I pretty strongly disagree with the notion that any country on Earth right now is “trying out MMT” merely by running deficits without concern for their impact on public debt levels. After all, traditional Keynesian macroeconomics recommends the exact same course during a recession. Moreover, MMT itself is far more radical than this. It recommends basically switching the functions of the fiscal taxation authority and the central bank. In particular, it holds that instead of raising taxes to finance new government programs, the central bank should simply buy government debt (i.e. print money) to pay for them, and when this inevitably leads to significant inflation, the fiscal authority (e.g. the national legislature) should address it by raising taxes (thus reducing aggregate demand) rather than having the central bank address it by raising interest rates. With respect to the long-run trajectory of interest rates, MMT’s proponents believe they should be maintained at a low, stable level (zero, according to some).
I want to emphasize how different this would look from what we’re seeing right now in developed economies facing COVID-19-induced recessions. To do this, I’ll walk through a hypothetical based on the U.S. Here (in the U.S.), MMT would likely require the legislative abolition of central bank independence. Then, it would involve the Chair of the Federal Reserve, the Secretary of the Treasury, and the Speaker of the House of Representatives (presumably) jointly announcing that from this point forward the Fed would no longer be adjusting interest rates to maximize the objectives of full employment and stable prices specified in its (now former) dual mandate. Instead, it would be (permanently) buying as much government debt as necessary to pay for fiscal programs passed by Congress while maintaining some specified target interest rate (presumably around zero — it’s hard to get much below that) irrespective of the amount that this would expand the money supply. The Speaker of the House, for her part, would announce that when this new policy caused inflation to exceed some specified rate (the Fed’s current symmetrical 2% target?), Congress would pass a tax increase to wrangle aggregate demand back in check, thus reducing demand-pull inflation and (hopefully) ensuring continued stable prices.
MMT’s proponents often argue that we can trust that what they’re defending would not cause runaway inflation because even as advanced economies around the world have radically expanded their money supplies over the last few decades (some by nearly an order of magnitude), none of those economies have struggled with elevated inflation (ostensibly indicating a persistent gap between the money supplies and the productive capacities of these economies). This is a basically fair observation, but it misses the point. Today, we understand far better than we did 20 years ago that the inflationary consequences of extraordinary monetary policy (at the zero lower bound — i.e. like now) are caused primarily by the ways in which policy affects expectations regarding the future trajectory of interest rates and inflation rates. That is to say: Policy changes that radically reconfigure market participants’ expectations regarding the long-run monetary trajectory will almost certainly yield substantial changes in the inflation rate.
This is a policy that would radically reconfigure market participants’ expectations regarding the long-run monetary trajectory (pretty much by definition). It’s true that in MMT proponents’ perfect world, we would simply raise taxes to combat this rising inflation, and if market participants trusted that would occur as described when the transition to MMT policy were first announced, then perhaps the effects of the policy announcement on market expectations would be less pronounced. However, basically no onebelieves that (for example) the U.S. Congress could be trusted to pass (inevitably unpopular) tax hikes merely for the purpose of mitigating the risk of runaway inflation sometime down the road (remember, good monetary policy is proactive, not reactive!), given that it has proven unable to pass tax hikes to finance overwhelmingly popular government programs for the last decade. As a result, a government declaring an MMT policy approach would wildly shift market inflation expectations, thus causing significant inflation (a self-fulfilling prophecy), demanding, presumably, an increase in taxes shortly thereafter. When this increase fails to materialize, the market’s suspicions that the national legislature was not fully committed to the elements of MMT that require economic discipline would be confirmed, and inflation would increase even further, potentially surpassing the ability of any relevant government authorities to control it.
Even if you don’t buy any of that, though, I think there are some other pretty serious problems with the notion of using the tax code to control inflation. Intuitively, it’s an idea with significant distributive implications (though, of course, controlling inflation by raising interest rates also has distributive implications, and I can imagine how a person might reasonably conclude that those distributive implications are worse). More importantly, however, it could have a pretty substantial impact on labor market incentives. Bear in mind that if your intent in raising taxes is to reduce aggregate demand, you can’t just raise the corporate rate and call it a day (at least in the U.S., that would spur some tax inversions and some deduction-hunting but probably would not have much of an impact on demand). Rather, to have the desired effect, you would need to raise income taxes. That’s fine for a while, but eventually, inflation could get so high that the tax rate required to rein it in would meaningfully reduce people’s interest in working, thus diminishing the economy’s productive capacity (and GDP, etc.). Moreover, you better trust that your tax code is well-designed; otherwise, differentially disincentivizing work in this manner could produce some pretty odd labor market distortions. (Yes, many arguments similar to this one have been grossly mis-used by conservative economists to defend empirically indefensible conclusions. As Paul Krugman himself notes, this is not such a case.)
