4 ms·

I studied economics and philosophy in college. Did a bit of economics research for a senior thesis. I was encouraged to study economics as a more "practical co

I studied economics and philosophy in college. Did a bit of economics research for a senior thesis. I was encouraged to study economics as a more "practical counterbalance" to philosophy. The funny thing is, I've found philosophy, in how it helped me improve the rigor of my thought, to be much more useful and practical than anything I learned in economics. Economics needs to rely more on computation and data than on abstract mathematical models if it ever hopes to have real predictive power.

stupidcar 9y ago
Yeah, it's weird. Early economists clearly didn't have the data or computing capability available to do analysis on the real world. So instead the took some inspiration from it and then built these simplified models, that were at least possible to study and make conclusions about. However, because real-world economic outcomes are so important, politically and socially, people ended up taking their conclusions and trying to apply them back to the real world. It's as if, after early AI researchers who took some inspiration from biology to create neural networks, the medical establishment then took their results and decided to use them as a guide to performing brain surgery.
- maxxxxx 9y ago
  "It's as if, after early AI researchers who took some inspiration from biology to create neural networks, the medical establishment then took their results and decided to use them as a guide to performing brain surgery. " I like that analogy.
- beberlei 9y ago
  If you read Hayek, Mises ("Human Action", very hard to follow) or later Rothbart ("Man, Economy, State"; easier to read), they all derived the same conclusions entirely based on philosophy rather than the "neoclassical economists" who use mathematical models. And all this was researched and known before math creeped so deeply into economics.
- IshKebab 9y ago
  Very true, however you make it sound like economists actually do caveat their hilariously simple economic models with "this is a wild guess of how the real world works". Whenever I've read it it has been presented as "this is how a market works" or whatever.
neilwilson 9y ago
It can't rely on data. The data is collected in a context of a policy arrangement and is therefore tainted by that policy. Economists think they can look beyond that and divine the deep truths. But they can't. They are looking for the philosophers stone. If you collect data about a man in a straitjacket you can only predict the movements of a man in a straitjacket. You can say nothing about what would happen if you took the straitjacket off. The purpose of economics is to embed the beliefs of those funding economists. Particularly the idea that there is no alternative - because data.
- 1_2__4 9y ago
  Policy is also data. I don't know why you don't acknowledge that. In fact most data sets are tethered to their context. So I don't actually get your point, other than that trying to analyze data sets contextlessly won't work.
  - smhcf8 9y ago
    His point is that policy data, unlike most sources of data in the sciences/computer-science world, is contaminated by behavioral adjustments. This is a point that was first brought up by the famous (and frequently misunderstood) Lucas critique. https://en.wikipedia.org/wiki/Lucas_critique https://en.wikipedia.org/wiki/Lucas_critique And the central tenet of the critique has been reaffirmed by decades of meticulous "natural experimental" studies and small-scaled field experiments which show that real-world economic causal effects do not often correspond to observational data on policy changes. Believe it or not, both of you, there are plenty of competent economists working on that exact methodological problem and resolving other statistical issues that are rarely encountered in non-behavioral sciences. James Heckman's works on this topic have been so widely influential, and lauded, that his papers are now frequently taught as canonical texts in graduate-level statistics courses. And this is a guy that started out only wanting to estimate returns to early-childhood education and basically carved out the best way to do it over 10 years. Here's a long review paper by Heckman on the literature on microeconomic policy evaluation: http://www.nber.org/papers/w11259.pdf http://www.nber.org/papers/w11259.pdf The language is clear, the use of statistical theory and technical concepts is brief but rigorous (the proofs are on p.54 onward), and literally all 80 pages are devoted to reviewing other recent papers on the best statistical methods to analyze policy data towards making accurate causal inferences. Realize the sheer size of that body of work? Maybe, just maybe, there are good academics in every discipline who aren't hopeless idiots that fit into your grandiose stereotypes. And maybe a random one-liner "omg they should look at policy data" isn't exactly a well-informed criticism in the context of the sheer amount of contemporary empirical work that you've never heard of in a discipline that you aren't familiar with.
digi_owl 9y ago
https://en.wikipedia.org/wiki/Steve_Keen#Minsky_software_project https://en.wikipedia.org/wiki/Steve_Keen#Minsky_software_pro...
beberlei 9y ago
There is a whole branch of econometrics that has endless methods of doing macroeconomic quantitative analysis using data. It just doesn't work on the macro level, the data is too aggregated, the system is too entangled with dependencies between collected data and incomparable between countries, results have only historic and not predictive value and even then they are usually more voodoo than anything else. So I disagree: Macro economics needs less computation, data and abstract mathematical models. The Hayek quote that CaseySoftware wrote into another toplevel comment sums it up perfectly: Only pretense of knowledge coming from this.
Mikeb85 9y ago
> Economics needs to rely more on computation and data than on abstract mathematical models if it ever hopes to have real predictive power. In industry it does. Every major bank and hedge fund employs economists, and generally they have a pretty good grasp of what's going on. Of course those models are kept secret, much like trading algorithms. That's why stuff like Econometrics is taught, and why (at least in the program I took) a decent amount of stats courses are required. There's many branches of Economics which look at different things; you have macro which is mostly theoretical and constructs broad models which attempt to model behaviour in a somewhat abstract way, micro which attempts to model individual behaviour, game theory which is practical but not necessarily for predictive power on a broad level, and Econometrics which is very practical, but doesn't necessarily provide much insight beyond medium-term predictions.
- yvdriess 9y ago
  > In industry it does. Every major bank and hedge fund employs economists, and generally they have a pretty good grasp of what's going on. Of course those models are kept secret, much like trading algorithms. How useful are those models, however? Without knowing much of the field, I do see the managed funds not outperforming index funds over the long term. Would the current run to the large unmanaged funds not demonstrate how limited their trade secret models are? In the case of automatic trading models, occasionally running wildly off the rails, there is probably still value, in that it does something we otherwise could not. A hedge fund manager that fails to outperform the index over the long term is not providing any value, except perhaps in providing a social function.
mvindahl 9y ago
And even then I think that it's important to realize that you can only get that far in terms of predictability. First off, it's a complex system; your sampled data set is incomplete by definition. Secondly, your theory is part of the system that it's measuring, at least if it ever becomes popular enough to influence policy. I think economics has some valuable insights into human nature -- not necessarily applicable to every human but applicable to enough of us to be of use -- but they should stop cargo culting mathematics. Or more accurately, everyone else should call their bluff.

