Friday 17 April 2015

A View on the Economic Model Debate from a Non-economist (but someone who builds models for a living)

Frances Coppola recently took up the attack on the macroeconomics profession. Like Steve Keen before her, she attacked the 'loanable funds model' which is widely used by macroeconomists and takes no account of the money creation powers of banks. Quite a large oversight on the part of the macroeconomics profession considering that a) they are completely and unambiguously wrong on this point and b) in 2008 there was a financial crash which proved how wrong they were and they seem to have not changed their models.

As far as I am concerned, this is an open and shut case. However Krugman changed the focus recently in his reply - pointing out the impracticality of the non-linear models that Coppola and Keen suggest using. Annoyingly (and I do think he did it in a very annoyingly superior manner) I find myself inclined to agree with him on this.

I build models with data for a living, and I am acutely aware of the problems with using non-linear models to make any sort of accurate predictions - even with huge volumes of data to calibrate it with.

It is not that the systems are linear. They are hugely complex. My problem is that they are too complex to model even with non-linear models. My belief is that linear models do have to be used but with a full understanding of the non-linearity of real life. Also, the whole building of macro-models from first principles, based on 'rational' agents, is a complete joke of a way to design a model that is supposed to be used in the real world.

In any case, I want to give an example to illustrate how knowledge of non-linearity is important but building a complete model is impossible.

Take a country. For an interesting example, let's call it Australia. Now put lots of commodities there. Now let's create a world trading system around it. Already we are getting complicated. Now put a country called China in that is industrialising very quickly and has a high demand for commodities. Now put in a non-linear commodity boom and bust price model in to determine the commodity prices. What is going to happen to Australia?

Well, as the price of commodities rises, their exchange rate is going to go up and exporters of other goods are going to find fewer and fewer markets abroad for their goods. The other export industries in Australia are going to atrophy.  It's not all bad though; lots of commodity money is flowing into the country and imports are cheap so people in Australia are happy for now.

Now, let's also model that China's boom creates a very high savings rate. And this money has to go somewhere. And Australia is nearby, stable and a safe place for rich Chinese princelings to put their hard earned cash. So the Chinese people buy a lot of Australian assets forcing a 4% current account deficit per year. Australia do not make a sovereign wealth fund to counteract this.

Now what's going to happen? On top of all of the commodity bounty coming in, there is now all of this unwanted Chinese money coming in. The high exchange rate makes imports even cheaper and exports even more expensive and the only way to keep domestic demand up is to take out debt. Hence private sector debt will go up. House price bubbles will ensue, but everything will still seem fine because the commodity prices are still high. Debt will rise and rise but times are good.

Next, the non-linear commodity model takes a downturn and commodity revenues drop. Chinese savings are still being parked in Australia so debt is needed to cope with that, but now more debt is needed to deal with the reduction in demand caused by lack of commodity income. There is not much export industry left and there is an economic downturn.

Eventually the house prices crash, debts become unsustainable and the economy enters downturn.

In this hypothetical example, the systems are hugely complicated. You have a model for commodity prices, a model for the Chinese savings rate and where the savings go, a model for domestic industry when the exchange rate is too high, a model for domestic credit, a model for house price bubbles. And fitting around all of that is a model for the entire Australian economy.

Making a model to do this is absolutely impossible. Or it is possible but it won't tell you anything more than you already know when you set the model up because further predictions are not going to have any level of accuracy in such a chaotic system.

However a knowledge of economic history enables one to piece together all of the non-linear parts of this and, with common sense, make a complete system.

It is a completely non-linear system outside of the bounds of the models of standard macroeconomics. But also one for which new models are possibly not needed as they are not necessarily going to add anything.

Update: Steve Keen wrote a response to this which can be found here.


  1. "And Australia is nearby, stable and a safe place for rich Chinese princelings to put their hard earned cash."

    Rich chinese princelings can't put their cash in Australia. Australia uses the Aussie Dollar and China the Yuan.

    That means there has to be an exchange of currencies, which means one person moves into Chinese currency for every person that moves out.

    So the model fails on the convertibility assumption.

    Which shows that you are modelling at too high an abstraction level and are missing things that are vitally important - like where the source of convertibility comes from that allows any sort of deficit or surplus to build up.

    For me the model has to be built up from atomic parts in a massively parallel simulation.

    Which of course is already a widely deployed technology in the online computer games arena.

  2. Neil, thanks for the comment.

    First, I would like to say that I am not an expert on the Australian economy - it was just an example of an economic system that is complex and, for me, it is an interesting example.

    Having said this, the converting of other currencies into Australian dollars is exactly my point. If Australia is running a 4% current account deficit then it means that foreigners are converting 4% worth of Australian GDP into Australian dollars, and keeping it as investment (be it in bonds, cash, property, stocks etc). This keeps the Australian dollar too high and interest rates too low. It takes demand worth 4% of GDP out of the economy. In order to counteract this effect on domestic demand, debt must be taken. This fuels bubbles in Australian assets

    Now, I am not sure exactly who is putting their money into Australia, although I did hear something recently about the Australians wanting to put a ban on Chinese buying houses there. Feel free to correct me here. But someone is.

    I am interested in your atomic parts model, but the tricky part here is modelling the interaction between the parts and the feedback loops contained. Simple models that do this, like Sugarscape (, are very enlightening and informative.

    Myonly issue is that it is difficult to use ideas like this to take quantitative decisions like 'what should the interest rate be?'.

  3. This is no more complicated than the modelling, simulation and analysis of a relatively simple analog electronic circuit. "Complex" is in the eye of the beholder. Engineers simulate things much more complex than this every day.


Sorry, I have had to moderate comments because of an annoying pest spammer who keeps posting American football matches links.