Suchergebnisse

Imagine my surprise when I see a statement "rates are now barely positive according to all official inflation and rate data" in an article titled "Will the Fed Elect Biden?" and the accompanying graph: Source: ZeroHedge. Notes: (Data: Federal Reserve Economic Data (FRED), St. Louis Fed; Chart: Jeffrey A. Tucker) Well, the graph looks this […]

By Shirin Malekpour and Jens Newig This research was a collaboration between Monash Sustainable Development Institute, Monash University (Melbourne, Australia), and the Research Group Governance and Sustainability, Leuphana University (Germany). Dr Shirin Malekpour (Monash) is the recipient of the Green Talents award, which enabled her to undertake a research sabbatical at Leuphana in 2019, conducting … Continue reading What it takes to exercise adaptive planning

'My interest in entangled colonial and imperial histories [has] fuelled questions about how Vincentian heritages are affected by the communities' relationship to land, foreign investment/settlement, and the more-than-human world.'

Die KI-Methoden, um Schüler individuell zu fördern und
Lehrkräfte zu entlasten, existieren. Doch damit sie den Unterricht wirklich verändern können, muss die Politik die Kooperation von Wissenschaft, EdTech und Bildungspraxis systematisch
unterstützen.
Ein Gastbeitrag von Detmar Meurers.







Detmar Meurers ist Professor für Computerlinguistik an der Universität Tübingen und leitet ab April die Arbeitsgruppe "Sprache und KI in der
Bildung" am Leibniz-Institut für Wissensmedien (IWM).
 Foto: IWM. 






KATHARINA GÜNTHER-WÜNSCH fasste es als KMK-Präsidentin Ende 2023 klar zusammen: "Die Ergebnisse der PISA-Studie 2022 sind besorgniserregend, sie
bestätigen die Befunde der IGLU-Studie sowie der IQB-Bildungstrends 2021 und 2022. Eine zunehmend heterogene Schülerschaft stellt das Schulsystem und auch die Lehrkräfte vor enorme
Herausforderungen. [....] Alle sind sich einig, dass es jetzt vor allem auf die Stärkung der Basiskompetenzen ankommt." 



 



Dass bei der Stärkung der Basiskompetenzen in einer heterogenen Schülerschaft digitale Systeme eine wichtige Rolle spielen können, liegt nach dem Strategiepapier "Bildung in der Digitalen Welt" nahe, in dem die
KMK bereits 2017 festhielt: "Digitale Medien halten ein großes Potential zur Gestaltung neuer Lehr- und Lernprozesse bereit, wenn wir allein an die Möglichkeiten zur individuellen Förderung von
Schülerinnen und Schülern denken."



 



Allerdings standen dann bei der Digitalisierung der Bildung in den vergangenen sechs Jahren vor allem die Hardwareausstattung und Internetverbindung von Schulen im Fokus sowie die
medienpädagogische Vorbereitung auf eine zunehmend digitale Welt. Dieser Beitrag soll ein Weckruf sein, neben Visionen von Digitalität tatsächlich gezielt digitale Methoden zu entwickeln und
systematisch zu nutzen, die konkret die eingangs identifizierten Herausforderungen der schulischen Bildung angehen. Hierfür bieten insbesondere adaptive intelligente Systeme in der Schulpraxis
effektive Möglichkeiten zur individuellen Förderung der Lernenden und Unterstützung der Lehrkräfte. 



 



Heterogenität
ernst nehmen



 



Lernende unterscheiden sich substanziell nicht nur in ihrem fachlichen Wissen und ihren Kompetenzen, sondern auch in ihren bildungssprachlichen Fähigkeiten, kognitiven Eigenschaften, Interessen
und Motivation und ihrem soziokulturellen Hintergrund. Um das Potenzial digitaler Medien zur adaptiven individuellen Förderung von Basiskompetenzen realisieren zu können, benötigt es das
Zusammenspiel von Lehr-Lernforschung, Fachdidaktik und KI-Methoden (i) zur Modellierung der individuellen Lernenden und der curricularen Ziele und (ii) zur Erstellung und Auswahl von vielfältig
parametrisierten Aufgaben für adaptive Lernpfade. Außerdem kommt es (iii) auf die automatische Verarbeitung von Sprache  an zur Analyse sprachlicher Komplexität und dem Generieren von
Feedback.