stupidcar 9y ago

Yeah, it's weird. Early economists clearly didn't have the data or computing capability available to do analysis on the real world. So instead the took some inspiration from it and then built these simplified models, that were at least possible to study and make conclusions about. However, because real-world economic outcomes are so important, politically and socially, people ended up taking their conclusions and trying to apply them back to the real world. It's as if, after early AI researchers who took some inspiration from biology to create neural networks, the medical establishment then took their results and decided to use them as a guide to performing brain surgery.

maxxxxx 9y ago
"It's as if, after early AI researchers who took some inspiration from biology to create neural networks, the medical establishment then took their results and decided to use them as a guide to performing brain surgery. " I like that analogy.
beberlei 9y ago
If you read Hayek, Mises ("Human Action", very hard to follow) or later Rothbart ("Man, Economy, State"; easier to read), they all derived the same conclusions entirely based on philosophy rather than the "neoclassical economists" who use mathematical models. And all this was researched and known before math creeped so deeply into economics.
IshKebab 9y ago
Very true, however you make it sound like economists actually do caveat their hilariously simple economic models with "this is a wild guess of how the real world works". Whenever I've read it it has been presented as "this is how a market works" or whatever.

neilwilson 9y ago

It can't rely on data. The data is collected in a context of a policy arrangement and is therefore tainted by that policy. Economists think they can look beyond that and divine the deep truths. But they can't. They are looking for the philosophers stone. If you collect data about a man in a straitjacket you can only predict the movements of a man in a straitjacket. You can say nothing about what would happen if you took the straitjacket off. The purpose of economics is to embed the beliefs of those funding economists. Particularly the idea that there is no alternative - because data.

