September 22, 2008 | Graham

Sub-prime and climate change

Is there a link between the demise of Lehman Brothers and global warming? Jennifer Marohasy certainly seems to think so, but she doesn’t say why. There is a spate of theories from her commenters.
I think that there is a link, but it’s not specific to Lehman Brothers, but rather systemic, and it has to do with computers and modelling.
When I first started in finance in 1983 I was issued with a Hewlett Packard HP12C calculator. This nifty little beast, which I still own, was programable, although we mostly used only the 5 or so buttons needed to work out lease rates. But this was before personal computers, so for most investors and developers cashflows were still laboriously produced using pencil and paper. Changing one variable meant rubbing out a lot of spreadsheet, so you made sure your assumptions were very robust.
Personal computers changed all of that. A few years later I was programming Lotus 123 spreadsheets to automatically predict development profits in which I could tweak the assumptions ad nauseum. I wasn’t the only one, and pretty shortly no valuation was complete without a discounted cashflow, with sensitivity analysis.
At the same time, as a result of the 1987 sharemarket crash, New York stockbroking firms were going one step further and hiring physics PhD graduates to construct computer programs to predict the direction and magnitude of stockmarket prices. These graduates shamelessly took their millions, even though they should have been aware that the task was the philosopher’s stone of sharemarket investment.
In the real estate investment and development industry computer models never really took over. Valuation practice meant that valuers had to check their calculations by using at least two, and preferably three methods for comparison. Cost of construction and direct market comparisons didn’t negate computerised discounted cashflow models, but they did mean banks wouldn’t lend to you on the digital blue-sky valuations. The models might be right, but few lenders were prepared to risk their shirts on them.
I know I soon realised that if it didn’t work on the back of an envelope, then making it work with a computer program was very dangerous.
The same thing can’t be said for equity and credit markets, where asset pricing models for risk have taken over at the large ticket end of things. Which brings us to the sub-prime mess.
Even though a cursory explanation of how the mortgage packages were structured sounds daft, the models said that they were fine. GIGO (garbage in garbage out) is the technical term for this. And the models were so complex, and the products they were used to produce so opaque, that no-one really knew the full risks of what they were “investing” in.
And at the bottom of the pile, making all of this possible with abstract computerised models, were undoubtedly a lot of physics and maths graduates.
Which is pretty much where we are with climate change. There are some “back of the envelope” models of what happens with increases in CO2, like this one. Or slightly more complex ones like this (thanks to Jennifer for both the links). Neither requires a computer, just a pencil and a sheet of paper, plus some advanced maths. While both predict temperature rises, there is nothing too scary about them.
To get the scary temperature rises requires GCM (general circulation models) which run on computers and are programmed with positive feedbacks. Like the models under-pinning sub-prime mortgage securitisation, you have to be a specialist to get a good handle on them, so most decision makers just trust the programmers, because, as they have PhDs in physics and maths, they must have sound judgement, musn’t they, just like those graduates programming the assets risk allocation models? Whoops.
Which is where Lehman Brothers legitimately comes into the argument. It didn’t go broke because it ran a program promoting carbon trading. They went broke because they, and many others, relied on smart people to construct models which ultimately didn’t reflect the real world. Undoubtedly many inside Lehman Brothers and the other Wall Street firms knew that they didn’t, but there was a conspiracy of silence. No-one wanted to risk this year’s bonus or next year’s promotion by pointing out that the only thing transparent about these opaque models were the emperor’s clothes!
So it is with global warming. Other methods of evaluating the risks, like reviewing the historical record, suggest that nothing that is happening at the moment that can’t be accommodated, but the models say otherwise, and the weight of money and ambition has gone behind the models. As a result, we in the West are caught up in an environmental bubble economy where everyone is spruiking cimate change.
Which is where Lehman Brothers comes in again. While these phenomena can persist long past their natural life, in the end, they burst. It’s costing perhaps a trillion dollars to clean-up the subprime modelling mess in the US. How much will it cost to clean-up the AGW modelling mess in the world?

Posted by Graham at 10:20 pm | Comments (17) |
Filed under: Environment


  1. I’m no expert, but I imagine that in order to test your theory you’d have to pick a random sample of cases where modelling has been used and work out a success/failure rate, wouldn’t you?
    Considering our reliance on modelling for just about everything that seems like a worthwhile exercise.
    If your sample was made up of a range that didn’t include Lehman Bros and climate change, what do you reckon the outcome would be?

    Comment by Lyn — September 23, 2008 @ 8:43 am

  2. Hi Lyn, I don’t think you need to test anything with the subprime mess – it’s obvious that the models failed to assess risk appropriately.
    I’m also not saying that models can’t be a good thing, but they need to be tested against alternative ways of evaluating things. I’m also saying that simple models are likely to be just as accurate as complex ones, but they have the virtue of being easier to understand, and therefore easier to interrogate.
    With climate modelling my suspicion is that all the increasingly complex models are giving us is a false sense of security that we know what is going to happen in the future, when the simple mathematical models are probably just as accurate.

