RealClimate logo

Why Levitt and Dubner like geo-engineering and why they are wrong

Filed under: — gavin @ 18 October 2009

Many commentators have already pointed out dozens of misquotes, misrepresentations and mistakes in the ‘Global Cooling’ chapter of the new book SuperFreakonomics by Ste[ph|v]ens Levitt and Dubner (see Joe Romm (parts I, II, III, IV, Stoat, Deltoid, UCS and Paul Krugman for details. Michael Tobis has a good piece on the difference between adaptation and geo-engineering). Unfortunately, Amazon has now turned off the ‘search inside’ function for this book, but you can read the relevant chapter for yourself here (via Brad DeLong). However, instead of simply listing errors already found by others, I’ll focus on why this chapter was possibly written in the first place. (For some background on geo-engineering, read our previous pieces: Climate Change methadone? and Geo-engineering in vogue, Also the Atlantic Monthly “Re-Engineering the Earth” article had a lot of quotes from our own Raypierre).

Paul Krugman probably has the main issue right:

…it looks like is that Levitt and Dubner have fallen into the trap of counterintuitiveness. For a long time, there’s been an accepted way for commentators on politics and to some extent economics to distinguish themselves: by shocking the bourgeoisie, in ways that of course aren’t really dangerous.


Clever snark like this can get you a long way in career terms — but the trick is knowing when to stop. It’s one thing to do this on relatively inconsequential media or cultural issues. But if you’re going to get into issues that are both important and the subject of serious study, like the fate of the planet, you’d better be very careful not to stray over the line between being counter-intuitive and being just plain, unforgivably wrong.

Levitt was on NPR at the weekend discussing this chapter (though not defending himself against any of the criticisms leveled above). He made the following two points which I think go to the heart of his thinking on this issue: “Why would anyone be against a cheap fix?” and “No problem has ever been solved by changing human behaviour” (possibly not exact quotes, but close enough). He also alluded to the switch over from horse-driven transport to internal combustion engines a hundred years ago as an example of a ‘cheap technological fix’ to the horse manure problem. I deal with each of these points in turn.

Is geo-engineering cheap?

The geo-engineering option that is being talked about here is the addition of SO2 to the stratosphere where it oxidises to SO4 (sulphate) aerosols which, since they are reflective, reduce the amount of sunlight reaching the ground. The zeroth order demonstration of this possibility is shown by the response of the climate to the eruption of Mt. Pinatubo in 1991 which caused a maximum 0.5ºC cooling a year or so later. Under business-as-usual scenarios, the radiative forcing we can expect from increasing CO2 by the end of the century are on the order of 4 to 8 W/m2 – requiring the equivalent to one to two Pinatubo’s every year if this kind of geo-engineering was the only response. And of course, you couldn’t stop until CO2 levels came back down (hundreds, if not thousands of years later) without hugely disruptive and rapid temperature rises. As Deltoid neatly puts it: “What could possibly go wrong?”.

The answer is plenty. Alan Robock discussed some of the issues here the last time this came up (umm… weeks ago). The basic issues over and above the costs of delivering the SO2 to the stratosphere are that a) once started you can’t stop without much more serious consequences so you are setting up a multi-centennial commitment to continually increasing spending (of course, if you want to stop because of huge disruption that geo-engineering might be causing, then you are pretty much toast), b) there would be a huge need for increased monitoring from the ground and space, c) who would be responsible for any unanticipated or anticipated side effects and how much would that cost?, and d) who decides when, where and how much this is used. For point ‘d’, consider how difficult it is now to come up with an international agreement on reducing emissions and then ponder the additional issues involved if India or China are concerned that geo-engineering will cause a persistent failure of the monsoon? None of these issues are trivial or cheap to deal with, and yet few are being accounted for in most popular discussions of the issue (including the chapter we are discussing here).

Is geo-engineering a fix?

