So what would it take to make some reliable economic predictions—however simple or easy? How far away is the field from being able to do any useful predictive work at all?
I would point out that it is perfectly possible to be a worse predictor than a random or max-ent predictor; if all economics does is—per the quote—remove our (Marxist/Communist/socialist/Keynesian/mercantilist/Maoist/Catholic/...) delusions that we can improve on the normal workings of the economy, it will have done us a useful service indeed. Aside from the very simplest predictions relating to supply-and-demand, obviously.
I don’t get your jump there—we don’t need to be able to forecast weather to be able to build a roof to shield ourself from rain. The same way, even if economists have no ability to forecast the market, we may still be able to devise social rules to soften the negative consequences of a market crash. Or we may not, but it’s a different issue.
I also don’t get what you call “normal workings” of the economy. Even most of the libertarian I know about still want to enforce and protect private property, using external force to do so. So they are already distorting the “normal working”. If you want to use public force to ban stealing, then you’re already thinking you can improve on the “normal workings” of the economy.
You certainly do need to be able to forecast weather to justify building a roof to shield yourself from rain! As opposed to blizzards, the monsoon season, sand storms, or any of the infinite varieties of weather which do or do not exist in your particular location.
I also don’t get what you call “normal workings” of the economy.
The normal workings are set by long tradition and experience and local experiments (see the Austrians like Hayek), which is data-driven stuff completely opposed to the top-down economic interventions I contrasted with.
My understanding of Austrian Economics is that it’s founded on Praxeology, which must be read very generously not to conflict with the findings of neuroeconomics, prospect theory, and heuristics and biases in general. Google “austrian economics,” and one of the first results is a long, detailed critique by Robin Hanson’s lunch buddy Bryan Caplan (quoted above).
Given that, I’m not sure it’s unambiguous to use the term “normal workings” to describe Austrian Economic theory, except in a non-normal sense.
Point taken, but my question stands: how far are we from improving on max-ent predictions?
I have no idea. To a considerable extent, economics shouldn’t be able to make many good predictions by the nature of the material; see efficient markets and “Markets are Anti-Inductive”.
I’m curious as to exactly what you mean.
The set of idiosyncratic norms and traditions developed over centuries by small groups solving economic problems, which frequently maximize value even while appearing either impossible or arbitrary; this is an old vein of libertarian economic thought, although the most recent work I’ve read is Seeing like a State (pretty good).
Hm. Thanks for the link; I’d somehow missed that post. It’s a very clear analysis of what it means for a market to be ‘efficient.’
Even given that markets are anti-inductive, though, I disagree with you that economics shouldn’t be able to make many good predictions. Perhaps finance shouldn’t be able to make many good predictions. But, in principle, it should be possible to delve past people’s financial expectations and reason about how changes in the supply of fundamental economic inputs—land, labor, capital, technology, the rule of law, risk-tolerance, patience, etc. -- will affect fundamental economic outputs—goods, services, credit, information, etc. For example, if the population grows by 10% over the next 2 years, and everything else stays constant, how much more or less will it cost me (in PPP-adjusted dollars) to hire a plumber to fix my toilet? The answer shouldn’t depend on what everyone thinks it will cost; it should depend on how real changes in the real economy affect people’s real behavior, i.e., whether they are willing to come over and muck about in my pipes in exchange for a given amount of resources.
You might think that making predictions of this sort is, as a practical matter, too challenging—but it’s not clear why it should be naturally or definitionally unlikely.
Finally, I object to your characterization of the belief that we can improve on the normal workings of the economy as “delusional.” Certainly, there are some extreme examples of top-down social planning that have failed catastrophically, but reversed stupidity is not intelligence, and I have not seen any evidence that the economic solutions that have been evolved by small groups over long periods of time are always or even usually better than those that are developed by modern experts over short periods of time. I will understand if you don’t want to debate this issue on Less Wrong; it seems political enough to be a mind-killer. However, “delusions” seems to me to be a strong word, which should not be used lightly.
