A small note: the fact that inflation might jump on impact isn't entirely uncomfortable. Some high-frequency proxies can give you an inflation jump on impact (e.g. Miranda-Aggripino and Ricco (2021) have that jump). However, the standard 3 equations model wants the peak inflation response to occur immediately and you just won't find that in the empirical literature as far as I know. So, as long as you can get to a small jump followed by a hump, you're probably okay.
Very interesting indeed! I can't help but wonder though if higher interest rates tend to spur the belief that politicians will feel more pressure to be fiscally responsible (and people believe politicians don't think too deeply about the difference between nominal and real interest rates).
This is a very interesting hypothesis. Chris Sims uses it to explain the 1980s. The Fed induced really high interest rates. That induced big interest costs on the debt, which induced Congress to get serious about fiscal reforms. A regime in which Congress responds quickly to interest costs on the debt but not so on other spending is an interesting one.
A major confounding factor is surely the extent to which "inflation" is extractive / monopolistic economic rent, which was largely not a problem since transport efficiency and free trade and other technology like refrigeration eliminated it in most products. That is, during many decades of the 2th century in most of the 1st world. In fact, not only has extractive / monopoly economic rent been eliminated, its reverse, consumer surplus, has increased steadily as supply chains for important inputs have become more efficient and competitive.
Modern urban planning fads that ration the supply of land to prevent "sprawl" have restored this worst form of economic rent in housing, and the changes that occur in property markets are systemic not merely cyclical. The REAL price of houses at least doubles within a single cycle, and furthermore as long as upzoning is occurring (almost always the case when Planning is attempting to reduce sprawl), there is a shrinkage in the size of new homes that does not reflect in median or average prices. The price of land inflates exponentially based on the ability to gouge or extract "the maximum that new entrants to the market can stand to pay" for the minimum the market is allowed to offer. There is no data that supports the assumption that denser cities have a lower real median or average house price; the reverse tends to be true.
Further complicating the impacts for the purposes of modeling, is that most of it is in mortgage repayments over a long time to come; the impact is not "upfront". It is like a creeping "reverse stimulus" in that household discretionary income is reduced for a long time to come by the added burden of paying off a home. Conversely, the few cities that have managed to resist these planning trends and keep their housing systemically affordable, start to demonstrate greater resilience in local spending as time goes on. Gradually over time, more and more mortgage holders are "new entrants to the market since the systemic change" so the creeping impounding of aggregate household discretionary income is gradual and slow.
And to further complicate things, this economic rent being extracted from households (it also tends to be extracted from businesses for their space) is capable of expansion to capture any reducing cost burdens elsewhere in the economy; if food, clothing, and important products and services fall in real cost, households will tend to bid up the price of their own housing with the added available funds. The same goes for "middle class welfare" by which well intentioned politicians attempt to assist with cost burdens on households. Generous payments for each child borne by a household, for example, have certainly contributed to house price inflation in Australia.
I believe this whole issue is "the elephant in the macro-economic room" and will sabotage all efforts to model the impacts of all the other, commonly assumed, important inputs.
I don't know if you can make it work in a formal model, but the question I have in my head is who is subsidizing what over which time period? And then how does that compare to prior expectations? My mental model is that over the short run when the fed raises their real interest rate in response to inflation they are paying a subsidy to banks to temporarily reduce the supply of money by parking it at the fed. In the long run the government has to either pay off the cost of that subsidy or inflate it away, but this long run effect is generally known, FTPL beliefs about future fiscal policy are generally not meaningfully impacted by a fed rate change that is more or less already priced into the existing long term rate of inflation, and any short term change in the fed rate only causes small updates on these long term beliefs (in and of itself without changes in belief about the government). In the short run though, the subsidy to remove money supply is highly leveraged with fractional reserves such that the economy will contract. And it is believed that if (real) fed rates go higher then the economy will contract further such that the short run negative effects on the economy of a small rate change have anticipatory effects on inflation, e.g. if the fed raises rates people expect that a recession is more likely in which case employees have less ability to negotiate an (inflation based) raise with their employer. And so in this version, in the long run you still have to pay off the cost of the subsidy or inflate it away a la FTPL, but that is already almost entirely priced in and in the short run that's just completely dwarfed by the 20x or more reduction in productive working capital banks will no longer be loaning per dollar parked at the fed to farm the temporary fed deposit subsidy. And so on the fed's happy path the immediate effect of a temporary rate hike is a short term subsidy that is sufficient to kill short run inflation without changing long term beliefs or requiring an escalatory spiral. Even on the unhappy path in this view the fed is not threatening explosive inflation off equilibrium because it can always decrease the temporary money supply and increase the price of capital at a relatively low price to itself and thus to the government. Inflation and rates spiral on the unhappy path occur because of a FTPL change in beliefs about the type of government, not because the fed goes nuclear. In this view, what the fed is doing on the happy path is already bad enough to force a reckoning and countries that end up with an inflationary spiral do so because their government is revealed to be significantly more irresponsible than people previously believed. More thoughts along these lines in a comment on your interest rates and exchange rates post: https://www.grumpy-economist.com/p/do-higher-interest-rates-raise-the/comment/69274872
It used to be that the fastest way to learn was to post something that was wrong on the internet and someone would quickly come along to correct you, but now you can also feed that into ChatGPT and it can tell you immediately that you’ve apparently missed a fundamental point from FTPL that all government debt including treasury debt is more or less equivalent and that post 2008 reserves are well in excess of requirements such that there is no real multiplier effect like I described. Is that true? Someone please correct me here or correct ChatGPT.
