Economics and Climatology

Following up from an interesting germ of an argument between Dan Olner and “Frank” in the comments to this article

I dislike it very much when people defend climatology from a point of view of credentials. Despite perceptions from some quarters, I am interested in engaging challenges to the legitimacy of physical climatology.

I am forced to this position by my severe doubts about economics as a discipline and as an academic culture.

Let me concede that there must be some substance, some value to economics. Nevertheless, it is my claim that the comparison between the disciplines is worth undertaking, and that economics comes up the loser by a wide margin.

This is doubly important. First of all, though both physical climatology and economics are frequently called into doubt, clear thought in both disciplines is a requirement for progress toward a sustainable future. From an engineer’s perspective, climate is a key part of the system under control, and economics is a key part of the controller. Without a good model of the system and a good design for the controller, it seems increasingly likely that the whole system will spin out of control.

Secondly, even leaving “green” issues off the table, economics makes grandiose claims to be the science of collective behavior, or even the science of collective happiness. Yet it dismisses any philosophically interesting aspects of these questions in favor of counting dollar-denominated transactions. Nevertheless, it claims for itself a unique position among the sciences, as the crux, the central weighing mechanism, for all public decision-making.

Oddly, one hears the same claim of hubris directed at climate science. If in fact things are as we climatologists claim, after all, this will have an enormous impact on all aspects of human behavior, even of procreation. Of course it is obvious to anyone (except perhaps an economist) that there are limits to growth in some sense imposed by our physical environment, so whether climate science is right or not, clearly there is some physical constraint that will trap us sooner or later. But it does matter that we get it right. If CO2 accumulation really is harmless, there is more time and there are fewer immediate constraints on what we do.

And so, I insist that the question is legitimate.

I also would not be one to throw away ALL of economics. Some of it is clear enough. And it would be good if our severest critics were more clear about what they think we are getting wrong as well. But the most severe critics of anything tend to be those furthest from it. I have tried to avoid dismissing economics, even as I disagree most strongly with what is often presented to me as economic consensus.

One place where climate science wins is in its connection to other disciplines. Indeed, climate science is either a client or a provider or both to most other disciplines. Economics is an island unto itself. Climate also understands the nature of regimes (this approximation for low Rossby number, this one for high Reynolds number, etc.) and how the right analysis tool for one situation is completely useless in another. Economists still seem to be trying to find the One Ring to Rule them All, despite the fact that the complexities of human preferences are buried underneath their abstractions. Both of these things disturb me and somehow don’t seem to bother others.

But above all, it bothers me that economics poses as a pure science when it is actually best construed as a particularly crude form of engineering. We should not be predicting what we do. We should be deciding what we do, and using economics as a tool to get us to the future, rather than as a measurement device that tells us by how much we are certainly going to fail.

Comments:

  1. Economics have failed us. I can't see any other way around it. Perhaps climatology did as well, but at least there were those, like mt, who understood what uncertainty within the science really meant, and that message was enough to engage the planet in a conversation on what to do about it.

    Continuing on economics, the failure of warning both before and during our current predicament will have more immediate consequences, depending on how that is defined. It's been 5 years since the 2007 crash folks, and we are still hanging on by a thread, buoyed by massive central band activity and price manipulation. I am not judging the utility of central bank activity, as it has extended the fallout, but this points to a fundamentally flawed system. Unfortunately, even those who recognize these facts are offering silly solutions, like returning to the gold standard. Recent economic data is pointing to another round of QE from the fed and another LTRO from the European CB. This will give us more time and everything will look hunky-dory to the short term investor and those who think we can keep printing until the models actually work and people get jobs again. As I see it, we are doing just enough to preserve a pile of dung, and the economics departments across the world are at fault.

  2. Michael,

    It's not clear to me what you are trying to say here with this post. You appear to be beating up strawmen of your own making and don't bother to back up any of your assertions with anything resembling evidence (e.g. 'Economics is an island unto itself'). The whole thing comes across as an unhinged rant. sorry, but that's my impression. Can't always hit it out of the park eh ;-)

    @ grypo,

    Economics hasn't failed us. politicians and the institutions that have been built over the last couple hundred of years have failed us. To try and pin this on a scientific discipline is ridiculous. I think it's probably more accurate to say that some people have put far too much faith in a particular strain of economic though and its champions (i.e. chicago school). Unfortunately, these people wield a disproportionately large degree of control over world affairs so the consequences have been substantial. If you haven't read it already, I highly recommend John Ralston Saul's book, Voltaire's Bastards: A Dictatorship of Reason in the West

    My general takeaway from Saul's work and that of others is that economics, as one embodiment of reason, can be a useful tool for decision making, but it is not the only , or even the most important tool. I'll end with one of my favorite quotes by Ulrich Beck that I think gets to the heart of the matter:

    • Hey Marlowe,

      What you say is actually what I mean. Te discipline itself, is awash in different types of economics thought, including Marxist(!). But these types of classes have moved primarily into Liberal Arts Colleges, ie philosophy, social etc. and away from the actual applicable science. This is certainly the work of misguided politics, as you state.

    • Not much really. Just wanted to put this topic on the front burner.

      A few random observations of my own to stir the pot.

      I may find time for a better thought out and constructed argument.

      But this comparison is a question that has long interested me.

  3. hmm...the quote that should have been appended to the last post is:

    "scientific rationality without social rationality remains empty, but social rationality without scientific rationality remains blind."

  4. Lots to respond to there, but for now, one of my current favourite economics quotes. Most of my favourites at the moment tend to be economists pointing out that, actually, they have thought about things quite hard for a long time. But then, so did middle-ages Religious orders, so I guess that's no guarantee of anything.

    Oh also, just on: "how the right analysis tool for one situation is completely useless in another" - MT, did you read that Friedman link I gave? If not, give it a go. "Everything depends on the problem" (Friedman 1953a p.7).

    Anyway -

    Lucas: economists are "programming robot imitations of people, and there are real limits on what you can get out of that" (Klamer
    1984 p.49 in Vriend 1994, Artificial Intelligence and Economic Theory, p.31).

  5. Can I just pose a question as well? I've just submitted a grant proposal (and will be going home to sleep shortly!), hoping to estimate trade flows between small-scale zones in the UK, and then think about how fuel cost changes may affect the choices facing people and organisations. MT - what sort of ideas would you suggest using, what tools, for thinking about or modelling how people and organisations make choices, given changing costs?

    Or is that just completely the wrong framework? Should we be designing systems able to respond appropriately to those changes, rather than attempting a top-down academic gaze, hoping to comprehend, judge and then 'feed into policy?'

    Apologies for bitty comments, brain's a bit zapped.

    • This whole thread is of great interest to me - as a "practicing" "ecological economist" it touches on a great many themes that I have studied and have opinions about. However due to time constraints (it's the last week of the semester and I have mas grading, plus the usual overdue reviews and manuscripts), I'll give only a very couple of very short comments, one to Dan's concrete question:

      I would start by thinking about the different scales that are in play, and about what one in economics might classify as the distinction between marginal and non-marginal changes. Keep in mind that agents face fuel cost changes all the time. I presume you're quite familiar with the concept of "elasticity" - there has been a great deal of study of the elasticity of responses to energy prices, but they obviously focus on marginal changes (although some of the historical studies address big changes like the oil shocks of (it's funny to say this) the last century).

      On average consumers in the same income classes have relatively similar energy expenditures, the differences being contingent inter alia on commuting distances and home construction. Businesses, on the other hand, have a great diversity, depending on, well, what their line of business is. It's not conceptually hard to compare the energy shares of expenditure of various types of businesses to see which ones can be expected to be impacted by fuel price changes.

      Did I say a couple of short comments? I'm just getting started. Dan, maybe we should take this off line. Besides you said you already submitted the grant - can you send it to me to take a look (after I submit grades on the 7th)?

      --pb

  6. Marlowe, meant to say: some really good stuff there. There is a lot of assertion in the article, but it's useful assertion, and gives us a chance to unpick a bunch of assumptions. I only really have a School of Gladwell take on things, with an addition of some arcane modelling work on getting economic agents to take asynchronous actions in space (turns out to be horribly messy, at least for me.) I'm outside the discipline looking in, and find myself defending it having spent several years digging through other human action modelling ideas that turned out to be... unhelpful.

