Pielke Jr on Models

The National Post in Toronto led off a week of columns on questionable science, leading up to the awarding of the prestigious Rubber Duck Award (named after a Canadian Environmental Defense campaign against homicidal rubber ducks) with a column by Pielke Jr. on models.

RP leads with a discussion of the role of models in the 2008 financial debacle and ends with reflections on hurricanes. He opens with a quote from Keynes.

In 1936, John Maynard Keynes warned that “the ideas of economists and political philosophers, both when they are right and when they are wrong, are more powerful than is commonly understood. Indeed the world is ruled by little else.” Today, almost 75 years later, the power of economics often manifests itself through sophisticated financial and risk models which — when they are right and when they are wrong — exert a powerful influence on many aspects of our daily lives. Understanding the role of these models and how to use them wisely is something that we are still learning how to do.

Here I have a mild criticism. Surely this would have been an ideal opportunity to cite Keynes’ own sharp commentary on econometric modeling (Keynes, Economic Journal, 1940). Keynes’ article is, of course, conveniently accessible online at CA here and is a commentary that still reads freshly today. I placed it online when we discussed (here) Hendry’s Econometrics – Alchemy or Science, in which Hendry also discussed using cumulative UK rainfall as a predictor (“proxy”) for the Consumer Price Index, a correlation so imposing that is puzzling that it was neglected by Steig et al 2009 and Mann et al 2008.

10 Comments

  1. jae
    Posted Jun 16, 2009 at 8:22 PM | Permalink

    I placed it online when we discussed (here) Hendry’s Econometrics – Alchemy or Science, in which Hendry also discussed using cumulative UK rainfall as a predictor (“proxy”) for the Consumer Price Index, a correlation so imposing that is puzzling that it was neglected by Steig et al 2009 and Mann et al 2008.

    LOL. You just cannot resist the rib, eh?

    • Geoff Sherrington
      Posted Jun 17, 2009 at 3:37 AM | Permalink

      Re: jae (#1),

      Why complain? The problem is explicitly stated and it has not been fixed. Would you continue to drive your car without further comment when you knew it had a flat?

  2. Tom C
    Posted Jun 17, 2009 at 7:42 AM | Permalink

    …a correlation so imposing that is puzzling that it was neglected by Steig et al 2009 and Mann et al 2008.

    Good grief! Don’t give them any more ideas.

  3. Kenneth Fritsch
    Posted Jun 17, 2009 at 3:01 PM | Permalink

    I am not as confident as some of you apparently are that model failure is the real culprit here. I suspect that modelers know better than advertised the risk involved at the extremes. Those in positions who had to make decisions based on these models had another risk issue and that being if things really went wrong at the extremes of the model what would be the consequences and was too big, or too important, to fail an input (official or unofficial) into that process. I think it was. Government enterprises in the US, like Fannie Mae and Freddie Mac, were often noted by government officials as not backed by the US government, but, in fact, were rated as though they were by investors and their borrowing powers and indeed, as finally reality showed, that was the case.

    I would suggest that models used for hurricane risk and hurricane event forecasting have an additional inherent risk factor that officially and/or unofficially considers the failure of over and under predicting these events and consequences. Given the initial choice of a model that might under predict or over predict, the over prediction, I judge, has the least adverse consequences if incorrect in the opposite direction.

    The confidences and narrow uncertainties that climate scientists claim for their climate models forecasts, and for that matter, the climate scientist consensus on AGW are predicated as much, in my view, on what they see as the consequences of being wrong as on any hard analysis on a scientific level. I would suggest that most of these climate scientists see little adverse unintended consequences of AGW mitigation (unlike myself) and many of them would judge the mitigation actions worthy even without the issue of AGW.

    Keynes may have been in tune with mass psychology in some areas, but that would not necessarily extent to his ability in being rather blind (I hope) to the reaction of politicians to what his economic model called for in good economic times when politicians were expected to balance their budgets. Keynes also may have seen the adverse effects of others models, but not his own. That is perhaps human nature. In my view stagflation broke the Keynesian model in the US in the 1970s for all to see, but there are no doubts that that model is being applied aggressively in current time.

  4. MikeN
    Posted Jun 18, 2009 at 1:52 PM | Permalink

    Regarding the IPCC not providing data, don’t their reports use the Mauna Loa CO2 levels? THat site has a license for use which requires all users to be reciprocal, and to provide all data and methods and be helpful to people who ask.

  5. gah
    Posted Jun 20, 2009 at 4:07 PM | Permalink

    I noticed people haven’t commented too much on this thread. Very understandable as the climate change deniers have no model of their own and they probably wouldn’t even know where to start.

    BTW, UK rainfall as a predictor of GDP is not a model so much as a correlation. Models usually come form first principles and unless you have something that follows a reasonable and logical causality, I wouldn’t count it as a model. Yet, similar inane statistical correlations is what the ClimateAudit seem to treat as their bread and butter, and it doesn’t surprise me that Hendry misses the point entirely.

    • Steve McIntyre
      Posted Jun 20, 2009 at 4:34 PM | Permalink

      Re: gah (#7),

      The greater part of the commentary at this site pertains to proxy reconstructions, rather than climate models. Proxy reconstructions of the Mannian type or the Steig climate reconstruction rely entirely on statistical correlation.

      If you choose to regard Steig et al 2009, Mann et al 2008 and similar literature as built on “inane statistical correlations”, then you are obviously free to do so and you will find that few readers here will disagree with you.

      I’m not sure what point you believe that Hendry was “missing”. Hendry’s article was written in the early 1980s, long before IPCC, and addressed the problem of spurious correlation and the difficulty that it presented for standard statistical tests. Hendry is a distinguished econometrician and your name-calling here shows more about you than about him.

    • Gerald Machnee
      Posted Jun 21, 2009 at 8:58 AM | Permalink

      Re: gah (#7),

      Very understandable as the climate change deniers have no model of their own and they probably wouldn’t even know where to start.

      “Climate change deniers” do not want a model, they just want facts. One would be the MEASUREMENT of the temperature increase or decrease caused by greenhouse gases.
      Contributors here are not “deniers” but “questioners” and researchers who seek answers. There are many knowledgeable contributors to this weblog, likely a higher number than most other blogs.
      Would you say that the statistical connection between CO2 and temperature is “inane”?

  6. pete m
    Posted Jun 22, 2009 at 7:24 PM | Permalink

    Steve, Pielke Jr has moved his site:

    linky

  7. Andrew
    Posted Jun 17, 2009 at 7:44 AM | Permalink

    Re: Hoi Polloi (#3), As much as I totally agree…isn’t allowing comments like these against blog policy?

    Steve:
    yes. It was deleted.

Follow

Get every new post delivered to your Inbox.

Join 3,332 other followers

%d bloggers like this: