of course increase CO2 emissions.

  3. Walter thanks for showing interest in this blog and this series. Yes, although I, like everyone, has some bias as to what kinds of scenarios may unfold in the future, it is rare to find studies where the assumptions are even articulated much the less tested by varying the assumptions. That was one of my motivations for starting this blog in the first place, and putting in the hard work that goes into each article.

    Hello Nick. It is not so much that I am ignoring coal liquefaction as a solution, no more than I am ignoring tar sands oil, shale oil, off-shore oil, arctic ocean oil, algae oil etc..... as “solutions.”

    I am not trying to guess what these or other specific potential sources of oil may or may not generate in the future.

    Rather, this is a trend analysis.

    I am extrapolating the present production and export/import trends into the future, and based on that extrapolation, I trying predict what will happen in the future, if those trends continue.

    I think that no matter where the oil comes from, burning it will increase CO2 emissions.


Your comments, questions and suggestions are welcome! However, comments with cursing or ad hominem attacks will be removed.