I see the point from the point of view of an individual investor choosing between assets that it’s really just a difference in maturity, but it still seems to me that roughly speaking the government plans to spend the principle plus interest when it issues treasuries, but it plans to only spend the interest when it pays interests on fed deposits. And while it also drives up the interest on other government debt, paying people to park money you are not actually doing anything with still seems like a special case. Sure, if the government paid people to just hide cash under a mattress when politicians have recently been irresponsible that could have a similar effect, but try getting politicians to do that. I am also still not entirely shaking the multiplier on deposits intuition: maybe the relevant thing is not that it is a federal reserve, but that it is money that you are paid so that principle can just sit there without any productive function? This seems different from borrowing to finance any other function of government, even entitlements, in a way that is multipliery. Please someone explain to me the error of my ways!
In a recent working paper, we explore a different mechanism - tied to another woefully under-appreciated variable in the macro literature - that delivers the correct shape and direction of inflation following MP shocks: variable capacity utilization. We basically show that when demand is uncertain and capacity takes time to raise (both obviously true), firms like keeping some "precautionary capacity" on hand. So when an expansionary MP shock raises demand, more capacity gets used up, leading to higher capacity utilization - again, obviously true from the data. But we show that this has two countervailing effects: first using the *same capacity* (since capacity can't be immediately raised) to produce *more output* implies that unit costs are lower. This "productivity effect" drives prices down. On the other hand, higher demand means that firms hit their capacity constraint more often - giving firms more pricing power. This leads them to raise prices - by raising markups - to eliminate excess demand. This "markup effect" raises prices. The interplay between these two effects generates the hump-shape. When there is a lot of slack in the economy (say, after a recession), lags are therefore very long; inflation rises gently with a hump shape; a "price puzzle" may even form. When excess capacity is already low in the economy (think post-Covid), transmission is near-instant leading to a sharp increase in inflation. The rest of the model is standard NK, so the linearized Phillips curve looks very similar except for an extra term that captures the state-dependency with capacity utilization.
Very interesting paper. How did you manage to eliminate all time-0 jumps in all variables? In NK Taylor rule models something has to jump to the saddle path.
Thanks! Variables do jump at time 0 in the DSGE model. What you see in Figure 6 is an artifact of the Bayesian impulse response matching we do to match the model to the data. The empirical impulse responses are based on a VAR identified under the assumption that the shock affects only the FFR at time 0 (as in Christiano, Eichenbaum, & Evans 2010, 2016, 2021). The other variables are zero by assumption at time 0, so the model is matched to the data only from time period 1 onwards. This is what the figure shows. You can see the instantaneous response of inflation in Figure 4 for example, and it jumps (either up or down depending on the amount of slack in the economy) at time 0.
I found it very interesting and it got me thinking.
I think much of the debate centres on the interest elasticity of money demand. As Tobin wrote, this elasticity 'is a key parameter in macroeconomic theory' (1993, p. 53). Is the demand for money elastic or inelastic in response to market interest rates? The answer is crucial for understanding the impact of interest rates on inflation.
[Tobin, J. (1993). Price flexibility and output stability: an old Keynesian view. Journal of Economic Perspectives, 7(1), pp. 45–65].
A small note: the fact that inflation might jump on impact isn't entirely uncomfortable. Some high-frequency proxies can give you an inflation jump on impact (e.g. Miranda-Aggripino and Ricco (2021) have that jump). However, the standard 3 equations model wants the peak inflation response to occur immediately and you just won't find that in the empirical literature as far as I know. So, as long as you can get to a small jump followed by a hump, you're probably okay.
Very interesting indeed! I can't help but wonder though if higher interest rates tend to spur the belief that politicians will feel more pressure to be fiscally responsible (and people believe politicians don't think too deeply about the difference between nominal and real interest rates).
This is a very interesting hypothesis. Chris Sims uses it to explain the 1980s. The Fed induced really high interest rates. That induced big interest costs on the debt, which induced Congress to get serious about fiscal reforms. A regime in which Congress responds quickly to interest costs on the debt but not so on other spending is an interesting one.
A major confounding factor is surely the extent to which "inflation" is extractive / monopolistic economic rent, which was largely not a problem since transport efficiency and free trade and other technology like refrigeration eliminated it in most products. That is, during many decades of the 2th century in most of the 1st world. In fact, not only has extractive / monopoly economic rent been eliminated, its reverse, consumer surplus, has increased steadily as supply chains for important inputs have become more efficient and competitive.
Modern urban planning fads that ration the supply of land to prevent "sprawl" have restored this worst form of economic rent in housing, and the changes that occur in property markets are systemic not merely cyclical. The REAL price of houses at least doubles within a single cycle, and furthermore as long as upzoning is occurring (almost always the case when Planning is attempting to reduce sprawl), there is a shrinkage in the size of new homes that does not reflect in median or average prices. The price of land inflates exponentially based on the ability to gouge or extract "the maximum that new entrants to the market can stand to pay" for the minimum the market is allowed to offer. There is no data that supports the assumption that denser cities have a lower real median or average house price; the reverse tends to be true.
Further complicating the impacts for the purposes of modeling, is that most of it is in mortgage repayments over a long time to come; the impact is not "upfront". It is like a creeping "reverse stimulus" in that household discretionary income is reduced for a long time to come by the added burden of paying off a home. Conversely, the few cities that have managed to resist these planning trends and keep their housing systemically affordable, start to demonstrate greater resilience in local spending as time goes on. Gradually over time, more and more mortgage holders are "new entrants to the market since the systemic change" so the creeping impounding of aggregate household discretionary income is gradual and slow.
And to further complicate things, this economic rent being extracted from households (it also tends to be extracted from businesses for their space) is capable of expansion to capture any reducing cost burdens elsewhere in the economy; if food, clothing, and important products and services fall in real cost, households will tend to bid up the price of their own housing with the added available funds. The same goes for "middle class welfare" by which well intentioned politicians attempt to assist with cost burdens on households. Generous payments for each child borne by a household, for example, have certainly contributed to house price inflation in Australia.
I believe this whole issue is "the elephant in the macro-economic room" and will sabotage all efforts to model the impacts of all the other, commonly assumed, important inputs.
I don't know if you can make it work in a formal model, but the question I have in my head is who is subsidizing what over which time period? And then how does that compare to prior expectations? My mental model is that over the short run when the fed raises their real interest rate in response to inflation they are paying a subsidy to banks to temporarily reduce the supply of money by parking it at the fed. In the long run the government has to either pay off the cost of that subsidy or inflate it away, but this long run effect is generally known, FTPL beliefs about future fiscal policy are generally not meaningfully impacted by a fed rate change that is more or less already priced into the existing long term rate of inflation, and any short term change in the fed rate only causes small updates on these long term beliefs (in and of itself without changes in belief about the government). In the short run though, the subsidy to remove money supply is highly leveraged with fractional reserves such that the economy will contract. And it is believed that if (real) fed rates go higher then the economy will contract further such that the short run negative effects on the economy of a small rate change have anticipatory effects on inflation, e.g. if the fed raises rates people expect that a recession is more likely in which case employees have less ability to negotiate an (inflation based) raise with their employer. And so in this version, in the long run you still have to pay off the cost of the subsidy or inflate it away a la FTPL, but that is already almost entirely priced in and in the short run that's just completely dwarfed by the 20x or more reduction in productive working capital banks will no longer be loaning per dollar parked at the fed to farm the temporary fed deposit subsidy. And so on the fed's happy path the immediate effect of a temporary rate hike is a short term subsidy that is sufficient to kill short run inflation without changing long term beliefs or requiring an escalatory spiral. Even on the unhappy path in this view the fed is not threatening explosive inflation off equilibrium because it can always decrease the temporary money supply and increase the price of capital at a relatively low price to itself and thus to the government. Inflation and rates spiral on the unhappy path occur because of a FTPL change in beliefs about the type of government, not because the fed goes nuclear. In this view, what the fed is doing on the happy path is already bad enough to force a reckoning and countries that end up with an inflationary spiral do so because their government is revealed to be significantly more irresponsible than people previously believed. More thoughts along these lines in a comment on your interest rates and exchange rates post: https://www.grumpy-economist.com/p/do-higher-interest-rates-raise-the/comment/69274872
It used to be that the fastest way to learn was to post something that was wrong on the internet and someone would quickly come along to correct you, but now you can also feed that into ChatGPT and it can tell you immediately that you’ve apparently missed a fundamental point from FTPL that all government debt including treasury debt is more or less equivalent and that post 2008 reserves are well in excess of requirements such that there is no real multiplier effect like I described. Is that true? Someone please correct me here or correct ChatGPT.