In a word, no. To be fair, if the planet was a single column with completely homogeneous properties from the surface to the top of the atmosphere and the only free variable was the surface temperature, it would be fine. Unfortunately, the real world (still) has an ozone layer, winds that depend on temperature gradients that cause European winters to warm after volcanic eruptions, rainfall that depends on the solar heating at the surface of the ocean and decreases dramatically after eruptions, clouds that depend on the presence of condensation nuclei, plants that have specific preferences for direct or diffuse light, and marine life that relies on the fact that the ocean doesn’t dissolve calcium carbonate near the surface.

The point is that a planet with increased CO2 and ever-increasing levels of sulphates in the stratosphere is not going to be the same as one without either. The problem is that we don’t know more than roughly what such a planet would be like. The issues I listed above are the ‘known unknowns’ – things we know that we don’t know (to quote a recent US defense secretary). These are issues that have been raised in existing (very preliminary) simulations. There would almost certainly be ‘unknown unknowns’ – things we don’t yet know that we don’t know. A great example of that was the creation of the Antarctic polar ozone hole as a function of the increased amount of CFCs which was not predicted by any model beforehand because the chemistry involved (heterogeneous reactions on the surface of polar stratospheric cloud particles) hadn’t been thought about. There will very likely be ‘unknown unknowns’ to come under a standard business as usual scenario as well – another reason to avoid that too.

There is one further contradiction in the idea that geo-engineering is a fix. In order to proceed with such an intervention one would clearly need to rely absolutely on climate model simulations and have enormous confidence that they were correct (otherwise the danger of over-compensation is very real even if you decided to start off small). As with early attempts to steer hurricanes, the moment the planet did something unexpected, it is very likely the whole thing would be called off. It is precisely because climate modellers understand that climate models do not provide precise predictions that they have argued for a reduction in the forces driving climate change. The existence of a near-perfect climate model is therefore a sine qua non for responsible geo-engineering, but should such a model exist, it would likely alleviate the need for geo-engineering in the first place since we would know exactly what to prepare for and how to prevent it.

Does reducing global warming imply changing human behaviour and is that possible?

This is a more subtle question and it is sensible to break it down into questions of human nature and human actions. Human nature – the desire to strive for a better life, our inability to think rationally when trying to impress the objects of our desire, our natural selfishness and occasionally altruism, etc – is very unlikely to change anytime soon. But none of those attributes require the emission of fossil fuel-derived CO2 into the atmosphere, just as they don’t require us to pollute waterways, have lead in gasoline, use ozone-depleting chemicals in spray cans and fridges or let dogs foul the sidewalk. Nonetheless, societies in the developed world (with the possible exception of Paris) have succeeded in greatly reducing those unfortunate actions and it’s instructive to see how that happened.

The first thing to note is that these issues have not been dealt with by forcing people to think about the consequences every time they make a decision. Lead in fuel was reduced because of taxation measures that aligned peoples preferences for cheaper fuel with the societal interest in reducing lead pollution. While some early adopters of unleaded-fuel cars might have done it for environmental reasons, the vast majority of people did it first because it was cheaper, and second, because after a while there was no longer an option. The human action of releasing lead into the atmosphere while driving was very clearly changed.

In the 1980s, there were campaigns to raise awareness of the ozone-depletion problem that encouraged people to switch from CFC-propelled spray cans to cans with other propellants or roll-ons etc. While this may have made some difference to CFC levels, production levels were cut to zero by government mandates embedded in the Montreal Protocols and subsequent amendments. No-one needs to think about their spray can destroying the ozone layer any more.

I could go on, but the fundamental issue is that people’s actions can and do change all the time as a function of multiple pressures. Some of these are economic, some are ethical, some are societal (think about our changing attitudes towards smoking, domestic violence and drunk driving). Blanket declarations that human behaviour can’t possibly change to fix a problem are therefore just nonsense.

To be a little more charitable, it is possible that what was meant was that you can’t expect humans to consciously modify their behaviour all the time based on a desire to limit carbon emissions. This is very likely to be true. However, I am unaware of anyone who has proposed such a plan. Instead, almost all existing mitigation ideas rely on aligning individual self-interest with societal goals to reduce emissions – usually by installing some kind of carbon price or through mandates (such as the CAFE standards).

