The post is very clear, except it would have been good to explain upfront what distinguishes finance from engineering.
You think historically, so you think action-across-time matters. Samuelsonian Physics-envy economics has trouble with time—and especially the power of subjectivity (aka expectations) that action-across-time entails—because, oddly, time is not very important in most physics. A default of complete information, completely rational actors with no computational limitations is a default without action across time: it is more a frozen moment analysis. It is easy to turn into equations. (Rational Expectations was turned into a mechanism to use equations to get rid of the messiness of limited information expectations.)
You think of people as boundedly-rational people with limited information pursuing a range of strategies (with markets tending to winnow strategies), because that is what history reveals us to be. That is not easy to turn into equations. Hence, economics as a discipline has some tendency to ignore analyses that take history seriously but do not make for congenial equations.
Robert Fogel wrote an entire book explaining how the mass migration let loose by steamships and railways fractured the American Republic across its fault-line of slavery. He has been completely ignored because, like The Midas Touch, Without Consent or Contract was a work of analytical history even though, like The Midas Touch, it was steeped in economics. (As one might expect from a Nobel memorial Laureate.)
Tyler, see if you can spot a sharp break in one of the nine following steps:
1. Pure gold standard. The dollar is defined as one gram of gold; otherwise, the government is not involved in the monetary system. Private banks are free to produce gold-backed banknotes.
2. Gold exchange standard. The government produces paper currency and promises to redeem each paper dollar for one gram of gold, on demand.
3. Gold price targeting. The government adjusts monetary policy in such a way as to keep the free-market price of gold at $1 per gram. But the government does not directly buy and sell gold at that price; rather, it targets gold prices by engaging in open market operations with other assets, such Treasury securities.
4. Symmetallism. Monetary policy targets the price of a basket of metals containing one gram of gold plus 20 grams of silver at a price of $1. Notice that I said “plus.” If I’d said “or,” the system would have been bimetallism. Symmetallism is less well-known than bimetallism but is a more important concept. It’s a key step toward inflation targeting.
5. Commodity price targeting. The government adjusts monetary policy in such a way as to stabilize the spot price of a basket of 37 actively traded commodities.
6. Commodity futures price targeting. The government adjusts monetary policy in such a way as to stabilize the price of a basket of 37 actively traded commodity price futures.
7. Consumer price index (CPI) futures targeting. The government adjusts monetary policy in such a way as to stabilize the price of one-year forward CPI contracts at 100.
8. Inflation futures targeting. The government adjusts monetary policy in such a way as to ensure that the price of one-year forward CPI contracts rises at 2 percent a year.
9. NGDP futures targeting. The government adjusts monetary policy in such a way as to ensure that the price of one-year forward NGDP contracts rises at 4 percent a year.
But you still need a model at #* and #9 to tell you which monetary policy levers to pull in order to control the CPI or the NGDP and at #1 to define what "not involved" means.
A pet peeve of mine is that the Fed acts as if forward guidance means saying a time frame for the position of an instrument (usually IOR but also QE/QT volumes). Then they continue obeying that promise well after it's clear that they need to adjust.
The 2-year rate-of-change in money flows, the volume and velocity of our means-of-payment money supply, peaked in February 2022. But since Vt is understated, the typical 3-month surge in the transaction's velocity of money, based on the G.6 release (since discontinued in Sept 1996), corresponds to the peak in inflation.
See: “Quantity Leads, and Velocity Follows” Cit. Dying of Money -By Jens O. Parson
I would say that the Fed does target the forecast--that is not a new idea. It was doing that when I was a research assistant there in 1976. But it didn't do a very good job of lining up policy to hit the desired numbers.
We remember Paul Volcker as a hero. But there were huge forecast errors in the early 1980s--it's just that NGDP came in much lower than expected then, as opposed to higher than expected in the late 1970s. So we say that he "broke the back of inflation." But the bond market kept long-term interest rates well above where the Fed would have liked in the 1980s, and that was a factor in the NGDP shortfall.