Regardless of how challenging, we are required to engage constantly in making economic predictions. My boss asks me to do a what I view as a stupid task because (a) I do not perceive the utility or (b) I predict, based upon previous experience, that my boss will, in fact, make no use of my output; and performing this task will occupy time that I believe could be spent more productively. How I respond, whether and how conscientiously I perform this task, requires several economic decisions that are as important to me personally as whether the Federal Reserve decides to increase the fed fund rate. My action will be based upon predictions such as the downside if I don’t do it or do a shoddy job; my future with the company; how others perceive my future in the company, how others perceive my boss, my prospects for employment with a different company, …. Many challenging assessments and predictions to make in a short period of time.
If I understand you correctly, Mass-Driver, I agree with you. I believe that competent, risk-based decision analysis is possible and when we employ such analysis, we make less disastrous economic decisions for ourselves and others.
My example is local, or personal if you will, to avoid political controversy. As a general matter, I believe that everything one does or does not do will have a consequence that is, in its totality, both unpredictable and unknowable. Nevertheless, we have to make plans based upon the best predictions of which we are capable.
Finally, I object to your characterization of the belief that we can improve on the normal workings of the economy as “delusional.” Certainly, there are some extreme examples of top-down social planning that have failed catastrophically, but reversed stupidity is not intelligence, and I have not seen any evidence that the economic solutions that have been evolved by small groups over long periods of time are always or even usually better than those that are developed by modern experts over short periods of time. I will understand if you don’t want to debate this issue on Less Wrong; it seems political enough to be a mind-killer. However, “delusions” seems to me to be a strong word, which should not be used lightly.
The examples exist throughout human history under regimes of every sort and with every ideology along the spectrum from fascist to communist to theocracies, and has a trivial evopsych justification; I don’t think the case is any less good for it being a delusion than any of the other cognitive biases or supernatural beliefs we dismiss.
There are also countless examples of “top-down social planning” that leaded to huge success—from sending men to the moon to eradicating smallpox to building the TGV (French high-speed train) network to eradicating illiteracy in some countries. We can argue for long if those results could have been achieved otherwise, or if they had more drawbacks than they are worth, … that would mean entering in a full-scale political discussion, which is not the purpose of Less Wrong. But calling it a “delusion” with a sleight of the hand like you do really looks like you’re victim of mind-killing on that issue. Rational political debate shouldn’t appear so blatantly one-sided.
There appears to be a definitional disconnect here. Although the Apollo program was top-down in many ways, it wasn’t what I would call social planning.
So what would it take to make some reliable economic predictions—however simple or easy? How far away is the field from being able to do any useful predictive work at all?
I would point out that it is perfectly possible to be a worse predictor than a random or max-ent predictor; if all economics does is—per the quote—remove our (Marxist/Communist/socialist/Keynesian/mercantilist/Maoist/Catholic/...) delusions that we can improve on the normal workings of the economy, it will have done us a useful service indeed. Aside from the very simplest predictions relating to supply-and-demand, obviously.
I don’t get your jump there—we don’t need to be able to forecast weather to be able to build a roof to shield ourself from rain. The same way, even if economists have no ability to forecast the market, we may still be able to devise social rules to soften the negative consequences of a market crash. Or we may not, but it’s a different issue.
I also don’t get what you call “normal workings” of the economy. Even most of the libertarian I know about still want to enforce and protect private property, using external force to do so. So they are already distorting the “normal working”. If you want to use public force to ban stealing, then you’re already thinking you can improve on the “normal workings” of the economy.
You certainly do need to be able to forecast weather to justify building a roof to shield yourself from rain! As opposed to blizzards, the monsoon season, sand storms, or any of the infinite varieties of weather which do or do not exist in your particular location.
The normal workings are set by long tradition and experience and local experiments (see the Austrians like Hayek), which is data-driven stuff completely opposed to the top-down economic interventions I contrasted with.
My understanding of Austrian Economics is that it’s founded on Praxeology, which must be read very generously not to conflict with the findings of neuroeconomics, prospect theory, and heuristics and biases in general. Google “austrian economics,” and one of the first results is a long, detailed critique by Robin Hanson’s lunch buddy Bryan Caplan (quoted above).
Given that, I’m not sure it’s unambiguous to use the term “normal workings” to describe Austrian Economic theory, except in a non-normal sense.
Bryan’s critique.
Point taken, but my question stands: how far are we from improving on max-ent predictions?
