The disappearance of macro models as something anyone takes seriously is SO refreshing. I've got a PhD in stats and have been baffled and infuriated forever that those can get published, much less taken seriously by anyone except those whose jobs depend on them being taken seriously. A bunch of made up equations based on baldly ridiculous assumptions that are just transparently mathified reactionary ideology. It's science! The plutocrats funded an entire movement and generation of owned academics and think tankers to embed as "common sense" the pretention that their massive rents and power were scientifically proven to be necessary. Still very successful and tons of damage to undo but it's nice the tide has finally turned.
I'll make it clearer: it's not that I won't pay for your newsletters, it's that I won't pay for any of them. I want a distributed model where my average views make my fees distributable over the authors, and limited by month. When that happens, I'll subscribe.
Unfortunately this is a bundle of experiments, and during a pandemic. So I expect in hindsight it won't be clear which caused what, or whether the lessons apply to other situations.
My take is the power of $1.9t is being overblown. It's actually mostly continuing existing spending: like unemployment insurance, SBA. The aid to local goverment mostly ensures they won't have to cut. The $1400 checks are against the backgound of the recent $600 and $1200 checks; bigger but with a signal that these are almost certainly the last.
For me the most interesting experiment is the largely accidental increase-most-people's-savings program. A huge portion of the population never lost their income, but found themselves spending much less under quarantine. And then the government sent them stimulus checks. So their savings have gone way up. Will they now all go roaring '20s as the clubs re-open? Yeeh, most of these people are 40+. I'm sure many will splurge in their own ways, but not necessarily on the sorts of commodities that are tracked by inflation indicators. Not a few are investing into personal businesses.
Will crypto survive it? The idea that government money creation inherently makes the rest of us poorer is the fundamental sales pitch of the crypto pyramid scheme. Once the incoming and outgoing balance, the system has to pay for itself, and no one involved has any real plan for that whatsoever.
You're a great person, Noah. I have good reason to know that. I'm not going to subscribe at this time. I'm holding out for a plan that pays for the 'payed' content I view from everyone and not a single person. You have great viewpoints, and I rarely disagree except with the technicalities in how equations do and do not relate to reality. I hope you do well here. You're a great writer and a great person.
Re DeLong's France in the 1920's point. That has everything to do with the gold standard and forcing Germany to make payments beyond what their industrial capacity to produce allowed, which triggered the hyperinflation.
Great post! While I'd love to see DSGE models relegated to the dustbin of history, I'm not so sanguine about this happening. The fact remains that they are the work horse in academic macroeconomics and so my bet is that decision makers will continue to reference them as a means of adding the legitimacy (and cover) to their policy plans.
The one thing I disagree with you on is the contention that macro models are hard to test. I think they are trivial to test: all you need to do is compare their predictions with what actually happens. They should be judged entirely by the predictive improvement they provide over a random walk. This seems to be the only way to address the serious overfitting issues caused by small sample sizes and extensive "calibration."
Yes. But to the point Noah made, at the macro level there's like a max n of 5 for any question. So you can formulate a ridiculously large set of contradictory models that will all predict those 5 relevant observations fine. The fact that lots of models don't pass even that test gives away the game.
But while predictive accuracy may be able to filter out some total joke models, it will never be able to distinguish between many mutually exclusive models that fit the tiny samples available. Hence theoretical macro being more or less a joke. Or maybe more accurately an intellectual parlor game.
gosh, I had no idea how big this problem is. how long will it be until we have a large enough sample size of natural macro events (not just synthesized from micro ones)?
I'm a subscriber. and have a suggestion--consult Paul Krugman before publishing on this subject. He's not always right, but on the big things he mostly is. Not saying you have to agree with him, but I can't take you seriously if you don't engage with his views. My 2 cents
Why no reference to Krugman, who predicts a burst of inflation like after the Great Inflation but no accelerating inflation like the 70s? I understand that he, too, has little to say about theory, but he makes a concrete prediction, unlike, say, Cochrane who was completely wrong about the Great Recession, Obama's stimulus, and QE, has never admitted as much, but has whined extensively about being called to account.
This is just off the top of my head, but a Keynesian would be more concerned about debt to GDP, growth rates, and interest rates as a borrowing constraint. MMT just basically says "only unemployment and inflation are real."
Keynesians still believe crowding out can occur, even if there are circumstances under which it does not occur, like a liquidity trap. But MMT argues that crowding out does not occur because the way we finance spending it's just a swap of safe assets (bonds and dollars) and that the Fed can just hold interest rates down.
MMT also seems to believe very strongly in a jobs guarantee, which isn't a normal part of a monetary theory. They also believe that the federal government should combat inflation, while the Fed holds interest rates down to reduce unemployment. I find the most unorthodox part to be using the federal government to drive down inflation with tax hikes while the fed keeps long term interest rates low, and think that would be pretty different from a Keynesian viewpoint.
