Modern Monetary Theory discussion

[El_Machinae]

I follow one Chicago School economist, and he very often complains that there's too much math in the macro publications. And a lot of the problem is p-hacking, where people compile datasets and then run endless regressions to get a paper.

I'd not know. I actually only listen to economists talk, where they're explaining. I don't know the literature that the Ph.D. students are struggling to produce.

 

[Lexicus]

The problem with macro isn't too much math per se, it's the fact that the whole discipline is framed by an inappropriate analogy to classical mechanics embodied in the concept of equilibrium.

 

[Hygro]

Point made by Noah Smith in a recent blog similar to what I've been saying.




https://noahpinion.substack.com/p/mmt-and-the-meme-ification-of-macro

I read it. Now go watch Steve Keen’s series modeling MMT’s stock flow consistent economy YouTube series.

 

[uppi]

I believe the distinction between soft and hard science is mostly propaganda in order to discredit other disciplines as "not really proper science". the main difference between what we call "soft science" and "hard science" is imho the amount of variables that have to be considered. working with human subjects makes this amount of variables approach infinity very, very fast. which is why, say, doing a clinical trial is so incredibly expensive and extensive. many experiments from the hard sciences may be more "in a vacuum" compared to those from soft sciences, but are obviously still influenced by the researchers conducting the experiment and all the other people involved.

"soft sciences" have the same standards (or should have, I can't speak for your favorite journal in terms of methodology, experimental design, sampling, hypotheses, and even theoretical models (if we believe Kuhn anyway).

There is a very important difference between "soft science" and "hard science", which is the amount of evidence generally expected in support of a claim. You need to be very brave to present a physics paper with a p-value of 0.05, which is often accepted in other discipline. This means that in those disciplines more than 1 in 20 claims of experimental evidence are expected to be wrong even in the most ideal case (from which we are far away).

The reason for this is that it is much easier to collect evidence in the "hard sciences". Measuring a few weeks longer is something different than subjecting another 1000 people to a trial.

the fact that many studies in economics "lack rigor" as you state may also be a result of that scientific discipline having completely different standards than your discipline does. just as a quick generalization: in humanities a researcher is usually asked to examine his/her own biases, and not doing that would be akin to a grave mistake in many contexts. I doubt this is stressed as much in mathematics, although it may also play a crucial role there, too. that is just not how the discipline of mathematics has historically evolved, thus this criteria is not stressed as relevant.

I would (reluctantly) concede the point if you were talking about physics, but mathematics? Either a proof is correct or it is not, there is no room for personal bias there.

 

[Samson]

The problem with macro isn't too much math per se, it's the fact that the whole discipline is framed by an inappropriate analogy to classical mechanics embodied in the concept of equilibrium.

I suspect I am just restating this, but it seems that the assumptions are just wrong. This graph is supposed to model reality:

Is there any goods for which this bears any relationship to reality? No manufactured goods follow the supply curve, in that if I wanted to say increase the global supply of phones by 10% that does not mean that each phone will cost 10% more to make. Food may follow the supply curve, in that to increase the supply of food by 10% means putting less productive land under the plough so the extra food will cost more to produce, but the demand curve of food does not look anything like this, as in if food was cheaper I would not eat more food. And that is before we get to intellectual property, where there is zero extra cost per extra supply.

 

[Lexicus]

I suspect I am just restating this, but it seems that the assumptions are just wrong.

More importantly, wrong in ways that cause the model to be unable to predict many situations that occur in reality.

 

[El_Machinae]

Supply and demand isn't so bad when it comes to The Chicago School. But they never seem to understand that dollars are legal tender, and are the cheapest thing to use to collapse debt.

Sometimes when I'm listening to them speak, I just substitute apples instead of using dollars. At that point, the conversation is much more sensible.

 

[Lexicus]

I suspect I am just restating this, but it seems that the assumptions are just wrong. This graph is supposed to model reality:

Is there any goods for which this bears any relationship to reality? No manufactured goods follow the supply curve, in that if I wanted to say increase the global supply of phones by 10% that does not mean that each phone will cost 10% more to make. Food may follow the supply curve, in that to increase the supply of food by 10% means putting less productive land under the plough so the extra food will cost more to produce, but the demand curve of food does not look anything like this, as in if food was cheaper I would not eat more food. And that is before we get to intellectual property, where there is zero extra cost per extra supply.

There is actually a whole chapter of Steve Keen's book Debunking Economics that discusses how the mathematical foundations of the supply and demand curves are extremely shaky - and that's without even getting into their empirical validity. I forget the details (and probably lack the competence to really grasp it) but the fact that the curves can, in any real-world situation, take almost any shape is a serious problem for the conventional modelling based on equilibrium from the interaction of supply and demand.

And that is before we get to intellectual property, where there is zero extra cost per extra supply.

And another classic example of a market that doesn't work according to classical logic is the labor market. How well do you think a model that can't describe a labor market is going to describe a modern economy where virtually everyone either participates in the labor market or is a dependent of someone who does?

 

[Cutlass]

My argument here would be "What evidence is presented that traditional economics is any more "unbiased search for truth" than MMT?" For example, they state "Keynesianism really works", and present as evidence this. If we compare this to an actual search for truth, eg. the pfizer vaccine protocol there are some big differences. The core elements that are missing is a falsifiable hypothesis, a description of what the data would look like in both the accept and reject situations, and some indication that these were defined before looking at the data. Without that, it is very hard for me to believe that there is good empirical evidence to show that one model is better than any other.

I am not at all an economist, but I spend most of my time these days building mathematical models that can process pre-existing data in a way that allows me to accept or reject hypotheses to a standard that gets my work published in peer reviewed journals. I have never seen an economic text with anything like the rigor that would get it published in a "real" science journal, and see no reason why considering the vast availability of economic data. My conclusion is that no economists care about truth as much as advocacy.