To give a clear example of the difference, let’s tackle the problem of leaving lights on in rooms where there is no-one around. This is a clear waste of energy and would be economically beneficial to reduce regardless of the implications for carbon emissions. We can take a direct moralistic approach – strong exhortations to people to always turn the lights off when they leave a room – but this is annoying, possibly only temporary and has only marginal success (in my experience). Alternatively, we can install motion-detectors that turn the lights out if there is no-one around. The cost of these detectors is much lower than cost of the electricity saved and no-one has to consciously worry about the issue any more. No-brainer, right? (As as aside, working out why this isn’t done more would be a much better use of Levitt and Dubner’s talents). The point is changing outcomes doesn’t necessarily mean forcing people to think about the right thing all the time, and that cheap fixes for some problems do indeed exist.

To recap, there is no direct link between what humans actually want to do and the emissions of CO2 or any other pollutant. If given appropriate incentives, people will make decisions that are collectively ‘the right thing’, while they themselves are often unconscious of that fact. The role of the economist should be to find ways to make that alignment of individual and collective interest easier, not to erroneously declare it can’t possibly be done.

What is the real lesson from the horse-to-automobile transition?

Around 1900, horse-drawn transport was the dominant mode of public and private, personal and commercial traffic in most cities. As economic activity was growing, the side-effects of horses’ dominance became ever more pressing. People often mention the issue of horse manure – picking it up and disposing of it, it’s role in spreading disease, the “intolerable stench” – but as McShane and Tarr explain that the noise and the impact of dead horses in the street were just as troublesome. Add to that the need for so many stables downtown taking up valuable city space, the provisioning of hay etc. it was clear that the benefits of the horse’s strength for moving things around came at a great cost.

But in the space of about 20 years all this vanished, to be replaced with electrified trolleys and subways, and internal combustion engine-driven buses and trucks, and cars such as the Model-T Ford. Almost overnight (in societal terms), something that had been at the heart of economic activity had been been relegated to a minority leisure pursuit.

This demonstrates very clearly that assumptions that society must always function the same economic way are false, and that in fact we can change the way we do business and live pretty quickly. This is good news. Of course, this transition was brought about by technological innovations and the switch was decided based on very clear cost-benefit calculations – while cars were initially more expensive than horses, their maintenance costs were less and the side effects (as they were understood at the time) were much less burdensome. Since the city had to tax the productive citizens in order to clear up the consequences of their own economic activity, the costs were being paid by the same people who benefited.

Levitt took this example to imply that technological fixes are therefore the solution to global warming (and the fix he apparently favours is geo-engineering mentioned above), but this is a misreading of the lesson here in at least two ways. Firstly, the switch to cars was not based on a covering up of the manure problem – a fix like that might have involved raised sidewalks, across city perfuming and fly-spraying – but from finding equivalent ways to get the same desired outcome (transport of goods and people) while avoiding undesired side-effects. That is much more analogous to switching to renewable energy sources than implementing geo-engineering.

His second error is in not appreciating the nature of the cost-benefit calculations. Imagine for instance that all of the horse manure and dead carcasses could have been easily swept into the rivers and were only a problem for people significantly downstream who lived in a different state or country. Much of the costs, public health issues, etc. would now be borne by the citizens of the downstream area who would not be benefiting from the economic prosperity of the city. Would the switch to automobiles have been as fast? Of course not. The higher initial cost of cars would only have made sense if the same people who were shelling out for the car would be able to cash in on the benefits of the reduced side effects. This is of course the basic issue we have with CO2. The people benefiting from fossil fuel based energy are not those likely to suffer from the consequences of CO2 emissions.

The correct lesson is in fact the same as the one mentioned above: if costs and benefits can be properly aligned (the ‘internalising of the externalities’ in economist-speak), societies and individuals can and will make the ‘right’ decisions, and this can lead to radical changes in very short periods of time. Thus far from being an argument for geo-engineering, this example is an object lesson in how economics might shape future decisions and society.