 



Entgegen dem medialen Hype um KI als unspezifisches Wundermittel werden KI-Methoden hier also gezielt genutzt. Dabei können traditionelle und generative KI-Methoden integriert, weiterentwickelt
und evaluiert werden, um adaptive individuelle Lernpfade im realen komplexen Schulkontext zu ermöglichen. Denn dieser ist von verschiedensten Lernvoraussetzungen, pädagogischen Möglichkeiten und
expliziten curricularen Lernzielen geprägt.



 



Feedback und Lernpfade
im realen Schulkontext



 



Die Wirkung von individuellem Feedback während der Aufgabenbearbeitung haben wir anhand des von uns entwickelten Intelligenten Tutorsystems (ITS) FeedBook untersucht, das Übungen für die 7. Klasse Englisch mit KI-generiertem Feedback anbietet. Die Abbildung
illustriert eine Rückmeldung des Systems zur Formulierung von Vergleichen, die erklärt, wie die relevanten Formen zu bilden sind. 

















Solche Rückmeldungen müssen nicht pro Aufgabe hinterlegt werden, sondern ergeben sich dank der KI-Methoden aus den allgemeinen Regularitäten der Sprache und dem pädagogischen Modell. Ob solches
Feedback während der Aufgabenbearbeitung auch im regulären Schulkontext wirkt, haben wir in der ersten randomisierten Feldstudie mit einem ITS in deutschen Schulen untersucht. Das System wurde von allen genutzt, jedoch erhielten die
in zwei Gruppen unterteilten Lernenden zu unterschiedlichen sprachlichen Mitteln – etwa zum Einsatz von Nebensätzen, Vergleichen oder Konditionalsätzen – spezifisches Feedback vom
System. Der Lernerfolg bei den sprachlichen Mitteln, zu denen sie Feedback erhielten, war in den ansonsten wie üblich unterrichten Klassen um 63 Prozent höher. 



 



Solche adaptiven Systeme können also Lehrkräfte von Routineaufgaben entlasten und der Heterogenität durch die adaptive individuelle Vorbereitung so begegnen, dass die Lernenden besser am
gemeinsamen Unterricht teilhaben können – wie bei einem Orchester, bei dem jede und jeder individuell entsprechend den Fähigkeiten geübt hat, sodass bei den Proben und im Konzert der Fokus auf
dem Zusammenspiel liegen kann. Die effektive Integration des individuellen Übens und des gemeinsamen Unterrichts haben wir in einer weiteren Feldstudie mit der "Interact4School" genannten Weiterentwicklung unseres ITS-Systems FeedBook
untersucht und konnten zeigen, dass eine explizite Motivation des Übens als Vorbereitung für kommunikative Aufgaben in der Klasse den Lernerfolg weiter stärkt. 



 



Weniger unter Druck, zufriedener
mit der eigenen Leistung



 



Um die Idee von adaptiven Lernpfaden konkret zu machen, sehen wir uns eine Umsetzung im deutschen Schulkontext an, die wir gemeinsam mit der Universität Lüneburg und dem IÖB Oldenburg in dem von
der Joachim Herz Stiftung geförderten ALEE-Projekt entwickelt und in einer
gerade abgeschlossenen randomisierten Feldstudie in zehn Schulen in drei Bundesländern im regulären Unterricht untersucht haben. Hierfür haben wir systematisch Lernaufgaben unterschiedlicher
fachlicher, sprachlicher, kognitiver Komplexität für das in den Bildungsplänen verankerte Thema "Markt und Preisbildung" entwickelt und in einer digitalen Plattform bereitgestellt. Die Hälfte
jeder Klasse erhielt Aufgaben nach einer festen Liste, wie sie üblicherweise von Lehrkräften festgelegt wird; die Aufgaben für die andere Klassenhälfte wurden vom System individuell adaptiv
ausgewählt. Die Abbildung links zeigt die Lösungswahrscheinlichkeiten für den Standardpfad und rechts zwei individuelle Pfade durch die vielfältig parametrisierte Aufgabenlandschaft.

















