Why does everyone hate mmt?

OrphanSlug said:

Change always comes from those that others brand and try to dismiss as "extreme," in this case all MMT did was explain a few things that conventional economic principles have been wrong about.

Best example we have from recent history is the years after the financial collapse, monetary and fiscal policy responses to our economic condition saw FoxNews run nonstop "inflation is coming" commentary with full time sellers of Gold filling up the advertising space. Not only did inflation not happen the way they claimed it would, too many people bought overpriced gold that will take generations to recoup those investments.

MMT started to explain things in terms of realized impacts in our fiat money system, that those who are stuck in the gold standard days could not explain.

So then the false information and misleading polls started to come out to do what conservative economics always does, invoke fear through "socialist" and "Venezuela" talk. All devoid of what MMT really says or why.

Click to expand...

Contemporary economic theory doesn't fail to account for massive amounts of QE not materializing as inflation. I doubt Bernanke enacted QE because he paid attention to a small subset of unusual economists who claim they revolutionize monetary economics. The man worked with structural VAR, factor models, and DSGE models in his research and is an accomplished academic first and foremost. He sure didn't expect inflation would just rise on a massive scale, yet his framework was as close to contemporary mainstream theory as you could get it. Some of his work on financial frictions in DSGE models (1999) and his joint work with Boivin and Eliasz (2005) on the use of factor models in structural vector autoregressions still stimulate research to this day. His views are clearly similar to those of the bulk of the profession and he wasn't panicking about inflation. On the other hand, Peter Schiff was panicking on Fox News in front of an audience which wants to hear about government causing problems. Bernanke has a Ph.D. and conducts research, while Schiff only has a bachelor degree in economics and does not conduct research. It's not a stab taken at Schiff, as much as a statement about whose opinion more closely captures what up-to-date research in economics implies. This is further buttressed by the fact Krugman, Delong, and Summers didn't panic either about inflation. They all predicted beforehand that it would never rise above 3% during the crisis, some even suggesting it might even become negative during some months of the crisis depending on the measure you use for inflation. They're all prominent macroeconomists who hold mainstream views on economic theory.

Nowadays, what I would do if I had to accommodate QE without going into the complex and explicit modeling of its dynamics is to follow the advice of Wu and Zhang (2016) who proposed a nice way to summarize all monetary policy in a "shadow" interest rate that can become negative. This way, you can use a much more typical way to model monetary policy in an already very complicated model such as the one in Ascari, Phaneuf and Sims (2019) (pricing and wage setting Calvo-style frictions, financial borrowing by firms, firms produce intermediate goods in a network, firms face variable utilization rates of capital, adjustment costs to new investments, there is a whole continuum of goods and labor types, etc.). That model wouldn't imply massive inflation response under the ZLB to large QE, nor would it imply monetary policy is 100% impotent in such conditions... So, clearly, explaining these things is absolutely not beyond the scope of mainstream methods.


Moreover, it might be hard for you or anyone here to appreciate how good are our methods in economics since you likely never read a single paper on the subject. A good example is Ascari, Sims and Phaneuf (2019). I'll bring your attention to the quality of the match between theory and data. Their simulated economy matches evidence from the structural vector autoregression literature, as well as produce incredibly good match for contemporary and cross-correlations of variables like consumption, investment, output, real wages, etc. on top of correct volatility values for all of their cyclical component. Here is the paper in question:

APS March 2019.pdf

Page 36 onwards show statistics about how well the simulated economy (from the model) matches empirical data, as well as showing impulse response functions. Clear quantitative predictions in a transparent and flexible framework... That's why DSGE are so popular, period.

TheEconomist said:

Contemporary economic theory doesn't fail to account for massive amounts of QE not materializing as inflation. I doubt Bernanke enacted QE because he paid attention to a small subset of unusual economists who claim they revolutionize monetary economics. The man worked with structural VAR, factor models, and DSGE models in his research and is an accomplished academic first and foremost. He sure didn't expect inflation would just rise on a massive scale, yet his framework was as close to contemporary mainstream theory as you could get it. Some of his work on financial frictions in DSGE models (1999) and his joint work with Boivin and Eliasz (2005) on the use of factor models in structural vector autoregressions still stimulate research to this day. His views are clearly similar to those of the bulk of the profession and he wasn't panicking about inflation. On the other hand, Peter Schiff was panicking on Fox News in front of an audience which wants to hear about government causing problems. Bernanke has a Ph.D. and conducts research, while Schiff only has a bachelor degree in economics and does not conduct research. It's not a stab taken at Schiff, as much as a statement about whose opinion more closely captures what up-to-date research in economics implies. This is further buttressed by the fact Krugman, Delong, and Summers didn't panic either about inflation. They all predicted beforehand that it would never rise above 3% during the crisis, some even suggesting it might even become negative during some months of the crisis depending on the measure you use for inflation. They're all prominent macroeconomists who hold mainstream views on economic theory.