Keynesian is a big hoary family that's been around for generations, with lots of branches of the family tree going this way and that. The main thing they all have in common is the belief in a public interest in spending to reduce high unemployment. It seems fair to say MMT is an offshoot. When MMT came out it was a radical proposal to regulate the economy by modulating fiscal spending and taxation, instead of central bank regulation of interest rates. But in practice most MMT writers seem to emphasize the fiscal spending part and avoid being pinned down on the taxing part or the precise mechanics or checks/balances angles of how it would work.
One thing most people don't realize is that what you might call limited MMT has been practiced since the early 90s in the form of counter-cyclical fiscal policy and inflation targeting. If during recession the government loosens fiscally, and the central bank suppresses interest rates by buying public debt, then the net effect is the same as MMT would recommend: to loosen fiscally and have the central bank finance it.
Don't know much about the differences between mmt and old keynesian economics, but one major difference between New Keynesian economics and MMT is that NK models have budget constraints for governments. In the optimization problem(s) upon which NK models are built, agents must satisfy certain "transversality conditions" which limit how much debt they can accumulate. These budget constraints are not in MMT theory.
Both boosters and attackers of the $1.9T seem so confident in their predictions about the size of the multiplier inflation, that it always feels refreshing for an author to concede, even if he has an opinion, that the outcome is uncertain. As Noah generally favor Biden’s rescue and infrastructure bills, is the idea that the primary downside risk in the ‘If Biden’s bet fails, we’re in trouble’ is high inflation, stemming from a larger-than-expected multiplier? If so, how bad would have inflation have to be to be riskier than the costs of inaction? Are there probabilities you would assign to a set of possible outcomes?
The risks of persistently high inflation are minimal. This is a one shot boost, and if the "shot" goes to paying down debt - the effect will be small.
Where it gets riskier is if you tack on another 2T in infrastructure borrowing without an accompanying tax hike. I'd raise taxes enough to pay for the additional interest on the infrastructure borrowing. It would raise GDP in the long run and debt to GDP would decline.
Risks are definitely asymmetric. Some inflation is actually good, it would be healthy for the economy to lift off 0 interest bound, so that monetary policy can become efficient again. Annual inflation of 2-4% is very fine.
The disappearance of macro models as something anyone takes seriously is SO refreshing. I've got a PhD in stats and have been baffled and infuriated forever that those can get published, much less taken seriously by anyone except those whose jobs depend on them being taken seriously. A bunch of made up equations based on baldly ridiculous assumptions that are just transparently mathified reactionary ideology. It's science! The plutocrats funded an entire movement and generation of owned academics and think tankers to embed as "common sense" the pretention that their massive rents and power were scientifically proven to be necessary. Still very successful and tons of damage to undo but it's nice the tide has finally turned.
https://twitter.com/paulkrugman/status/1374755728620142601
I'll make it clearer: it's not that I won't pay for your newsletters, it's that I won't pay for any of them. I want a distributed model where my average views make my fees distributable over the authors, and limited by month. When that happens, I'll subscribe.
Check out the Brave browser, they have an interesting model
Very interesting, question, are you on twitter? Found a couple Noah Smith and Noah Pinion
Unfortunately this is a bundle of experiments, and during a pandemic. So I expect in hindsight it won't be clear which caused what, or whether the lessons apply to other situations.
My take is the power of $1.9t is being overblown. It's actually mostly continuing existing spending: like unemployment insurance, SBA. The aid to local goverment mostly ensures they won't have to cut. The $1400 checks are against the backgound of the recent $600 and $1200 checks; bigger but with a signal that these are almost certainly the last.
For me the most interesting experiment is the largely accidental increase-most-people's-savings program. A huge portion of the population never lost their income, but found themselves spending much less under quarantine. And then the government sent them stimulus checks. So their savings have gone way up. Will they now all go roaring '20s as the clubs re-open? Yeeh, most of these people are 40+. I'm sure many will splurge in their own ways, but not necessarily on the sorts of commodities that are tracked by inflation indicators. Not a few are investing into personal businesses.
Will crypto survive it? The idea that government money creation inherently makes the rest of us poorer is the fundamental sales pitch of the crypto pyramid scheme. Once the incoming and outgoing balance, the system has to pay for itself, and no one involved has any real plan for that whatsoever.
You're a great person, Noah. I have good reason to know that. I'm not going to subscribe at this time. I'm holding out for a plan that pays for the 'payed' content I view from everyone and not a single person. You have great viewpoints, and I rarely disagree except with the technicalities in how equations do and do not relate to reality. I hope you do well here. You're a great writer and a great person.
Re DeLong's France in the 1920's point. That has everything to do with the gold standard and forcing Germany to make payments beyond what their industrial capacity to produce allowed, which triggered the hyperinflation.
Great post! While I'd love to see DSGE models relegated to the dustbin of history, I'm not so sanguine about this happening. The fact remains that they are the work horse in academic macroeconomics and so my bet is that decision makers will continue to reference them as a means of adding the legitimacy (and cover) to their policy plans.
