Growth, Climate Change and the War of Numbers
- Author: Chris Smaje
- Date: 08 Dec 2015
- Copyright: Image appears courtesy of Getty Images
If politics is ‘war by other means’ then perhaps social science could be described as ‘war by means of data analysis’. Contending interpretations of the world and how to resolve its problems fill the pages of academic journals, usually invoking social data of one kind or another to support their case but seldom so compellingly as to make that case unarguable. The worlds of politics, policy and journalism do the same, generally with more tendentious use of data and less attention to the methodological niceties behind it. And all of this is compounded by the fact that these contending interpretations often involve value judgments and presentiments about the future where data analysis cannot go.
Take ecomodernism. The latest incarnation in the long-running contest between moderate, reformist environmentalism and more radical voices within the movement, ecomodernism is associated with the following sorts of claims, among others:
• economic growth, growth in energy availability and ‘modernization’ are positive forces for the improvement of human wellbeing
• the debilities of poverty arise from the lack of growth and modernization, afflicting people who have been ‘left behind’ by modernity (1)
• economic development can contribute to environmental problems such as climate change, but ultimately it fosters technical innovation that allows these problems to be solved
• A key example of this is nuclear power, which should in time provide limitless clean energy
Ecomodernism represents a broad constellation of environmentalist thinking rather than an organised school of thought – an influential statement of its themes came in Stewart Brand’s 2009 book Whole Earth Discipline (2). Earlier this year, Brand and others – including analysts from the Breakthrough Institute, which is strongly identified with ecomodernism – produced a more programmatic statement in the form of the Ecomodernist Manifesto (3), while other writers have argued along similar lines from a variety of perspectives, such as Viscount Ridley’s right-wing take in his book The Rational Optimist (4) and Leigh Phillips’s highly charged Marxist rendering in Austerity Ecology & the Collapse Porn-Addicts (5). Phillips focuses more strongly on economic growth and class inequality, which I examine below, than other ecomodernist writings which more typically emphasise energy growth.
I’ve written various critiques of ecomodernism (6), so I make no claims to neutrality in assessing it. Very briefly, my counter-argument to the points above is that there is always an uneven geography of economic growth – a matter of cores and peripheries, in which economic development in one place usually creates economic under-development in others. So poverty is not a failure of economic development in any simple sense, but in some measure a consequence of it. I argue too that the ‘problems of growth → solutions to growth’ trajectory isn’t foreordained, and the role of nuclear power for a clean energy future remains moot.
But the main point of this article isn’t for me to wage war with ecomodernism. It’s more about data and methods in clarifying one’s position around such issues, and more from a lay perspective rather than that of the academic or the professional statistician. Ecomodernism is just one of many contentions about how best to shape the world with which we’re confronted daily in the avalanche of fact and opinion served up by print and digital media – contentions that are hard for the layperson or even the professional to judge empirically because they are typically quite general, involve complex causalities, and often involve untestable claims about the future. The temptation is to give up and leave it to the experts to argue over. But there are data sources freely available to everyone that we can use to produce simple summary univariate and bivariate statistics. Doing so isn’t going to resolve the ecomodernism or any other such debate definitively, but it can help us to inform our own lay hypotheses about the shape of the world we live in and empower us to be data analysts for ourselves rather than just helpless consumers of other people’s statistics.
The two main data sources I use here are the World Bank’s world development indicators (7) and the USA’s Energy Information Administration’s international energy statistics (8). Both are free, downloadable and searchable data sets containing a plethora of information at country level for every country in the world over the last 30-50 years. Using a data management and statistics package such as SPSS (preferable), or even just a simple spreadsheet package (as I’ve done here), it’s possible to aggregate and manipulate these data in order to probe at some of the issues involved in the ecomodernism debate. In the analysis below I barely scratch the surface of these datasets, but hopefully at least illustrate a few possibilities.