Trotz der sehr unterschiedlich langen Lernpfade empfand die adaptive Gruppe das Lernen einer ersten Auswertung zufolge als interessanter, die Lernenden fühlen sich weniger unter Druck, waren mit
der eigenen Leistung zufriedener und kamen weiter im Lernstoff. Ein solcher Ansatz kann also gerade die zentrale Herausforderung adressieren, für eine heterogene Gruppe die grundlegenden
Kenntnisse sicherzustellen, auf denen dann die gemeinsame Arbeit in der Klasse aufbauen kann. 



 



Zusammenfassend kann die gezielte Entwicklung und der systematische Einsatz adaptiver digitaler Werkzeuge konkrete Lösungen bieten für die Stärkung grundlegender Kompetenzen einer zunehmend
heterogenen Schülerschaft. Für eine erfolgreiche Implementierung und kontinuierliche Verbesserung sollte die Politik Ansätze unterstützen, die die Verbindung zwischen Wissenschaft, Unternehmen
und Bildungspraxis stärken. Zugleich sollte sie eine Finanzierung adaptiver Bildungsmedien sicherstellen und die Nutzung von Bildungsdaten für adaptive Lernförderung und dadurch eine verbesserte
Bildungsgerechtigkeit ermöglichen.



 



Der Gastbeitrag von Detmar Meurers basiert auf dem ebenfalls von ihm verfassten Text "KI-Methoden für konkrete
Herausforderungen in der Bildung“, der heute in der Reihe "Analysen & Argumente" der Konrad-Adenauer-Stiftung erscheint.



Kostenfreien Newsletter abonnieren




In eigener Sache: Es geht so nicht mehr





Dieser Blog hat sich zu einer einschlägigen Adresse der Berichterstattung über die bundesweite Bildungs- und Wissenschaftspolitik entwickelt. Doch
wirtschaftlich steht die Idee seiner freien Zugänglichkeit vor dem Scheitern. 














 



 Mehr lesen...

Bottom line: No, farmers are not elk. However, could they adapt to climate change like elk?

A new study from UC Berkeley examined how annual elk migrations from valleys to the high country and back again are triggered by proximate environmental cues such as emergence of spring vegetation. Since climate change is shifting the timing and geography of those environmental cues, the researchers expect the elk to shift their ranges in order to adapt. These adaptive strategies can help the elk population keep up with climate change, although there will be ripple effects through the broader ecosystem given the importance of elk in the overall set of ecological interactions in places like Yellowstone National Park.

Could farmers and agriculture follow a similar adaptive strategy?

As difficult as policing vigorous dissent may be, having a flexible and adaptive strategy that draws from best practice and lessons learned will be the most effective way for police to plan, execute, adapt, and resolve the scenes to which they are called.

Facing a persistent housing crisis, Los Angeles is doubling down on converting unused commercial buildings into residential properties. But high interest rates make conversions more costly to finance. To mitigate this risk, L.A. might consider fiscal policy that would tip the scale more convincingly toward adaptive reuse.

Botnets are networks of computers infected with malware that an attacker controls and uses to fulfill malicious cyber activities. Decisionmakers could incorporate a complex adaptive systems perspective to assess if their organization and immediate network, as well as the overall ecosystem, are adaptable and resilient enough to respond to botnet activities.

As the debate surrounding AI regulation continues, it is crucial for policymakers, industry leaders, and civil society to engage in constructive dialogue and collaboration to develop a nuanced and adaptive regulatory approach that can keep pace with the breakneck speed of AI advancement
The post An AI Healthcare Coalition Suggests a Better Way of Regulating AI appeared first on American Enterprise Institute - AEI.

