An important new study concludes that reducing population growth will reduce greenhouse gas emissions. There is less here than meets the eye
by Ian Angus
Can a computer model prove that population growth is an important cause of climate change? An October 11 press release from the International Institute for Applied Systems Analysis (IIASA) says yes.
Headed Population change: another influence on climate change, the release says that IIASA researchers have found that “a slowing of … population growth could contribute to significantly reducing greenhouse gas emissions.” Following the UN’s lowest plausible population growth path could, all by itself, “provide 16 to 29 percent of the emission reductions thought necessary to keep global temperatures from causing serious impacts.”
If true, this would be an important breakthrough, and it would give immense credibility to those in the climate crisis movement who argue that one of our main goals should be improved birth control services for Third World women. Two prominent populationist groups in the U.S. have already published a policy brief citing this study in support of their views.
But there is less here than meets the eye.
Further down the page, the release says:
Scientists have long known that changes in population will have some effect on greenhouse gas emissions, but there has been debate on how large that effect might be.
The researchers sought to quantify how demographic changes influence emissions over time, and in which regions of the world. They also went beyond changes in population size to examine the links between aging, urbanization, and emissions.
In short, the researchers didn’t prove that population growth causes greenhouse gas emissions to grow. They assumed that it does, and then tried to determine how various demographic factors might affect the process.
That’s an important distinction. No computer model can prove facts about the real world. It can only assume facts to be valid and test their implications over time, under a given set of assumptions.
For example, the IPCC’s computer models don’t prove that greenhouse gas emissions cause atmospheric temperatures to rise. That fact has been proven by decades of scientific research, and confirmed by theoretical studies that clearly show exactly how the warming process works. The models then show the implications of that information under various assumptions about economic growth, technology development, and so on.
No model is any better than the data and assumptions that go into it.
Much to admire
There is much to admire about this latest study, which was conducted by a team led by Brian C. O’Neill, one of the most respected scientists working on demography and climate change, and reported in the Proceedings of the National Academy of Science (PNAS).
In particular, unlike previous studies, this one recognized that using aggregated global statistics can be misleading: instead, they divided the world into nine economic regions, and they used economic and demographic statistics from 34 countries containing 61% of the world’s population. They ran multiple simulations, using the high, medium and low population projections made by the UN and two of the scenarios used in IPCC climate studies. They considered not just the number of people in each region, but also the urban/rural distribution, household size, and age structure.
In the end, they concluded that emissions rise with population, that population structure is important but not decisive, and that “emissions would decrease significantly” if population growth slows. The actual effect of slowing population growth varies substantially according to various scenarios, but they estimate that by 2050 population-related emissions reductions could produce between 16% and 29% of the reductions that scientists believe are needed “to limit future climate change to safe levels.”
This study is a vast improvement on the simplistic “big is bad, bigger is worse” arguments we hear from so many populationists. It confirms again that O’Neill and his team are doing serious work that deserves careful attention from everyone who is concerned about climate change – even if we question their assumptions and conclusions.
Evaluating a study of this type poses almost insuperable difficulties for those of us who are not experts in computer modeling. We don’t have access to the model itself, and even if we did, we don’t have the expertise or the resources to determine whether it actually does what the authors claim.
Nevertheless, what the authors tell us in their paper raises significant questions about the reliability of the results.
For example, studying the world as nine regions rather than one big unit is an improvement, but the real world has several hundred countries, each composed of multiple economic and climatic zones. Even if we assume that the regions adequately represent such differences, it seems that quantity and quality of data varied substantially from region to region. For example, the researchers had “insufficient data” about households in sub-Saharan Africa – an important issue, because that region has the world’s highest birth rates.
The economic model is also oversimplified. It assumes that each region has just four industries producing energy (oil & natural gas, coal, electricity, and refined fuels), one industry producing materials for other industries, and four industries producing consumer goods (energy, food, transport, and other). A real capitalist economy with thousands of firms competing for sales and profits is likely to evolve very differently than one controlled by nine hypothetical monopolies.