    I ended up back with utility as a completely neutral way to discuss and program welfare, unhindered by any related value problems. It's utterly abstract, and better for it. Hence why I have problems with this line: it "dismisses any philosophically interesting aspects of these questions in favor of counting dollar-denominated transactions."

    What's interesting to you MT? I've found ending up back at utility again after an early sociological hatred of it a very interesting path. I could be wrong about a great many things, but it simply isn't the case that economics has nothing to offer there. Utility is a way of thinking about what 'better' and 'worse' means in a way that allows you to stick numbers on it. It's obviously ridiculous, pyschologically. Back to Lucas again: if the economist forgets they're "programming robot imitiations of people and there are real limits on what you can get out of that", they'll be in trouble.

    • My personal favorite, other than the standard one offered by Oscar Wilde, was given by by James Buchanan in 1964:

      "Economists should cease proffering policy advice as if they were employed by a benevolent despot, and they should look to the structure within which political decisions are made."

  7. danolner:

    Lucas: economists are "programming robot imitations of people, and there are real limits on what you can get out of that"

    There's nothing wrong in itself with "programming robot imitations of people". But the real problem is _how_ to do it properly.

    The way it's being done in AI tasks, from what I can tell, is to first come up with a general model of a particular task, but estimate any specific parameters in the model from real-world data. As an example, consider Hidden Markov models and the procedures used to train them.

    My impression of economics is that it generally doesn't try to do this, but instead tries to cogitate out an entire model of human behaviour _a priori_.

    Most of my favourites at the moment tend to be economists pointing out that, actually, they have thought about things quite hard for a long time.

    I guess the point is that it's folly for economists to look only at what other economists have been thinking about.

    Because much of the time, many people in _other fields_ have also been thinking about many of the same problems that economists are grappling with, even if the problems are often stated in different ways.

    -- frank

    • Going back to the Lucas quote, I'm more interested in a few snippets from Vriend's paper itself:

      The final objective of such type of analysis is not to become wise with respect to artificial worlds. but to understand what is going on in _real_ decentralized economies.

      Unfortunately I'm not sure Vriend takes his own advice that seriously:

      we could follow the approach of mapping actions and outcomes directly to new actions, leaving the mental processes implicit. This mapping, in order to determine the agents' new actions, is not fixed a priori, but kept flexible. Competing hypotheses are tested and their perceived usefulness is updated in parallel. Reinforcement of hypotheses takes place on the basis of payoffs experienced in the market.

      No, no, no... this is just wrong-headed. Hypotheses should be kept or discarded according to how well they agree with what's happening in the real wide world.

      The thinking that 'we must build a model of a human agent who seeks to optimize such and such an outcome!' needs to go. Forget about optimization; the goal should be to build models of human agents that agree with reality.

      -- frank

    • "The thinking that ‘we must build a model of a human agent who seeks to optimize such and such an outcome!’ needs to go. Forget about optimization; the goal should be to build models of human agents that agree with reality."

      But isn't that one of the fundamental differences between economic and climate models, that in modelling human behaviour there is a potential feedback between the model and what it is modelling? An economic model can influence behaviour, or behaviour can be influenced to fit the model, so it is at least possible to conceive of a situation in which an economic model could validly be selected based on its ability to optimise some outcome rather than its agreement with reality.

    • But isn't that one of the fundamental differences between economic and climate models, that in modelling human behaviour there is a potential feedback between the model and what it is modelling?

      Yes and no.

      I suspect that, at the micro level, people buy and sell things according to rather simple rules of the thumb -- even the most seasoned economist probably won't consult his econometric model before deciding which groceries to buy.

      As for policy-making, policy tends to change very slowly and in very discrete steps, so they can be handled specially. (Compare with "scenarios" in climatology.)

      -- frank

      • Scenarios are to climate science as I would like policies to be to economic theory: boundary conditions imposed on the domain of discourse.

        It is the range of emissions scenarios which determines the climatological outcome. Our predictions are conditional on our collective behavior. That is exactly where climatology ought to pass the buck to economics. But economics is not there to catch the buck. (whatever that means...)

        What economics should tell us is what policies lead tobgiven scenarios ( conditional, presumably, on availability of technologies. This IS a wicked problem after all.) I do not kow that a politician with a proposed policy can get estimates of the impact of such policy on aggregate behavior.

        I realize this is a tall order and perhaps it can never be done very well, but in the absence of a useful theory we are flying blind. No wonder the simplistic regulation-is-bad taxation-is-bad foolishness has so many adherents. At least it manages to proffer some advice, however simpleminded and foolish.

  8. "We should not be predicting what we do. We should be deciding what we do, and using economics as a tool to get us to the future, rather than as a measurement device that tells us by how much we are certainly going to fail."
    To the mind of this ethicist, it sounds like what you're saying is that we need ethics, an account of an ordering of social goods and the human desires that lead to or thwart them, rather than assuming that ethics can be safely subsumed into economic calculations.

  9. "For The Common Good: Redirecting the Economy toward Community, the Environment, and a Sustainable Future" by Herman E. Daly & John B. Cobb Jr. offers not only devestating criticisms of Adam Smith & David Ricardo (with regrad to the modern econmy) but also offers pertinent suggestions related to the title and subtitle.

    Daly & Cobb aren't your rationalist, mathematical economists talking about robots.

  10. So I would posit

    1) there is very little real consensus in macroeconomics

    2) just as "climate science" is one thing to the public and a great many things to it's practitioners, so it would seem to be about economics - that's a point of similarity .

    3) a key difference is that the many sciences comprising the body of knowledge about climate are not in fundamental disagreement (with an asterisk) - they address different aspects of the domain using different evidence and different tools, but tend to functionally converge on a coherent picture. On the other hand, economics consists of competing schools with little overlap and little cross-pollination, because it starts from different assumptions.

    THe asterisk in climate is, of course, when climate touches on economics, where we get wildly differing and competing claims.

    Psychology had similar problems until recently, but many disputes are now being broken by objective techniques in neurophysiology. There is no sign of a comparable emerging connection between economics and any mature objective empirical discipline.

    4) of all disciplines with a mathematical component, aside from pure mathematics itself, economics distinguishes itself by a massive and systematic aversion to data. (somewhere in the InIt archives is a pointer to an economist making this assertion quite convincingly)

    5) there are simple arguments based on plausible assumptions within economics that purport to show that any regulation is a net cost. If these arguments were correct it would be possible to get places faster if the government didn't intervene to tell you what side of the road to drive on. This conclusion is obviously false. Consequently the plausible assumptions cannot all be correct. In general if a model leads to an absurd conclusion, a scientist will refine the model before claiming it is applicable. Demonstrably, economists prefer to avoid the counter evidence

    6) economists waste a lot of computer cycles on general equilibrium models, which scale terribly with dimension. Given a billion fold increase in computer power, some will take it as an opportunity to double the dimensionality of their problem space, but a 30 dimensional general equilibrium model is not a billion times more useful than a 15 dimensional model; indeed it is hardly more informative at all.

    7) agent models show promise, but that would require economists to talk to computer scientists, a prospect they loathe

    8) economists correctly claim that collective decision making , and thus, sustainability, is crucially an economic problem. But then they refuse to solve the problem they are presented with, preferring instead to defend their cloister.

    9) climate science has a similar problem, one much pounced upon by the naysayers and iconoclasts. But it is nowhere near as severe, because climate science is far more severely constrained by empirical realty. Nevertheless, resistance to vigorous replacement of fossil carbon fuels is commonly based on arguments from the authority of economists.

    10) Ironically, there is a B-school prof called Anderson who writes about "forecasting principles" who has aligned himself with the naysayers specifically on the ground that economic forecasts cannot usefully be made more than fa few years out. This is true. Applying it to field with different time constants makes not much sense.

    11) forecasting is not really the issue. Encouraging and rewarding useful behaviors, discouraging and assigning costs to unrewarding behaviors, is really the issue. in short, forecasting is the wrong use ce for economic theory in dealing with sustainability. We have to look for ways to identify goals we share in common and gently encourage their fulfillment in individual choices, while minimizing coercion.