Would you taboo “normal,” please? I’m curious as to exactly what you mean.
I have no idea. To a considerable extent, economics shouldn’t be able to make many good predictions by the nature of the material; see efficient markets and “Markets are Anti-Inductive”.
The set of idiosyncratic norms and traditions developed over centuries by small groups solving economic problems, which frequently maximize value even while appearing either impossible or arbitrary; this is an old vein of libertarian economic thought, although the most recent work I’ve read is Seeing like a State (pretty good).
Hm. Thanks for the link; I’d somehow missed that post. It’s a very clear analysis of what it means for a market to be ‘efficient.’
Even given that markets are anti-inductive, though, I disagree with you that economics shouldn’t be able to make many good predictions. Perhaps finance shouldn’t be able to make many good predictions. But, in principle, it should be possible to delve past people’s financial expectations and reason about how changes in the supply of fundamental economic inputs—land, labor, capital, technology, the rule of law, risk-tolerance, patience, etc. -- will affect fundamental economic outputs—goods, services, credit, information, etc. For example, if the population grows by 10% over the next 2 years, and everything else stays constant, how much more or less will it cost me (in PPP-adjusted dollars) to hire a plumber to fix my toilet? The answer shouldn’t depend on what everyone thinks it will cost; it should depend on how real changes in the real economy affect people’s real behavior, i.e., whether they are willing to come over and muck about in my pipes in exchange for a given amount of resources.
You might think that making predictions of this sort is, as a practical matter, too challenging—but it’s not clear why it should be naturally or definitionally unlikely.
Finally, I object to your characterization of the belief that we can improve on the normal workings of the economy as “delusional.” Certainly, there are some extreme examples of top-down social planning that have failed catastrophically, but reversed stupidity is not intelligence, and I have not seen any evidence that the economic solutions that have been evolved by small groups over long periods of time are always or even usually better than those that are developed by modern experts over short periods of time. I will understand if you don’t want to debate this issue on Less Wrong; it seems political enough to be a mind-killer. However, “delusions” seems to me to be a strong word, which should not be used lightly.
Responding to the idea of economic predictions:
Regardless of how challenging, we are required to engage constantly in making economic predictions. My boss asks me to do a what I view as a stupid task because (a) I do not perceive the utility or (b) I predict, based upon previous experience, that my boss will, in fact, make no use of my output; and performing this task will occupy time that I believe could be spent more productively. How I respond, whether and how conscientiously I perform this task, requires several economic decisions that are as important to me personally as whether the Federal Reserve decides to increase the fed fund rate. My action will be based upon predictions such as the downside if I don’t do it or do a shoddy job; my future with the company; how others perceive my future in the company, how others perceive my boss, my prospects for employment with a different company, …. Many challenging assessments and predictions to make in a short period of time.
If I understand you correctly, Mass-Driver, I agree with you. I believe that competent, risk-based decision analysis is possible and when we employ such analysis, we make less disastrous economic decisions for ourselves and others.
That’s not the sort of planning and prediction that gwern thinks is doomed, since it’s “local.”
My example is local, or personal if you will, to avoid political controversy. As a general matter, I believe that everything one does or does not do will have a consequence that is, in its totality, both unpredictable and unknowable. Nevertheless, we have to make plans based upon the best predictions of which we are capable.
The examples exist throughout human history under regimes of every sort and with every ideology along the spectrum from fascist to communist to theocracies, and has a trivial evopsych justification; I don’t think the case is any less good for it being a delusion than any of the other cognitive biases or supernatural beliefs we dismiss.
There are also countless examples of “top-down social planning” that leaded to huge success—from sending men to the moon to eradicating smallpox to building the TGV (French high-speed train) network to eradicating illiteracy in some countries. We can argue for long if those results could have been achieved otherwise, or if they had more drawbacks than they are worth, … that would mean entering in a full-scale political discussion, which is not the purpose of Less Wrong. But calling it a “delusion” with a sleight of the hand like you do really looks like you’re victim of mind-killing on that issue. Rational political debate shouldn’t appear so blatantly one-sided.
There appears to be a definitional disconnect here. Although the Apollo program was top-down in many ways, it wasn’t what I would call social planning.