I see the point from the point of view of an individual investor choosing between assets that it’s really just a difference in maturity, but it still seems to me that roughly speaking the government plans to spend the principle plus interest when it issues treasuries, but it plans to only spend the interest when it pays interests on fed deposits. And while it also drives up the interest on other government debt, paying people to park money you are not actually doing anything with still seems like a special case. Sure, if the government paid people to just hide cash under a mattress when politicians have recently been irresponsible that could have a similar effect, but try getting politicians to do that. I am also still not entirely shaking the multiplier on deposits intuition: maybe the relevant thing is not that it is a federal reserve, but that it is money that you are paid so that principle can just sit there without any productive function? This seems different from borrowing to finance any other function of government, even entitlements, in a way that is multipliery. Please someone explain to me the error of my ways!
In a recent working paper, we explore a different mechanism - tied to another woefully under-appreciated variable in the macro literature - that delivers the correct shape and direction of inflation following MP shocks: variable capacity utilization. We basically show that when demand is uncertain and capacity takes time to raise (both obviously true), firms like keeping some "precautionary capacity" on hand. So when an expansionary MP shock raises demand, more capacity gets used up, leading to higher capacity utilization - again, obviously true from the data. But we show that this has two countervailing effects: first using the *same capacity* (since capacity can't be immediately raised) to produce *more output* implies that unit costs are lower. This "productivity effect" drives prices down. On the other hand, higher demand means that firms hit their capacity constraint more often - giving firms more pricing power. This leads them to raise prices - by raising markups - to eliminate excess demand. This "markup effect" raises prices. The interplay between these two effects generates the hump-shape. When there is a lot of slack in the economy (say, after a recession), lags are therefore very long; inflation rises gently with a hump shape; a "price puzzle" may even form. When excess capacity is already low in the economy (think post-Covid), transmission is near-instant leading to a sharp increase in inflation. The rest of the model is standard NK, so the linearized Phillips curve looks very similar except for an extra term that captures the state-dependency with capacity utilization.
The paper is here: https://vasudeva-ram.github.io/assets/pdf/JobMarketPaper.pdf
A twitter thread: https://x.com/VasudevaRamasw1/status/1861821431723512180
Very interesting paper. How did you manage to eliminate all time-0 jumps in all variables? In NK Taylor rule models something has to jump to the saddle path.
Thanks! Variables do jump at time 0 in the DSGE model. What you see in Figure 6 is an artifact of the Bayesian impulse response matching we do to match the model to the data. The empirical impulse responses are based on a VAR identified under the assumption that the shock affects only the FFR at time 0 (as in Christiano, Eichenbaum, & Evans 2010, 2016, 2021). The other variables are zero by assumption at time 0, so the model is matched to the data only from time period 1 onwards. This is what the figure shows. You can see the instantaneous response of inflation in Figure 4 for example, and it jumps (either up or down depending on the amount of slack in the economy) at time 0.
Markets economy, complex system, is closer to the cat than the washing machine, but tends to be mistaken for washing machines :) https://open.substack.com/pub/themonetaryfrontier/p/the-magical-moniac-water-machine?r=1z5qli&utm_medium=ios
I found it very interesting and it got me thinking.
I think much of the debate centres on the interest elasticity of money demand. As Tobin wrote, this elasticity 'is a key parameter in macroeconomic theory' (1993, p. 53). Is the demand for money elastic or inelastic in response to market interest rates? The answer is crucial for understanding the impact of interest rates on inflation.
[Tobin, J. (1993). Price flexibility and output stability: an old Keynesian view. Journal of Economic Perspectives, 7(1), pp. 45–65].