This submit is from a set of feedback I gave on the NBER Asset Pricing convention in early November at Stanford. Convention agenda right here. My full slides right here. There was video, however sadly I took too lengthy to write down this submit and the NBER took down the convention video.
I used to be requested to touch upon “Downward Nominal Rigidities and Bond Premia” by François Gourio and Phuong Ngo. It is a very good clear paper, so all I might suppose to do as discussant is reward it, then transfer on to greater points. These are actually feedback about entire literatures, not about one paper. One can admire the play however complain concerning the sport.
The paper implements a model of Bob Lucas’ 1973 “Worldwide proof” remark. Costs are much less sticky in excessive inflation international locations. The Phillips curve extra vertical. Output is much less affected by inflation. The Calvo fairy visits each night time in Argentina. To Lucas, excessive inflation comes with variable inflation, so individuals perceive that worth modifications are principally mixture not relative costs, and ignore them. Gourio and Ngo use a new-Keynesian mannequin with downwardly sticky costs and wages to specific the thought. When inflation is low, we’re extra usually within the more-sticky regime. They use this concept in a mannequin of bond threat premia. Occasions of low inflation result in extra correlation of inflation and output, and so a unique correlation of nominal bond returns with the low cost issue, and a unique time period premium.
I made two factors, first about bond premiums and second about new-Keynesian fashions. Solely the latter for this submit.
This paper, like lots of earlier than it, provides just a few elements on prime of a regular textbook new-Keynesian mannequin. However that textbook mannequin has deep structural issues. There are recognized methods to repair the issues. But we frequently construct on the usual mannequin, somewhat than incorporate recognized methods or discover new methods to repair its underlying issues.
Downside 1: The signal is “unsuitable” or not less than unconventional.
.tiff)
The essential signal is unsuitable — or not less than counter to the usual perception of all coverage makers. Within the mannequin, greater rates of interest trigger inflation to leap down instantly, after which rise over time. Everybody on the Fed uniformly believes that greater rates of interest trigger inflation to go nowhere instantly, after which gently decline over time, with “lengthy and variable lags.”
Larry Ball pointed this out 30 years in the past. The habits comes straight from the forward-looking Phillips curve. Decrease output goes with decrease inflation, relative to future inflation. I.e. inflation rising over time.
To be clear, perhaps the mannequin is true and the beliefs are unsuitable. It is wonderful that a lot modeling and empirical work has gone in to massaging principle and information to evolve to Milton Friedman’s 1968 proclamation of how financial coverage works. The “lengthy and variable lags” specifically are a bother to trendy economics. If you already know costs are going up tomorrow, you elevate costs as we speak. However that is for one more day. This mannequin doesn’t behave the best way most individuals suppose the financial system behaves, so if you are going to use it, not less than that wants a significant asterisk.
Nicely, we all know repair this. You possibly can see that sneaking lagged inflation into the Phillips curve goes to be an enormous a part of that.
.tiff)
Christiano Eichenbaum and Evans, 20 years in the past, produced a extensively cited mannequin that “fixes” this downside. It has a variety of elements. Most of all, it assumes that wages and costs are listed. Corporations and staff that do not get tapped by the Calvo fairy to vary their worth or wage nonetheless elevate by noticed inflation. This offers a Phillips curve with lagged inflation. Furthermore, in preferences, funding, and this Phillips curve, CEE modify the mannequin to place development charges rather than ranges. (Extra assessment in a three half collection on new-Keynesian fashions right here.)
Nice. However discover that neither Gourio and Pho nor just about anybody else builds on this mannequin. We cite it, however do not use it. As an alternative, 20 extra years of NK theorizing research totally different extensions of the essential mannequin, that do not clear up the central conundrum.
Downside 2: Fed induced explosions
.tiff)
The usual new-Keynesian mannequin says that if the Fed holds rates of interest fixed, inflation is secure — will go away by itself — however indeterminate. There are a number of equilibria. The usual new-Keynesian mannequin thus assumes that the Fed intentionally destabilizes the financial system. If inflation comes out greater than the Fed needs, the Fed will lead the financial system to hyperinflation or hyper deflation. Below that menace, individuals bounce to the inflation that the Fed needs to see.
However the Fed does no such factor. Central bankers resolutely state that their job is to stabilize the financial system, to carry inflation again from wherever it would go. Regardless of 1000’s of papers with new-Keynesian equations written at central banks, if anybody had been ever to actually describe these equations within the introduction, “we assume that the central financial institution is dedicated to reply to inflation by hyperinflation or deflation in an effort to choose from a number of equilibria” they might be laughed out of a job.
As an alternative, we go on with the identical mannequin and its underlying extensively counterfactual assumption about coverage.
Downside 3: The match is horrible
.tiff)
A mannequin consists of a set of equations, with the factor you need to decide (say, inflation) on the left, the financial causes described by the mannequin on the fitting, plus “shocks,” that are issues your mannequin cannot seize. Within the clarification half, there are parameters ((sigma, beta, kappa, phi)), that management how a lot the issues on the fitting have an effect on the issues on the left.
The match of new-Keynesian fashions is normally horrible. In accounting for financial variables ((x_t,) (pi_t, ) (i_t ) right here), the error phrases ((varepsilon)) are a lot bigger than the mannequin’s financial mechanisms (the (x,) (pi) on the fitting hand facet). Forecasts — predicting (pi), (x) forward of time — is worse. For instance, the place did inflation come from and why did it go away? Anticipated inflation hasn’t moved a lot, and the financial system simply plugged alongside. A lot of the rise and fall of inflation got here from inflation shocks.
Associated, the match of the fashions is about the identical quantity of horrible for various values of the parameters. Meaning the parameters are “poorly recognized” if recognized in any respect. That signifies that the mechanisms of the mannequin — say, how a lot greater rates of interest decrease output, after which how a lot decrease output impacts inflation — are weak, and poorly understood.
Partially this is not usually observed as a result of we received out of the behavior of evaluating fashions by match within the Eighties. Most fashions are evaluated, as I confirmed above for CEE by matching choose “recognized” impulse response features. However as these response features additionally clarify small variances of output and inflation, it is attainable to match response features properly, but nonetheless match the info badly, i.e. match the info solely by including massive shocks to each equation.
I do not know of excellent fixes right here. Quaint ISLM fashions had comparable issues (See Sims 1980). However it’s a undeniable fact that we simply ignore and go on.
The Phillips curve is a central downside, which has solely gotten worse currently. Unemployment was excessive and declining all through the 2010s, with secure inflation. Inflation got here with excessive unemployment in 2021. And inflation fell with no excessive actual rates of interest, no unemployment, and robust development in 2022-2023. However what is going to change it?
So the place are we?
.tiff)
Macro is surprisingly un-cumulative. We begin with a textbook mannequin. Individuals discover some shortcomings and recommend a repair. However somewhat than incorporate that repair, the subsequent paper provides a unique repair to the identical textbook mannequin. One would suppose we’d observe the trail on the fitting. We do not. We observe the trail on the left.
That is frequent in economics. The actual enterprise cycle literature adopted a lot the identical path. After the King Plosser Rebelo stochastic development mannequin grew to become the usual, individuals spent a decade with one extension after one other, every properly motivated to repair a stylized reality. However by and enormous the subsequent paper did not construct on the final one, however as a substitute supplied a brand new variation on the KPR mannequin.
Posteriors observe priors in keeping with Bayes’ rule, after all. So one other manner of placing the remark, individuals appear to place a reasonably excessive prior on the unique mannequin, however do not belief the variations in any respect.
Wę want a digestion. Which of the brand new elements are dependable, sturdy, and belong as a part of the brand new “textbook” mannequin? That is not simple. Dependable and sturdy could be very onerous to search out, and to steer individuals. There are such a lot of to select from — CEE’s smorgasbord, capital, monetary frictions, heterogeneous brokers, totally different expectation formation tales, totally different pricing frictions, and so forth. What is the minimal simple set of those to make use of?
A part of the difficulty lies in how publishing works. It is almost unimaginable to publish a paper that removes previous elements, that digests the mannequin right down to a brand new textbook model. The rewards are to publishing papers that add new elements. Even when, like CEE, everybody cites them however would not use them.
I’ve requested many economists why they construct on a mannequin with so many recognized issues, and why they do not embody recognized fixes. (Not simply fiscal principle!) The reply is normally, sure, I find out about all these issues, however no one will trouble me about them since each different paper makes the identical assumptions, and I have to get papers printed.
I went on a little bit of a tear right here as I referee plenty of nice papers like this one. Each a part of the paper is nice, besides it builds on a mannequin with massive flaws we have recognized about for 30 years. It feels unfair to complain concerning the underlying mannequin, because the journal has printed and can publish 100 different papers. However at what level can we, collectively, scream “Cease!”
The brand new-Keynesian mannequin has been the usual mannequin for an astonishing 30 years. None of ISLM, monetarism, rational expectations, or actual enterprise cycles lasted that lengthy. It is much more wonderful that it’s so unchanged in all this time. It’s positively time for a greater textbook model of the mannequin! Perhaps this can be a plea for Woodford, Gali or one of many different NK textbook authors, which a lot better command of all of the variations than I’ve, to bless us a brand new textbook mannequin.
Or, maybe it is time for one thing completely new.
That is not fiscal principle per se. Fiscal principle is an ingredient, not a mannequin. You possibly can marry it to new-Keynesian fashions, as I, Leeper, Sims, and others have achieved. However you too can marry it to previous ISLM or anything you need. Given the above, perhaps there is not an present modification however a brand new begin. I do not know what that’s.
(My feedback even have some comparable feedback about time period premiums and the way to consider them, however this submit is lengthy sufficient.)
Replace:
The previous is a VAR with excessive frequency measurement of the financial coverage shock. And..
 |
| Supply: Miranda-Agrippino and Ricco |
The worth stage in addition to the inflation charge can bounce down instantly when the rate of interest rises! (I believe the graph plots the extent of CPI, not development charge.) That is even stronger than the baseline mannequin by which the worth stage, being sticky, doesn’t transfer, however the inflation charge jumps on the rate of interest rise.
The latter is a pleasant theoretical paper. It provides a variety of the CEE assumptions. I overstated an important deal that others haven’t used these elements. They’re utilized in these “medium scale” fashions, simply not in “textbook” fashions. Nonetheless, it removes listed costs and wages with purely ahead trying Phillips curves. It provides intermediate items nonetheless. This makes costs modifications work by the community of suppliers including fascinating dynamics, which has all the time struck me as a vital ingredient. And…
 |
| Supply: Phaneuf, Sims ,and Victor |
The principle estimate is the darkish line. Right here you see a mannequin with the traditional response: inflation doesn’t transfer on influence, and will increase a while after the rate of interest rise.