Figure 1 presents per capita Gross Domestic Product (at constant 2005 US$) plotted year on year from 1990-2013 for ten country aggregations. An individual plot for each of the world’s 200+ countries would provide a bewildering information overload – by aggregating them into bigger regional groupings it becomes possible to see a broader patterning, albeit at the cost of obscuring variation within each group. The groupings are mostly contiguous geographic ones which hopefully make intuitive geopolitical sense. The two exceptions are:
• the ‘wealthy country’ group, comprising the G7 countries (Canada, France, Germany, Italy, Japan, the UK and the USA) plus Australia, New Zealand and most of the countries of western Europe
• the ‘BRICS’ countries (Brazil, Russia, India, China and South Africa), identified as strong emerging economies which are becoming regional or global economic core areas
The adequacy of GDP has been trenchantly criticised as a measure of economic activity, but it’s still the most widely used measure of economic output and the figure that most people have in mind when they talk about ‘economic growth’.
The most striking feature of Figure 1 is the massive and persistent gulf between the GDP per capita of the wealthy country group and almost all the others. Despite the existence of steady economic growth over time among all the groupings, the relative gulf between the developed economies and the others barely changed. This would seem to be consistent with the ‘core-periphery’ thesis – national/regional differences in economic strength are persistent and structural. Of course, this doesn’t imply that economic growth is unnecessary for the smaller economies, but it may suggest that attempting to grow their economies within the existing global economic order is unlikely to enable them to achieve parity with the larger ones and grow their way into relative prosperity.
The only grouping that approaches the developed country one in size is Southern Europe – perhaps a ‘periphery within the core’? Even Southern Europe’s per capita GDP is only a little more than half the size of the wealthy countries’, and in turn is more than three times larger than the next one. At 184%, the BRICS economies experienced the greatest economic growth of any group, but there’s still more than a tenfold difference between them and the wealthy countries. There’s diversity within the BRICS group, the laggard being India which has a GDP/capita of just $1,235 despite the prodigious recent economic developments in the country, perhaps suggesting there can be ‘cores’ and ‘peripheries’ within countries as well as between them, a point indeed emphasised by economic historians (9). In general, the data presented here don’t seem supportive of the notion that an emphasis on economic growth within countries while neglecting structural economic inequality between countries is likely to create a more uniformly prosperous world.
Figure 1 considered the wealth of nations, as measured by GDP, but the wealth or poverty of the individuals within nations is equally germane to the idea of ‘development’. Poverty can be defined in two ways. Relative poverty refers to the proportions of total wealth commanded by distinct subgroups of the population, whereas absolute poverty refers to a population that falls below some specifically defined level of resource access (which to some degree will always be arbitrary). Both absolute and relative poverty can affect life chances, and while it’s easy to think of absolute poverty as a ‘harder’ measure, influential research suggests that relative poverty can be equally or more damaging (10). In situations where there is both economic growth and population growth, it’s possible for relative and absolute poverty measures to move in opposite directions. Relative poverty may decline while absolute poverty increases, or vice versa.
The World Bank indicators contain various measures of both relative and absolute poverty. The relative poverty indicator examined here is income share held by the poorest 10% of the population. In a situation of relative equality each population decile ordered by income would hold around a 10% income share, whereas with increasing inequality the lower deciles will hold relatively less. The absolute poverty measure is the number of people earning less than US$3.10 per day, adjusted for inflation.
It would be interesting to plot changes in poverty levels with changes in GDP. Unfortunately, for most countries there are many missing values year by year in the dataset – some countries report no poverty data at all, whereas others report data only for a handful of years. How to deal with missing values is always a tricky data analysis question. An imperfect solution in this instance might be to split the time series into two parts (1990-2002 and 2003-2015, for example) and compute average poverty levels in the two time periods for countries where there is at least one datum in each period.
I did such an analysis for two of the aggregate groups – sub-Saharan Africa, and the BRICS countries. In sub-Saharan Africa, the income share of the bottom 10% increased slightly from 2% in 1990-2002 to 2.2% in 2003-2015. But the number of people in absolute poverty also increased by about 90 million. Conversely, in the BRICS countries the income share of the bottom 10% declined slightly from 2.2% to 1.9%, but the number of people in absolute poverty declined by more than 700 million.
If poverty reduction is a policy aim then direct attention to the interpersonal distribution of wealth and income – in other words, to social inequality – is likely to be more effective. But this is an issue on which ecomodernism is mostly silent.
