3 ms·

More of a social science It is as much a science as any other study of a complex system. Meaning, yes, it is a science. Economics can be deeply mathematical

More of a social science It is as much a science as any other study of a complex system. Meaning, yes, it is a science. Economics can be deeply mathematical once you consider the game theory aspect, analysis etc.

pmiller2 7y ago
The problem with mathematical economics is that the models are incomplete, and people aren’t perfectly rational actors. Math got us into the financial crisis in 2008.
- cameldrv 7y ago
  Yes, bond ratings were a big cause of 2008. Adverse selection/the market for lemons was a big part of that. Suppose the bond rating agency has a set of criteria to rate a bond AAA. You figure out how to structure your product to meet the letter of those criteria in one tranche, but the characteristics of the product's statistical distribution is very different than what the person who made the criteria was expecting. Or, you convince the bond rater into making some more statistical assumptions about the product that the product doesn't actually meet, and then you pass the AAA test with those assumptions in place. There is a blurry line between #2 and #3 which is basically just paying off the bond rater. Ultimately modern structured products creation and sales is basically just like old-school used car sales. Put sawdust in the motor, roll back the odometer, and tell the customer that it was driven by a little old lady that only took it to church and to the supermarket. Caveat emptor.
- codesushi42 7y ago
  Math got us into the financial crisis in 2008 That is a terrible misunderstanding of the 2008 financial crisis.
  - JudgeWapner 7y ago
    there's a charitable way to read that, which is that they did ample statistical modelling/risk analysis of CDS and "concluded" they were safe enough to bundle mortgage debt under. Then you could sort of "blame math" (or the abuse thereof) in that circumstance. I know CDS is only like 1/10 chapters of the whole crisis, however.
  - arethuza 7y ago
    I thought using Gaussian copula for CDO pricing was often regarded as being an important factor in facilitating the crisis?
- rumanator 7y ago
  > The problem with mathematical economics is that the models are incomplete, and people aren’t perfectly rational actors. I don't understand this criticism. All sciences are based on models that lie somewhere between incomplete and utterly wrong, but still these models are in fact useful and provide value. Take for instance physics, and how in some applications gravity is still modelled as a constant acceleration of 9.8m/s^2 pointing down. Although that model is very wrong it still works well enough to be useful.
- klodolph 7y ago
  The models are incomplete in chemistry, too, but that doesn’t make it “not a science”. In general you must choose a model on some scale between “accurate but too complex to be useful” and “easy to use but too inaccurate”. In the middle, hopefully, lies some theory which is simple enough to use in analysis and accurate enough to have applications. For example, you wouldn’t see a civil engineer design a bridge with quantum field theory. They would generally use a less accurate, incomplete model of materials physics. Something simple enough that they can parameterize and run simulations. Science is there in order for us to understand the limitations of these theories. This is why you can still e.g. use Lewis diagrams and get work done. They didn’t become “not science” just because they are an incomplete model of how bonding works, but through scientific experiment we have an understanding of how useful the model is.

pmiller2 7y ago

The problem with mathematical economics is that the models are incomplete, and people aren’t perfectly rational actors. Math got us into the financial crisis in 2008.

cameldrv 7y ago
Yes, bond ratings were a big cause of 2008. Adverse selection/the market for lemons was a big part of that. Suppose the bond rating agency has a set of criteria to rate a bond AAA. You figure out how to structure your product to meet the letter of those criteria in one tranche, but the characteristics of the product's statistical distribution is very different than what the person who made the criteria was expecting. Or, you convince the bond rater into making some more statistical assumptions about the product that the product doesn't actually meet, and then you pass the AAA test with those assumptions in place. There is a blurry line between #2 and #3 which is basically just paying off the bond rater. Ultimately modern structured products creation and sales is basically just like old-school used car sales. Put sawdust in the motor, roll back the odometer, and tell the customer that it was driven by a little old lady that only took it to church and to the supermarket. Caveat emptor.
codesushi42 7y ago
Math got us into the financial crisis in 2008 That is a terrible misunderstanding of the 2008 financial crisis.
- JudgeWapner 7y ago
  there's a charitable way to read that, which is that they did ample statistical modelling/risk analysis of CDS and "concluded" they were safe enough to bundle mortgage debt under. Then you could sort of "blame math" (or the abuse thereof) in that circumstance. I know CDS is only like 1/10 chapters of the whole crisis, however.
- arethuza 7y ago
  I thought using Gaussian copula for CDO pricing was often regarded as being an important factor in facilitating the crisis?
rumanator 7y ago
> The problem with mathematical economics is that the models are incomplete, and people aren’t perfectly rational actors. I don't understand this criticism. All sciences are based on models that lie somewhere between incomplete and utterly wrong, but still these models are in fact useful and provide value. Take for instance physics, and how in some applications gravity is still modelled as a constant acceleration of 9.8m/s^2 pointing down. Although that model is very wrong it still works well enough to be useful.
klodolph 7y ago
The models are incomplete in chemistry, too, but that doesn’t make it “not a science”. In general you must choose a model on some scale between “accurate but too complex to be useful” and “easy to use but too inaccurate”. In the middle, hopefully, lies some theory which is simple enough to use in analysis and accurate enough to have applications. For example, you wouldn’t see a civil engineer design a bridge with quantum field theory. They would generally use a less accurate, incomplete model of materials physics. Something simple enough that they can parameterize and run simulations. Science is there in order for us to understand the limitations of these theories. This is why you can still e.g. use Lewis diagrams and get work done. They didn’t become “not science” just because they are an incomplete model of how bonding works, but through scientific experiment we have an understanding of how useful the model is.