    Comment by Graham Young — September 23, 2008 @ 9:05 am

  3. “With climate modelling my suspicion is that all the increasingly complex models are giving us is a false sense of security that we know what is going to happen in the future”
    I’m glad you admit it’s a “suspicion” that confirms that you have no evidence, have done no reasearch and have no expertise on which to base you comments. Perhaps you could include these caveats in you next anti-AGW piece?
    I find this kind of comment particularly galling. The assumption by the anti-AGW faithful is that there is no emperical research informing the modelling, which is just bollocks. Models are projections of the real world not contructions of alternative realities, that’s the province of anti-AGW.

    Comment by patrick_b — September 23, 2008 @ 9:26 am

  4. “Suspicion” is exactly the right term Patrick. The modellers don’t even know whether their models are producing relevant results, which is why the IPCC says they are not forecasts.
    If they were forecasts, then they would fail because they haven’t predicted the last 10 years of temperature plateaus.
    I suspect, that word again, that I have a lot more experience modelling things than you do. And that experience says that the more complex your model the more likely you are to make mistakes. Sometimes the mistakes cancel themselves out, and sometimes they shake the model to pieces.
    A “check valuation” on AGW modelling against the real world says that a temperature rise of 2 degrees is well within natural variability in the last 1000 years, therefore not threatening to life on earth. To get much more than that you have to postulate positive forcings, but as they haven’t existed in the past, what is the basis for thinking they will exist in the future?
    Some modeller’s suspicion?
    The biggest problem in this area is uninformed people like you who try to make out that the speculations of some scientists are in fact certainties, and to then suppress any dissent from these speculations.

    Comment by Graham Young — September 23, 2008 @ 9:46 am

  5. Blind faith has never served the human race well – be it theological or secular. Our blind faith in the capacity of models to give us a reliable glimpse into the future makes climate change more problematic than it need be.
    All we need to know is that there is a finite possibility that an increase in C02 in the atmosphere will create problems.
    We do not need any modelling however to work out that the assumptions underpinning economic theory are intellectually bankrupt. All one needs is a basic understanding of the impact of exponential growth. Climate change is merely a manifestation of a far deeper problem – our failure to ensure that those physics PhDs factored in the second law of thermodynamics in their models.

    Comment by John Tons — September 23, 2008 @ 9:46 am

  6. Hi Graham
    I agree with the general argument, but not the specifics. You are comparing valuation models to general circulation models. It would more appropriate to compare general equilibrium models to GCMs. The valuation modelling process has been abused and thier users have been sloppy. But valuation models have an external reference point. In a GEM the modeller can make up their own whole economy without necessarily referencing to the external world. So too in a GCM. So, I think the problem is even worse than you suggest.

    Comment by Sinclair Davidson — September 23, 2008 @ 10:33 am

  7. With climate modelling my suspicion is that all the increasingly complex models are giving us is a false sense of security that we know what is going to happen in the future, when the simple mathematical models are probably just as accurate.
    Graham, you said in the post that modelling is the problem with both the sub-prime thing and the climate change thing. That modelling itself, or complex modelling, is the problem. So it seems to me that since we’re reliant on complex modelling for just about everything, it would be sensible to investigate modelling generally.
    If your suspicions on complex modelling of climate change is right, then it would apply to modelling of a lot of other things, wouldn’t it?

    Comment by Lyn — September 23, 2008 @ 10:42 am

  8. Lyn the point that you make about being reliant on modelling is precisely the problem. There is nothing wrong with modelling per se, it is an effective tool to use when investigating possibilities. The problem is that it is used as a substitute for thinking by people who do not fully understand the nature of modelling. (and to be fair even some who should know) But I guess that is politics.

    Comment by John Tons — September 23, 2008 @ 1:15 pm

  9. I agree with John. I don’t have a problem with modelling per se, but it has its limits and should never be a substitute for thinking. You need to have systems which check the models, and you shouldn’t just assume that a model is right.
    Much theoretical physics is really about building a model to explain the universe. But in good science they then go away and construct experiments to test the model. The Large Hardron Collider is an attempt to check the model of the universe that we now use to understand our world. If they don’t find the Higgs Boson, then we will know that the model is not as good as we thought. That’s a $7(?)Billion investment in checking a model.
    What is strange about Climate Research is that there appears to be little research money being spent robustly putting the models to the test. When someone comes up with an objection, they are treated like a crank.