To their credit, the authors highlight a particularly important problem – the fact that their model doesn’t include any explanation of why the population growth rate might increase or decrease. That’s significant because many demographers believe that economic growth is the most important cause of falling birth rates.
If it were assumed that increases in economic growth rates were driving fertility decline, our results would differ: faster economic growth would have an upward effect on emissions, offsetting the emissions reductions caused by slower population growth to some degree.
The only alternative they suggest is that population might decline as a result of improved family planning programs. That is at best a weak response to a problem that seriously undermines the model’s credibility.
Serious as those concerns are, they are secondary. The problem is not whether the model calculates properly or has sufficient data, but whether it makes appropriate assumptions about what data needs to be included and what calculations are appropriate.
There was an important discussion of just these issues following publication of The Limits to Growth in 1972. The authors of that landmark study also claimed to have developed a computer model of the global economy; they said it predicted that if then-current trends continued, “the limits to growth on this planet will be reached sometime within the next one hundred years,” and that the most likely result would be “a rather sudden and uncontrollable decline in both population and industrial capacity.”
The Limits to Growth, a popularly-written book by scientists at MIT, was a monster bestseller. Millions of people read it, and its conclusions became part of the accepted wisdom of many environmentalists.
Far less attention was paid to Thinking About the Future, a much drier study published 10 months later, in which thirteen specialists in different disciplines from the University of Sussex carefully dissected The Limits to Growth and found it wanting, to say the least. They showed in detail that the MIT computer model was seriously flawed and that the data it used to make predictions was inadequate.
Most importantly, they argued that using a computer model to predict what were essentially social trends gave the study a spurious appearance of objectivity, while concealing political, economic and social biases of which even the scientists concerned might not have been aware.
In the opening essay, “Malthus With A Computer,” economist Christopher Freeman wrote:
The nature of their assumptions is not a purely technical problem. It is essential to look at the political bias and the values implicitly or explicitly present in any study of social systems. The apparent detached neutrality of a computer model is as illusory as it is persuasive. Any model of any social system necessarily involves assumptions about the workings of that system, and these assumptions are necessarily coloured by the attitudes and values of the individual or groups concerned….
It cannot be repeated too often that the validity of any computer calculation depend entirely on the quality of the data and the assumptions (mental models) which are fed into it. Computer models cannot replace theory. 
Freeman exempted The Limits to Growth from the common accusation of, “garbage in, garbage out,” because the authors had obviously gone to a great deal of effort to get data, adopt reasonable assumptions and test the model.
Although it would be quite wrong to talk of “garbage” in the MIT model, there is a real point in the description: “Malthus in, Malthus out.” … what is on the computer print-out depends on the assumptions which are made about real-world relationships, and these assumptions in turn are heavily influenced by those contemporary social theories and values to which the computer modelers are exposed. 
Forty years later, computers are much more powerful and computer models can be much more complex than anything imagined by the authors of The Limits to Growth – but Freeman’s arguments retain their full force. Indeed, given the increased complexity of the models – and thus the increased possibility of error – it is arguably much more important today that the modelers be fully aware of their assumptions and make those assumptions as explicit as possible.
Like the authors of The Limits to Growth, O’Neill and his team are “heavily influenced by … contemporary social theories and values.”
One such theory was expressed clearly by Brian O’Neill in an interview with the Los Angeles Times on October 10: “As the economy grows faster, it raises the income for everybody, and people are spending more money and consuming more and emitting more.”
Or, more formally in the PNAS paper: “In the PET model, households can affect emissions either directly through their consumption patterns or indirectly through their effects on economic growth …”
The assumption that economic expansion is driven by consumer demand, so more consumers equals more growth is a fundamental component of the economic theories that underlie O’Neill’s PET model.
In other words, his conclusions are built into his assumptions.
