Economic modelling

KyussBishop

There's a good discussion on the Irish Economy forum relating to economic forecasting/modelling, and I think andrew's post there is very good, but makes the mistake of looking at modelling economics from a microfoundations level (individual people), instead of a macro level.

When you compare that to weather forecasting, that (micro level modelling) would be a bit like trying to model every individual atom in the atmosphere in order to build the model (which would make it seem impossible), whereas taking a macro approach (examining on a larger scale, how the overall system will or is likely to function), it allows much more room for putting together a useful picture and predictions of what may happen (even if human psychology is still a totally unpredictable factor).

There's a very good article on exactly this topic, from Steve Keen, who is working on dynamically modelling economies with his Minsky program:
http://www.businessspectator.com.au/article/2013/1/14/resources-and-energy/blame-it-bad-weathermen

This is a notable contrast to how it is done today with existing models: Those models, by and large, are based on static, not dynamic, analysis - specifically, assuming that the economy is constantly returning to 'equilibrium' (even stuff like DSGE, is actually based on this 'static'/equilibrium assumption).


I think that what is far more important than providing accurate medium/long-term predictions though (which is arguably not possible, as e.g. a completely unpredictable natural disaster, like from the weather , could severely disrupt economies unpredictably) - far more important is providing a model and description of what is happening, or what will happen given specific inputs, as this allows economics to take on a much more scientific grounding, making predictions that can be held up to empirical scrutiny and refutation (which is something economic theory avoids like the plague right now, in my view for political/ideological reasons - which is part of why it is the 'dismal science').

I highly recommend picking up Steve Keens Debunking Economics for a good overview of how economics is stuck in 'static'/equilibrium thinking, as opposed to more reality-based dynamic/disequilibrium thinking (and just for a general overview of how breathtakingly wrong most leading theory and economic teaching is).

andrew

KyussBishop wrote: »

There's a good discussion on the Irish Economy forum relating to economic forecasting/modelling, and I think andrew's post there is very good, but makes the mistake of looking at modelling economics from a microfoundations level (individual people), instead of a macro level.

That's not a 'mistake'; I was trying to answer the OP's question. The OP was asking whether it'd be possible to model an economy in the same way as weather is modeled; my post points out the current difficulties in creating such detailed 'micro' level models.

When you compare that to weather forecasting, that (micro level modelling) would be a bit like trying to model every individual atom in the atmosphere in order to build the model (which would make it seem impossible), whereas taking a macro approach (examining on a larger scale, how the overall system will or is likely to function), it allows much more room for putting together a useful picture and predictions of what may happen (even if human psychology is still a totally unpredictable factor).

Are you saying that if we could somehow model every atom in the atmosphere, that this would lead to a less useful picture of the climate? How? Similarly, if we could model individuals, (and if we understood fully how their actions aggregate into the macro economy) how would this be a less detailed less useful model then the current models we use? Macro models allow for 'more room?'

I think that what is far more important than providing accurate medium/long-term predictions though (which is arguably not possible, ...far more important is providing a model and description of what is happening, or what will happen given specific inputs

Isn't "what will happen given specific inputs" exactly the same as "providing...predictions." All economic models take specific inputs and then spit out an assessment of what will happen as a result. This is one of the main reasons for creating models.

as this allows economics to take on a much more scientific grounding, making predictions that can be held up to empirical scrutiny and refutation (which is something economic theory avoids like the plague right now, in my view for political/ideological reasons - which is part of why it is the 'dismal science').

What's your basis for saying Economics isn't held up to empirical scrutiny and refutation? The empirical scrutiny of Economic theories is the reason that an entire branch of statistics called 'Econometrics' exists. Every single undergrad in Economics learns (or at the very least has the opportunity to learn) Econometrics. You can't do anything beyond undergraduate level without learning Econometrics. Economists try constantly to obtain data which might enable them to use Econometics to test economic theories. Take the 'Permanent income hypothesis,' as a random example. Here's a Google scholar search for 'testing the Permanent income hypothesis.' Paper after paper of studies in which Economists try to see if the Permanent income hypothesis holds up to scrutiny, by using Econometrics. This is true for pretty much every Economic theory for which data can be obtained. The idea that Economists are content to ignore empirical data and are content to just sit around making stuff up without checking whether it has real world relevance; It simply isn't true.

I highly recommend picking up Steve Keens Debunking Economics for a good overview of how economics is stuck in 'static'/equilibrium thinking, as opposed to more reality-based dynamic/disequilibrium thinking (and just for a general overview of how breathtakingly wrong most leading theory and economic teaching is).

Before reading a book like that, though, it seems that it'd make sense for people to read an Economics textbook, or something. It's hard to judge whether an idea has actually been 'debunked' if you don't know anything about that idea in the first place.

KyussBishop

andrew wrote: »

Are you saying that if we could somehow model every atom in the atmosphere, that this would lead to a less useful picture of the climate? How? Similarly, if we could model individuals, (and if we understood fully how their actions aggregate into the macro economy) how would this be a less detailed less useful model then the current models we use? Macro models allow for 'more room?'