    12) unfortunately, we have dilly-dallied long enough that we are well into sub-optimal territory already, and the discouragement of fossil fuel consumption needs to become so severe it might as well be coercion. And yet Obama not only says he is for low gasoline prices, he has twisted Chu's arm to pretend the same. In a sense it is already too late for the sort of economics that I would prefer to see.

    13) Yes, there are economists who account for resource depletion and pollution (which really is the depletion of the commons as a dumping ground) But they still miss the point that I am trying to make. Economic policy is how we control the system while avoiding coercion and maximizing the capacities of private creativity and enterprise. We know it is powerful stuff. But our politics is too unhinged to make any use of it. This is partly because the people who claim to "do" economics, well, don't.

    • In recent years some academic econmists have turned away from what I'll call the selfish robot model. Without difficulty a quick search turns up
      Journal of Economic Psychology,
      Journal of Neuroscience, Psychology, and Economics,
      M. Rabin at UCB doing Psychcology and Economcs,
      C Camerer doing Behavioral economics,
      MIT OpenCourseWare on Economics and Psychology, ...

      There is a similar list for Environmental Economics. For example, a program at Harv ard by that name.

      However, none of this seems to have made its way into standard Econ 101 textbooks. And then again, how many politcians take even that course? So I suppose politicians might listen to economists, without knowing which 'brand' is closest to reality [if they actually care].

      ----
      I still think you might do quite well to have a visit with a senior economist over in your Dept. of Economics.

    • Things like Behavioural Economics and Environmental Economics should've become the mainstream paradigm long ago already. The fact that "Adam Smithist Economics" and reasoning via aphorisms are still considered _the_ standard economic paradigm -- even after more than a century of inquiry -- is indicative of a huge problem.

      -- frank

  11. It's good to read some provoking ideas, MT, but - again - it would be nice to see some evidence to back up your arguments. You're not wrong about a lot of things but I don't think you're right either. I'm not in any great position of expertise, mind, but the bullet points feel a little like watching a jazz band fall down the stairs.

    "economics distinguishes itself by a massive and systematic aversion to data." Do you follow Krugman's blog? Have you seen his massive and systematic reference to data? Each time, repeatedly pointing out that it clearly supports well-established macro-economic principles, not actions currently being taken in Washington and Europe? I presume you're making a point about economic modelling methods and how they're used?

    "There are simple arguments based on plausible assumptions within economics that purport to show that any regulation is a net cost. If these arguments were correct it would be possible to get places faster if the government didn’t intervene to tell you what side of the road to drive on. This conclusion is obviously false. Consequently the plausible assumptions cannot all be correct."

    Who makes these arguments? Largely (though not exclusively) a small set of economists-for-hire, buyable from the same dealers who will provide you with some top-notch, well-oiled climate FUD. There's a small overlap with Krugman's Very Serious People, particularly the ones peddling 'nasty medicine' solutions to our current economic woes.

    I think you'd be very, very hard pressed to find many economists who don't understand that some costs need to be internalised and the only way to do that is through institutions. You might then want to go on and criticise the idea of internal costs - fine. I'd then suggest we bring (Nobel-prize-winning economist) Elinor Ostrom in and talk about other institutional structures for resource management. That'd be good for me, because my knowledge of her work is first-few-chapter-skimming only at the moment.

    I mean, change your statement to: "There are simple arguments based on plausible assumptions within climatology that purport to show that variation in solar output, not carbon, is responsible for recent warming." You do then go on to say, well, the assumptions can't be correct - correct! But: straw.

    "Agent models show promise, but that would require economists to talk to computer scientists, a prospect they loathe".

    My PhD used a form of agent-modelling. Plenty of economists are in there with both feet. Start with the 'handbook of computational economics: agent-based computational economics' 2006. I think there are some severe problems with agent modelling as it stands - of the "all fur coat and no knickers" variety. Mainly in how the uses of modelling are understood, and that many really need to stop and take on board that Friedman article I posted. This sort of thing -

    “Agent-based models potentially present a way to model the financial economy as a complex system, as Keynes attempted to do, while taking human adaptation and learning into account, as Lucas advocated. Such models allow for the creation of a kind of virtual universe, in which many players can act in complex - and realistic - ways” (Farmer and Foley 2009 p.685-6).

    - is, I think, really false-headed. There's this notion that we make better economic models by making them 'more realistic'. What the hell does that mean? To what end? Frank, this relates to your point about the `real wide world', and you're right to say that's vital. One can make a realistic-looking model, so what? You've made sim city. You can also make a realistic model of atoms bouncing around in a container. Sometimes that's useful, but more often you can stick to models using temperature and pressure. "Everything depends on the problem."

    "Economic policy is how we control the system / forecasting is the wrong use for economic theory in dealing with sustainability / From an engineer’s perspective, climate is a key part of the system under control, and economics is a key part of the controller."

    "Economics" is the study of how people react to cost changes. That doesn't involve putting a dollar sign on every f*cking thing on this planet (though I'm with Bill on marketing, on the whole) any more than modelling evolution does. (Since, to model evolution, you need to have some notion of cost. WOOP WOOP: horrible overlap there with some folk finding too many evolution/market forces parallels. Well, yes, in the same way there are parallels between `lions eating gazelles' and `table tennis'.)

    I've posted this before, but here's Marshall's list of questions. Any surprising ones there? A lot's fallen out of favour, but its broad sweep isn't too far wrong. Note also his take on mathematics.

    So it's a field of inquiry into human behaviour. MT, you're talking about that in the same breath as economic policy, which is a bit like saying climatology is all nonsense because we don't have an effective global carbon trading system. Think we need to be clear what we're talking about.

    The interesting parallel for me (and MT's pointed at it): what's occurred as climatology has got itself tangled in policy and governance? The resulting confusion might, e.g., explain Pielke Jr's take. He seems to have done what I just described: from his perspective, "society is a key part of the system under control, and climatalogy is a key part of the controller." To the extent that policy is created based on climatology, is he wrong? What does that mean for climate science - anything?

    A while back he cited Scott's Seeing Like a State in his list of top 5. That's exactly the stuff we're talking about, and it's far from straightforward. When an enquiring science finds itself coming back out into the world through the hugely powerful gaze of state institutions, all sorts of problems can occur. We probably need some sort of political settlement on a par with the establishment of the separation of powers. Perhaps not coincidentally, in the US in particular at the minute, that separation is under sustained attack. Where does physical science fit into that? What about the study of human behaviour?

    • We should probably be discussing why the current economic system makes it impossible to deal with the climate change and sustainability issues, rather than putting the science itself under scrutiny. Pielke Jr's advice is always interesting, and it is only through understanding it that we can effectively show it to be wrong, outside of neo-liberal economic models. The IRON LAW:

      [P]resents a boundary condition on policy design that is every bit as limiting as is the second law of thermodynamics, and it holds everywhere around the world, in rich and poor countries alike. It says that even if people are willing to bear some costs to reduce emissions (and experience shows that they are), they are willing to go only so far...

      So what we have is a physics problem that needs fixing, and economic system that will not allow for it. So what is adjustable? Well, if we learn what we are told, economics is an IRON LAW. You cannot stop economic growth (the political problem), an economics problem (current economic serves power over humanity and is therefore unchangeable), and a physical limits problem (which always harms the poorest and most vulnerable populations).

      To me, it is a power issue. It is the only element that is malleable. We cannot expect the physics the change, and we cannot expect the poor to suffer economic hardships. Solutions and ideas must favor the weakest (progressive carbon taxes! high luxury consumption taxes!) and transfer the power over economics away from those who own everything to those who don't.

      We did it during and following WWII for certain purposes. The law wasn't all that iron back then.

    • excellent post dan, especially this part:

      "So it’s a field of inquiry into human behaviour. MT, you’re talking about that in the same breath as economic policy, which is a bit like saying climatology is all nonsense because we don’t have an effective global carbon trading system. Think we need to be clear what we’re talking about."

      on a side note, I'm heading up to Ottawa this afternoon. anyone want me to pick up a Sens jersey for them? I'm looking at you Michael ;-)

    • danolner:

      Frank, this relates to your point about the 'real wide world', and you're right to say that's vital. One can make a realistic-looking model, so what? You've made sim city. You can also make a realistic model of atoms bouncing around in a container. Sometimes that's useful, but more often you can stick to models using temperature and pressure. "Everything depends on the problem."