1_2__4 9y ago
Policy is also data. I don't know why you don't acknowledge that. In fact most data sets are tethered to their context. So I don't actually get your point, other than that trying to analyze data sets contextlessly won't work.
- smhcf8 9y ago
  His point is that policy data, unlike most sources of data in the sciences/computer-science world, is contaminated by behavioral adjustments. This is a point that was first brought up by the famous (and frequently misunderstood) Lucas critique. https://en.wikipedia.org/wiki/Lucas_critique https://en.wikipedia.org/wiki/Lucas_critique And the central tenet of the critique has been reaffirmed by decades of meticulous "natural experimental" studies and small-scaled field experiments which show that real-world economic causal effects do not often correspond to observational data on policy changes. Believe it or not, both of you, there are plenty of competent economists working on that exact methodological problem and resolving other statistical issues that are rarely encountered in non-behavioral sciences. James Heckman's works on this topic have been so widely influential, and lauded, that his papers are now frequently taught as canonical texts in graduate-level statistics courses. And this is a guy that started out only wanting to estimate returns to early-childhood education and basically carved out the best way to do it over 10 years. Here's a long review paper by Heckman on the literature on microeconomic policy evaluation: http://www.nber.org/papers/w11259.pdf http://www.nber.org/papers/w11259.pdf The language is clear, the use of statistical theory and technical concepts is brief but rigorous (the proofs are on p.54 onward), and literally all 80 pages are devoted to reviewing other recent papers on the best statistical methods to analyze policy data towards making accurate causal inferences. Realize the sheer size of that body of work? Maybe, just maybe, there are good academics in every discipline who aren't hopeless idiots that fit into your grandiose stereotypes. And maybe a random one-liner "omg they should look at policy data" isn't exactly a well-informed criticism in the context of the sheer amount of contemporary empirical work that you've never heard of in a discipline that you aren't familiar with.

digi_owl 9y ago

https://en.wikipedia.org/wiki/Steve_Keen#Minsky_software_project https://en.wikipedia.org/wiki/Steve_Keen#Minsky_software_pro...

beberlei 9y ago

There is a whole branch of econometrics that has endless methods of doing macroeconomic quantitative analysis using data. It just doesn't work on the macro level, the data is too aggregated, the system is too entangled with dependencies between collected data and incomparable between countries, results have only historic and not predictive value and even then they are usually more voodoo than anything else. So I disagree: Macro economics needs less computation, data and abstract mathematical models. The Hayek quote that CaseySoftware wrote into another toplevel comment sums it up perfectly: Only pretense of knowledge coming from this.

Mikeb85 9y ago

> Economics needs to rely more on computation and data than on abstract mathematical models if it ever hopes to have real predictive power. In industry it does. Every major bank and hedge fund employs economists, and generally they have a pretty good grasp of what's going on. Of course those models are kept secret, much like trading algorithms. That's why stuff like Econometrics is taught, and why (at least in the program I took) a decent amount of stats courses are required. There's many branches of Economics which look at different things; you have macro which is mostly theoretical and constructs broad models which attempt to model behaviour in a somewhat abstract way, micro which attempts to model individual behaviour, game theory which is practical but not necessarily for predictive power on a broad level, and Econometrics which is very practical, but doesn't necessarily provide much insight beyond medium-term predictions.

yvdriess 9y ago
> In industry it does. Every major bank and hedge fund employs economists, and generally they have a pretty good grasp of what's going on. Of course those models are kept secret, much like trading algorithms. How useful are those models, however? Without knowing much of the field, I do see the managed funds not outperforming index funds over the long term. Would the current run to the large unmanaged funds not demonstrate how limited their trade secret models are? In the case of automatic trading models, occasionally running wildly off the rails, there is probably still value, in that it does something we otherwise could not. A hedge fund manager that fails to outperform the index over the long term is not providing any value, except perhaps in providing a social function.

mvindahl 9y ago

And even then I think that it's important to realize that you can only get that far in terms of predictability. First off, it's a complex system; your sampled data set is incomplete by definition. Secondly, your theory is part of the system that it's measuring, at least if it ever becomes popular enough to influence policy. I think economics has some valuable insights into human nature -- not necessarily applicable to every human but applicable to enough of us to be of use -- but they should stop cargo culting mathematics. Or more accurately, everyone else should call their bluff.