What I meant was more that weather modelling isn't very 'micro', but built on higher level macro generalizations/models about how the weather tends to work - so pointing out the difficulties of modelling economics from a micro level, isn't a good analogy to modelling the weather - it's more appropriate to look at modelling economics from a macro level.

The same way you can't model or predict the location of every particle in a weather system, you can't model or predict the psychology of all the individuals in an economy either - so this makes a true micro-level model impractical for both (exceptionally useful in both cases, if it could be done, though just is unfortunately impractical).

This doesn't impede useful economic modelling though, in the same way that it doesn't impede useful weather modelling.

andrew wrote: »

Isn't "what will happen given specific inputs" exactly the same as "providing...predictions." All economic models take specific inputs and then spit out an assessment of what will happen as a result. This is one of the main reasons for creating models.

Yes, pretty much, I guess my comment there was a bit redundant - you just can't predict how those inputs will change (or rather, that gets increasingly difficult as time goes on), so that limits how far into the future you can make useful predictions (but doesn't preclude it of course).

I guess more what I meant, is that the models should (first and foremost) have their accuracy judged on consistently describing all past events and possibilities, before it even gets to the point of being judged based on predictions.

andrew wrote: »

What's your basis for saying Economics isn't held up to empirical scrutiny and refutation? The empirical scrutiny of Economic theories is the reason that an entire branch of statistics called 'Econometrics' exists. Every single undergrad in Economics learns (or at the very least has the opportunity to learn) Econometrics. You can't do anything beyond undergraduate level without learning Econometrics. Economists try constantly to obtain data which might enable them to use Econometics to test economic theories. Take the 'Permanent income hypothesis,' as a random example. Here's a Google scholar search for 'testing the Permanent income hypothesis.' Paper after paper of studies in which Economists try to see if the Permanent income hypothesis holds up to scrutiny, by using Econometrics. This is true for pretty much every Economic theory for which data can be obtained. The idea that Economists are content to ignore empirical data and are content to just sit around making stuff up without checking whether it has real world relevance; It simply isn't true.

Most economics as it is taught today, does not properly take into account an accurate portrayal of the monetary system, debt, or banks (and the accounting behind this, tends to get left out of models).

Obsolete theories, such as the idea that banks are reserve-constrained (when in reality they are capital-constrained), that money is created exogenously (when it is created endogenously by bank loans), and broader ideas, such as the idea that economies 'tend towards equilibrium' (which even DSGE still represents a form of) instead of disequilibrium, are still mainstream and considered credible, even when the evidence does not support this.

A lot of economics is corrupted by politics in this way - it is politically/economically advantageous for many people, to keep bad economic theory mainstream.


Econometrics is a useful field alright, but it is only as good as the assumptions underlying the economic theory used in modelling - a lot of econometric modelling is based on incorrect static/linear or equilibirum assumptions, rather than using more dynamic/disequilibrium modelling, and not many of them properly factor in the role of the banking/monetary system (that requires modelling the actual accounting of the monetary system - which, to my understanding, is rarely done).

With Minsky, Steve Keen is specifically developing it as a program for dynamically modelling economies, in a way which takes dynamics/disequilibrium and accounting of the monetary/banking system into account - he knows much more, and can explain way better, the issues with economic modelling, than I can, so it's worth keeping an eye on his blog or checking his book.

andrew wrote: »

Before reading a book like that, though, it seems that it'd make sense for people to read an Economics textbook, or something. It's hard to judge whether an idea has actually been 'debunked' if you don't know anything about that idea in the first place.

I agree, the trouble is that it's exactly some of the misinformation in textbooks which is what Keen's book tackles - reading it and a textbook alongside each other, might make them useful companions (otherwise there's the risk of accepting some of the bad information in the textbook uncritically).

In any case, this author has done a good overview of the book, in a long series of blog posts (some of the ending blog posts - response to criticism of Keen's book - touch on some of what I mention above too):
https://unlearningeconomics.wordpress.com/tag/debunking-economics/

KyussBishop

I mentioned Steve Keen's Minsky program in my OP here, and this just got made 'Project of the Month' on SourceForge; there is a good interview with the Minsky team here, which gives a very good/brief overview of the state of macroeconomic modelling in mainstream economics, and what Minsky is trying to achieve:
https://sourceforge.net/blog/january-2014-potm/

A good snippet from the first line, which pretty much highlights how mainstream economics is basically non-reality-based:

Bizarre as it may seem, maintream economic theory ignores banks, debt and money completely, and imagines that the dynamic, innovative and crisis-prone social system we call capitalism can be modelled as if it is almost always in equilibrium. It’s little wonder therefore that most economists didn’t see the financial crisis coming back in 2007.

For those who do much programming, the project (being on SourceForge) is open-source, and (being a very specific area of interest) the team have a difficult time finding more coders - if people find it interesting, worth considering helping out.

Personally, I think it's one of the few software projects with a particularly high potential for having a large positive impact, due to it's potential to reform economic teaching and practice; would be working on it myself, if I didn't have to keep up with coding already in my day-job