      What if you were one day told that this "sim city" may open the way to solving many useful problems in economics? What if, for example, the naive models using "temperature and pressure" offer no way to study economic bubbles or predict them, and a more realistic "sim city" model might well hold the promise to do these things?

      Let's look at climate models again. Probably, once upon a time, some people thought, 'What's the darn point of creating realistic models of climate? Don't we already have weather prediction techniques passed down from the ancients?'

      But now we see that climate models allow people to tackle some really hard and really useful problems, such as (1) attribution of recent climate change (2) estimation of climate sensitivity (3) forecasting of future climatic states.

      (1) and (2) are particularly interesting, because to do these, the climatologist can't just take the output of climate models; he has to go back to the _latent parameters_ used in the models, and see which combinations of parameters produce realistic results, and which don't.

      And (3) is where the policy stuff begins. From model-based forecasts, we know that climate change is likely to be disastrous if we carry on our current CO2 trajectory. Surely this kind of result is useful.

      (Since, to model evolution, you need to have some notion of cost. [...])

      That looks like another misconception. One of the key insight of evolution theory is that mutations to genes happen _randomly_. It's natural selection -- through the simple mechanisms of survival and death -- that determines which genes make it and which don't.

      The point is, at no point do genes consciously decide which genes they'll mutate into based on some fitness criteria. It's all random.

      -- frank

    • "A jazz band falling down the stairs"... Awesome metaphor! Not, however, what I got from MT's comment.

      There is indeed a great deal of heterodoxy within economics, and more emerging all the time. And the problem is not that it doesn't get into econ 101 textbooks - environmental economics is now mainstream, and I'm sure behavioral economics is getting new chapters all the time. The critical problems are (1) what gets incorporated is dependent on being - or being seen as - non-threatening to the primary conclusions; and (2) econ 101 is almost all micro-economics and no macro-economics, and the problems of global sustainability are macroeconomic problems. We all know that if we raise the price of fuel or the disposal of GHGs we will get decreases in demand. We don't know how to organize a global system of production/distribution/consumption in such a way that production can be constrained and distribution/consumption made fair and sustainable (e.g., the problem of delinking income from employment), or more particularly we don't know how to engineer the transition from here to there.

  12. The unfortunate jazz band is not a terrible analogy. I'm expressing my discomfort with economics in a way that is analogous to how many uninformed people discuss climate science. I'm very aware of how wrong they are inmost of heir points, though not all. I am suspicious of my own suspicions by analogy.

    But I agree that such suspicions can easily be raised in good faith, and that is is often difficult to find coherent answers that don't smack of defensiveness. Again we are in a ralm where analogies are helpful. I know how easy it is to hurl accusations at climate science, and how difficult it is to balance accessibility, completeness, and fitness to purpose in the answer.

    Indeed, part of my point is that both disciplines and indeed any policy-relevant discipline, should be prepared to allocate an enormous fraction of their efforts in educating the interested public and in defending their respective approaches. The requisite patience is going to be hard to come by, but the required public communications resources - human skill sets and a business model for those exercising it - even harder because broadly unprecedented.

    And currently even the idea of such outreach is under attack in the form of. A glib and superficial analysis from social science alleging that "facts don't matter".

    Well, if we achieve nothing else at P3, let's try to build a community where facts do matter.

    I very much appreciate the efforts of those who are willing to take up my I'll-structured discomfort with economics and point to potentially useful readings. I think this is the whole idea actually. We won't attain to sustainability by winning arguments. We have to actually link knowledge across a broad range of ideas. There is plenty that is discouraging about that prospect but the good news is that it is amazingly interesting...

  13. Just-before-sleep-after-weekend-away comment: what's interesting about this thread is that there's actually quite a lot that's potentially offensive or provocative, but that it's also obviously committed, engaged and sincere. I think that's an important contrast to threads where it can appear 'provocative' and controversial, but it's just derailing and pointless. Lots of people here at P3 probably have the experience of trying to engage at e.g. WUWT, thinking that such a thread might become productive. Is it possible to tell, behind the text, what kind of thread it is? Are our heuristic devices good enough to tell one from the other before we've spent several hours arguing pointlessly?

    On the 'offensive' part, apologies for the jazz band analogy. I'd had only a few hour's sleep, woke up at 5, checked phone, wrote comment... lesson = don't write comments at 5am having just woken up! Thanks for the thread, MT. PB, I'll get you the proposal shortly...

  14. Sorry again for the essay-length comment, I aim for brevity and always fail! A nice Krugman example pointing to this interesting piece: the recent natural economic experiment appears to be coming down on one side. Now, I'm not anywhere NEAR knowledgeable enough to muster the appropriate data/model/analysis combo to test this for myself (I'm pretty macro-ignorant) but this at least highlights an opportunity to dig into what evidence and hypothesis-testing in economics looks like.

    MT, apologies, it's useless me just saying "go read so-and-so handbook" without adding some commentary about it. In this case it was just to highlight that a literature does exist - that in itself is sometimes interesting enough.

    Frank, I was actually going to add: there are plenty of roles for sim-city-type approaches, not least in allowing a user to learn how a particular system functions. If only educational games were better (though I haven't played any of the recent climate ones.)

    The quote I posted before from Farmer and Foley: here's the full Nature article. They say some of what you're implying I think. They even suggest the kind of model coupling seen in, say, ocean-atmosphere modelling and make the climate connection explicitly:

    "Creating a carefully crafted agent-based model of the whole economy is, like climate modelling, a huge undertaking. It requires close feedback between simulation, testing, data collection and the development of theory."

    My problem with this is that it sounds very persuasive: more 'realistic' models must be better, right? Again, that's the whole point Friedman's trying to deal with. What does 'more realistic' mean? How do you know whether an element you're adding is serving a purpose or creating unnecessary adhocery?

    I'm not very good with my complexity maths, but I'm reasonably sure there are sound reasons why no-one will ever have predictive power over certain elements of the economy. If we kept the same market structure for 2000 years, perhaps we'd be able to get solid aggregate stats on financial crises, but that's very unlikely to happen. Predictability in climate models comes through predictable outcomes at the macro level, regardless of the chaos underneath. (Still a point that skeptics attempt to exploit when they say, "predict the climate? Hah, can't even predict the weather in two weeks!") That said, I don't understand the role that adding more detail has in climate models for constraining outcomes, or sub-macro prediction work.

    Also: prediction is often assumed to be the only worthy model goal. We've mentioned another: user education. Epstein has a good list of others (though I don't agree with Epstein's 'generative' approach to modelling: 'if you didn't grow it, you didn't explain it'). The most often-overlooked one (not on this list, but sort-of-mentioned) is ontological prediction, e.g. the prediction of Neptune's existence, or Einstein's work implying light bends near the sun: predicting something we didn't think to look for, but was always there.

    Tying back to MT's point about economics-as-controller: the predictability question is important because, if the hope is held out that predictability might one day be met through 'more realism' then better control is implied. But what if that's impossible? What if it's the market structure involved that needs changing - the control elements? Now we're getting into cybernetics stuff, and e.g. Stafford Beer pointing out that our methods for controlling the economy are a bit like trying to steer a car with a 30 second delay between steering and the car actually reacting. If the system is inherently unpredictable and dangerous, either adapt to deal with that (build hurricane shelters) or, if it's not an actual force of nature (despite what some say; Network 1976 spoiler!) and we have some planning ability, redesign the fecking system so it's not dangerous! Start by making your data as close to allowing intrinsic control as possible. (Now I must be demonstrating my ignorance to MT, hope to be corrected...)

    Frank, on the last point about evolution and cost: mutations happen randomly, selection is not random (or not entirely). If it was, there'd be no evolution. Example: I've coded a genetic algorithm version of boids with predator and prey boids. They start off completely randomly reacting to each other's positions, so some prey leap into the jaws of waiting predators and some predators scarper from tasty food. They don't last long: their failure costs them their virtual lives. Successful genetic crosses produce better hunters and better runners-away.

  15. Reading another Krugman blog entry, this time on the possibility of Economist editorials being autogenerated. He quotes Henry Farrell:

    "I wonder whether even the writer of this editorial would be able to define 'reform' or 'change' if he were asked, beyond appealing to some sort of 'social protection bad, market good' quasi-autonomic reflex embedded deep in his lizard brain."

    Anyway, it made clear an obvious point I completely failed to emphasise: there would be no way for climate science to ever end up like economics. Entirely contradictory economic worldviews can co-exist in academia. All of them can get enough lifeblood to thrive.

    Yet, I wonder what might happen to climate science given another 50 years? It's not beyond the bounds of reason to imagine a number of states becoming so anti-science that currently marginalised contrary climate theory gains solid foundations.

    So, is climatology different from economics because physics is the ultimate arbiter - and economics doesn't have that arbiter? Or is it simply that climatology hasn't, up until recently, been exposed to the full, reality-shaping glare of political and economic power?

  16. Many of the same folks that mistrust climate models so much are the same ones who would extrapolate a simplistic economic model years into the future to predict disaster.

    In that vein, what is the GDP sensitivity to unemployment? To taxes?

    If an extrapolated empirical, uncertain compound growth model is good enough to make significant policy decisions, then the physics based models are at least as sound.

  17. David Deutsch, in his "The Beginning of Infinity", briefly treats the notion of a 'good explanation'. To be good, the explanation leaves little possibility of revision while still explaining the phenomena in question. I opine that by this criterion climatology is a good explanation while seemsly economic thought is not.

    I noticed the other day that in the University of Texas at Austin, the Department of Economics is a part of the College of Liberal Arts. I opine that is appropriate and certainly implies something important about economics.

  18. danolner, sorry for the belated reply...

    My problem with this is that it sounds very persuasive: more 'realistic' models must be better, right? Again, that's the whole point Friedman's trying to deal with. What does 'more realistic' mean? How do you know whether an element you're adding is serving a purpose or creating unnecessary adhocery?

    Again, people have thought about such problems before, in the context of other fields such as statistics, AI, and (!) climatology. See cross-validation. The idea is to partition the available data into a "training" set and a "validation" set, and then test the model computed from the "training" set against the empirical data from the "validation" set, and do that with different partitionings of the data. If I remember correctly, Mann et al.'s paleoclimatic work also uses some form of validation (though perhaps not cross-validation).

    The idea behind is that any model elements which are crucial to accurate modeling will also result in greater predictive power w.r.t. the "validation" set, while model elements which are _ad hoc_ will lead to worse prediction -- overfitting.

    I guess the conclusion to be drawn is this: economists may have thought about many of the big questions in economic modeling -- but many in other disciplines have already found ways to _solve_ these problems to some extent, often decades ago. Surely this means that economists should modify their methods in the light of such advances, instead of repeatedly throwing their hands up and saying, 'this is too hard'.

    mutations happen randomly, selection is not random (or not entirely).

    Indeed, but the key problem is that selection doesn't act according to some "utility" function which depends only on an organism's genes. Because a lot also depends on the environment: for example, a polar bear would have a hard time surviving in a rainforest. The way to tackle this problem is to model not "utility", but life and death.

    Thus, I continue to be mystified by mainstream economists' all-important quest to find some "utility" function that applies equally to everyone and will explain everything about buying and selling. To me, "utility" is at best a useless distraction, and at worst an obstacle in the way of truth.

    e.g. Stafford Beer pointing out that our methods for controlling the economy are a bit like trying to steer a car with a 30 second delay between steering and the car actually reacting.

    Actually I'd say is more like steering a car, when we have zero idea about how our steering actions will affect the car's movement (either now or in the future), and people try to fill in this gap of knowledge with quasi-religious doctrines. The idea, again, is that we should instead fill in this gap with some actual knowledge.

    -- frank

  19. Sorry, keep on coming here and reposting Krugman: a NY times column talking about the effect of wealth in the US on economic ideas.

    There are regularities in how people react to cost changes - enough, at any rate, to make it worthwhile thinking about how they work. But, again, Krugman's column shows what happens when a powerful section of society can megaphone their own version of reality through the body politic.

    Frank, on utility in particular: I started my PhD thinking the same as you and listening to a lot of agent modellers promising the moon on a stick. Realistic, reactive, built on plausible psychology blah. I've ended it completely seeing what the idea of utility is good for. It's not meant to be a 'realistic' take on human psychology, it's a neutral framework for thinking through how cost changes shift people's responses. E.g. the fairly unremarkable fact that people will tend to cut back on other spending first as fuel costs increase can be described in terms of elasticities. It's tautological, but then all theories are, by themselves. It doesn't stop it being very useful: if you know that increasing fuel costs will actually reduce spending in the rest of the economy more than it affects fuel spending, that's important - not least for working out tax revenues.

    It doesn't explain *why* those choices are being made, and for a lot of purposes it doesn't need to. It depends on what level of explanation you're after. Others might want to ask: well, is the inelasticity of fuel prices amenable to change in other ways, if we break it down? What role, for example, does the physical structure of cities have?

    Compare to another of my favourite utility-based findings: "savings in walking and waiting times are valued at between two and three times savings in on-vehicle time - parameters that have proved to be remarkably robust over the years." (Button, Transport Economics, p.104)

    So: more people will be - for instance - willing to drive for an hour each way to work, when they would never consider walking for that amount of time. Of course, one can come up with many theories to explain the underlying facts: anything from the structure of towns/cities/transport design to people's (relative) aversion to physical exercise (I commute about an hour each way, with some walking. I'm fairly fit, I run, but I'm not sure I'd want to walk to work for 2 hours a day...!) But *at a given level of explanation* it is very useful to know that people value transport time differently for different modes.

    So taking that simple example: if it appears that people will only walk or cycle for much shorter time stretches than under power, what can you do with that info? Several possible things. Find out the impact of physical/urban structure. Try and pull out the underlying factors affecting people's choices, separate from those structures. If you want to reduce carbon output, you could either berate people for being so weak and tell them they should drive less - or perhaps think about other ways behaviour might be amenable to change, given what we know about how people react to those changes.

    I'm not saying utility is the only way to think about people's choices, and certainly not that it's a description of the way people "really are". But it seems to be clearly a useful tool for thinking about cost change, either monetary, time or some other cost imposed by our environment.

    And just going back to the 'level of explanation' thing: "I have not been able to discover the causes of those properties... and I frame no hypothesis". That's Newton on gravity. He came up with what's probably the first exemplar of a scientifically robust, generalised theory, but understood that deeper levels of explanation awaited. We're still working on that.

    I think physicists actually have less of a problem with this difference between levels of explanation than economists, who tend to assume that whole collections of people act like a single person (the 'representative agent' idea). But we have to get comfortable with what level of explanation we're talking about. People are not utility-maximisers. Utility is useful for thinking about, describing and even predicting what impact people's reaction to cost changes will have. What's the problem? Pinning utility onto rationality is not necessary - any more than the theory of gravity requires planets to love each
    other.

    • danolner:

      People are not utility-maximisers. Utility is useful for thinking about, describing and even predicting what impact people's reaction to cost changes will have. What's the problem? Pinning utility onto rationality is not necessary – any more than the theory of gravity requires planets to love each other.

      I think you might have described the crux of the problem. The problem with the use of a "utility function" is that -- in actual practice -- it tends to become treated as The Function That Must Always Be Maximized No Matter What. Even Vriend's paper mentioned above, despite giving a nod to the idea of modeling reality, seems to take this view.

      It's possible, as you've done, to posit a weaker view, where such a "utility function" isn't the be-all and end-all, but merely another factor in the huge complex of decision making. But then it's no longer a "real" utility function -- as least not a utility function in the sense used in the mainstream.

      In fact this formulation becomes not that much different from more general problem in statistics of modeling dependent variables (decisions to buy and sell) in terms of input variables (observable prices, 'latent' properties of things, etc.).

      (In this sense, a "utility function" in the usual sense means that the dependent variables are assumed _a priori_ to be the arg max of some function over the space of decisions, if you get my drift.)