The key to hitting a policy target is accurate knowledge of the feedback loop from the policy instrument (call it open market operations) to the result (call it NGDP). Whoever is doing the forecasting has to have an idea of how open market operations will affect NGDP.
The econometric models are bad at that. But my guess is that the financial futures markets aren't any better. That makes hitting a nominal target more a matter of luck rather than skill.
We are not far apart, as I also believe that forecast errors was not the main problem. The problem in the 1970s was that they were targeting an excessively high NGDP growth rate. I recall when 10% or 11% NGDP growth was fully expected. And yes, Volcker got lucky in 1981-82 when NGDP slowed more than expected.
Yes. I ordered a T-shirt with "Never reason from a price change" a few years ago---should arrive any day now.
Economists, and crackpots like me who read blogs like this, will fixate on inflation rates and get grumpy if a graph isn't adjusted to real dollars or a benchmarked timeline. Ordinary people, and I'm willing to bet a lot of people in finance, fixate on nominal prices because that information is what is most relevant to our lives. While it's amazing that I'm using a computer that's 1000x more powerful than anything the U.S. military had in the 1960's it doesn't make me feel any better about the price I'll have to pay for the next computer.
I do not understand how either an "engineering" model or a "finance" model of inflation can work without building in a Fed reaction function. I take it that one difference is what kind of exogenous variables are used, but don't know what an "engineering" variable is and a "finance" variable is. Is it just what's available in real time? But for either, where is the central bank? And even in an "engineering model" movement in monetary policy instruments could act like "finance" variables so that past settings have no information content about present decisions.
The idea that “understanding” means “path predictability” is specially ridiculous after the Lorentz attractor was discovered: even having the full set of equations of a deterministic system does not mean you can make path prediction.
Here's an example I like. Imagine an economy where velocity is an absolute constant. (The "simple" quantity theory.) Now imagine the money supply is a random walk. In that case, changes in M are 100% correlated with changes in NGDP, they perfectly "explain" NGDP. But M is 0% predictive of future NGDP.
No money stock figure, standing alone, is adequate as a guidepost for monetary policy.
"The rate of change - RoC - is the speed at which a variable changes over a specific period of time. RoC is often used when speaking about momentum, and it can generally be expressed as a ratio between a change [ Δ, first derivative f’], in one variable relative to a corresponding change, Δ in another [second derivative f” ].” - Investopedia.
It seems to me that there is a general problem with this approach to modeling: it doesn't provide any evidence that the central bank does have the ability control inflation in the short run. If we attribute deviations from the 2% target to policy mistakes, we are implicitly assuming that the central bank can always hit its target. We are not providing any evidence that it can, in fact, do this. The situation would be different if we had a good structural model of the price level (say, if we found a monetary aggregate for which the velocity is constant). In that case, we could confidently say that by affecting the money supply, the central bank can influence the inflation rate in the short run. But without such a model, we simply do not know if the central bank can do this, and therefore we do not know if it could have prevented deviations form its target.
I'm inspired by your approach to rebutting Tyler, avoiding the he-said-she-said standard mostly used for these things. Instead you step back and take accountability for being misunderstood, endeavoring to build a better explanation rather than the easier point-by-point refutation. May I suggest you write the book? And aim it at a wider educated lay audience, not (just) the economists you're trying to persuade. Nothing is a more powerful tonic for improved explanation than being obligated to slow down and explain at the junior high school level.
** I'm reposting this since I somehow placed it at the wrong hierarchy earlier today, as though my comment was in response to another comment rather than to Scott's article :-\
Maybe that proves my point. Stop writing to your peers and start writing the same message but reworked to be understood and appeal to a lay audience, the same folks who read, say, "Freakanomics" or "The Black Swan." Especially in the latter example, where new ground was being defined (rather than the former's simple popularizing), the author used the larger bully pulpit afforded by something in wide readership to advance ideas that would otherwise be ignored by the "anointed" consensus view. For sure, writing to a wider audience is a harder kind of writing (and it's hard enough to write anything already!), but you have a good message and a good voice. If you still have the ambition (you're not that old!), then a best seller might be the way in..
