4 ms·

Mathiness: not just a problem of macro-economics (2015)

46 points by jimsojim 10y ago

nerdponx 10y ago
Glad this issue is getting attention. It's one of the things that got me to reconsider applying for an economics PhD.
matt4077 10y ago
People have been complaining for 20 years+ that economics chose the wrong tool. It needs psychology, not math. (Or, more accurately: more and less, respectively – you'll still need math). The old joke about physicists starts "Consider a spherical chicken". It's funny because it gets at the reductionism at the core of physics, and it's even funny for physicists because they actually have been pretty successful with that model. It just doesn't work that way in economics. Even worse, the reductionism in economics is unethical, because when it registered (somewhere) that humans are not "objective, utility-maximising spheres" they more or less started advocating that they /should/ be. That got us to "corporations are people" and 24-step daily checklists to "make better use of your time". It also made fast-food the superior choice (time is money!), so at least the "spherical" part is coming along. After this short rant, which was obviously about microeconomics and therefore off-topic, something different for macro: While he's right that the "growth" debate has had more or less the same trajectory as the Grand Unifying Theory in physics, I wouldn't discount the improvements we've made in other areas. Inflation, for example, is pretty well under control. People don't appreciate that too much, but I remember how my Grandmother talked about inflation. It was basically /the/ catastrophe of her life (Germany in the 20ies – WW2 was just a result of it). So we seem to be doing something right – maybe economics, maybe politics, but certainly better. I also disagree with the very last paragraph of the article where he seems to advocate dropping the scientific method for a model closer to that of the social sciences. An improved discourse would obviously be great, but he seems to be equating math with the scientific method. We can try to get at the complexities of economics with new tools and still verify the results. Ultimately, math should be able to make a comeback when it develops the tools to describe systems of this complexity. It's a lot like the brain: it is obviously possible to describe the brain using physics & math (because it's a physical system). BUT we're not there yet, and until we are, we get better results with the collection of heuristics called "psychology"
- yummyfajitas 10y ago
  It just doesn't work that way in economics. Even worse, the reductionism in economics is unethical, because when it registered (somewhere) that humans are not "objective, utility-maximising spheres" they more or less started advocating that they /should/ be. That got us to "corporations are people" and 24-step daily checklists to "make better use of your time". It also made fast-food the superior choice (time is money!), so at least the "spherical" part is coming along. Some citations are clearly needed here. In particular about the normative claims you believe economics makes - as far as I know the field is very strictly positive [1]. Economists will say that fast food is better because people are choosing to consume it, and most of those same economists will discuss the topic over a lunch of artisinal kale salad locally grown by vegan hipsters. [1] There are cases where economists are clearly motivated by normative concerns to draw a particular positive conclusion - e.g. that the law of demand doesn't work for low skill labor. But that's a different critique. http://cafehayek.com/2016/12/41844.html http://cafehayek.com/2016/12/41844.html
  - matt4077 10y ago
    I guess the scientific term is https://en.wikipedia.org/wiki/Homo_economicus https://en.wikipedia.org/wiki/Homo_economicus. The classic experiment is probably the Ultimatum game. And that's so old, you're tempted to say: see – they did notice the flaws in their theories! Yes, economists will readily admit to these flaws and encourage you to keep it in mind when evaluating their predictions and prescriptions. Unfortunately, people aren't just marginally different from the models. Just yesterday I violated transitivity of preferences when choosing lunch. I did an evaluation of GDO growth forecasts for my bachelor thesis. None of the institutions (Central banks etc.) managed to outperform a well-adjusted random number generator.
    - jernfrost 10y ago
      I don't think economist always are that frank about the limits of their models. E.g. Milton Friedman I think seemed to think that the world worked mostly has homo economicus. He had some rather ridiculous examples. E.g. he was touring a factory in Hong Kong and claimed workers had picked noisy and hazardous work environments over more pleasant ones, because it paid better. That strikes me as a guy with the head in the cloud lost his his theories. Jobs don't exist in endless varieties like that. If I'd rather have a better health insurance but lower paid job, I can't easily change to an almost identical job, with just those differences. I also remember him arguing why it was rational for Ford or GM to install a cheap fix for dangerous fuel tanks. His straw man argument was that an infinitely safe car would cost infinitely amounts of money. With such arguments you can argue in favor of no minimum standards for anything. Never mind that he failed to see the blindingly obvious problem. It simply didn't make any economic sense because any car company making such a cynical calculation would severely damage their reputation and thus their sales. It only works for companies who manage to cover up. And Milton Friedman is one of our most celebrated economists. Sure he create some great theories, but he seemed also to not heed to typical advice of not taking models too literal. He basically advocated policies based on rather fundamentalists interpretations of economic models. If he could do that, then surely lots of other economists would fail in the same manner.
jernfrost 10y ago
Nice to see some focus on this, as I believe so many of the ills of the present days has been caused by this mathiness. I don't think math or mathematical models are bad per say, but they have clearly been used in the wrong way in too many instances. Ones has placed far too much faith in simplified models of the real world. We don't like the messy world of human interactions and so we have tried to turn it all into math, while overstating the utility or applicability of the models. It is a bit like trying to create a mathematical models for determining the quality of a persons character rather than simply relying on people's subjective characterizations. We forget that humans have brains which are very good at processing massive amounts of messy data. Instead of relying on human brains why have outsourced the task to mathematical models which can't easily deal with messy data, and thus simplify the world to the point where quite essential characteristics are lost. Also as a software developer I think we could gain a lot by employing more agent based simulations to the economics field. I think too much of the dynamics of a system over time is lost by relying so exclusively on mathematical equations. Too much of economic science also seems to have turned into a political ideology for many on the right. The fact that a mathematical model predicts inefficiencies from taxation in a free market, seems to make some people believe it is inherently bad with taxation and it must be limited at all cost. Such models should always be compared with real world data, where there has never been found a clear evidence that higher taxes really retards economic growth the way a simple model predicts. Naturally because an economy is extremely complicated with many interdependent variables. Higher taxes might cause worse allocation of resources but it might also allow for better schools and thus better skills among workers increasing efficiency more than is lost through distortions of taxations. Not claiming it is like that, but giving food for thought about the idea that it is hard to look at these models in isolation. Mathematical models were used in the late 70s to argue in favor of high CEO salaries. We know today that there is no empirical evidence that it worked. Western economies are not growing faster and companies are not better run. There is no provable difference between highly paid CEOs and lower paid CEOs. This shows the danger of such mathematical models. They become self fulfilling prophecies. Once people start using them, they keep using them because of inertia and the tendency to conform and stick to traditions.