I'd say that there is both more academic rigor in economics than many people think, and more politics masquerading as economics than most people think.

 

[Samson]

I'd say that there is both more academic rigor in economics than many people think, and more politics masquerading as economics than most people think.

Can you explain to me how the document they present as evidence that "Keynesianism really works" shows that? it looks to me like a parameterization exercise, and while I have built models that would not parameterise, the fact that it is possible is not at all proof of a models validity.

 

[Hrothbern]

Supply and demand isn't so bad when it comes to The Chicago School. But they never seem to understand that dollars are legal tender, and are the cheapest thing to use to collapse debt.

Sometimes when I'm listening to them speak, I just substitute apples instead of using dollars. At that point, the conversation is much more sensible.

I use apples and working hours

 

[amadeus]

Is there any goods for which this bears any relationship to reality?

I’d say most of them do. There are other variables for each individual good, but as a general principle I would say it’s sound.

No manufactured goods follow the supply curve, in that if I wanted to say increase the global supply of phones by 10% that does not mean that each phone will cost 10% more to make.

I wouldn’t infer a 1:1 relationship on this single graph alone. If you’re going to produce more phones, the producers of components are likely (again, contingent on many variables) to increase their production to meet your demand.

[...] if food was cheaper I would not eat more food.

That’s where (in)elasticity comes into play. But let’s take one item out of the grocery: apples. Your demand for apples is probably somewhat elastic, in that I would guess you wouldn’t pay £100 for one, but you might be inclined to get one for a farthing.*

Spoiler fun* fact :

A farthing was 1/4 of a penny in pre-decimal British currency.

4 farthings to a penny,
12 pennies to a shilling, and
20 shillings to the pound.

So a farthing is 1/960 of the old pound sterling.

 

[Samson]

I’d say most of them do. There are other variables for each individual good, but as a general principle I would say it’s sound.

Care to give an example of one?

I wouldn’t infer a 1:1 relationship on this single graph alone. If you’re going to produce more phones, the producers of components are likely (again, contingent on many variables) to increase their production to meet your demand.

Yes, and the cost per unit will go down when that happens, not up.

That’s where (in)elasticity comes into play. But let’s take one item out of the grocery: apples. Your demand for apples is probably somewhat elastic, in that I would guess you wouldn’t pay £100 for one, but you might be inclined to get one for a farthing.*

Spoiler fun* fact :

A farthing was 1/4 of a penny in pre-decimal British currency.

4 farthings to a penny,
12 pennies to a shilling, and
20 shillings to the pound.

So a farthing is 1/960 of the old pound sterling.

There may be a few luxury items of food that this applies to, but very little and not apples.

 

[amadeus]

Care to give an example of one?

Oranges, since we’re bound to start comparing them anyway.

Yes, and the cost per unit will go down when that happens, not up.

Yes, and I’m not surprised. What part do you find inconsistent with the graph you posted?

 

[Samson]

Oranges, since we’re bound to start comparing them anyway.

As something like a luxury food, this is may be as close as it comes to matching the demand curve. However without evidence we cannot be sure, would you eat more oranges if they were 10% cheaper? Looking at the numbers, I see no evidence this really work (most UK oranges come from Spain). Also, are they going to follow the demand curve? If the next land that would be put to orange cultivation is not worse than that currently used how would that work, and is there evidence that this representation of the cost function is useful?

Yes, and I’m not surprised. What part do you find inconsistent with the graph you posted?

The graph has a linear and positive relationship between volume and unit cost to produce. If the cost to produce goes down as the volume goes up that is absolutely inconsistent with this graph, and will likely make conclusions based on the model this graph describes unreliable.

 

[amadeus]

The graph has a linear and positive relationship between volume and unit cost to produce. If the cost to produce goes down as the volume goes up that is absolutely inconsistent with this graph, and will likely make conclusions based on the model this graph describes unreliable.

The graph is an abstraction, it’s only linear in its conception. The unaccounted variables create the drop in price, which looks to contradict the graph.

 

[Samson]

The graph is an abstraction, it’s only linear in its conception. The unaccounted variables create the drop in price, which looks to contradict the graph.

Then we get back to the original point. If reality does not match the assumptions of the model, and we do not have empirical evidence of the power of the models predictions, what use is it in the real world?

 

[amadeus]

Then we get back to the original point. If reality does not match the assumptions of the model, and we do not have empirical evidence of the power of the models predictions, what use is it in the real world?

I believe it tends to be truer than not true, even if I can’t prove it.

Thankfully, I’m not an economist and I’m not paid to empirically prove these concepts.

 

[El_Machinae]

The supply/demand curve IS an abstraction. The fact that mini economies-of-scale happens is a 2nd layer of complication. Increase demand more, and you'll find increased supply with diminishing returns of capital investment.
The criticism that the models predict a drive to equilibrium is valid, but the models also predict a drive to an average. Not really the same thing.

I dunno, this is a bit like calling physics 'wrong' because you're dropping a real cow from an airplane and not a spherical, frictionless, cow.

Many of the models are useful. They're also handy because they have pretty nicely defined terms and concepts, which speeds up discussion. The fact that we're people and not homo economicus can be used within the discussion framework. They just don't be able to seem to flit between using currency as an accounting short-hand and as a legal entity.

Economics has a huge psychology component, and the field is only just beginning to flirt with realizing that. And unlike, say, animal psychology experiments, even telling people

 

[Lexicus]

I dunno, this is a bit like calling physics 'wrong' because you're dropping a real cow from an airplane and not a spherical, frictionless, cow.

If you want to draw a comparison with physics, then mainstream macro models are like physics models that don't conserve energy.