      I started my PhD thinking the same as you and listening to a lot of agent modellers promising the moon on a stick. Realistic, reactive, built on plausible psychology blah.

      Well, if it's any comfort, it seems to me that some of the most interesting paradigm shifts come from papers with such nondescript titles as "An inequality with applications to statistical estimation for probabilistic functions of Markov processes and to a model for ecology" (yep, indeed). They may not promise the moon, but they work well in their own quiet ways. I suspect that some of the good work in the area of Behavioural Economics may be of this variety.

       * * *

      Also, Krugman is right.

      -- frank

      • The Vriend thing is something I find muy simpatico, thanks, and it's almost 20 years old! Did this work actually go anywhere?

        In the work I do in tuning climate models, we know that the utility function is somewhat arbitrary. But we tend to find that for a given problem there is a more or less intuitive set of "reasonable" utility functions that give similar results. This would be very expensive to investigate with full-blown GCMs, but I think it is worth pursuing. You end up quickly in a horrible recursion of finding the utility function with the maximum utility if you think about it too hard. But that is okay.

        Suppose we find a way to enumerate people's preferences about what a just and sustainable world would look like into a set of objective measures. We might imagine as many utility functions as there are people, but let's simplify by clustering into, say, a thousand affinity groups of five to ten million people each. Then, conceptually at least, if we could model economic behavior in a rule-based system, we could see which set of rules best fulfilled the preferences of each group, and do some sort of constrained optimization.

        I do not seriously propose that this problem is practicably computable in time to decide the serious problems of our century. We will have to make educated guesses at the answer. But I do seriously propose that we conceptually model our decision-making in this way.

        I think this is William's valid point in the discussion we are having over at Stoat. We need some sort of objectivity, and that means a utility function. No cost metric, no cybernetics. That is the long and short of it.

        Oh, and Krugman is wrong, because we cannot grow our way out of our problems forever.

    • "Oh, and Krugman is wrong, because we cannot grow our way out of our problems forever."

      Well, I disagree, we can, it's just that the results aren't favorable to most people. Capital can be remade, reshaped in many different ways. Think gentrification. Think the destruction of the labor class. When capital cannot make its 3 percent compound it immediately goes into crisis. It looks for different ways to grow, different places to put the surplus capital and this must keep happening constantly. And it finds ways. Always. Even Marx marveled at its awesomeness. It just sucks when ethics, humanity, environment, labor, land, etc get in the way, as those are not concerns for capital.

      The sooner we get a hold of these notions the better off we'll be.

  20. I haven't read all this, sorry. But I'm stuck at:

    > it claims for itself a unique position among the sciences, as the crux, the central weighing mechanism, for all public decision-making.

    You appear to be objecting to that (and other stuff, like dollar-denomination). I think you need to clarify your position. You're (a) rejecting the idea of having a weighing mechanism, or you (b) reject the idea that economics is it?

    I don't think (b) is plausible - economics is, pretty well by definition, the weighing mechanism.

    So I'll assume you mean (a). I think that's a bad choice - the only alternative to a central mechanism is a decentralised mechanism, where everyone uses their own prejudices. Which is the bad bits of the current political mess. I think you want more economics, not less, but done properly.

    • I don't think (b) is plausible - economics is, pretty well by definition, the weighing mechanism. Assuming you believe in one.

      Of course there is another option: practical reason, phronesis, wise judgement. Not every judgement can be translated into quantities. Sometimes, careful conceptual analysis and the exercise of practical reason are paramount. Indeed, insofar as economics is of use, it is as a tool of practical reason, to help calculate those aspects of public deliberation that are indeed amenable to calculative reasoning.

  21. Hopefully you don't have a really snappy answer, cos I just wrote http://scienceblogs.com/stoat/2012/05/economics_and_climatology.php

  22. Pingback: Economics and Climatology? : Stoat

  23. Pingback: Another Week of GW News, April 22, 2012 [A Few Things Ill Considered]

  24. Couple of questions.

    1. What tools should we use to assess whether continued co2 accumulation is a risk to human society? What tools can we use to assess the extent of the risk, and what our options are? Can we do all or most of that with physics alone? If not, what else?

    2. Presuming we can get answers to 1, what tools should we use to then change co2 accumulation? (Note: if the answer's that we need net zero human carbon output, that still leaves us needing to get to net zero. Again, what tools?)

    Here's a slightly related quote - in that what I'm implying above is, we need some form of modelling, however diverse, however cybernetic, however threaded through organic human institutions. Neary talking about abstract models used in economics, and might apply to some physics/engineering/logistics/statecraft too:

    "They are, as the Roman poet Propertius said of his lover, difficult to live with, impossible to live without."

    (Neary, J.P., 'Of Hype and Hyperbolas: Introducing the New Economic Geography'. Journal of Economic Literature 39, no. 2 (June 2001): 536–561.)

  25. Dan, thanks for this.

    I also appreciated this list, especially as it got past the merely descriptive and into the prescriptive around the fourth paragraph. I remain a bit befuddled by whether there is a universally accepted definition of "economics", though, and this only exacerbates it. After all, you say

    ""Economics" is the study of how people react to cost changes."

    a very modest claim indeed (presuming you mean 'cost' in the colloquial sense of money). And yet the list of questions by a founding economist to which you happily refer proffers questions such as "How should we act so as to increase the good and diminish the evil influences of economic freedom, both in its ultimate results and in the course of its progress?" clearly you are asking about something different than this narrow psychological question to which you may propose to attach a simple empirical formula of some sort.

    (which in my experience usually reduces to Butter has so much marginal utility and Guns so much, so the equilibrium production (the problem is generally grossly simplified by an assertion of convexity) is where these two lines coincide, but never mind whether that is amazingly silly or not for now.)

    I am interested in whether we know to what extent the free market can satisfy our needs. I am reasonably convinced that it will continue to systematically destroy natural capital. I suspect that it ends up destroying natural capital at a rate comparable to the discount rate, so that every generation we have huger bank accounts and live in a substantially impoverished world, but I don't have a theorem handy. I'm pretty sure the discount rate has something to do with the rate constant though.

    The reason Stern asks for a vanishingly small discount rate is because he doesn't have a theory for a zero or negative discount rate which would correspond to a shrinking economy. But the lack of a theory in no way reveals the lack of a consequence. At some point the world will be so naturally impoverished that society will be fragile, and will break.

    This seems to me just an instance of something that is ambiguously an economic question, or a meta-economic one, or an extra-economic one depending on whether you choose a narrow or a broad definition of the topic. I seem to be getting in trouble on nomenclature. It matters little to me whether I ask questions that economists have long thought about, or questions that economists have dismissed, or questions that economists think outside their bailiwick.

    They certainly are related to how the world, including the money part, can be structured to keep as many of our current advantages as possible without actually systematically robbing future generations. Let's think about that. I can use whatever words you like, just stop weaseling (sorry William) about what they mean or don't mean.

    I like these questions:

    1. What tools should we use to assess whether continued co2 accumulation is a risk to human society? What tools can we use to assess the extent of the risk, and what our options are? Can we do all or most of that with physics alone? If not, what else?

    2. Presuming we can get answers to 1, what tools should we use to then change co2 accumulation? (Note: if the answer’s that we need net zero human carbon output, that still leaves us needing to get to net zero. Again, what tools?)

    I think these are among the questions we should answer. At this point I see that I need to slow down and make my points one by one.

    But CO2 is not the whole problem. I propose that we cannot keep discounting the future without destroying it. Some countervailing pressure needs to be baked into the system, perhaps ethically, perhaps as legal proscription, perhaps as incentives. Things which can be used up cannot be discounted. I don't know how to make this happen realistically. But it seems obviously both necessary and absent.

    As for this: "One can make a realistic-looking model, so what? You've made sim city. You can also make a realistic model of atoms bouncing around in a container. Sometimes that's useful, but more often you can stick to models using temperature and pressure. "Everything depends on the problem.""

    Well, as a climate guy I've thought about this class of question a lot.

    I would say that just as the existence of convincing CGCMs indicates that our knowledge of the climate system is mature, the lack of convincing agent models of economics indicates that our knowledge of economics is premature. As does the existence of numerous schools of thought. Unfortunately, we do not have the luxury of shrugging and walking away. We will have to monkey with the system in a fairly significant way in the best of scenarios. So we need the best theory of long term economy monkeying we can get, and as far as I know, we haven't got one.