To close, I’d just note that MMT has substantial factual and theoretical faults entirely unrelated to anything I discussed here but nonetheless worthy of consideration. For a cogent run-through of just a few, I’d recommend thesethreecolumns by Krugman (one of which is also linked above).
Thanks! This response makes a lot more sense to me than many of the other materials I’ve been able to find online (including the various Krugman pieces I’d found, and the debate between Summers and Kelton on Bloomberg). Appreciate the time you’ve taken to write it.
It’s still astonishing to me that we seem to have such an imperfect understanding of how monetary systems, which humans created, work. I guess it’s just a product of these being very complex systems, somewhat akin to a weather system. I’d be interested to see though how well the mainstream macroeconomic theories fare in terms of predictive power? That is, not with the benefit of hindsight, but whether from the outset the impact of different macroeconomic interventions can be reliably predicted by leading economists.
It’s also kind of interesting that so much rests on markets’ subjective expectations rather than the objective intervention undertaken. Seems like, if the government can satisfy markets that the present time is extraordinary, then they might get a free kick at printing extra money without paying the price in terms of inflation.
Thanks again. I’m going to try to keep learning more about all of this, as it’s pretty fascinating.
I actually don’t think it’s that surprising that we have so much difficulty modeling the macroeconomy with high fidelity. In part, this is because of large degrees of endogeneity and general equilibrium effects (the shocks you are trying to model may alter key parameters of the models themselves), and in part, it is because contemporary macroeconomic models (i.e. DSGE/New Keynesian models) must necessarily make assumptions about human psychology in establishing their “microfoundations.” For a long time, for the sake of simplicity, there was little focus among macroeconomists on ensuring that those assumptions about human psychology were accurate. More recently, significant focus has been dedicated to that question, and more sophisticated, hopefullymoreaccurate New Keynesian models have emerged. That said, even if these models better reflect the expert consensus in psychology and behavioral economics, they are still reliant on empirical findings in those fields for their accuracy, and given the replication crisis in psychology, that reliance may be compromising. It seems that there is no easy way around the difficulty of modeling human behavior.
Regarding the reliability of macroeconomic predictions, I think it’s safe to say that macroeconomists have developed a poor reputation as predictors. Particularly, in the aftermath of the 2008 financial crisis, there was a lot of highly critical discussion about why “no one saw it coming,” where “no one,” more often than not, referred to macroeconomists. While this is not completely accurate (some macroeconomists, like Robert Shiller, did see it coming), macroeconomists, by and large, failed rather grandly to understand what was going on in markets. There are many reasons for this, but a large one, arguably, especially among experts at the U.S. Federal Reserve, was that they were looking at DSGE models based on implausible and largely discredited assumptions both about people (e.g. rational expectations, the permanent income hypothesis, etc.) and about markets (e.g. perfect competition, no asymmetric information, no price stickiness, no financial or labor market frictions, etc.). This inspired a substantial backlash against DSGE models in general (typified in work like chapter three of John Quiggin’s book Zombie Economics).
This criticism, I think, has proven misguided (and not only because the critics never seemed to be equipped with a superior alternative to DSGE). In the years since the crisis, as I mentioned earlier, DSGE models have been updated and improved, and many authoritative New Keynesian approaches today are (to the best of our knowledge) realistic in precisely the same areas in which their predecessors were unrealistic. The Federal Reserve Bank of New York actually now publishes its standard DSGE model on GitHub. If you’re curious about it, you can take a look here. Only time will tell whether the best models of today materially outperform the models of the early 2000s, but I will say that thus far, it looks like the Federal Reserve has done a vastly better job of responding to the COVID-19 economic crisis than it did in responding to the 2008 housing crisis, so I am optimistic about the progress the field has made in the last ten years.
Finally a minor clarification based on your comment: While the impact of monetary policy is mediated just about exclusively through the expectations channel when an economy is at the zero lower boundand nonetheless lacks full employment (like now, in most advanced economies), this is generally not the case when interest rates are well above zero, and monetary policy is reliant upon traditional rather than extraordinary tools (e.g. open market operations rather than quantitative easing).
I found your discussion very interesting. Curious to know if you have any updated thoughts on this two years on. For me, I tend to agree with Wei Dai’s comment below, that the federal reserve and government working closely together is a synthetic form of MMT. Would you say that Japan’s “yield curve control” (pegging interest rates with unlimited money printing) is demonstrating that they are going down the MMT route? And what about Europe’s transmission protection instrument to peg Italian bonds?
Curious to know if the massive inflation numbers have changed your mind about whether the central banks have handled covid better than 2008.