    Comment by Graham Young — September 23, 2008 @ 1:45 pm

  10. A good observation Graham.The Merrycans[Bush speak] are no longer the happy can do nation.It will take years to sort out this financial mess.
    Things must be really tough there.NSW has been in the doldrums for a few yrs now especially in the building industry.Guess what,the morons want to tax us even more to pay for their mistakes!
    The big challenge will be bring confidence back so people will again feel confident enough to spend.Otherwise,if this continues,AGW will be the least of our worries.

    Comment by Arjay — September 23, 2008 @ 7:04 pm

  11. It is commonly understood programming that acknowledges that in complex programs there are going to be ‘latent’ (compatibility) bugs may only occur in singularly rare and unconsidered (untested) circumstances even then they may remain “undetected” until the ‘right’ conditions. There are a number of historical incidents that a small bug caused a cascade of failures or aberrant errors.
    The more complex the data input, the number aggregated inputs in place of missing data (temperatures, air pressures etc) et al the more the ‘model’ must tend towards GIGO.
    Add to this the choice of language, programming style, nature of the programmer’s individual “took kit routines” and assorted compatibility issues add to the vulnerability of the model.
    Despite this models (and statistics) do serve very useful purposes, if only to provide tools for specialists in particular fields.
    To me the problems seem to be:
    • Incompatibilities of comprehension of goals of the program between the designer/programmer/user (different domains)
    • Programming can’t compensate for unknowns with precision.
    • Climate dynamics are precise not an amalgam of “near enoughs.” Our knowledge of the impacts of all related issues is still far too embryonic to be able to claim ‘absolutes’. Ergo models are indicative not predictive. Notwithstanding this the evidence of global environmental stress is obvious.
    • Comparisons between valuing models and climate models are invalid for a number of reasons (some technical) except in a theoretical context.
    • It seems to me that there is sufficient evidence that our current modus operandi is both short termed (exploitative pejorative meaning) and ultimately limited. Therefore reliance on an incomplete understanding of “climate dynamics” and our scientific ability is unwise at best.

    Comment by examinator — September 25, 2008 @ 12:02 pm

  12. Hello Modellers,
    I firstly wanted to mention a concept I found in science fiction. Time travelling and guarding against committing a ‘minimum neccessary change’.
    That is mistakenly interfering enough with history so that the whole path of history would be irrevocably changed – So would smothering Hitler at birth change history – or would all you needed to do was hold him up for fifteen seconds at a street corner so that he never met Hermann Goering??
    A slick concept and worthy of cogitation.
    However all you guys have been talking about modelling – or believe you have.
    Actually you’ve been talking about planning and mapping abstractions. Don’t get me wrong. Nothing wrong with that.
    But a model is something created in three dimensions that exists and operates in the fourth dimension of time. Depending on how well the model is designed it may emulate the thing that it is modelled upon reasonably well or be a complete flop and either refuse to begin working, work partly successfully but not sustainably – or operate well but in some fashion differently from that object it was intended to emulate.
    Hey. Just ask several generations of schoolboys who started out making things fly with rubberband power even before the Wright Brothers got into the air. Many of these schoolboys are old farts now and still playing with their models.
    Models that have become so complex and sophisticated that they require special certification and specially alloted airspace in order to fly.
    Why do I mention this?
    Simply to humbly suggest that to ‘model’ Great Gaia in all her beauty and enthralling complexity we risk stuffing up immensely if we try to modify her systems when we only have a few scrappy blueprints – when we completely lack the skill or means to make a true ‘model’ – when even if we tried the model would be, like those toys of a century ago, completely unguided and propelled by something that at best would have a ten second burst of power followed, at best, by a controlled crash.
    I rest my case for not stupidly ‘stepping outside the envelope’!

    Comment by A NON FARMER — September 26, 2008 @ 9:22 pm

  13. The purpose of modelling is to match the results with empirical results. Last I heard, the models were getting ever better at showing what has happenend, and predicting the future.
    But, as I think I’ve pointed out before, climate change is just one element in a global impending environmental collapse. Consider the disappearance of many species of reptiles and birds, just for starters.
    But there surely is a link between the errors of relying on modelling for climate change and money markets: a failure to acknowledge that values and evidence are the most important factors. Greed drove the big and many of the little players in the financial drama; any reality check would have forced them to acknowledge they were living in lala land. Likewise, anyone who refuses to acknowledge that we are all living much too high on the environmental hog risks the hubris when reality hits.