The BRICS figures are mostly driven by China, and aren’t unexpected on the basis of what we know about the country. From a situation in the 1970s where almost everyone was extremely poor, the liberalisation of the economy has created wider disparities between rich and poor, so relative poverty has increased. But the larger national wealth, even though shared less equally, has pulled most people out of absolute poverty, despite a considerable rise in total population. Similar, but much weaker, relationships have occurred in India. There, 80% of the population were in absolute poverty in 1993, whereas this figure had dropped to 58% by 2011 (the corresponding figures for China are 82 and 27%). But India’s population increased by 35% (323 million) over that period, resulting in a reduction of only about 13 million people out of absolute poverty.
At the wealthy end of the spectrum, almost nobody in the developed countries falls into absolute poverty as defined by the global $3.10/day figure. In the Western European countries for which data are available, relative poverty as measured by the income share of the poorest 10% decreased slightly from a 2.6% share in 2004 to a 2.7% share in 2012 – a period which saw GDP growth of 6% across the whole period in those countries.
Perhaps we can conclude that the rapid economic growth of the BRICS countries is associated with a decline in absolute, though not relative, poverty and in this sense find evidence to concur with certain strands of ecomodernist thinking: economic growth can get people out of absolute poverty. But at the same time, we might raise a number of caveats. The BRICS countries have experienced exceptional economic growth, whereas poverty (absolute and relative) is fairly static or increasing in regions, rich or poor, with unexceptional growth such as sub-Saharan Africa or Western Europe. GDP growth in the BRICS countries has been prodigious – over 600% in China and 250% in India over the last 20 years or so. Yet these countries have varied considerably in their abilities to get people out of absolute poverty – which, at $3.10 per day, sets the bar fairly low as a standard of living – with the ratio of poverty reduction lagging far behind the ratio of economic expansion in both of these key countries.
So I’m tempted to say that, yes, simply growing the economy can get people out of absolute poverty, but it’s probably only a truly effective poverty reduction strategy in that handful of countries which are experiencing exceptionally fast economic growth, and even then the results aren’t terribly impressive for some countries like India. But generally the poor don’t seem to gain a proportionate benefit from GDP growth. If poverty reduction is a policy aim then direct attention to the interpersonal distribution of wealth and income – in other words, to social inequality – is likely to be more effective. But this is an issue on which ecomodernism is mostly silent.
Energy and Climate Change
Perhaps the main environmental issue facing the world today is climate change caused by greenhouse gas (GHG) emissions from human activity, principally the combustion of fossil fuels. The ecomodernists argue that economic growth can be decoupled from GHG emissions by substituting clean for polluting energy sources and by increasing the economic output achieved by each unit of energy input. Without pursuing a full discussion of these issues, here I simply broach a few questions about the relationship between GHG emissions and economic growth, and the kind of actions that will be required of humanity in the future.
Figure 2 plots percentage change in global GDP against a 1990 base year and percentage change in global emissions of carbon dioxide, the most important GHG. Though the two lines track each other quite closely, it’s a statistical cliché that ‘association does not prove causality’. A plot of any two unrelated variables which coincidentally increase over time would produce a similar graph. But in the real world of data analysis one usually has a prior theory that links two variables, in this case that increased economic activity will broadly be associated with increased fuel consumption and GHG emissions – plotting them together is an initial exploratory test of the theory.
The kink in the GDP plot around 2008-9 corresponds with the global downturn associated with the fiscal crisis of that period. The fact that the emissions plot shows an identical downturn is a piece of evidence that the two variables indeed are causally and not just contingently connected. Another point worth remarking is the divergence in the slope of the two lines around the turn of the millennium – GDP growth was occurring more rapidly than CO2 emissions growth, possibly suggestive of the decoupling propounded by the ecomodernists. But more recently the relationship switched, with CO2 emissions growing more rapidly than GDP.
Another exercise is to break per capita CO2 emissions down into the aggregate groupings described above, shown in Figure 3. The pattern displayed in Figure 1 in relation to GDP is broadly repeated, albeit not quite as strikingly. In Figure 1 there’s a 55-fold difference between the highest and lowest of the groups, whereas in Figure 2 it’s only 36-fold, and the figure for the highest group (wealthy countries) is reducing. So perhaps this could be taken to support the ecomodernist view that increasing national wealth permits greater mitigation of environmental problems. On the other hand, the absolute figure for the wealthy countries is still much higher than for all other aggregations (and the absolute rather than the per capita figure globally is rising, which is the ultimate arbiter of greenhouse gas abatement). It may turn out that marginal emissions per unit of GNP decrease but emissions still increase with increasing GNP, thereby negating the value of growing the economy for tackling emissions. And it’s worth noting that while carbon dioxide emissions for the wealthy countries have increased in absolute terms by 1.3 billion tonnes between 1990 and 2011, they’ve increased by 10 billion tonnes in the BRICS countries. Given the relative decline of manufacturing industry in the wealthy countries and the rise of export-oriented manufacture in the BRICS, it’s possible that the true footprint of the wealthy countries is even larger. For all these reasons, the evidence for a positive association between GDP and carbon abatement does not seem compelling.