In Innovation for the Masses: How to Share the Benefits of the High-Tech Economy, Neil Lee proposes abandoning the Silicon Valley-style innovation hub, which concentrates its wealth, for alternative, more equitable models. Emphasising the role of the state and the need for adaptive approaches, Lee makes a nuanced and convincing case for reimagining how we “do” … Continued

In Activist Affordances: How Disabled People Improvise More Habitable Worlds, Arseli Dokumacı argues that in the adaptive ways they improvise everyday tasks, disabled people demonstrate how all people can create a more habitable planet. Connecting ideas from the fields of ethnography, psychology, disability studies and performance studies, Dokumacı's original work challenges normative, ableist conceptions of activism and environmental protection, writes Kostadin Karavasilev. … Continued

I just finished a new draft of "Expectations and the neutrality of interest rates," which includes some ruminations on inflation that may be of interest to blog readers. A central point of the paper is to ask whether and how higher interest rates lower inflation, without a change in fiscal policy. That's intellectually interesting, answering what the Fed can do on its own. It's also a relevant policy question. If the Fed raises rates, that raises interest costs on the debt. What if Congress refuses to tighten to pay those higher interest costs? Well, to avoid a transversality condition violation (debt that grows forever) we get more inflation, to devalue outstanding debt. That's a hard nut to avoid.  But my point today is some intuition questions that come along the way. An implicit point: The math of today's macro is actually pretty easy. Telling the story behind the math, interpreting the math, making it useful for policy, is much harder. 1. The Phillips curveThe Phillips curve is central to how the Fed and most policy analysts think about inflation. In words, inflation is related to expected future inflation and by some measure if economic tightness, factor \(x\). In equations, \[ \pi_t = E_t \pi_{t+1} + \kappa x_t.\] Here \(x_t\) represents the output gap (how much output is above or below potential output), measures of labor market tightness like unemployment (with a negative sign), or labor costs. (Fed Governor Chris Waller has a great speech on the Phillips curve, with a nice short clear explanation. There are lots of academic explanations of course, but this is how a sharp sitting member of the FOMC thinks, which is what we want to understand. BTW, Waller gave an even better speech on climate and the Fed. Go Chris!)  So how does the Fed change inflation? In most analysis, the Fed raises interest rates; higher interest rates cool down the economy lowering factor x; that pushes inflation down. But does the equation really say that? This intuition thinks of the Phillips curve as a causal relation, from right to left. Lower \(x\) causes lower inflation. That's not so obvious. In one story, the Phillips curve represents how firms set prices, given their expectation of other's prices and costs. But in another story, aggregate demand raises prices, and that causes firms to hire more (Chris Waller emphasized these stories). This reading may help to digest an otherwise puzzling question: Why are the Fed and its watchers so obsessed with labor markets? This inflation certainly didn't start in labor markets, so why put so much weight on causing a bit of labor market slack? Well, if you read the Phillips curve from right to left, that looks like the one lever you have. Still, since inflation clearly came from left to right, we still should put more emphasis in curing it that way. 2. Adjustment to equilibrium vs. equilibrium dynamics. But does the story work? Lower \(x_t\) lowers inflation \(\pi_t\) relative to expected future inflation \(E_t \pi_{t+1}\). Thus, it describes inflation that is rising over time.  This does not seem at all what the intuition wants. So how do we get to the intuition that lower \(x_t\) leads to inflation got goes down over time?  (This is on p. 16 of the paper by the way.) An obvious answer is adaptive expectations: \(E_t \pi_{t+1} = \pi_{t-1}\).  Then lower \(x_t\) does mean inflation today lower than it was in the past. But the Fed and most commenters really don't want to go there. Expectations may not be "rational," and in most commentary they are either "anchored" by faith in the Fed, or  driven by some third force. But they aren't mechanically last year's inflation. If they were, we would need much higher interest rates to get real interest rates above zero. Perhaps the intuition comes from remembering these adaptive expectations dynamics, and not realizing that the new view that expectations are forward looking, even if not rational, undermines those dynamics. Another answer  may be confusion between adjustment to equilibrium and movement of equilibrium inflation over time. Lower \(x_t\) means lower inflation \(\pi_t\) than would otherwise be the case. But that  reduction is an adjustment to equilibrium. It's not how inflation we observe -- by definition, equilibrium inflation -- evolves over time. This is, I think, a common confusion. It's not always wrong. In some cases, adjustment to equilibrium does describe how an equilibrium quantity changes, and in a more complex model that adjustment plays out as a movement over time. For example, a preference or technology shock might give a sudden increase in capital; add adjustment costs and capital increases slowly over time. A fiscal shock or money supply shock gives a sudden increase in the price level; add sticky prices and you get a slow increase in the price level over time. But we already have sticky prices. This is supposed to be the model, the dynamic model, not a simplified model. Here, inflation lower than it otherwise would be is not the same thing as inflation that goes down slowly over time. It's just a misreading of equations. Another possibility is that verbal intuition refers to the future, \[ E_t \pi_{t+1} = E_t \pi_{t+2} + \kappa E_t x_{t+1} .