The one thing I disagree with you on is the contention that macro models are hard to test. I think they are trivial to test: all you need to do is compare their predictions with what actually happens. They should be judged entirely by the predictive improvement they provide over a random walk. This seems to be the only way to address the serious overfitting issues caused by small sample sizes and extensive "calibration."
Yes. But to the point Noah made, at the macro level there's like a max n of 5 for any question. So you can formulate a ridiculously large set of contradictory models that will all predict those 5 relevant observations fine. The fact that lots of models don't pass even that test gives away the game.
But while predictive accuracy may be able to filter out some total joke models, it will never be able to distinguish between many mutually exclusive models that fit the tiny samples available. Hence theoretical macro being more or less a joke. Or maybe more accurately an intellectual parlor game.
gosh, I had no idea how big this problem is. how long will it be until we have a large enough sample size of natural macro events (not just synthesized from micro ones)?
I'm a subscriber. and have a suggestion--consult Paul Krugman before publishing on this subject. He's not always right, but on the big things he mostly is. Not saying you have to agree with him, but I can't take you seriously if you don't engage with his views. My 2 cents
See my post above.
I see us as in need of a private debt reset. The young are so saddled with debt they cannot take risks. Its a bad idea.
For my family, the stimulus wiped out our medical debt and put me in a position to pay off my student loans a lot quicker.
Why no reference to Krugman, who predicts a burst of inflation like after the Great Inflation but no accelerating inflation like the 70s? I understand that he, too, has little to say about theory, but he makes a concrete prediction, unlike, say, Cochrane who was completely wrong about the Great Recession, Obama's stimulus, and QE, has never admitted as much, but has whined extensively about being called to account.
Any chance this will help macro-bros on the job market next year? :P
Can anyone tell me the big differences between mmt and keynesian economics. Seems like mmt is just a subset of keynesian thinking?
This is just off the top of my head, but a Keynesian would be more concerned about debt to GDP, growth rates, and interest rates as a borrowing constraint. MMT just basically says "only unemployment and inflation are real."
Keynesians still believe crowding out can occur, even if there are circumstances under which it does not occur, like a liquidity trap. But MMT argues that crowding out does not occur because the way we finance spending it's just a swap of safe assets (bonds and dollars) and that the Fed can just hold interest rates down.
MMT also seems to believe very strongly in a jobs guarantee, which isn't a normal part of a monetary theory. They also believe that the federal government should combat inflation, while the Fed holds interest rates down to reduce unemployment. I find the most unorthodox part to be using the federal government to drive down inflation with tax hikes while the fed keeps long term interest rates low, and think that would be pretty different from a Keynesian viewpoint.
Keynesian is a big hoary family that's been around for generations, with lots of branches of the family tree going this way and that. The main thing they all have in common is the belief in a public interest in spending to reduce high unemployment. It seems fair to say MMT is an offshoot. When MMT came out it was a radical proposal to regulate the economy by modulating fiscal spending and taxation, instead of central bank regulation of interest rates. But in practice most MMT writers seem to emphasize the fiscal spending part and avoid being pinned down on the taxing part or the precise mechanics or checks/balances angles of how it would work.
One thing most people don't realize is that what you might call limited MMT has been practiced since the early 90s in the form of counter-cyclical fiscal policy and inflation targeting. If during recession the government loosens fiscally, and the central bank suppresses interest rates by buying public debt, then the net effect is the same as MMT would recommend: to loosen fiscally and have the central bank finance it.
Don't know much about the differences between mmt and old keynesian economics, but one major difference between New Keynesian economics and MMT is that NK models have budget constraints for governments. In the optimization problem(s) upon which NK models are built, agents must satisfy certain "transversality conditions" which limit how much debt they can accumulate. These budget constraints are not in MMT theory.
Whose the "we" in "we're in trouble"? America, Macro believers, or some other group?
Both boosters and attackers of the $1.9T seem so confident in their predictions about the size of the multiplier inflation, that it always feels refreshing for an author to concede, even if he has an opinion, that the outcome is uncertain. As Noah generally favor Biden’s rescue and infrastructure bills, is the idea that the primary downside risk in the ‘If Biden’s bet fails, we’re in trouble’ is high inflation, stemming from a larger-than-expected multiplier? If so, how bad would have inflation have to be to be riskier than the costs of inaction? Are there probabilities you would assign to a set of possible outcomes?
The risks of persistently high inflation are minimal. This is a one shot boost, and if the "shot" goes to paying down debt - the effect will be small.
Where it gets riskier is if you tack on another 2T in infrastructure borrowing without an accompanying tax hike. I'd raise taxes enough to pay for the additional interest on the infrastructure borrowing. It would raise GDP in the long run and debt to GDP would decline.
Risks are definitely asymmetric. Some inflation is actually good, it would be healthy for the economy to lift off 0 interest bound, so that monetary policy can become efficient again. Annual inflation of 2-4% is very fine.