In his Marxist take on ecomodernism, Leigh Phillips argues that emissions are mostly associated with the rich within countries as well as between them – quoting the figure that the top 20% of income earners in the USA are responsible for 70% of consumption, and drawing the conclusion that “phrases such as “the greenhouse gas emissions of the average American” or “per capita consumption” contain absolutely no useful information” (11). To address this point, I calculated a per capita emissions figure for the USA in which 70% of the country’s emissions were imputed to 20% of the population and excluded from the analysis. This scenario reduces US emissions from 17.5 tonnes per capita (the 10th highest in the world, out of 193 countries) to 6.4 tonnes per capita, which would place it 50th out of 193 and still well over twice the global median emission figure of 2.5 tonnes per capita. Other research appears to confirm the importance of country-level income on individual emissions profiles. For example, the carbon footprint of the poorest 50% of US citizens is 20 times higher than the footprint of the poorest 50% of Indian citizens (12). Perhaps we might conclude that, in the case of carbon dioxide emissions, the distribution of emissions at country-level is such that mean per capita figures do still tell us something useful about the probable usage levels of individual citizens across the income spectrum.
So it seems probable that economic growth is a driver of increased emissions and that rich countries must shoulder a large part of the responsibility for climate change, however emissions are distributed by wealth within their populations. Nevertheless, the ecomodernists argue that economic growth can safely be maintained by switching from GHG-emitting energy sources to clean ones. The main clean energy sources are nuclear power and renewables such as hydroelectricity, photovoltaics and wind turbines. All of these sources are methods of generating electricity, but electricity generation currently only accounts for 14% of the world’s total primary energy consumption, and clean generation only accounts for about 30% of electricity generation, so a clean energy switch actually requires two major transformations: from fossil fuels to electricity, and from fossil fuel electricity generation to clean electricity generation.
Breakthrough Institute analysts suggest that in order to achieve stabilisation of atmospheric CO2 at 450 parts per million it will be necessary to install 1-2 Gigawatts of clean electricity capacity per day from now until 2050 (13). Presumably it will also be necessary for this new capacity to substitute for fossil fuel generated electricity. Figure 4 shows global installed electricity generating capacity from 1980 to 2012 (the last year for which figures are available), broken down into fossil fuel, nuclear, hydroelectricity and other renewable generation. It then projects installed capacity forwards to 2050 on the basis of 1.5 GW clean capacity installed daily and substituting for fossil fuel sources. Industry projections suggest that it won’t be possible to increase hydroelectric capacity more than twofold between now and 2050 (14), so this doubling is included in the projection, the remainder of the 1.5 GW/day being split evenly for the sake of the projection between nuclear and non-hydroelectric renewables.
A few points can be made about the figure. First, another statistical cliché is that it’s unsafe to project past trends into the future. Still, the break in the trend from 1980-2012 that will have to occur in order to achieve a clean energy transition is starkly apparent from Figure 4 and is perhaps worth presenting graphically in this way in order to demonstrate the scale of the challenge. Will it be possible? Of course, nobody knows. But it’s worth pausing for a moment to consider how plausible the prospect is of switching so thoroughly out of the present fossil fuel based energy economy within the next 35 years while increasing total generating capacity in order to continue funding economic growth.
Another question that arises from studying Figure 4 is the strong ecomodernist emphasis on nuclear power. One reason for this emphasis is that the ecomodernists argue, probably rightly, that few other clean sources of energy are scalable to existing electricity demands. This is the case for example with hydroelectricity, as mentioned above. But the low level of existing installed capacity for nuclear power (and other clean renewable sources) points to the need for a massive expansion of these technologies in so short a timescale as to be wholly unprecedented in human history. And the cost of installing nuclear power stations is such that at present it’s only feasible in relatively wealthy countries.