To conclude, the reasons why Levitt and Dubner like geo-engineering so much are based on a misreading of the science, a misrepresentation of proposed solutions, and truly bizarre interpretations of how environmental problems have been dealt with in the past. These are, in the end, much worse errors than their careless misquotes and over-eagerness to shock highlighted by the other critiques. Geo-engineering is neither cheap, nor a fix, and the reasons why it is very likely to be a bad idea are ethical and legal, much more than its still-uncertain scientific merits.

406 Responses to “Why Levitt and Dubner like geo-engineering and why they are wrong”

  1. 401
    Patrick 027 says:

    “If a workable (i.e. economically feasible) CCS comes along it would do so without a tax on Carbon. ”

    For what reason if not a tax on carbon? Only alternative is PR benifits. I don’t know if that’s strong enough.

    You just aren’t getting this.

  2. 402
    Chris says:


    I reviewed again your general predictions for mitigation spending and I do not think that they are realistic. Empirically, you need economic growth to develop new technologies and not the other way around. If advances are to be made at the pace at which everyone claims they can be then a carbon tax will be ruinous to this effort because it will cause a major contraction in our economy. I do not believe that you can have government spur the kind of technological advances or adopt the solutions that you are putting forward (even if everything goes perfectly) without risking a massive “green bust” where we have too many seemingly promising solutions that get funded but go nowhere. Just look at all of the Hydro power projects we had during the Great Depression that failed. If this happens again, given our current dependence on electricity, we will be unable to produce our own power efficiently (i.e. without subsidies). At that point we will either have to scrap our entire grid or import dirty power from Mexico.
    My general solution to this would be to remove all of the legal restrictions on Nuclear power, to deregulate our utilities so they are forced to run efficiently and allow Nuclear plants to borrow at government interest rates. My guess is that since the long term costs of Nuclear are lower all the new plants will be nuclear and have no emissions. PR will impact coal and oil plants to reform as dictated by the market, which may not be “good enough” but it is a fair bit better than causing an economic catastrophe every time emissions levels are ratcheted down.

  3. 403
    Patrick 027 says:

    Chris –

    If you have the government spend x1 on geoengineering to cool the earth but with potential other side-effects, and if those side-effects cost x2, you’re taxing the economy x1+x2, for a benifit z.

    If mitigation pathways are pursued, with an additional cost to the economy of y1, for a benifit z, then the economy is being taxed y1. But what if y1 is less than x1+x2 and less than or equal to z?

    Why would the private sector pay y1 more than otherwise if there is no public policy, such as a tax on the emissions?


    “to remove all of the legal restrictions on Nuclear power,”

    I really have not made up my mind regarding nuclear power, but I most definitely think some regulation is necessary there.

    People have been known to try to get away with murder, you know.


    “Empirically, you need economic growth to develop new technologies and not the other way around. ”

    I agree that having resources to invest in invention is necessary for invention, but this is affected by both supply and demand. There may be a tendency for economic growth to help technological progress, but I don’t see why economic growth would be necessary or sufficient in an absolute way for technological progress – futhermore, technological progresss should strongly tend to stimulate economic growth, or reduce economic decay.

  4. 404
    CM says:

    Chris (#402) said: “My guess is that since the long term costs of Nuclear are lower all the new plants will be nuclear and have no emissions.”

    I don’t want to start another discussion of nuclear. But it seems pretty clear that your guess is wide of the mark. Unless the industry gets a lot better at building nukes on time and budget, the only way new nukes can compete with new coal or gas if the market is “distorted”, as you would see it, by hefty carbon price through tax or cap-and-trade.

    MIT study update suggests $25/ton CO2 might almost but not quite do the trick for nukes versus coal: (p. 6)

    Not likely under Boxer-Kerry anytime soon:

  5. 405
    Patrick 027 says:

    Re myself re Chris:

    More precisely:

    emitting economic pathway: public cost z1, which is somewhat hard to measure but is understood to exist and be significant.