I pretty strongly disagree with the notion that any country on Earth right now is “trying out MMT” merely by running deficits without concern for their impact on public debt levels. After all, traditional Keynesian macroeconomics recommends the exact same course during a recession. Moreover, MMT itself is far more radical than this. It recommends basically switching the functions of the fiscal taxation authority and the central bank. In particular, it holds that instead of raising taxes to finance new government programs, the central bank should simply buy government debt (i.e. print money) to pay for them, and when this inevitably leads to significant inflation, the fiscal authority (e.g. the national legislature) should address it by raising taxes (thus reducing aggregate demand) rather than having the central bank address it by raising interest rates. With respect to the long-run trajectory of interest rates, MMT’s proponents believe they should be maintained at a low, stable level (zero, according to some).
I want to emphasize how different this would look from what we’re seeing right now in developed economies facing COVID-19-induced recessions. To do this, I’ll walk through a hypothetical based on the U.S. Here (in the U.S.), MMT would likely require the legislative abolition of central bank independence. Then, it would involve the Chair of the Federal Reserve, the Secretary of the Treasury, and the Speaker of the House of Representatives (presumably) jointly announcing that from this point forward the Fed would no longer be adjusting interest rates to maximize the objectives of full employment and stable prices specified in its (now former) dual mandate. Instead, it would be (permanently) buying as much government debt as necessary to pay for fiscal programs passed by Congress while maintaining some specified target interest rate (presumably around zero — it’s hard to get much below that) irrespective of the amount that this would expand the money supply. The Speaker of the House, for her part, would announce that when this new policy caused inflation to exceed some specified rate (the Fed’s current symmetrical 2% target?), Congress would pass a tax increase to wrangle aggregate demand back in check, thus reducing demand-pull inflation and (hopefully) ensuring continued stable prices.
MMT’s proponents often argue that we can trust that what they’re defending would not cause runaway inflation because even as advanced economies around the world have radically expanded their money supplies over the last few decades (some by nearly an order of magnitude), none of those economies have struggled with elevated inflation (ostensibly indicating a persistent gap between the money supplies and the productive capacities of these economies). This is a basically fair observation, but it misses the point. Today, we understand far better than we did 20 years ago that the inflationary consequences of extraordinary monetary policy (at the zero lower bound — i.e. like now) are caused primarily by the ways in which policy affects expectations regarding the future trajectory of interest rates and inflation rates. That is to say: Policy changes that radically reconfigure market participants’ expectations regarding the long-run monetary trajectory will almost certainly yield substantial changes in the inflation rate.
This is a policy that would radically reconfigure market participants’ expectations regarding the long-run monetary trajectory (pretty much by definition). It’s true that in MMT proponents’ perfect world, we would simply raise taxes to combat this rising inflation, and if market participants trusted that would occur as described when the transition to MMT policy were first announced, then perhaps the effects of the policy announcement on market expectations would be less pronounced. However, basically no one believes that (for example) the U.S. Congress could be trusted to pass (inevitably unpopular) tax hikes merely for the purpose of mitigating the risk of runaway inflation sometime down the road (remember, good monetary policy is proactive, not reactive!), given that it has proven unable to pass tax hikes to finance overwhelmingly popular government programs for the last decade. As a result, a government declaring an MMT policy approach would wildly shift market inflation expectations, thus causing significant inflation (a self-fulfilling prophecy), demanding, presumably, an increase in taxes shortly thereafter. When this increase fails to materialize, the market’s suspicions that the national legislature was not fully committed to the elements of MMT that require economic discipline would be confirmed, and inflation would increase even further, potentially surpassing the ability of any relevant government authorities to control it.
Even if you don’t buy any of that, though, I think there are some other pretty serious problems with the notion of using the tax code to control inflation. Intuitively, it’s an idea with significant distributive implications (though, of course, controlling inflation by raising interest rates also has distributive implications, and I can imagine how a person might reasonably conclude that those distributive implications are worse). More importantly, however, it could have a pretty substantial impact on labor market incentives. Bear in mind that if your intent in raising taxes is to reduce aggregate demand, you can’t just raise the corporate rate and call it a day (at least in the U.S., that would spur some tax inversions and some deduction-hunting but probably would not have much of an impact on demand). Rather, to have the desired effect, you would need to raise income taxes. That’s fine for a while, but eventually, inflation could get so high that the tax rate required to rein it in would meaningfully reduce people’s interest in working, thus diminishing the economy’s productive capacity (and GDP, etc.). Moreover, you better trust that your tax code is well-designed; otherwise, differentially disincentivizing work in this manner could produce some pretty odd labor market distortions. (Yes, many arguments similar to this one have been grossly mis-used by conservative economists to defend empirically indefensible conclusions. As Paul Krugman himself notes, this is not such a case.)
To close, I’d just note that MMT has substantial factual and theoretical faults entirely unrelated to anything I discussed here but nonetheless worthy of consideration. For a cogent run-through of just a few, I’d recommend these three columns by Krugman (one of which is also linked above).
Thanks! This response makes a lot more sense to me than many of the other materials I’ve been able to find online (including the various Krugman pieces I’d found, and the debate between Summers and Kelton on Bloomberg). Appreciate the time you’ve taken to write it.
It’s still astonishing to me that we seem to have such an imperfect understanding of how monetary systems, which humans created, work. I guess it’s just a product of these being very complex systems, somewhat akin to a weather system. I’d be interested to see though how well the mainstream macroeconomic theories fare in terms of predictive power? That is, not with the benefit of hindsight, but whether from the outset the impact of different macroeconomic interventions can be reliably predicted by leading economists.
It’s also kind of interesting that so much rests on markets’ subjective expectations rather than the objective intervention undertaken. Seems like, if the government can satisfy markets that the present time is extraordinary, then they might get a free kick at printing extra money without paying the price in terms of inflation.
Thanks again. I’m going to try to keep learning more about all of this, as it’s pretty fascinating.
I’m happy to hear that what I wrote was helpful!
I actually don’t think it’s that surprising that we have so much difficulty modeling the macroeconomy with high fidelity. In part, this is because of large degrees of endogeneity and general equilibrium effects (the shocks you are trying to model may alter key parameters of the models themselves), and in part, it is because contemporary macroeconomic models (i.e. DSGE/New Keynesian models) must necessarily make assumptions about human psychology in establishing their “microfoundations.” For a long time, for the sake of simplicity, there was little focus among macroeconomists on ensuring that those assumptions about human psychology were accurate. More recently, significant focus has been dedicated to that question, and more sophisticated, hopefully more accurate New Keynesian models have emerged. That said, even if these models better reflect the expert consensus in psychology and behavioral economics, they are still reliant on empirical findings in those fields for their accuracy, and given the replication crisis in psychology, that reliance may be compromising. It seems that there is no easy way around the difficulty of modeling human behavior.
Regarding the reliability of macroeconomic predictions, I think it’s safe to say that macroeconomists have developed a poor reputation as predictors. Particularly, in the aftermath of the 2008 financial crisis, there was a lot of highly critical discussion about why “no one saw it coming,” where “no one,” more often than not, referred to macroeconomists. While this is not completely accurate (some macroeconomists, like Robert Shiller, did see it coming), macroeconomists, by and large, failed rather grandly to understand what was going on in markets. There are many reasons for this, but a large one, arguably, especially among experts at the U.S. Federal Reserve, was that they were looking at DSGE models based on implausible and largely discredited assumptions both about people (e.g. rational expectations, the permanent income hypothesis, etc.) and about markets (e.g. perfect competition, no asymmetric information, no price stickiness, no financial or labor market frictions, etc.). This inspired a substantial backlash against DSGE models in general (typified in work like chapter three of John Quiggin’s book Zombie Economics).
This criticism, I think, has proven misguided (and not only because the critics never seemed to be equipped with a superior alternative to DSGE). In the years since the crisis, as I mentioned earlier, DSGE models have been updated and improved, and many authoritative New Keynesian approaches today are (to the best of our knowledge) realistic in precisely the same areas in which their predecessors were unrealistic. The Federal Reserve Bank of New York actually now publishes its standard DSGE model on GitHub. If you’re curious about it, you can take a look here. Only time will tell whether the best models of today materially outperform the models of the early 2000s, but I will say that thus far, it looks like the Federal Reserve has done a vastly better job of responding to the COVID-19 economic crisis than it did in responding to the 2008 housing crisis, so I am optimistic about the progress the field has made in the last ten years.
Finally a minor clarification based on your comment: While the impact of monetary policy is mediated just about exclusively through the expectations channel when an economy is at the zero lower bound and nonetheless lacks full employment (like now, in most advanced economies), this is generally not the case when interest rates are well above zero, and monetary policy is reliant upon traditional rather than extraordinary tools (e.g. open market operations rather than quantitative easing).
I found your discussion very interesting. Curious to know if you have any updated thoughts on this two years on. For me, I tend to agree with Wei Dai’s comment below, that the federal reserve and government working closely together is a synthetic form of MMT. Would you say that Japan’s “yield curve control” (pegging interest rates with unlimited money printing) is demonstrating that they are going down the MMT route? And what about Europe’s transmission protection instrument to peg Italian bonds?
Curious to know if the massive inflation numbers have changed your mind about whether the central banks have handled covid better than 2008.