    Comment by Ronda Jambe — September 27, 2008 @ 10:06 am

  14. Hello again people and forgive me Ronda – models are not plans or predictions or artificially generated scenarios.
    Models are devices that stand in realtime/realspace with the more sophisticated models configured to operate some way.
    The more sophisticated the model and the closer to full-scale the more closely MAY it emulate the original.
    I don’t want to get away from the thread too much but a few key words like modelling and the Emperor’s clothes happen to tickle me for all sorts of reasons.
    I don’t know how many of you are into serendipity – but I happened to read the following this afternoon – UAV challenge Toowoomba
    Yes its that good Aussie underdog story where a ‘model’, but for a broken wire, could have cleaned up all opposition in a very public place against very serious equipment being fielded in fair competition by the world’s experts.
    I submit, the ‘experts ain’t expert at all and they usually cost too much while they thrash about in committee reinventing the same old mousetrap.
    In the meantime while they (the oligarchy) stand firmly in the path of progress by sucking up all the funding they are consequently denying ordinary people their natural skills.
    Is it fair of me to mention modelling, the Emperor’s new clothes and oligarchy in the same tome?
    Any comments?

    Comment by A NON FARMER — September 27, 2008 @ 6:30 pm

  15. I think it is disingenuous to compare models of physical reality with models of financial markets.
    If I were to model a purely physical system, the out come and units of measure do not depend on my “feeling” towards them.
    However, I can walk into two identical houses and come out with two very different views of the “value” of each house if one “felt funny” and the other “felt nice” or some such similarily difficult to quantify property…
    Until economics has a proper theory or model of human behaviour, instead of its assumptions then these models are likely to be very flawed.
    No matter how I feel about a meter/ speed of light/ permittivity of free space/ Enthalpy of combustion of petrol, these values do not change… the same is true of other fundamental units of measure in the physical world.
    This is not true of abstractions of “value” like currencies which fluctuate depending on how market players “feel” about them… and these “feelings” then influence other market players.
    This might explain why we make mathematical models of gravity (al la Newtons laws) much more easily than mathematical models of stock market behaviour.
    The problem is not modelling per se. It is some operators idealistic or unreasonable expectations of what a model is, what it can teach us, and how we can use that information.
    A model is not reality. If you wanted that then why are you modelling? A lot of people seem to forget that a model is only a representation of reality.
    The situation gets even worse when models become iterative.
    Models have limits… but I would argue that the predictive ability of a climate model will be greater than attempts to predict the stock market… for the simple reason that gas molecules do not change there behaviour if I shout at them but my banker might!

    Comment by SP — October 4, 2008 @ 1:57 am

  16. SP, you make some valid points, but I was never making a straight comparison between financial and climate models. However I think you miss the point of these climate models. They don’t just embody known physical relationships, but make huge assumptions as to how other little understood factors might interact with the known factors.
    In making these assumptions subjectivity, which is what you assert applies to financial models, but not ones of physical occurrences, intrudes.
    The models can’t accurately model aerosols, cloud formation or energy transfers from the oceans. The figures that are put into them to represent these things are therefore quite subjective.
    This isn’t the same thing as modelling the universe, which can be done using a pen and paper.
    I think there is also an issue as to whether climate is an equilibrium system, as the solar system is. If it is not an equilibrium system, then it is closer to the economy than it is to the solar system.

    Comment by Graham Young — October 4, 2008 @ 10:32 pm

  17. Oh dear SP, Modelling.
    Why get so sophisticated?
    If a model is a model of a model – then it might be best if he/she/it still functions well.
    Yes. I risk being sexist for good reason.
    Consider. I’m a bloke but have discussed the scenario with Lady friends and found a common sentiment.
    We were all once young and gormless and wanted to be ‘seen out’ with someone who would ‘Wow’ our peers.
    Sometimes that ideal person, ‘that Model’, which seemed a good idea back then might cause us nausea in our maturity.
    But let’s face it – there are those of us who are straitened by circumstance and finance who simply cannot afford the real thing.
    Here I come to the nub of the issue –
    They may choose to ‘model’ in small scale those things they hold precious. Superbly crafted models of all sorts of things. Works of art that reflect the love of their creators.
    Then there are those who just decide that they’ve ‘been there and done that’ and make a decision that technology provides an outlet for them to expand their knowledge or past career in a manageable way.
    In this regard it is amazing the number of airline pilots, fighter jocks, naval captains, auto and motorcycle racers, astronauts – you name it – who enjoy spending big money by replicating what they used to do before retirement – but in miniature.
    Now these ‘modellers’ do not become involved with these activities to make money or prove anything.
    They become engaged because they want to and make a personal decision to follow their nose without any regard for the rest of society (except maybe for the kids next door).
    If such a model, say of a Boeing 17 is of about 24 foot wingspan, has four engines, is radio controlled from the ground with quarruple redundancy telemetry and takes three people attending the controls to fly the beast – is that a model or a slightly smaller replication of the original?
    If hobbyists go that far – just for the hell of it; then where does that leave uni graduates with no trade skills or life experience.
    I suggest absolutely nowhere.
    I’d suggest they get out there and at least experiment with a few chuck-gliders, some Tamiya kits, or some embroidery before they opened their mouths.

    Comment by A NON FARMER — October 4, 2008 @ 10:55 pm

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.