    Net public cost is z1.

    Geoengineering: pubic cost (via government funding) g1 in order for a public benifit z2, plus more public cost g2.

    Net public cost is g1 + g2 – zg.

    There are some different ways to assign effects to either g2 or zg; in the case of aerosols, letting g2 be the net adverse side effects of global cooling to counteract global warming by aerosols (changes in precipitation, regional), plus more direct effects on the ozone layer, then g2 is equal (or proportional) to the portion of z1 that is global warming; the rest of z1 is ocean acidification (for CO2), … etc. Note that there is no obvious way to get zg as large as z1 by cooling alone, because making the climate too cold is not a net benifit. Thus the only way to get zg as large as z1 is to do at least something of net public benifit in addition to the cooling, such as distribution CaCO3 and CaSiO3, etc, around the oceans to counteract the acidification, or ________. Of course, this actually has that effect plus a CO2 sequestration effect, and depending on how the material is dispersed, could have an aerosol cooling effect itself, and the sequestering and aerosol cooling which would reduce the necessary other aerosols for cooling to achieve the same zg…

    Sequestration: cost (public funding or private) s1, public benifit zs, plus more public cost s2

    net public cost = s1 + s2 – zs

    Same issues as with geoengineering, except if zs can be brought nearly equal to z1 and s2 can be made small or negative (olivine/etc. dispersal (land or ocean or air to ocean?), etc, or other carbonate mineral sequestration, biochar).


    Mitigation: cost of replacing emitting pathways with non- or reduced-emitting pathways to either private sector or public sector is m1, public benifit zm, public cost m2

    net cost = m1 + m2 – zm

    zm can generally approach or be proportional to z1; options exist that make m2 small and manageable.


    mitigation_cost = m1 + m2 – zm
    sequestration_cost = s1 + s2 – zs
    geoengineering_cost = g1 + g2 – zg

    Best path to pursue is the one with the least (likely) cost.

    There is no way besides PR to get the private sector to take on the costs of m1, s1, or g1, without a public policy that directly or otherwise puts a price or cap on emissions. Even if direct public funding is eschewed in all cases, so that emitters pay directly for offsets as s1 or g1 or else reduce emissions (m1), there has to be some incentive to do so, and some regulation to keep people from just making stuff up (and to watch out for poor approximations).

    So you’ve got government involved whichever way.

    And even if scarcity of emitting pathway resources eventually makes m1 zero or negative, this actually shifts the best trajectory even more strongly towards mitigation, or mitigation + sequestration + geoengineering, because the private sector without public policy still wouldn’t do enough and as soon as justified by BOTH scarcity AND externalities – it would only tend (ideally) towards optimum performance as judged by scarcity; the externality issue would still leave room for improvement from there, although it would tend to shrink. But timing is important.

    Uncertainty is arguably greater for zg than for zs and zm, zs perhaps being more uncertain than zm; g2 and s2 may be more uncertain in absolute (vs relative) terms than m2. Evaluation of m1 has some basis in real-world data – of course, it may be very easy to project s1 and g1 for some cases – are those the cases with the best overall outcomes?

  6. 406
    Patrick 027 says:


    mitigation_cost = m1 + m2 – zm
    sequestration_cost = s1 + s2 – zs
    geoengineering_cost = g1 + g2 – zg
    none of the above = z1

    Best path to pursue is the one with the least (likely) cost.

    In the above, zg, zs, and zm are the net public benifits that occur by changes in the opposite directions, along various dimensions (including global average surface temperature, ocean pH), as the changes that occur that produce the net public cost z1.

    g2, s2, and m2 are the net public costs caused by changes in other dimensions or changes in the same direction as those which cause z1 along some dimensions.

    (for ‘incomplete’ geoengineering plans, such as SO2 injections:)
    “Note that there is no obvious way to get zg as large as z1 by cooling alone, because making the climate too cold is not a net benifit. ”

    Well, given prior emissions, it is possible that future actions could cancel out more than the warming caused by future emissions and lead to a greater benifit, within limits.