Nassim Nicholas Taleb is a con man. I predicted the May 6th crash 6 months in advance and within one day (likewise the October 15th, 2024 bond price chaos).
We don't know what we don't know. I love the idea of viewing macroeconomics as how to plan around mistakes rather than perfectly making predictions. It still leaves room for improving predictions while making much more effective policy recommendations. I hope this reform happens.
Entrepreneurs/businesses discover what the market demands and will supply it if they make a profit (see Nvidia).
Central banks will discover the quantity of money to supply to satisfy market demand for money that keeps forecast ngdp growth stable. That's the discovery process, the results of which we see in hindsight money supply figures (or in Nvidia's growth)
No one knows now what the market will demand in the future, or what quantity of money will be demanded to ngdp growth forecasts stable.
Does this sum things up: "Use engineering to control simple things, use finance to control complicated things—and inflation and business cycles are complicated"?
I feel like there's a powerful idea here about using finance as a way of controlling variables in general, not just macroeconomic ones. Is there a reason you couldn't control every variable of interest with finance? How about a vision of cybernetics done via finance?
I didn't respond to all of your points because I'm not an expert on exactly where the finance approach works best. Clearly you need sufficient interest in a asset price for that price to be informative. I suspect you are right that the engineering approach works better for simple things, I'd have to give that more thought. We know that the field of physics is good at simple systems like planetary orbits, and we struggle with complex systems like the weather, earthquakes, and the macroeconomy.
I'm interested in examples of unconventional economies, like economies where the economic actors are cells instead of humans. It's still very speculative, but I think there's reasons to believe that diseases in interacting systems of cells, i.e., organisms, can be sometimes explained by macroeconomic phenomena. If we can discover or implement financial solutions in those kinds of economies, this may provide valuable evidence for where a financial approach works best.
"Dr. Richard G. Anderson's research focuses on the velocity of broad money (M2) from the Great Depression through the Great Recession. The model considers factors such as changes in uncertainty and risk premia, financial innovation, and major banking regulations"
"Remember this statement from the 1810 Bullion Committee’s report to Parliament: “The most detailed knowledge of the actual trade of the Country, combined with the profound science in all the principles of Money and Circulation, would not enable any man or set of men to adjust, and keep always adjusted, the right proportions of circulating medium in a country to the wants of trade.” In essence, the committee said that central banking was impossible. Consequently, money is always in disequilibrium." (From a post by Russell Napier)
I agree that money is always in disequilibrium, at least to some extent. I think of it this way: It's better to have a regime that minimizes disequilibrium, such as NGDPLT, rather than a regime that generates a large amount of disequilibrium.
Instead of thinking of it as solving the problem, think of it as minimizing the problem.
Great post. As a non professional economist with a fair amount of formal education both in economics and finance, i agree with taking a bit more of a finance approach to the topic. You are right with regard to understanding the logic of how someone comes to a conclusion, forecast/prediction not only in economics, but in all areas of life.
As an aside i bought NVDA stock in 2015, not a lot, as i saw the changing business model and what GPU's would portend going forward. Did sell about have after a 12X gain and that was a mistake as AI came down the pike, something i had not seen to take place as fast as it did.
Have disliked the dual mandate of the Fed for as long as i can remember and have felt Fed should be concerned with only one thing: price stability.
But politics and economics make for a toxic mix.
Lastly, i always find your views to be well reasoned!
I think this was a good post and the right post to respond to Tyler with, but it does raise a conundrum. Your argument is basically that policymakers shouldn't rely on inflation forecasting models and should instead rely on market-based expectations of inflation (or nominal GDP). But how should the market participants forecast inflation (or nominal GDP)? I don't think nominal GDP is high because the Fed let nominal GDP be high is satisfactory for this purpose (maybe if the Fed actually had the objective to target NGDP).
Under your original NGDP targeting proposal, fixing the futures price one year ahead for NGDP controls the expectations of NGDP. Realized NGDP may differ from expectations. Level targeting can help ensure that there aren’t significant cumulative forecast errors.
Under interest rate targeting, the argument is implicitly that a path of interest rates implies a path of expected NGDP (or inflation). But again realized NGDP (or inflation) can behave differently than expectations.
For instance, look at the inflation forecasts post COVID vs realized inflation. Inflation was way higher than forecast. Similarly, NGDP was way higher than forecast (I believe, without checking the numbers). So is the argument that policy was set such that it was consistent with a higher expected NGDP rate, but no one realized that NGDP should be expected to be higher? My problem is that this is all backward looking. It doesn’t let you know ex ante about the policy stance.
Whether NGDPLT, inflation, or price level targeting, the target needs to have an "F" in front of its name. Targets make sense in the context of an average size and frequency of shocks. Extraordinary shocks require flexible, temporary above- target inflation or NGDP growth. So, _F_AIT, _F_NGDPLT,
The post is very clear, except it would have been good to explain upfront what distinguishes finance from engineering.
You think historically, so you think action-across-time matters. Samuelsonian Physics-envy economics has trouble with time—and especially the power of subjectivity (aka expectations) that action-across-time entails—because, oddly, time is not very important in most physics. A default of complete information, completely rational actors with no computational limitations is a default without action across time: it is more a frozen moment analysis. It is easy to turn into equations. (Rational Expectations was turned into a mechanism to use equations to get rid of the messiness of limited information expectations.)
You think of people as boundedly-rational people with limited information pursuing a range of strategies (with markets tending to winnow strategies), because that is what history reveals us to be. That is not easy to turn into equations. Hence, economics as a discipline has some tendency to ignore analyses that take history seriously but do not make for congenial equations.
Robert Fogel wrote an entire book explaining how the mass migration let loose by steamships and railways fractured the American Republic across its fault-line of slavery. He has been completely ignored because, like The Midas Touch, Without Consent or Contract was a work of analytical history even though, like The Midas Touch, it was steeped in economics. (As one might expect from a Nobel memorial Laureate.)
Tyler, see if you can spot a sharp break in one of the nine following steps:
1. Pure gold standard. The dollar is defined as one gram of gold; otherwise, the government is not involved in the monetary system. Private banks are free to produce gold-backed banknotes.
2. Gold exchange standard. The government produces paper currency and promises to redeem each paper dollar for one gram of gold, on demand.
3. Gold price targeting. The government adjusts monetary policy in such a way as to keep the free-market price of gold at $1 per gram. But the government does not directly buy and sell gold at that price; rather, it targets gold prices by engaging in open market operations with other assets, such Treasury securities.
4. Symmetallism. Monetary policy targets the price of a basket of metals containing one gram of gold plus 20 grams of silver at a price of $1. Notice that I said “plus.” If I’d said “or,” the system would have been bimetallism. Symmetallism is less well-known than bimetallism but is a more important concept. It’s a key step toward inflation targeting.
5. Commodity price targeting. The government adjusts monetary policy in such a way as to stabilize the spot price of a basket of 37 actively traded commodities.
6. Commodity futures price targeting. The government adjusts monetary policy in such a way as to stabilize the price of a basket of 37 actively traded commodity price futures.
7. Consumer price index (CPI) futures targeting. The government adjusts monetary policy in such a way as to stabilize the price of one-year forward CPI contracts at 100.
8. Inflation futures targeting. The government adjusts monetary policy in such a way as to ensure that the price of one-year forward CPI contracts rises at 2 percent a year.
9. NGDP futures targeting. The government adjusts monetary policy in such a way as to ensure that the price of one-year forward NGDP contracts rises at 4 percent a year.
That's great! I wish I'd thought of that. :)
Super duper! Your bullet list reminds me of Robert Nozick's steps about The Tale of the Slave in Anarchy, State and Utopia. Very nice.
But you still need a model at #* and #9 to tell you which monetary policy levers to pull in order to control the CPI or the NGDP and at #1 to define what "not involved" means.
Is "#*" supposed to be "#8"?
Fantastic post.
A pet peeve of mine is that the Fed acts as if forward guidance means saying a time frame for the position of an instrument (usually IOR but also QE/QT volumes). Then they continue obeying that promise well after it's clear that they need to adjust.
Yes, that's a problem. That's one reason they moved too slow in 2022.
And kept doing QE into March 2022 even as 5 year breakevens got over 3.5%.
The 2-year rate-of-change in money flows, the volume and velocity of our means-of-payment money supply, peaked in February 2022. But since Vt is understated, the typical 3-month surge in the transaction's velocity of money, based on the G.6 release (since discontinued in Sept 1996), corresponds to the peak in inflation.
See: “Quantity Leads, and Velocity Follows” Cit. Dying of Money -By Jens O. Parson
Yes! Forward guidance should be about expected outcomes, not expected movements in policy instruments.
Excellent clarifications, Scott.
I would say that the Fed does target the forecast--that is not a new idea. It was doing that when I was a research assistant there in 1976. But it didn't do a very good job of lining up policy to hit the desired numbers.
We remember Paul Volcker as a hero. But there were huge forecast errors in the early 1980s--it's just that NGDP came in much lower than expected then, as opposed to higher than expected in the late 1970s. So we say that he "broke the back of inflation." But the bond market kept long-term interest rates well above where the Fed would have liked in the 1980s, and that was a factor in the NGDP shortfall.
The key to hitting a policy target is accurate knowledge of the feedback loop from the policy instrument (call it open market operations) to the result (call it NGDP). Whoever is doing the forecasting has to have an idea of how open market operations will affect NGDP.
The econometric models are bad at that. But my guess is that the financial futures markets aren't any better. That makes hitting a nominal target more a matter of luck rather than skill.
We are not far apart, as I also believe that forecast errors was not the main problem. The problem in the 1970s was that they were targeting an excessively high NGDP growth rate. I recall when 10% or 11% NGDP growth was fully expected. And yes, Volcker got lucky in 1981-82 when NGDP slowed more than expected.
"the real problem is nominal"
We need t-shirts with that on it!
In rocket launch lingo, "nominal" means no problem. :)
Yes. I ordered a T-shirt with "Never reason from a price change" a few years ago---should arrive any day now.
Economists, and crackpots like me who read blogs like this, will fixate on inflation rates and get grumpy if a graph isn't adjusted to real dollars or a benchmarked timeline. Ordinary people, and I'm willing to bet a lot of people in finance, fixate on nominal prices because that information is what is most relevant to our lives. While it's amazing that I'm using a computer that's 1000x more powerful than anything the U.S. military had in the 1960's it doesn't make me feel any better about the price I'll have to pay for the next computer.
Never reason from an endogenous variable. :)
I do not understand how either an "engineering" model or a "finance" model of inflation can work without building in a Fed reaction function. I take it that one difference is what kind of exogenous variables are used, but don't know what an "engineering" variable is and a "finance" variable is. Is it just what's available in real time? But for either, where is the central bank? And even in an "engineering model" movement in monetary policy instruments could act like "finance" variables so that past settings have no information content about present decisions.
Yes, but what does the HK monetary authority's "reaction function" look like when they peg the HK dollar at 7.8?
I don’t know. Presumably it’s telling their trading desk to buy and sell KH dollars to keep it in line.
Scott, probably you will like my little essay on prediction and control:
https://www.theseedsofscience.pub/p/prediction-and-control-in-natural
Yes, that's very good. I do see some links to what I wrote.
The idea that “understanding” means “path predictability” is specially ridiculous after the Lorentz attractor was discovered: even having the full set of equations of a deterministic system does not mean you can make path prediction.
Here's an example I like. Imagine an economy where velocity is an absolute constant. (The "simple" quantity theory.) Now imagine the money supply is a random walk. In that case, changes in M are 100% correlated with changes in NGDP, they perfectly "explain" NGDP. But M is 0% predictive of future NGDP.
No money stock figure, standing alone, is adequate as a guidepost for monetary policy.
