[JumpCrisscross]

No, it would cover the models that work. The reason any respectable economics text starts with frictionless markets of rational and then immediately talks about market failure is because to recognise the latter as a divergence you need the former as a baseline. Looking at things and asking "how does this make you feel" isn't scientific. Learning models that, while not great have shown some predictive power, is better. Economics pedagogy is far from perfect, but backing into the model by studying deviations seems like the wrong way to do it.

[littlestymaar]

> Learning models that, while not great have shown some predictive power, is better.

The problem is that theses models have a predictive power approximatly the same that Aristotle Law of free fall. They describe a world that seems intuitive but is totally wrong in most cases. For instance the diminishing return hypothesis comes from Ricardo's studies of agriculture in the UK. Since then this hypothesis is almost always used when studying the law and supply and demand. The problem is : it's almost always false unless you're harvesting a natural ressource.

Teaching a model which is known to be a pretty terrible representation of reality is not a good thing because it gives a false sense of understanding, which is dangerous from a democratic point of view.

[JumpCrisscross]

> The problem is that theses models have a predictive power approximatly the same that Aristotle Law of free fall

Having done exercises of rolling carts down inclines and comparing the measured velocity to predicted velocity, and having taken supply schedules and use those to predict price variation, I can say one set of calculations was more accurate than the other. (No doubt this was due to data quality.) Simple regressions of supply and demand curves have uncanny predictive power.

They are not a terrible representation of reality in the same way that rigid bodies are not a terrible representation. They are a limited representation that naturally extends itself. Don't confuse modern macroeconomic models with the terrifically-successful microeconomics early econ tends to focus on.

[littlestymaar]

> and having taken supply schedules and use those to predict price variation, […] Simple regressions of supply and demand curves have uncanny predictive power.

Do you have access to the actual data you worked with ? I personally attempted to study real world data to get an idea of the actual elasticity of basic products (gas, real estate, tobacco) and I've never found a single relevant measurement. I'd be really interested to see a real life example of this phenomenon.

[JumpCrisscross]

> gas, real estate, tobacco

Gas is refined, real estate is inscrutable and tobacco highly regulated. That makes them hard. (I don't think we have any good models for real estate.) Try thermal coal or crude oil or pork. The EIA, U.S. DoE and CME have loads of data.