Nowadays, what I would do if I had to accommodate QE without going into the complex and explicit modeling of its dynamics is to follow the advice of Wu and Zhang (2016) who proposed a nice way to summarize all monetary policy in a "shadow" interest rate that can become negative. This way, you can use a much more typical way to model monetary policy in an already very complicated model such as the one in Ascari, Phaneuf and Sims (2019) (pricing and wage setting Calvo-style frictions, financial borrowing by firms, firms produce intermediate goods in a network, firms face variable utilization rates of capital, adjustment costs to new investments, there is a whole continuum of goods and labor types, etc.). That model wouldn't imply massive inflation response under the ZLB to large QE, nor would it imply monetary policy is 100% impotent in such conditions... So, clearly, explaining these things is absolutely not beyond the scope of mainstream methods.


Moreover, it might be hard for you or anyone here to appreciate how good are our methods in economics since you likely never read a single paper on the subject. A good example is Ascari, Sims and Phaneuf (2019). I'll bring your attention to the quality of the match between theory and data. Their simulated economy matches evidence from the structural vector autoregression literature, as well as produce incredibly good match for contemporary and cross-correlations of variables like consumption, investment, output, real wages, etc. on top of correct volatility values for all of their cyclical component. Here is the paper in question:

APS March 2019.pdf

Page 36 onwards show statistics about how well the simulated economy (from the model) matches empirical data, as well as showing impulse response functions. Clear quantitative predictions in a transparent and flexible framework... That's why DSGE are so popular, period.

Click to expand...

It is true that DSGE models are popular but popularity does not always imply solidarity.

The DSGE Model Quarrel (Again) | Bruegel

TheEconomist said:

APS March 2019.pdf

Page 36 onwards show statistics about how well the simulated economy (from the model) matches empirical data, as well as showing impulse response functions. Clear quantitative predictions in a transparent and flexible framework... That's why DSGE are so popular, period.

Click to expand...

DSGE models are popular and do explain behaviors under normal conditions, but suffer from at least two severe flaws.

First, "stationary" does not really exist. No matter if at economic peak or trough, there will be shift to measurements of aggregate demand. Equilibrium conditions altered to look like stationary terms for monetary policy then become problematic. Reactionary instead of planning, and the Central Bank does this repeatedly.

Second, underlying economic conditions (such as distributions and trade implications) can shift unexpectedly which result in changes to how central banking expects money creation to exist in our modern fiat money system. Assumptions of growth, willingness to take on debt at costs that fluctuate, etc. all suggest that not only do stationary terms not exist but the principles of economic continuance suggest medium term and long term corrective action.

The reality is in terms of teaching the principles of central banking DSGE models have a role in explaining expectations, but they are at best shaky mathematical models that entirely ignore changes in behavior based on the very bubble and pop economics we know our own economic model is prone to. Because of this decisions by economic players of all impact (all income quintiles) are subject to periods that are not routine, not predictable, and tend to show why we look to indicators of uncertainty. Something that DSGE models also do not bother with very much.

Trying to forecast causality between all the inputs to DSGE modeling as constants is its ultimate failure. You rip, or even alter, the variable in any part of any of the equations for future expectation and the models fall apart.

But back to my point, QE was not based solely on an argument for or against MMT. QE was about dealing with the cost of debt in the face of aggregate demand fault, and in the balance was a needed removal of toxic assets no one really wanted to hold in the short term. The FoxNews driven intention to panic everyone over monetary expansionary policy ended up unrealized. MMT explained why in a way that most conventional economic principles could not... and still do not.

You are welcome to attack my economic background, claim what I have and have not read, be by guest. But at the end of the day MMT is making suggestions that mean more to our modern fiat money system in terms of both behavior across the income quintiles and why we see such economic turbulence.

Vadinho said:

It is true that DSGE models are popular but popularity does not always imply solidarity. The DSGE Model Quarrel (Again) | Bruegel

Click to expand...

You ignored the entire post, as well as the paper I references, focusing your entire attention on a single line at the very end of my post. I never even went close to making the claim their popularity justified their use. I merely stated what I believe was an interesting advantage of DSGE models: everything is transparent in a mathematical model. You can take any DSGE model, change any subset of the mechanisms involved in how it explains large scale behavior, and compute the consequences. This is why I called them transparent: it is literally impossible to hide changes in assumptions.

It wasn't an apology of DSGE modeling as the only way to think through a problem. Nobody in social science has ground to make that claim about any theoretical methodology in any field. My point merely was that it's demonstrably useful, that it has major advantages and that most people don't even know how we work with them, although they do care to criticize. I'm not trying to shield DSGEs. I am trying to expose their working and their results so that nonexperts do not rely solely on critiques to know what is a DSGE and how it works. I'm trying to make this a discussion. If you notice, moreover, I didn't make any accusation at MMT: I didn't read about it, so I do not feel peculiarly in a good position to say what it can and cannot do.

OrphanSlug said:

You are welcome to attack my economic background, claim what I have and have not read, be my guest. But at the end of the day, MMT is making suggestions that mean more to our modern fiat money system in terms of both behavior across the income quintiles and why we see such economic turbulence.

Click to expand...

My intention was not to attack your credentials. I'm not trying to cloak economic theory in a veil of jargon and technicalities, although I understand I might have been too clumsy to make it obvious in my previous post. If possible, I'd rather give you key bits of information so you have a better idea of what economists do in reality and not solely collect your understanding of what they do from their critiques. If you want to know my specific attitude, I contend that all models are wrong in some ultimate sense, but some might be good enough to be useful for some purposes. This suggests pursuing many different ways to do things is probably a sane thing to do.

OrphanSlug said:

First, "stationary" does not really exist. No matter if at economic peak or trough, there will be a shift to measurements of aggregate demand. Equilibrium conditions altered to look like stationary terms for monetary policy then become problematic. Reactionary instead of planning, and the Central Bank does this repeatedly. Second, underlying economic conditions (such as distributions and trade implications) can shift unexpectedly which result in changes to how central banking expects money creation to exist in our modern fiat money system. Assumptions of growth, willingness to take on debt at costs that fluctuate, etc. all suggest that not only do stationary terms not exist but the principles of economic continuance suggest medium term and long term corrective action.

Click to expand...

There is some truth to your comment here, but it is not exactly correct because DSGE models often explicitly model nonstationary behavior. Let me try to explain what they really do, hopefully in a way that everyone understands. This one is hard to spot because in sufficiently simple DSGE models with sufficiently simple modeling of long term dynamics, ignoring the long term behavior is without incidence on your results which means that, whenever possible, people skip it altogether. Now, the modeling of growth over the long run in DSGE models depend on what is the purpose of the model. We usually design a model to fit a specific subset of questions. What I am acquainted with are cyclical models: they are designed to capture some of the dynamics involved in movements that last anywhere between 8 to 32 quarters. What they really say is that variables are stationnary around a trend. The "trend" they posit is equally simple. It says some factors of growth like population, technological factors and factors involving the transformation of investment into the capital are on average constant. This can be tested: the DSGE model will imply that the logarithm of output, consumption, and investment will behave like a straight line, with small ups and downs.

Obviously, as you pointed out, things change and we can in fact patch together datasets to go back to 100 or even 150 years for the US. There are technical issues involved since merging datasets usually means considering as equivalent measures that are in fact different, but it is the best we can do and it will confirm your suspicion. Over enough time, the simpler model of trend behavior is not very good. However, I invite you to look at the data for the last 50 years in the US and plot their logarithm. You will notice immediately the assumption is far from stupid, although manifestly imperfect. The question is not whether any given model is perfect, or if we should only use DSGE models. The question is whether the approximation is good enough for the much more modest purpose of helping us to talk about a specific set of problems. We fudge long term dynamics when we don't think a major part of explaining a certain set of phenomena doesn't hinge on changes in the determinants of long terms behavior. It's not even a big fudging either for a 40 to 50 year period.

More to the point and, much to the ignorance of most people, you can write DSGE models that are explicitly designed to explain some phenomena as resulting from changes in the determinants of long term behavior. A large part of the macroeconomic literature with which I unfortunately only have a course-based acquaintance relies on studying the transition dynamics. In these studies, cross-sections are usually more interesting than the time dimension of the problem. That would include studies on the determinants of income, wealth and consumption inequalities such as in Quadrini (2000), or of various programs (severance pay, firing penalties, unemployment benefits, etc.) on wages, unemployment, duration of employment and unemployment, or the size of firms.

OrphanSlug said:

The reality is in terms of teaching the principles of central banking DSGE models have a role in explaining expectations, but they are at best shaky mathematical models that entirely ignore changes in behavior based on the very bubble and pop economics we know our own economic model is prone to. Because of this, decisions by economic players of all impact (all income quintiles) are subject to periods that are not routine, not predictable, and tend to show why we look to indicators of uncertainty. Something that DSGE models also do not bother with very much.

Click to expand...

Most DSGE models make use of a representative household which is the same thing as arguing it is sufficient to look at the mean behavior to determine aggregate results. I'll give you an example so you can guess when your comment is more relevant on your own.

There is a literature which insists on modeling entire distributions of people: the population in the model might contain people who are employed, some who are unemployed, some who different levels of income, asset, productivity, occupations, etc. We'll take a simple example: people face an individual risk of unemployment and it's not possible to take some perfect insurance against it. The sought-after interest of this diversity at a theoretical level is in part that people with low wealth and who cannot borrow freely respond very strongly to transfer payments from the government and other types of policies. They spend 100% of windfall gains at the margin. The problem is that absent other forces, everyone agrees on the optimal amount of savings so you can get a huge mass of people around the mean value of wealth and the insanely more complicated modeling of entire distributions of people becomes essentially a waste of time: the mean behavior is enough to characterize everything. That's the Kruskell and Smith (1998) paper if some people want to check.

For the fact that people are different in some dimensions to be of any significance, the unusual behavior cannot be too isolated. Moreover, you do need these unusual people, or these normal people facing unusual circumstances, to all move in a way that reinforces their individual influence on aggregate outcomes. If their respective irregularity cancels out or partly offset each other, the difficulty you raise would be inconsequential. That, by the way, is how people talk in a seminar on macroeconomic theory. We all have in our heads potentially relevant mechanisms and a DSGE is just a formalized version of what happens when you package a few of them together. They're useful, though not correct.

There is ample irony in the criticism that is directed at macroeconomic theory.

The way macroeconomic models are designed involves a near paranoia with the fear of misspecification and an obsession with the empirical congruence of the theory. Although the assumptions are abstract and technical, each of them is made with the intention of making certain mechanisms emerge in the model. Each of those mechanisms is what macroeconomists quarrel about using intuition, existing theory, theories from other fields, as well as extensive microeconomic data. It is a large part of the job of macroeconomists to figure out simple tricks that can be demonstrated to capture very real aspects of a phenomenon: every trick needs to be motivated by facts and often undergoes decade long debates before becoming established practice. And, if it wasn't enough, the models in question are systematically compared with statistics drawn from the data. The model is calibrated in various ways, estimated in various ways and compared to both simple and complicated statistics so that everyone can notice its strengths and weaknesses. Obviously, we even often insist on formal statistical tests to be done on top of less formal evaluations. All of this with the very modest goal of explaining a subset of economic activity. Only a handful of more vocal economists who are obviously looking after their personal finances make sweeping claims about these phenomena on the basis of these models (or lofty abstract claims about why the assumptions of those models are wrong). It's admittedly not meant to be a perfect description, but a useful toolbox for talking openly about what is going on.

The standards I described above for judging DSGE models are high standards for social sciences. Just about every bit of evidence of multiple kinds is brought to bear on the matter before a model is deemed worth publishing. The claim I am making and which I believe most practitioners would make is that it earns the methodology a place at the table for talking about policy. Maybe not the only place, but at least a place.

The irony I claimed above is that critiques must impose harsher standards still to reject DSGE wholesale while applying invariably less stringent criterion when it comes to their favored explanation or methods. It is only convenient that these theories help them diversify their products as intellectuals, or fall on a specific side of a political dispute they privilege as politicized nonexperts. Either way, the argument is not limited to questioning, or challenging economists to justify their models and requiring them to adapt their models if they fail to account for relevant facts. The argument is usually made by uninformed parties and carried to the extreme of wholesale rejection of an entire methodology they neither understand nor try to understand in the first place.


On the other hand, did you see me make an uninformed critique of MMT? No. Did you see me say DSGE is the only way to work? No. I only said the methodology had advantages that make it suitable for open discussions. If you're going to fixate madly as some do on the maximum operator and point at laboratory experiments that provide evidence at least some of these formulations will have trouble explaining all human behavior under all circumstances, you're going to face a more serious problem of congruence in your beliefs elsewhere: the same standard of imperfect fit outside the scope of application of a theory can be used to reject all theory of human behavior, perception, psychology, etc., provided you put that bar high enough. To me, that's the ridiculous thing to do and what I encountered more than my share of time. Abstractions are tools, no more and no less.

OrphanSlug said:

But at the end of the day, MMT is making suggestions that mean more to our modern fiat money system in terms of both behavior across the income quintiles and why we see such economic turbulence.

Click to expand...

That comparison can only be made by reference to both MMT and the more common DSGE methodology. I am not acquainted with MMT and my knowledge of DSGE models is limited. My own research is more empirical and statistical than theoretical. Regardless, I am not in any position to say what advantage or disadvantage one might have over the other. In your case, the situation is reversed, so I don't see how you can compare them more than I can.