"The rate of change - RoC - is the speed at which a variable changes over a specific period of time. RoC is often used when speaking about momentum, and it can generally be expressed as a ratio between a change [ Δ, first derivative f’], in one variable relative to a corresponding change, Δ in another [second derivative f” ].” - Investopedia.
It seems to me that there is a general problem with this approach to modeling: it doesn't provide any evidence that the central bank does have the ability control inflation in the short run. If we attribute deviations from the 2% target to policy mistakes, we are implicitly assuming that the central bank can always hit its target. We are not providing any evidence that it can, in fact, do this. The situation would be different if we had a good structural model of the price level (say, if we found a monetary aggregate for which the velocity is constant). In that case, we could confidently say that by affecting the money supply, the central bank can influence the inflation rate in the short run. But without such a model, we simply do not know if the central bank can do this, and therefore we do not know if it could have prevented deviations form its target.
I'm inspired by your approach to rebutting Tyler, avoiding the he-said-she-said standard mostly used for these things. Instead you step back and take accountability for being misunderstood, endeavoring to build a better explanation rather than the easier point-by-point refutation. May I suggest you write the book? And aim it at a wider educated lay audience, not (just) the economists you're trying to persuade. Nothing is a more powerful tonic for improved explanation than being obligated to slow down and explain at the junior high school level.
** I'm reposting this since I somehow placed it at the wrong hierarchy earlier today, as though my comment was in response to another comment rather than to Scott's article :-\
I'm done writing books, as my previous books were mostly ignored.
Maybe that proves my point. Stop writing to your peers and start writing the same message but reworked to be understood and appeal to a lay audience, the same folks who read, say, "Freakanomics" or "The Black Swan." Especially in the latter example, where new ground was being defined (rather than the former's simple popularizing), the author used the larger bully pulpit afforded by something in wide readership to advance ideas that would otherwise be ignored by the "anointed" consensus view. For sure, writing to a wider audience is a harder kind of writing (and it's hard enough to write anything already!), but you have a good message and a good voice. If you still have the ambition (you're not that old!), then a best seller might be the way in..
Nassim Nicholas Taleb is a con man. I predicted the May 6th crash 6 months in advance and within one day (likewise the October 15th, 2024 bond price chaos).
We don't know what we don't know. I love the idea of viewing macroeconomics as how to plan around mistakes rather than perfectly making predictions. It still leaves room for improving predictions while making much more effective policy recommendations. I hope this reform happens.
Maybe if you told Tyler macroeconomics is a discovery process, like the key insight of Hayekian micro, he'd get what you are on about.
Entrepreneurs/businesses discover what the market demands and will supply it if they make a profit (see Nvidia).
Central banks will discover the quantity of money to supply to satisfy market demand for money that keeps forecast ngdp growth stable. That's the discovery process, the results of which we see in hindsight money supply figures (or in Nvidia's growth)
No one knows now what the market will demand in the future, or what quantity of money will be demanded to ngdp growth forecasts stable.
Ha, William Bretz, of Juncture Recognition in the Stock Market, corrected Ed Fry's G.6 numbers.
Does this sum things up: "Use engineering to control simple things, use finance to control complicated things—and inflation and business cycles are complicated"?
I feel like there's a powerful idea here about using finance as a way of controlling variables in general, not just macroeconomic ones. Is there a reason you couldn't control every variable of interest with finance? How about a vision of cybernetics done via finance?
You might want to check out Robin Hanson's writing on "futarchy", using prediction markets for all sorts of decisions.
I didn't respond to all of your points because I'm not an expert on exactly where the finance approach works best. Clearly you need sufficient interest in a asset price for that price to be informative. I suspect you are right that the engineering approach works better for simple things, I'd have to give that more thought. We know that the field of physics is good at simple systems like planetary orbits, and we struggle with complex systems like the weather, earthquakes, and the macroeconomy.