\]Now, perhaps, raising interest rates today lowers future factor x, which then lowers future inflation \(E_t\pi_{t+1}\) relative to today's inflation \(\pi_t\). That's still a stretch however. First, the standard new-keynesian model does not have such a delay. \[x_t = E_t x_{t+1} - \sigma(i_t - E_t \pi_{t+1})\]says that higher interest rates also immediately lower output, and lower output relative to future output. Higher interest rates also raise output growth. This one is more amenable to adding frictions -- habits, capital accumulation, and so forth -- but the benchmark model not only does not have long and variable lags, it doesn't have any lags at all.  Second, maybe we lower inflation \(\pi_{t+1}\) relative to its value \(\pi_t\), in equilibrium, but we still have inflation growing from \(t+1\) to \( t+2\). We do not have inflation gently declining over time, which the intuition wants. We are left -- and this is some of the point of my paper -- with a quandary. Where is a model in which higher interest rates lead to inflation that goes down over time? (And, reiterating the point of the paper, without implicitly assuming that fiscal policy comes to the rescue.) 3. Fisherian intuitionA famous economist, who thinks largely in the ISLM tradition, once asked me to explain in simple terms just how higher interest rates might raise inflation. Strip away all price stickiness to make it simple, still, the Fed raises interest rates and... now what? Sure point to the equation \( i_t = r + E_t\pi_{t+1} \) but what's the story? How would you explain this to an undergraduate or MBA class?  I fumbled a bit, and it took me a good week or so to come up with the answer. From p. 15 of the paper, First,  consider the full consumer first-order condition \[x_t = E_t x_{t+1} - \sigma(i_t -E_t \pi_{t+1})\] with no pricing frictions.  Raise the nominal interest rate \(i_t\).  Before prices change, a higher nominal interest rate is a higher real rate, and induces people  to demand less today \(x_t\) and more next period \(x_{t+1}\).  That change in demand pushes down the price level today \(p_t\) and hence current inflation \(\pi_t = p_t - p_{t-1}\), and it pushes up  the expected price level next period \(p_{t+1}\) and thus expected future inflation \(\pi_{t+1}=p_{t+1}-p_t\). So, standard intuition is correct, and refers to a force that can lower current inflation. Fisherian intuition is correct too, and refers to a natural force that can raise expected future inflation. But which is it, lower \(p_t\) or higher \(p_{t+1}\)? This consumer first-order condition, capturing an  intertemporal substitution effect, cannot tell us. Unexpected inflation and the overall price level is determined by a wealth effect. If we pair the higher interest rate with no change in surpluses, and thus no wealth effect, then the initial price level \(p_t\) does not change [there is no devaluation of outstanding debt] and the entire effect of higher interest rates is a rise in \(p_{t+1}\).  A concurrent rise in expected surpluses leads to a lower price level \(p_t\) and less current inflation \(\pi_t\). Thus in this context standard intuition also implicitly assumes that fiscal policy acts in concert with monetary policy. In both these stories, notice how much intuition depends on describing how equilibrium forms. It's not rigorous. Walrasian equilibrium is just that, and does not come with a price adjustment process. It's a fixed point, the prices that clear markets, period. But believing and understanding how a model works needs some sort of equilibrium formation story. 4. Adaptive vs. rational expectations The distinction between rational, or at least forward-looking and adaptive or backward-looking expectations is central to how the economy behaves. That's a central point of the paper.  It would seem easy to test, but I realize it's not. Writing in May 2022, I thought about adaptive (backward-looking) and rational (forward-looking), and among other points that under adaptive expectations we need nominal interest rates above current inflation  -- i.e. much higher -- to imply real interest rates, while that isn't necessarily true with forward-looking expectations. You might be tempted to test for rational expectations, or look at surveys to pronounce them "rational" vs. "behavioral," a constant temptation. I realize now it's not so easy (p. 44): Expectations may seem adaptive.  Expectations must always be, in equilibrium, functions of variables that people observe, and likely weighted to past inflation. The point of "rational expectations'' is that those forecasting rules are likely to change as soon as a policy maker changes policy rules, as Lucas  famously pointed out in his "Critique."  