Moreover, there is a reason I keep talking about DSGE models (plural): it's a methodology, a way to talk about theories and to work on theories. It's not a theory, but a set of potential theories. It's hard to tell what it can and cannot do. The paper I showed here was a response to an earlier belief that something was impossible in DSGE models (you can read about the Barro-King curse if you like). The dispute involved experts and it took 30 years to show them wrong with a counterexample. Imagine how wrong you or I could be if we tried to delineate the limits of the methodology when actual users can be wrong about something quite fundamental. You can try, but that requires knowledge of the minute details of the methodology and it's not obvious that the criticism will be useful, let alone justified in any serious sense.

However, you don't need to accept the theories of economists wholesale either. It's not because it's hard to attack a method that you cannot try to see if it's easy to poke holes in a specific theory. I'm telling you it's impossible to avoid having at least some holes, so you'll have to judge for yourself if the model does explain the set of facts it claims to explain in a way that lends itself to be used in a discussion.

As for MMT itself, it's possible that it attracts people who are interested in different things or provide interesting insights into human behavior that are seldom covered in one form or another. Would you care to pick an example of something MMT explains, how it is evaluated empirically by its users, etc.? Examples are better than lofty claims here.


I have no idea what MMT can and cannot do, although I am somewhat worried about the peculiar political divide it seems to create here. One notable feature of mainstream economics is that it attracts an unusually politically balanced set of researchers. The field still leans around 3 to 1 to the left in surveys, but that's way more balanced than other social science departments and it can be compared with departments in technical fields like mathematics, biology, etc. It seems to sell a story that goes enough to something of substance that people almost everywhere on the political spectrum find it relevant enough to study all the way to the Ph.D. in this field. By contrast, it seems like people attracted to MMT definitely lean to the left. I am wrong? Is it just an artifact of this forum?

My problem is basically based on our tendency to be self-serving and to seek confirmatory evidence: a biased field is suspicious, regardless of what it says. It would also be curious if the theory lent itself to systematically favoring intervention programs and intentional designs of such programs by groups of experts. Intellectuals are just as interested parties to political discussions as are oil companies sending marketing representatives to talk about climate change, even though people seldom think about it. If your research is heavily financed by an agency dealing with affirmative action, your opinion on affirmative action might not be as objective as it might seem.

TheEconomist said:

My intention was not to attack your credentials. I'm not trying to cloak economic theory in a veil of jargon and technicalities, although I understand I might have been too clumsy to make it obvious in my previous post. If possible, I'd rather give you key bits of information so you have a better idea of what economists do in reality and not solely collect your understanding of what they do from their critiques. If you want to know my specific attitude, I contend that all models are wrong in some ultimate sense, but some might be good enough to be useful for some purposes. This suggests pursuing many different ways to do things is probably a sane thing to do.

Click to expand...

I am not going to say that DSGE models are useless, but I will say they are not everything and we have enough economic commentary on the subject to suggest scrutiny is not out of line.

Again, I will not pretend to be the world’s expert at economics. It is just that all of my study, reading on the subject, and writings on the matter for whatever college course reason all takes me back to the same thing... the study of behavior.

I have never taken issue with economic models that on a long enough timeline are ultimately wrong simply because there is too much unknown in how every participant (by any bracketed means) reacts to economic turbulence, and generally speaking it is based on some aggregate demand fault.

Any economic modeling, at any point in the economic cycle, is based on what is known and assumptions on what is not known. Applied behavior based observations mashed into on all the relevant conditions at the time the model is being put in place.

I will not say all of these economists who put high stock in DSGE modeling are entirely wrong either, but I will point out the risks with these models on a long enough timeline and usually based on what MMT does explain. Does that mean MMT is beyond criticism? Of course not, but when I see conventional wisdom get a few things dead wrong I tend to migrate to economic principles based on what we can see. Namely it involves when and why inflation should occur.

I have no choice but to consider what set of economic principles tells me why things resulted in what they did, and that means expanding the scope of models that could apply.

The good news is we agree far more than we do not.

TheEconomist said:

Would you care to pick an example of something MMT explains, how it is evaluated empirically by its users, etc.? Examples are better than lofty claims here..

Click to expand...

Why inflation did not occur as most right leaning economists suggested would happen post all the rounds of QE.

TheEconomist said:

You ignored the entire post, as well as the paper I references, focusing your entire attention on a single line at the very end of my post. I never even went close to making the claim their popularity justified their use. I merely stated what I believe was an interesting advantage of DSGE models: everything is transparent in a mathematical model. You can take any DSGE model, change any subset of the mechanisms involved in how it explains large scale behavior, and compute the consequences. This is why I called them transparent: it is literally impossible to hide changes in assumptions.