I'm interested in examples of unconventional economies, like economies where the economic actors are cells instead of humans. It's still very speculative, but I think there's reasons to believe that diseases in interacting systems of cells, i.e., organisms, can be sometimes explained by macroeconomic phenomena. If we can discover or implement financial solutions in those kinds of economies, this may provide valuable evidence for where a financial approach works best.
see:
https://www.hoover.org/sites/default/files/research/docs/16111_-_money_and_velocity_during_financial_crises_-_anderson_bordo_and_duca.pdf
"Dr. Richard G. Anderson's research focuses on the velocity of broad money (M2) from the Great Depression through the Great Recession. The model considers factors such as changes in uncertainty and risk premia, financial innovation, and major banking regulations"
Scott, does NGDPLT solve this problem?
"Remember this statement from the 1810 Bullion Committee’s report to Parliament: “The most detailed knowledge of the actual trade of the Country, combined with the profound science in all the principles of Money and Circulation, would not enable any man or set of men to adjust, and keep always adjusted, the right proportions of circulating medium in a country to the wants of trade.” In essence, the committee said that central banking was impossible. Consequently, money is always in disequilibrium." (From a post by Russell Napier)
I agree that money is always in disequilibrium, at least to some extent. I think of it this way: It's better to have a regime that minimizes disequilibrium, such as NGDPLT, rather than a regime that generates a large amount of disequilibrium.
Instead of thinking of it as solving the problem, think of it as minimizing the problem.
Complexity is a bummer for many!
Great post. As a non professional economist with a fair amount of formal education both in economics and finance, i agree with taking a bit more of a finance approach to the topic. You are right with regard to understanding the logic of how someone comes to a conclusion, forecast/prediction not only in economics, but in all areas of life.
As an aside i bought NVDA stock in 2015, not a lot, as i saw the changing business model and what GPU's would portend going forward. Did sell about have after a 12X gain and that was a mistake as AI came down the pike, something i had not seen to take place as fast as it did.
Have disliked the dual mandate of the Fed for as long as i can remember and have felt Fed should be concerned with only one thing: price stability.
But politics and economics make for a toxic mix.
Lastly, i always find your views to be well reasoned!
I include comments on the post in
https://thomaslhutcheson.substack.com/p/refereeing-sumner-vs-cowen
I think this was a good post and the right post to respond to Tyler with, but it does raise a conundrum. Your argument is basically that policymakers shouldn't rely on inflation forecasting models and should instead rely on market-based expectations of inflation (or nominal GDP). But how should the market participants forecast inflation (or nominal GDP)? I don't think nominal GDP is high because the Fed let nominal GDP be high is satisfactory for this purpose (maybe if the Fed actually had the objective to target NGDP).
"I don't think nominal GDP is high because the Fed let nominal GDP be high is satisfactory for this purpose".
I actually do think it is satisfactory. I am traveling over the holidays, but may do a post on this when I return--maybe at Econlog.
Under your original NGDP targeting proposal, fixing the futures price one year ahead for NGDP controls the expectations of NGDP. Realized NGDP may differ from expectations. Level targeting can help ensure that there aren’t significant cumulative forecast errors.
Under interest rate targeting, the argument is implicitly that a path of interest rates implies a path of expected NGDP (or inflation). But again realized NGDP (or inflation) can behave differently than expectations.
For instance, look at the inflation forecasts post COVID vs realized inflation. Inflation was way higher than forecast. Similarly, NGDP was way higher than forecast (I believe, without checking the numbers). So is the argument that policy was set such that it was consistent with a higher expected NGDP rate, but no one realized that NGDP should be expected to be higher? My problem is that this is all backward looking. It doesn’t let you know ex ante about the policy stance.
It's not "backward looking", as the distributed lag effect of money flows demonstrates. Monetary lags are mathematical constants.
Whether NGDPLT, inflation, or price level targeting, the target needs to have an "F" in front of its name. Targets make sense in the context of an average size and frequency of shocks. Extraordinary shocks require flexible, temporary above- target inflation or NGDP growth. So, _F_AIT, _F_NGDPLT,
"I don't think nominal GDP is high because the Fed let nominal GDP [or inflation] be high is satisfactory for this purpose."
It depends on the purpose, if one is critiquing policy or execution.