    Some people seem to have strong ideas about the short term, and I find myself squarely in the Krugman/Stiglits camp insofar as that goes. But I find their reasoning doesn't eliminate the consistent lemming march of resource squandering and nature-mining.

  26. Krugman featured your graphic on media bias (nb, May 12th):
    http://krugman.blogs.nytimes.com/2012/05/12/whats-really-scary

    My apologies, wasn't sure where to put it, but I think this is great news, and I hope it brings him to Planet3.0!

  27. > The reason Stern asks for a vanishingly small discount rate is because he doesn’t have a theory for a zero or negative discount rate which would correspond to a shrinking economy

    Are you sure? It sounds to me like you've just made that up.

    IMO the reason Stern uses a small discount rate (rather than a realistic one) is because he needs a very low rate, in order to generate high future costs that weigh, now.

    And... all the numbers for Stern are predicated on increases use of fossil fuels. This only happens in a growing economy - it makes no sense otherwise. With a static or shrinking economy there is no problem for Stern to solve, so assuming a static economy would make no sense at all.

    • The lower the discount rate, the more valuable a resource is in the future compared to the present.

      It basically takes a market-centric view, that everything is exchangeable for everything else, and that therefore a single discount rate applies, which is the same as the anticipated growth rate. A resource worth a dollar in ten years is only worth V = 1/((1+G)**10) < 1 now because I could "invest the V now at a return of G and buy it ten years from now for a dollar". This is pretty basic stuff. The lower the discount rate, the more we care about the future. Which is exactly Stern's trick.

      But Stern doing a trick? Or is it a weak compromise between the tools he has at hand and the real result?

      If growth in emissions and gross economic activity now leads to an actual economic collapse (say an abrupt reduction to 50% of present activity) in fifty years, the discount rate over the fifty year period would be negative. That means everything in the future is worth more than anything now, and one should stop spending altogether, which of course is silly. The discount rate is a key to many aspects of economic theory and practice, and all of it breaks if it goes below zero. That is why Stern made it very small - to get as near as he could to the right answer.

      As for your point that increased energy use only happens in a growing economy, that is not exactly right. It depends on whether you measure energy before or after it is shipped from the energy producer. As EROEI gets worse, upstream energy can increase even for otherwise static economy. But that's a minor point.

      Energy can conceivably grow for a while and then be followed by a collapse which that energy growth caused. The discount rate is the wrong tool, because the discount rate is not necessarily independent of the climate problem and is not necessarily positive. Small signal thinking again.

      Finally, this "With a static or shrinking economy there is no problem for Stern to solve," is surprisingly naive about the climate problem coming from you. You surely know better. With a static economy based on fossil fuels even at fixed or improving EROEI, the climate problem keeps getting worse. The economic impact trails the emissions by a good long while and is almost entirely dependent on cumulative emissions. This is another reason that conventional economic thinking falls so completely flat on the question. We care almost nothing about the emissions rate per se. We care much more about how long those emissions are maintained. The usual methods cannot account for that, I think.

  28. > The lower the discount rate, the more valuable a resource is in the future compared to the present.

    Yes. That's what I said. That is why Stern needs a low discount rate.

    > therefore a single discount rate applies, which is the same as the anticipated growth rate

    I don't think so. It might be equal to the investment rate, but I don't think that's the same thing. And anyway, you've left out the discounting-uncertainty bit.

    > It basically takes a market-centric view, that everything is exchangeable for everything else

    Yes, I think that's right. Rather like the laws of physics, as we write them, always assume that the laws of physics have a mathematical form. What else can we do?

    > If ... leads to an actual economic collapse... the discount rate over the fifty year period would be negative... That is why Stern made it very small – to get as near as he could to the right answer.

    Err hold on. You've jumped from an *If* - which many would disagree with - to the bald assertion that a negative discount rate is the right answer. You can't do that.

    I find it very confusing talking to you in this mode - you have a whole pile of what appear to me to be hidden assumptions, many of them, in my view, implausible. It would be a good idea for you to write them all down in a nice numbered list, so we all knew what they were.

    > surprisingly naive about the climate problem

    I was basing my assumption on the std climatology: that we don't get large, significantly large, temperature and/or climate shifts, without some pretty heavy increases in fossil fuel use. If CO2 emissions were to stabalise at todays levels, the climate problem becomes much less interesting.

    • "If CO2 emissions were to stabalise [sic] at todays levels, the climate problem becomes much less interesting." How so?

      See fig 22 here: http://www.ccrc.unsw.edu.au/Copenhagen/Copenhagen_Diagnosis_LOW.pdf e.g.

    • OK, "That is why Stern made it very small – to get as near as he could to the right answer." is a bit glib.

      The point is that we cannot reasonably assume continued growth into the future if we try very hard to break the planet now. Ray P says he is not convinced that very high sensitivities are excluded. And that does lead to very nasty scenarios. And the nasty scenarios should dominate the risk-weighting in matters like this. Which means the case of a reversal of long term growth cannot be eliminated.

      Look, if something meaningful is guaranteed to grow, we don't have a sustainability problem, do we? To say that "negative is the right answer" is too strong. But to say that negative can be neglected simply begs the question. Now how do we average across cases where the theory works and cases where the theory breaks down?

    • We have to get to zero net emissions as quickly as feasible. Stabilized is better than increasing but nowhere near adequate. This is generally accepted; it's not as if I were saying anything outside the conventional wisdom on that score. I honestly don't know what you mean by "interesting". The correct target for carbon emissions is, as Caldeira says, the same as the correct target for the number of little old ladies being beaten up and robbed in the street. Is this really news to you?

  29. The discussion would be better if one abandoned standard economic thought. I consider that highly defective, largely but not solely due to mt's writings. However "The correct target for carbon emissions is, as Caldeira says, the same as the correct target for the number of little old ladies being beaten up and robbed in the street" isn't good enough. What is needed is net negative carbon emissions until atmospheric CO2 is below some value to be detmined, but I opine is less than 300 ppm. That's about 500 gigatonnes of carbon (not CO2) to be reburied.

    Doing so requires not only net zero carbon emissions but a burial plan, such as
    Irrigated afforestation of the Sahara and Australian Outback to end global warming
    http://www.springerlink.com/content/55436u2122u77525/
    with the pdf free to all.

    • The less than 300ppm value is the lowest I have heard anyone claim is what is needed. Hansen for example says we should aim for less than 350ppm, and many think his value is much to low.

  30. Apols MT, cheers for the comment, I'll get onto that. But - William: "If it isn’t obvious that the problem is much less interesting with current emissions than with 1%/y growth, we have indeed grown far apart."

    People better at sums than me will hopefully confirm/correct, but given co2 ppm growth rates, stopping at what have now would only be a slight improvement. In 2011, yearly ppm was 391.57. From NOAA yearly data I get this ten-year average yearly change in ppm. So we're currently lurking around 1.8 to 2ppm a year. At 1.8ppm, that's a round 30 years for 450ppm; 57 years for 500; 85 years for 550ppm. The change in the growth rate since 1970 has been (OLS through the 10 yr av) +0.0241ppm/+0.0052% per yr...?

    Maths-wise, the growth-rate is more than exponential I think: from 1959 to 2011, the rate was e^0.00412533; from 1990: 0.00475615; from from 2000: 0.00526904. Assuming it stays at the 2000-2011 level, that would mean: 26 years for 450; 46 years for 500; 80 for 600ppm. Up to the end of the century the difference starts really showing itself. Earlier on, though, either path looks pretty, um, interesting. We're talking about the difference between 460 and 478ppm by 2050 - neither of which is very appealing.

    I'm not sure what dept these questions should be asked in, but that does highlight the two sides of the problem: what societal impact would those two paths have? How do we decide what an 'appropriate' target level of ppm is - or is that the wrong question? I agree with MT, it's obvious we need at least zero net emissions. Assuming that, how do we go about making it happen? Or, rather, is there actually any way of making it happen? Are we facing the equivalent of trying to stop a tanker by rowing up to it in a dinghy and pushing?