It wasn't an apology of DSGE modeling as the only way to think through a problem. Nobody in social science has ground to make that claim about any theoretical methodology in any field. My point merely was that it's demonstrably useful, that it has major advantages and that most people don't even know how we work with them, although they do care to criticize. I'm not trying to shield DSGEs. I am trying to expose their working and their results so that nonexperts do not rely solely on critiques to know what is a DSGE and how it works. I'm trying to make this a discussion. If you notice, moreover, I didn't make any accusation at MMT: I didn't read about it, so I do not feel peculiarly in a good position to say what it can and cannot do.

Click to expand...

I read your post and spent several minutes watching various economists discuss the model and why it was very useful and popular. I am not disputing your point at all, I am merely pointing out that even that model has its faults according to economists. I wonder why supporting that model implies that MMT is invalid or wrong. From my study of MMT as a layman, no one in the MMT field challenges this model or suggests an MMT based model to replace it. Steve Keen does have serious questions about modeling but he gets so far into the weeds that it is hard to follow his criticisms without having intimate knowledge of the econometrics involved in the models he challenges and suggests as alternatives. Modeling is extremely complex, it demands an almost lifetime of study, advanced mathematics and a hell of a lot of data that must be updated and collected constantly. Before you go into another long post meant to establish your bona fides as an "economist", lets see you reference a video or lecture by an MMT economist that has something to do with DSGE modeling.

Vadinho said:

I read your post and spent several minutes watching various economists discuss the model and why it was very useful and popular. I am not disputing your point at all, I am merely pointing out that even that model has its faults according to economists. I wonder why supporting that model implies that MMT is invalid or wrong.

Click to expand...

I only tried to point out the problems with some of the criticism that has been raised here regarding DSGE models. For all I know, it might even be possible to frame much of what MMT academics are saying in a DSGE model. It wouldn't the first time ideas that sound very different turned out to be compatible. I also made it clear DSGE models are just one of many useful tools. We use an entire array of more or less theoretically informed methods ranging from very empirical to more theoretical in macroeconomics -- and I think we're all the better for using so many of them.

Vadinho said:

From my study of MMT as a layman, no one in the MMT field challenges this model or suggests an MMT based model to replace it. Steve Keen does have serious questions about modeling but he gets so far into the weeds that it is hard to follow his criticisms without having intimate knowledge of the econometrics involved in the models he challenges and suggests as alternatives.

Click to expand...

I don't know if proponents of MMT reject or not the DSGE methodology, though it's the unsurprising first thing almost every single person who proposes unusual theories or methods in economics seem to do. However, I wasn't criticizing Steven Keen; I was answering to people on this forum.

OrphanSlug said:

I am not going to say that DSGE models are useless, but I will say they are not everything and we have enough economic commentary on the subject to suggest scrutiny is not out of line.

Again, I will not pretend to be the world’s expert at economics. It is just that all of my study, reading on the subject, and writings on the matter for whatever college course reason all takes me back to the same thing... the study of behavior.

I have never taken issue with economic models that on a long enough timeline are ultimately wrong simply because there is too much unknown in how every participant (by any bracketed means) reacts to economic turbulence, and generally speaking it is based on some aggregate demand fault.

Any economic modeling, at any point in the economic cycle, is based on what is known and assumptions on what is not known. Applied behavior based observations mashed into on all the relevant conditions at the time the model is being put in place.

I will not say all of these economists who put high stock in DSGE modeling are entirely wrong either, but I will point out the risks with these models on a long enough timeline and usually based on what MMT does explain. Does that mean MMT is beyond criticism? Of course not, but when I see conventional wisdom get a few things dead wrong I tend to migrate to economic principles based on what we can see. Namely it involves when and why inflation should occur.

I have no choice but to consider what set of economic principles tells me why things resulted in what they did, and that means expanding the scope of models that could apply.

The good news is we agree far more than we do not.

Click to expand...

In case you were curious about it, DSGE models aren't the only tools we use to make sense of phenomena. We also have more statistically inclined models such as the whole literature on structural vector autoregressions (SVAR), instrumental methods, as well as other methods more typical of microeconomics. All these methods, however, share the need to impose some assumptions to derive causal estimates. DSGE happen to impose the most structure on the data, but you can go in with a lot fewer presumptions about how things work. If you're curious, you can get what we call impulse response functions (IRF) (how variables react to things like policy changes) just by imposing adequate long term or short term restrictions. One example would be to agree with Milton Friedman that monetary policy hits GDP with a few months of delay -- so you force the contemporary impact of a variation in the Feds Funds Rate to 0. These things have complications of their own, but it's worth mentioning to nonexperts that we use more than one tool.