Adaptive expectations may even be model-consistent [expectations of the model equal expectations in the model] until you change the model.That observation is important in the current policy debate. The proposition that interest rates must be higher than current inflation in order to lower inflation assumes that expected inflation equals current inflation -- the simple one-period lagged adaptive expectations that I have specified here. Through 2021-2022, market and survey expectations were much lower than current (year on year) inflation. Perhaps that means that markets and surveys have rational expectations: Output is temporarily higher than the somewhat reduced post-pandemic potential, so inflation is higher than expected future inflation (\(\pi_t = E_t \pi_{t+1} + \kappa x_t\)). But that observation could also mean that inflation expectations are a long slow-moving average of lagged inflation, just as Friedman speculated in 1968 (\(\pi^e_t = \sum_{j=1}^\infty \alpha_j \pi_{t-j}\)). In either case, expected inflation is much lower than current inflation, and interest rates only need to be higher than that low expectation to reduce inflation. Tests are hard, and you can't just look at in-sample expectations to proclaim them rational or not. Rational expectations change when policy deviates from a rule, or when the policy rule changes. That's their key feature. We should talk perhaps about rational vs. exogenous expectations. 5. A few final Phillips curve potshotsIt is still a bit weird that so much commentary is so focused on the labor market to judge pressure on inflation. This inflation did not come from the labor market! Some of this labor market focus makes sense in the new-Keynesian interpretation of the Phillips curve: Firms set prices based on expected future prices of their competitors and marginal costs, which are largely labor costs. That echoes the 1960s "cost push" view of inflation (as opposed to its nemesis "demand pull" inflation). But it begs the question, well, why are labor costs going up? The link from interest rates to wages is about as direct as the link from interest rates to pries. This inflation did not come from labor costs, maybe we should fix the actual problem? Put another way, the Phillips curve is not a model. It is part of a model, and lots of equations have inflation in them. Maybe our focus should be elsewhere. Back to Chris Waller, whose speech seems to me to capture well sophisticated thinking at the Fed.  Waller points out how unreliable the Phillips curve is What do economic data tell us about this relationship? We all know that if you simply plot inflation against the unemployment rate over the past 50 years, you get a blob. There does not appear to be any statistically significant correlation between the two series.In more recent years, since unemployment went up and down but inflation didn't go far, the Phillips curve seemed "flat," the Phillips curve was very flat for the 20-plus years before the pandemic, You can see this in the decline of unemployment through 2020, as marked, with no change in inflation. Then, unemployment surged in 2021, again with no deflation.  2009 was the last time there was any slope at all to the Phillips curve. But is it "flat" -- a stable, exploitable, flat relationship -- or is it just a stretched out "blob", two series with no stable relationship, one of which just got stable? In any case, as unemployment went back down to 3.5 percent in 2022, inflation surged. You can forgive the Fed a bit: We had 3.5% unemployment with no inflation in 2020, why should we worry about 3.5% unemployment in 2022? I think the answer is, because inflation is driven by a whole lot more than unemployment -- stop focusing on labor markets! A flat curve, if it is a curve, is depressing news:  Based on the flatness of the Phillips curve in recent decades, some commentators argued that unemployment would have to rise dramatically to bring inflation back down to 2 percent. At best, we retrace the curve back to 2021 unemployment. But (I'll keep harping on this), note the focus on the error-free Phillips curve as if it is the entire economic model. Waller views the new Phillips curve as a "curve," that has become steeper, and cites confirming evidence that prices are changing more often and thus becoming more flexible.   ... considering the data for 2021... the Phillips curve suddenly looked relatively steep.. since January 2022, the Phillips curve is essentially vertical: The unemployment rate has hovered around 3.6 percent, and inflation has varied from 7 percent (in June) to 5.3 percent (in December).Waller concludes A steep Phillips curve means inflation can be brought down quickly with relatively little pain in terms of higher unemployment. Recent data are consistent with this story.Isn't that nice -- from horizontal to vertical all on its own, and in the latest data points inflation going straight down. Still, perhaps the right answer is that this is still a cloud of coincidence and not the central, causal, structural relationship with which to think about how interest rates affect inflation. If only I had a better model of inflation dynamics...