46 points by jimsojim 10y ago

Glad this issue is getting attention. It's one of the things that got me to reconsider applying for an economics PhD.

People have been complaining for 20 years+ that economics chose the wrong tool. It needs psychology, not math. (Or, more accurately: more and less, respectively – you'll still need math). The old joke about physicists starts "Consider a spherical chicken". It's funny because it gets at the reductionism at the core of physics, and it's even funny for physicists because they actually have been pretty successful with that model. It just doesn't work that way in economics. Even worse, the reductionism in economics is unethical, because when it registered (somewhere) that humans are not "objective, utility-maximising spheres" they more or less started advocating that they /should/ be. That got us to "corporations are people" and 24-step daily checklists to "make better use of your time". It also made fast-food the superior choice (time is money!), so at least the "spherical" part is coming along. After this short rant, which was obviously about microeconomics and therefore off-topic, something different for macro: While he's right that the "growth" debate has had more or less the same trajectory as the Grand Unifying Theory in physics, I wouldn't discount the improvements we've made in other areas. Inflation, for example, is pretty well under control. People don't appreciate that too much, but I remember how my Grandmother talked about inflation. It was basically /the/ catastrophe of her life (Germany in the 20ies – WW2 was just a result of it). So we seem to be doing something right – maybe economics, maybe politics, but certainly better. I also disagree with the very last paragraph of the article where he seems to advocate dropping the scientific method for a model closer to that of the social sciences. An improved discourse would obviously be great, but he seems to be equating math with the scientific method. We can try to get at the complexities of economics with new tools and still verify the results. Ultimately, math should be able to make a comeback when it develops the tools to describe systems of this complexity. It's a lot like the brain: it is obviously possible to describe the brain using physics & math (because it's a physical system). BUT we're not there yet, and until we are, we get better results with the collection of heuristics called "psychology"