DSGE also allow computing these IRF, so you can compare with those obtained with SVAR. In fact, you can estimate DSGE models by trying to minimize a distance function between the DSGE's IRF and the IRF of some SVAR(s) of your choice if you believe the SVAR IRF is what your theory should explain in detail (Martin Eichenbaum, for example, does that a lot).

It is probably clear right now that I agree with you on using many methods is advantageous.



PS: Nice Pen Jillette quote, by the way. It's a rather profound bit of wisdom stated colloquial terms. We're all the weird kid in some way, or at some point.

TheEconomist said:

That comparison can only be made by reference to both MMT and the more common DSGE methodology. I am not acquainted with MMT and my knowledge of DSGE models is limited. My own research is more empirical and statistical than theoretical. Regardless, I am not in any position to say what advantage or disadvantage one might have over the other. In your case, the situation is reversed, so I don't see how you can compare them more than I can.

Moreover, there is a reason I keep talking about DSGE models (plural): it's a methodology, a way to talk about theories and to work on theories. It's not a theory, but a set of potential theories. It's hard to tell what it can and cannot do. The paper I showed here was a response to an earlier belief that something was impossible in DSGE models (you can read about the Barro-King curse if you like). The dispute involved experts and it took 30 years to show them wrong with a counterexample. Imagine how wrong you or I could be if we tried to delineate the limits of the methodology when actual users can be wrong about something quite fundamental. You can try, but that requires knowledge of the minute details of the methodology and it's not obvious that the criticism will be useful, let alone justified in any serious sense.

However, you don't need to accept the theories of economists wholesale either. It's not because it's hard to attack a method that you cannot try to see if it's easy to poke holes in a specific theory. I'm telling you it's impossible to avoid having at least some holes, so you'll have to judge for yourself if the model does explain the set of facts it claims to explain in a way that lends itself to be used in a discussion.

As for MMT itself, it's possible that it attracts people who are interested in different things or provide interesting insights into human behavior that are seldom covered in one form or another. Would you care to pick an example of something MMT explains, how it is evaluated empirically by its users, etc.? Examples are better than lofty claims here.


I have no idea what MMT can and cannot do, although I am somewhat worried about the peculiar political divide it seems to create here. One notable feature of mainstream economics is that it attracts an unusually politically balanced set of researchers. The field still leans around 3 to 1 to the left in surveys, but that's way more balanced than other social science departments and it can be compared with departments in technical fields like mathematics, biology, etc. It seems to sell a story that goes enough to something of substance that people almost everywhere on the political spectrum find it relevant enough to study all the way to the Ph.D. in this field. By contrast, it seems like people attracted to MMT definitely lean to the left. I am wrong? Is it just an artifact of this forum?

My problem is basically based on our tendency to be self-serving and to seek confirmatory evidence: a biased field is suspicious, regardless of what it says. It would also be curious if the theory lent itself to systematically favoring intervention programs and intentional designs of such programs by groups of experts. Intellectuals are just as interested parties to political discussions as are oil companies sending marketing representatives to talk about climate change, even though people seldom think about it. If your research is heavily financed by an agency dealing with affirmative action, your opinion on affirmative action might not be as objective as it might seem.

Click to expand...

Exactly.

Vadinho said:

I read your post and spent several minutes watching various economists discuss the model and why it was very useful and popular. I am not disputing your point at all, I am merely pointing out that even that model has its faults according to economists. I wonder why supporting that model implies that MMT is invalid or wrong. From my study of MMT as a layman, no one in the MMT field challenges this model or suggests an MMT based model to replace it. Steve Keen does have serious questions about modeling but he gets so far into the weeds that it is hard to follow his criticisms without having intimate knowledge of the econometrics involved in the models he challenges and suggests as alternatives. Modeling is extremely complex, it demands an almost lifetime of study, advanced mathematics and a hell of a lot of data that must be updated and collected constantly. Before you go into another long post meant to establish your bona fides as an "economist", lets see you reference a video or lecture by an MMT economist that has something to do with DSGE modeling.

Click to expand...

Economics is a lot like the weather, easy to forecast after the fact. But, before the fact all you have is models that predict and often those predictions are wrong. I don't personally believe economics will ever be any different than the weather. Anyone claiming to have all the answers is arrogant, biased, and prejudiced in their beliefs and, as TheEconomist pointed out, MMT is a theoretical model used by liberals and socialists to further their agenda.