Perhaps where data analysis can help is in clarifying the nature of the challenge involved in the various positions adopted. I’d argue that the data I’ve presented here suggest that poverty will not be easily ameliorated simply by growing GDP, that the responsibility for carbon emissions is a rich country as well as a rich person phenomenon which will not be easily redressed by enhancing prosperity worldwide in accordance with standard models of economic growth, and that the scale of the energy transition required to fund ‘green growth’ is historically unprecedented.
China is one of the few countries with both the money and the political system in place to effect a rapid transformation to nuclear power. According to Phillips, it’s aiming to install up to 500 GW of capacity by 2050 (15) – a 22-fold increase on its present nuclear capacity and in excess of 30% more nuclear capacity than exists in the entire world at present. But even this extremely ambitious programme, which it’s unlikely any other country in the world could match, would achieve only 2% of the necessary global electrical capacity by 2050 projected in Figure 4. And this huge expansion in electrical generation capacity still only equates to about 60% of current total primary energy production.
Whatever happens with the future energy supply it seems clear, to use another cliché, that ‘business as usual is not an option’. But the data I’ve presented above do lead me to question the strong emphasis on nuclear power found in ecomodernist writings such as the Manifesto or Brand’s Whole Earth Discipline. The EIA’s report of 2010 Energy Technology Perspectives, projects that only 6% of the necessary emissions reductions will come from a switch to nuclear power, with 17% coming from renewables and 38% – the largest single contribution – from end-use fuel and electricity efficiency (16). In this respect, the polarised debate about nuclear power between ecomodernists and more traditional anti-nuclear greens seems something of a sideshow compared to other drivers of the energy transition. The green emphasis on degrowth, so pilloried by Phillips in particular but also by other ecomodernists, would also involve a huge and historically unprecedented transformation. But looking at what’s involved in stabilising the climate at existing levels of energy use, I can’t help feeling that it might be the easier of the two routes to follow.
It isn’t possible to resolve the ecomodernism debate definitively with the kind of data I’ve presented here, or perhaps with any kind of data at all in view of the wide-ranging and future-focused nature of the claims involved. In the rough-and-tumble world of political debate, data analysis quickly gives way to invective: ‘Cassandras’, ‘pessimists’, ‘neo-Malthusians’ and ‘misanthropes’ are words often employed by ecomodernists, and are liberally scattered throughout Phillips’ book. Perhaps where data analysis can help is in clarifying the nature of the challenge involved in the various positions adopted. I’d argue that the data I’ve presented here suggest that poverty will not be easily ameliorated simply by growing GDP, that the responsibility for carbon emissions is a rich country as well as a rich person phenomenon which will not be easily redressed by enhancing prosperity worldwide in accordance with standard models of economic growth, and that the scale of the energy transition required to fund ‘green growth’ is historically unprecedented.
In a recent article in the Boston Globe about climate activism, academics Joshua Goldstein and Steven Pinker argue “The movement should hit “Pause,” do the math, and work for the combination of policies that can actually solve the problem” (17), which in their view involves large-scale government funding for scientific breakthroughs in batteries, nuclear energy, liquid biofuels, and carbon capture. But however worthwhile that commitment, there’s little ‘math’ in their article to demonstrate the scale and urgency of the challenge. I hope that this article has provided a bit of ‘math’, however limited, that helps to do so. It leads me personally towards a different focus – degrowth, energy descent, social redistribution. But I’m not so naïve as to think that ‘math’ alone ever leads anyone anywhere. Goldstein and Pinker argue that “climate change must transcend ideology”, but a solid finding from social science – perhaps its only truly solid finding – is that nothing ever transcends ideology.
2. Brand, S. (2009). Whole Earth Discipline, Atlantic Books.
4. Ridley, M. (2010) The Rational Optimist, Fourth Estate.
5. Phillips, L. (2015) Austerity Ecology and the Collapse-Porn Addicts, Zero Books.
9. Heller, H. (2011) The Birth of Capitalism, Pluto.
10. Pickett, K. and Wilkinson, R. (2010) The Spirit Level, Penguin.
11. Reference 5, pp.55-6
15. Reference 5, p.180