yummyfajitas 10y ago
It just doesn't work that way in economics. Even worse, the reductionism in economics is unethical, because when it registered (somewhere) that humans are not "objective, utility-maximising spheres" they more or less started advocating that they /should/ be. That got us to "corporations are people" and 24-step daily checklists to "make better use of your time". It also made fast-food the superior choice (time is money!), so at least the "spherical" part is coming along. Some citations are clearly needed here. In particular about the normative claims you believe economics makes - as far as I know the field is very strictly positive [1]. Economists will say that fast food is better because people are choosing to consume it, and most of those same economists will discuss the topic over a lunch of artisinal kale salad locally grown by vegan hipsters. [1] There are cases where economists are clearly motivated by normative concerns to draw a particular positive conclusion - e.g. that the law of demand doesn't work for low skill labor. But that's a different critique. http://cafehayek.com/2016/12/41844.html http://cafehayek.com/2016/12/41844.html
- matt4077 10y ago
  I guess the scientific term is https://en.wikipedia.org/wiki/Homo_economicus https://en.wikipedia.org/wiki/Homo_economicus. The classic experiment is probably the Ultimatum game. And that's so old, you're tempted to say: see – they did notice the flaws in their theories! Yes, economists will readily admit to these flaws and encourage you to keep it in mind when evaluating their predictions and prescriptions. Unfortunately, people aren't just marginally different from the models. Just yesterday I violated transitivity of preferences when choosing lunch. I did an evaluation of GDO growth forecasts for my bachelor thesis. None of the institutions (Central banks etc.) managed to outperform a well-adjusted random number generator.
  - jernfrost 10y ago
    I don't think economist always are that frank about the limits of their models. E.g. Milton Friedman I think seemed to think that the world worked mostly has homo economicus. He had some rather ridiculous examples. E.g. he was touring a factory in Hong Kong and claimed workers had picked noisy and hazardous work environments over more pleasant ones, because it paid better. That strikes me as a guy with the head in the cloud lost his his theories. Jobs don't exist in endless varieties like that. If I'd rather have a better health insurance but lower paid job, I can't easily change to an almost identical job, with just those differences. I also remember him arguing why it was rational for Ford or GM to install a cheap fix for dangerous fuel tanks. His straw man argument was that an infinitely safe car would cost infinitely amounts of money. With such arguments you can argue in favor of no minimum standards for anything. Never mind that he failed to see the blindingly obvious problem. It simply didn't make any economic sense because any car company making such a cynical calculation would severely damage their reputation and thus their sales. It only works for companies who manage to cover up. And Milton Friedman is one of our most celebrated economists. Sure he create some great theories, but he seemed also to not heed to typical advice of not taking models too literal. He basically advocated policies based on rather fundamentalists interpretations of economic models. If he could do that, then surely lots of other economists would fail in the same manner.

jernfrost 10y ago

Nice to see some focus on this, as I believe so many of the ills of the present days has been caused by this mathiness. I don't think math or mathematical models are bad per say, but they have clearly been used in the wrong way in too many instances. Ones has placed far too much faith in simplified models of the real world. We don't like the messy world of human interactions and so we have tried to turn it all into math, while overstating the utility or applicability of the models. It is a bit like trying to create a mathematical models for determining the quality of a persons character rather than simply relying on people's subjective characterizations. We forget that humans have brains which are very good at processing massive amounts of messy data. Instead of relying on human brains why have outsourced the task to mathematical models which can't easily deal with messy data, and thus simplify the world to the point where quite essential characteristics are lost. Also as a software developer I think we could gain a lot by employing more agent based simulations to the economics field. I think too much of the dynamics of a system over time is lost by relying so exclusively on mathematical equations. Too much of economic science also seems to have turned into a political ideology for many on the right. The fact that a mathematical model predicts inefficiencies from taxation in a free market, seems to make some people believe it is inherently bad with taxation and it must be limited at all cost. Such models should always be compared with real world data, where there has never been found a clear evidence that higher taxes really retards economic growth the way a simple model predicts. Naturally because an economy is extremely complicated with many interdependent variables. Higher taxes might cause worse allocation of resources but it might also allow for better schools and thus better skills among workers increasing efficiency more than is lost through distortions of taxations. Not claiming it is like that, but giving food for thought about the idea that it is hard to look at these models in isolation. Mathematical models were used in the late 70s to argue in favor of high CEO salaries. We know today that there is no empirical evidence that it worked. Western economies are not growing faster and companies are not better run. There is no provable difference between highly paid CEOs and lower paid CEOs. This shows the danger of such mathematical models. They become self fulfilling prophecies. Once people start using them, they keep using them because of inertia and the tendency to conform and stick to traditions.