What the model actually tries to do is to use neoclassical economic theory to predict how much economic growth will result from various levels of population growth, and then to estimate the emissions growth that would result.
Surely the experience of recent years should cause us to question any use of neoclassical economics to predict future economic activity! As Steve Keen, a longtime critic of mainstream economic theory wrote recently, “The most important thing that global financial crisis has done for economic theory is to show that neoclassical economics is not merely wrong, but dangerous.”
Or, as Yves Smith says about financial economics, a computer model based on mainstream economic theory “rests on a seemingly rigorous foundation and elaborate math, much like astrology.”
Malthus In, Malthus Out …
The PNAS article does not include a list of the researchers’ basic assumptions, and it’s possible that they haven’t explicitly considered the issue, but on the basis of what they do tell us, we can reasonably conclude that they include the following.
- That population growth causes economic growth and that economic growth causes emissions growth.
- That population reduction won’t cause (or be caused by) other changes that could increase emissions.
- That neo-classical economic theory – including its basic assumption that economic growth is driven by consumer demand – provides a valid methodology for predicting economic activity decades into the future.
- That the United Nations’ predictions of population growth are reasonably accurate and won’t have to be adjusted every few years as they have in the past.
- That the only relevant social and economic changes are those related to population structure. Capitalism is here to stay.
- That there will be no significant changes in the distribution of wealth between rich and poor, within or between countries, other than those resulting from urbanization.
- That the data collected from 34 countries is both accurate and representative of the nine regions, and that aggregating countries into nine regions isn’t oversimplifying.
- That the economy consists of a very small group of industries that only grow in response to population growth.
- That steady economic growth will continue until at least 2050, uninterrupted by bubbles, crashes, depressions and wars.
- That the world won’t experience catastrophic “tipping point” climate changes long before any population program has any impact at all.
And undoubtedly many more assumptions are explicitly or implicitly embedded in the model.
What can we truly conclude from this study?
Obviously, that if your computer model assumes that population growth causes emissions growth, then it will tell you that fewer people will produce fewer emissions. Malthus In, Malthus Out.
More generally, if a computer model is based on questionable assumptions, then its forecasts should be viewed as equally questionable.
As a long-ago reviewer wrote about The Limits to Growth, what you will have is “essentially an analysis of the quirky behavior of a complicated set of equations … [with] no assurance at all that the model says anything about the world.”
 Population and the environment: Where we’re headed and what we can do (2010) Population Action International and Population Justice Project.
 O’Neill, Brian C. et al. (2010, October 12) Global demographic trends and future carbon emissions. Proceedings of the National Academy of Sciences. 107 (41) 17521-17526. See also Supporting information.
 Meadows, Donella, Jorgen Randers, Dennis Meadows & William W. Behrens. (1972) The Limits to Growth. New York: Universe Books p. 23
 Cole, H.S.D, Christopher Freeman, Marie Jahoda, K.L.R. Pavitt. (Eds.) (1973) Thinking about the future: A critique of The limits to growth. London: Sussex University Press pp. 7-8
 Ibid. pp. 8-9
 In another article, O’Neil says that PET “employs a neoclassical economic growth model.” See O’Neill, Brian C. (2010) Climate change and population growth. In Mazur (Ed.) A pivotal moment: Population, justice and the environmental challenge. (pp. 81-94) Washington: Island Press p. 87
 Keen, Steve (2009, March 25) Neoclassical Economics: mad, bad, and dangerous to know. Steve Keen’s Debtwatch. http://www.debtdeflation.com
 Smith, Yves. (2010) Econned: How unenlightened self interest undermined democracy and corrupted capitalism. New York: Palgrave MacMillan p. 20
 Koehler, John E. (1973) [Review of the book The limits to growth by D.H. Meadows, D.L. Meadows, J. Randers, W.W. Behrens] The Journal of Politics 35(2) 513-514 p. 514