TheEconomist said:

In case you were curious about it, DSGE models aren't the only tools we use to make sense of phenomena. We also have more statistically inclined models such as the whole literature on structural vector autoregressions (SVAR), instrumental methods, as well as other methods more typical of microeconomics. All these methods, however, share the need to impose some assumptions to derive causal estimates. DSGE happen to impose the most structure on the data, but you can go in with a lot fewer presumptions about how things work. If you're curious, you can get what we call impulse response functions (IRF) (how variables react to things like policy changes) just by imposing adequate long term or short term restrictions. One example would be to agree with Milton Friedman that monetary policy hits GDP with a few months of delay -- so you force the contemporary impact of a variation in the Feds Funds Rate to 0. These things have complications of their own, but it's worth mentioning to nonexperts that we use more than one tool.

DSGE also allow computing these IRF, so you can compare with those obtained with SVAR. In fact, you can estimate DSGE models by trying to minimize a distance function between the DSGE's IRF and the IRF of some SVAR(s) of your choice if you believe the SVAR IRF is what your theory should explain in detail (Martin Eichenbaum, for example, does that a lot).

It is probably clear right now that I agree with you on using many methods is advantageous.

Click to expand...

I never intended to suggest DSGE models are used in a vacuum, just pointing out flaws other economists offer as to why DSGE models can be problematic. Also, the way I was educated on SVARs comes down to linear representations of a vector of observable values within ranges, in terms of their own lags. But it is assumptions, be it minimal, that are still compatible with large class modeling efforts. The other thing to mention is there are competing SVAR methods. Mathematically natural, longer ran schemes tend to absorb shocks and unpredictability in the variables in a way that isolated or standard schemes do not.

Truth be told it does not necessarily matter to me that *some* assumptions are made, what does matter is when the assumptions produce a flawed model.

On a macroeconomic observance level what matters most to me is the reality of these economic indications that policy makers tend to pay more attention to, be it some combination of fiscal policy and monetary policy the indicators driving decision tend to use these models but observe the indications (ideally, even if they are wrong every so often and odds are because of political decision often devoid of economic basis.)

But I am still left with the same conclusions.

Conventional economics got wrong... What the rounds of QE would do to inflation, that everyone would rush to precious metals exploding their values, before the crash concluding that excesses in housing assets was not a problem, that the risks associated with rises in interest only and subprime mortgages could be mitigated by closed market valuations, that ending regulation on credit default swaps and other derivatives was safe.

What conventional economics still gets wrong... just about everything related to aggregate demand, that it is acceptable for the income quintiles to persistently get further apart, and labor rates competing on the international markets can be mitigated with tariffs.

If we are really going to talk about macroeconomic impacts and the observance of behavior then we have no alternative but to use the models but observe the indicators.

TheEconomist said:

PS: Nice Pen Jillette quote, by the way. It's a rather profound bit of wisdom stated colloquial terms. We're all the weird kid in some way, or at some point.

Click to expand...

Something about Pen Jillette, one of the more ignored but still great thinkers of our time. Some of his quotes and the reasons for them are nothing short of brilliant.

OrphanSlug said:

What conventional economics still gets wrong (...) that it is acceptable for the income quintiles to persistently get further apart, ...

Click to expand...

A few very important things many people push under the rug when it comes to measuring income inequality:
1. When household income data is used, one of the problems is that households in the top quantile tend to include more people, which exaggerates the differences. I think it was for 2010 that we're talking about some 69 million more Americans in top quintile households than in the bottom quintile;

2. The statistical agency responsible for income statistics in the US, unfortunately, conflate capital gains with other types of income. The problem is that this means someone who has bonds, stocks, housing, businesses, etc. and sells them can see their income for one year get yanked up very severely, even though the reality is that the gains correspond to an investment made often over years. A lot of the very high income derives from capital gains whereas almost all of the income of people in lower quintiles derive from labor income. That again exaggerates the differences across flesh-and-blood people;

3. Income quintiles, just as wealth quintiles, are not stable groups of people. A study whose authors escape me tracked a few thousand US workers over the 1975-1993 span. Of those who started in the bottom quintile, 95% ended up in higher quintiles. The turnover rates are also increasingly high as you move up.

It is dubious at best that what you would like to capture by looking at the evolution of the distribution of household income in the US is even remotely related to the picture you see. It's very easy to show examples where you get complete nonsense conclusions out of this -- e.g., you can devise examples that make you think people are getting poorer even when all of them are in fact getting richer.

And the study of the impact of the distribution of income on macroeconomic dynamics is rather recent. One strand of research is that of Benjamin Moll from Princeton if you are curious. Needless to say, this makes things quickly way more complicated to understand, which is probably one good reason why it took so long for us to look into it. As far as I know, economics is almost mute on this matter because it is insanely complicated. There is one caveat to this, but it doesn't come from a very quantitative field of economics, so it would only give you reasons to suspect "some" inequality is good without knowing what "too much" would look like.