3 ms·

> Remember that all models are wrong, but some are useful. I can't count the number of times I've heard that trotted out as a tired defense of bad modelling. A

> Remember that all models are wrong, but some are useful. I can't count the number of times I've heard that trotted out as a tired defense of bad modelling. All models are wrong: some are useful, some are dangerous. Are you sure you can distinguish between those categories? > Possibly the main missing point is that the book appears to be uninformed that there are many economic scholars researching inequality, race[1], feminism[2], environmentalism[3], and other progressive areas[4]. I don't think either the reviewer or the book itself are claiming that there are no economists that study those causes. It sounds like the book claims that postwar economists have enjoyed unaccountable policy influence, and their methodology has blinded them to certain consequences of their advice.

tomrod 7y ago
> Are you sure you can distinguish between those categories? Typically yes, through appropriate risk assessment. Cue "Antifragile" here.
- soVeryTired 7y ago
  So what does "appropriate risk assessment" look like? Vary some parameters and add confidence bounds to your outputs? If you're building an economic model and they economy enters a regime that you haven't seen before in your data, your model calibration is going to crumble. So maybe good models can be distinguished from bad when predicated on the assumption that the near future looks like the past. But when that changes (i.e. when you need your models most) is when your models become least reliable.
  - SantalBlush 7y ago
    It's really not a question of how bad an economic model is, but whether there exists a better one we can use in its place. When people criticize (often rightly) economic models and point out their limitations, there is an implication that we should abandon them because they are so flawed. But this is wrong. It is not enough to avoid flawed models; we need something demonstrably better to replace them with, and this is where the critics tend to come up short. Suppose we get rid of economists and their models; then what do we use in their place to guide policy? Our intuition? No thanks. If someone has a superior theory of the economy, they must show it with logic and empirical observation. Until they do, we are stuck with our more flawed models.
    - soVeryTired 7y ago
      > Suppose we get rid of economists and their models; then what do we use in their place to guide policy? Our intuition? Controlled small-scale experiments? Ethnography? Historical analysis? Debate among stakeholders? Why do decisions have to be justified using mathematical models? Especially when those models have to be simplified to the point where they're really just "stories" (Paul Krugman's words) about economies? Why not just tell the stories in English? In a number of fields where the object of study is particularly complex - biology and medicine being notable examples - mathematical models are ineffective. What makes you think the economy is different?
      - QuesnayJr 7y ago
        Because you can say things in mathematics that are hard to say in English. The stories are clearer when expressed in mathematical form. It's harder to bullshit your way through. Before the 70s/80s, everyone was somewhat confused about what they were arguing, particularly in macroeconomics. Now it's much clearer the kinds of claims we are making about the world.
        soVeryTired 7y ago
        I absolutely disagree. As an exercise, try translating a representative agent / rational expectations / DSGE model into English and see how ridiculous it sounds. Mathematics might force you to be internally consistent, but it also forces you to make bad assumptions. You can prove anything by reasoning from a faulty premise.
        QuesnayJr 7y ago
        Everyone who stays away from writing down a mathematical model is either confused themselves, or trying to confuse their readers. The advantage of a model with a representative agent, or rational expectations, is that I can see the assumption right there. They made it perfectly clear. They didn't try to fool me with a barrage of words. If I think their explanation is wrong because people are irrational, or because the agent heterogeneity is important, I can take their model and show how it falsifies their results. And one important thing we learned from those models (which I agree do not work) is that we don't get business cycles from irrationality or from agent heterogeneity -- we get them from the failure of markets to clear. Economists were confused about this for like a century, and thanks to mathematical modeling it became clear.
- sifar 7y ago
  Antifragile thinking is not really mainstream economics, is it ?

tomrod 7y ago

> Are you sure you can distinguish between those categories? Typically yes, through appropriate risk assessment. Cue "Antifragile" here.

soVeryTired 7y ago
So what does "appropriate risk assessment" look like? Vary some parameters and add confidence bounds to your outputs? If you're building an economic model and they economy enters a regime that you haven't seen before in your data, your model calibration is going to crumble. So maybe good models can be distinguished from bad when predicated on the assumption that the near future looks like the past. But when that changes (i.e. when you need your models most) is when your models become least reliable.
- SantalBlush 7y ago
  It's really not a question of how bad an economic model is, but whether there exists a better one we can use in its place. When people criticize (often rightly) economic models and point out their limitations, there is an implication that we should abandon them because they are so flawed. But this is wrong. It is not enough to avoid flawed models; we need something demonstrably better to replace them with, and this is where the critics tend to come up short. Suppose we get rid of economists and their models; then what do we use in their place to guide policy? Our intuition? No thanks. If someone has a superior theory of the economy, they must show it with logic and empirical observation. Until they do, we are stuck with our more flawed models.
  - soVeryTired 7y ago
    > Suppose we get rid of economists and their models; then what do we use in their place to guide policy? Our intuition? Controlled small-scale experiments? Ethnography? Historical analysis? Debate among stakeholders? Why do decisions have to be justified using mathematical models? Especially when those models have to be simplified to the point where they're really just "stories" (Paul Krugman's words) about economies? Why not just tell the stories in English? In a number of fields where the object of study is particularly complex - biology and medicine being notable examples - mathematical models are ineffective. What makes you think the economy is different?
    - QuesnayJr 7y ago
      Because you can say things in mathematics that are hard to say in English. The stories are clearer when expressed in mathematical form. It's harder to bullshit your way through. Before the 70s/80s, everyone was somewhat confused about what they were arguing, particularly in macroeconomics. Now it's much clearer the kinds of claims we are making about the world.
      - soVeryTired 7y ago
        I absolutely disagree. As an exercise, try translating a representative agent / rational expectations / DSGE model into English and see how ridiculous it sounds. Mathematics might force you to be internally consistent, but it also forces you to make bad assumptions. You can prove anything by reasoning from a faulty premise.
        QuesnayJr 7y ago
        Everyone who stays away from writing down a mathematical model is either confused themselves, or trying to confuse their readers. The advantage of a model with a representative agent, or rational expectations, is that I can see the assumption right there. They made it perfectly clear. They didn't try to fool me with a barrage of words. If I think their explanation is wrong because people are irrational, or because the agent heterogeneity is important, I can take their model and show how it falsifies their results. And one important thing we learned from those models (which I agree do not work) is that we don't get business cycles from irrationality or from agent heterogeneity -- we get them from the failure of markets to clear. Economists were confused about this for like a century, and thanks to mathematical modeling it became clear.
sifar 7y ago
Antifragile thinking is not really mainstream economics, is it ?