  31. I think I was comparing the "standard" of 1% growth. Does

    http://www.wolframalpha.com/input/?i=plot+y+%3D+391.57+%2B+1.8x%2C+y+%3D+391.57+*+%281.01%29%5Ex+for+x+%3D+0+to+88&dataset=

    work I wonder? That looks to be about a factor of two difference in CO2 by 2100. Is that even the right calculation? Certainly not in detail: the effect comes from CO2 levels, but emissions get affected by ocean and land uptake (or emission).

    But overall, saying "much less" may have been overkill on my part.

  32. Also, just some economics brainfood/assumption-checking. Via Dani Rodrik from a few years back: NYT on breaking economic taboos. It mentions heterodox economics; a related association (I get its newsletter but nothing has really jumped out at me; I find the one-foot-in folk like Krugman and Rodrik making better arguments.)

    Anyone from a social science background (at least in Europe) will know that economics is seen a little more than the priesthood of neoliberalism. That's where I started out. These can range from detailed, sophisticated critiques to kneejerk left wing rejection - but at any rate, there is no shortage of analysis. There's a rich economic sociology; see e.g: Granovetter & Swedberg. The Sociology of Economic Life. 2001, which includes classics like Geertz on the Bazaar Economy, which is the stuff I like: studies of actual markets. This stuff starts from the assumption that 'economic institutions are social constructions' and that "while interests are central to any explanation of economic activities, a purely interest-driven model is unacceptably distorted" (Granovetter/Swedberg p.9).

    I wonder if many physical scientists are simply unlikely ever to see the value in anything this broad? As with this comment, perhaps disciplines are ranked in order of their r^2 values. I know of a cognitive psychologist who sets up teaching to explore the "idea of research as a process of engaging with uncertainty, rather than as something which produces authoritative answers". This is the problem any analysis of social systems faces.

  33. Relevant to the discussion of discounting, some quotes from Stern's 'Blueprint for a Safer Planet'. Reads to me like 'economic modelling as a guide only to social action, negotiated and discussed openly, using all of our ethical wherewithall.'

    “The aggregate modelling of the stern review was presented in just one of the 13 chapters on mechanisms, impacts and targets that made up the first half of the review. In the other 12 chapters the probabilistic approach to outcomes and impacts were set out in some detail, and the costs of different kinds of action were examined. Nevertheless, chapter six attracted a lot of attention because it provided an aggregate statement of costs of action vs. costs of inaction.”

    “... It is crucial to emphasise, however, that the precise quantitative results of such models should not be taken too seriously. They are very sensitive to assumptions and leave out altogether crucial details such as conflict and different ethical approaches, and are usually week on the treatment of other issues such as risk or biodiversity. As the modelling section of the review concluded, ‘we therefore urge the reader to avoid an over-literal interpretation of these results. Nevertheless, we think that they illustrate a very important point: the risks involved in a business as usual approach to climate change are very large.” [93-4]

    On discounting:

    "This kind of issue cannot be settled by tight logical argument and evidence, but there is absolutely no excuse for avoiding them, as they are central to the challenge of formulating policy on climate change. Openness and clarity of discussion is vital and economists do themselves and their profession a disservice by not confronting them. We can, indeed, talk about these issues in ways that are helpful for suggesting which values might be appropriate in making social decisions. One way of doing this, which moral and political philosophers continually use, is the ‘thought experiment’... If we take simple circumstances which we can understand and then cross question ourselves and each other, we can make progress in clarifying these issues." [84]

    • > there is absolutely no excuse for avoiding them

      I'm surprised to see Stern saying that - my impression (http://scienceblogs.com/stoat/2006/11/nordhaus_on_stern.php) was that there was little discussion of discount rates, and avoiding the issue was pretty well what he does. My belief was that he just uses a near-zero rate; he doesn't give (which he trivially could) a re-run of his models at different discount rates.

      • This is only half-true, and this is my point. The models break down at negative discount rate. He chooses epsilon not because it is the right value, but because any acknowledgement that climate change could dramatically shrink the economy implies that a negative discount rate is the correct one, and the machinery of economics cannot handle that. The models would simply fail.

        Growth is built in to economic modeling. The presumption that this growth is actually desirable is taken for granted, but is not obvious. But even if it were desirable, I suspect it is the wrong way to think about valuing the services provided by natural systems. My suspicion is that discounting the atmosphere makes numerous extrapolations about the applicability of economic theory that are (as far as I know) nowhere justified and that are (I would guess) not justifiable.

        Of course, as I've said for decades, if, in extremis, the atmosphere were somehow to become unbreathable, this would be an excellent stimulus for oxygen tank and gas mask manufacturers. But it wouldn't be a better place to live.

  34. Dan Moutal --- The relevant Hansen et al. paper actually stated "... 350 ppm and probably less" as a maximum. I choose less than 300 ppm to avoid ice melt and rebuilding some of the glaciers. The exact value will surely be contentious and the climate modelers will probably want to have their say as well.

  35. MT, finally getting back to your comment:

    Me: "Economics is the study of how people react to cost changes." Hmm, you're right - I need to be more careful with nomenclature myself. I'm sure some of the heterodox econs would say the term is 'essentially contested'. Thinking about it, though, I think my def still works: economics concerns itself with the costs and benefits involved in human choices. I don't see that Marshall's questions contradict that: everything from how we structure institutions, what sectors should be state-owned, the ethics of wealth distribution - is all covered.

    "presuming you mean 'cost' in the colloquial sense of money". No, absolutely not. The magic washing machine is a pretty perfect example: what are the costs/benefits of how we use our time? How does technology and societal structure alter that? Rosling very cleverly illustrates precisely this point by pulling books out of the washing machine: it's a machine that produces women's education as well as clean clothes. I see no problem in thinking about time this way while also thinking about how time is socially constructed (see this classic E.P.Thompson article, PDF) and how economic definitions themselves may alter us.

    The transition from a mostly agricultural society to what we have today can be thought of in the same way. Two things have happened simultaneously: agri technology has improved, meaning waaay less people can produce massively more (put aside for now 'but it's all just eating fossil fuels'...) The rest of the economy has grown in a feedback process: enabling both increased agri output and freeing up people's time to work elsewhere. As a result, we've also seen a massive morphology change as food processing has moved out of the household/community and into today's sophisticated global industrial networks. A key part of that is how people value their time: we could all be growing food in community gardens and cooking it at home. We have the time to do that. Mostly, we don't. Why not? We prefer to work in paid jobs and access relatively cheaper food, as well as a set of other things we like. Computers, electricity, transport, beer, time to sit and stare at the wall...

    Again, there's a whig history danger here: it was meant to be thus, and is natural and good (while forgetting small matters like kicking people off their land when sheep became more profitable, much as we're doing now because car-food is more profitable than people-food [my blog] in many places). But there's also a lot of value in thinking about these changes through the prism of how we value the costs and benefits of our time.

    I've found myself looking at my own 'revealed preference' and changing my views. I used to be a lot more fervent about local food growing, until I realised what my shopping habits were telling me: I actually prefer to earn a living in academia and spend time I'd be putting into agriculture on other things, like commenting at P3. If other people feel differently, fine. But I don't think the existence of supermarkets is necessarily a sign of collective moral failure. I also reserve the right to a) reflect on that and change in the future but b) not to have anyone else actually force me to change, unless I've taken part in a democratic process to enforce it ->

    Cos maybe supermarket are evil, and individually we're too vulnerable. Marshall lists this in his questions: "what are the proper relations of individual and collective action in a stage of civilization such as ours? How far ought voluntary association in its various forms, old and new, to be left to supply collective action for those purposes for which such action has special advantages?" In the case of supermarkets - and some other market structures - perhaps we should not trust the emergent result of all our collective value-judgements. Instead, maybe we need to get together and decide a set of constraining rules: those we agree are needed, but that we recognise individual actions will tend to corrode over time. That would be democracy. It's also why people who claim that money represents the zenith of democracy [me again] are talking nonsense. If individually we are incapable of making the right carbon choices, collectively we can decide to restrict our choice set.

    Um, I appear to have only got as far as your first sentence, I'll go away for now...


Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>