This post takes up from two previous posts (part 1;  part 2), asking just what do we (we economists) really know about how interest rates affect inflation. Today, what does contemporary economic theory say? As you may recall, the standard story says that the Fed raises interest rates; inflation (and expected inflation) don't immediately jump up, so real interest rates rise; with some lag, higher real interest rates push down employment and output (IS); with some more lag, the softer economy leads to lower prices and wages (Phillips curve). So higher interest rates lower future inflation, albeit with "long and variable lags." Higher interest rates -> (lag) lower output, employment -> (lag) lower inflation. In part 1, we saw that it's not easy to see that story in the data. In part 2, we saw that half a century of formal empirical work also leaves that conclusion on very shaky ground. As they say at the University of Chicago, "Well, so much for the real world, how does it work in theory?" That is an important question. We never really believe things we don't have a theory for, and for good reason. So, today, let's look at what modern theory has to say about this question. And they are not unrelated questions. Theory has been trying to replicate this story for decades. The answer: Modern (anything post 1972) theory really does not support this idea. The standard new-Keynesian model does not produce anything like the standard story.  Models that modify that simple model to achieve something like result of the standard story do so with a long list of complex ingredients. The new ingredients are not just sufficient, they are (apparently) necessary to produce the desired dynamic pattern. Even these models do not  implement the verbal logic above. If the pattern that high interest rates lower inflation over a few years is true, it is by a completely different mechanism than the story tells. I conclude that we don't have a simple economic model that produces the standard belief. ("Simple" and "economic" are important qualifiers.) The simple new-Keynesian model The central problem comes from the Phillips curve. The modern Phillips curve asserts that price-setters are forward-looking. If they know inflation will be high next year, they raise prices now. So Inflation today = expected inflation next year + (coefficient) x output gap.  \[\pi_t = E_t\pi_{t+1} + \kappa x_t\](If you know enough to complain about \(\beta\approx0.99\) in front of \(E_t\pi_{t+1}\) you know enough that it doesn't matter for the issues here.)Now, if the Fed raises interest rates, and if (if) that lowers output or raises unemployment, inflation today goes down. The trouble is, that's not what we're looking for. Inflation goes down today, (\(\pi_t\))relative to expected inflation next year (\(E_t\pi_{t+1}\)). So a higher interest rate and lower output correlate with inflation that is rising over time.    Here is a concrete example: The plot is the response of the standard three equation new-Keynesian model to an \(\varepsilon_1\) shock at time 1:\[\begin{align} x_t &= E_t x_{t+1} - \sigma(i_t - E_t\pi_{t+1}) \\ \pi_t & = \beta E_t \pi_{t+1} + \kappa x_t \\ i_t &= \phi \pi_t + u_t \\ u_t &= \eta u_{t-1} + \varepsilon_t. \end{align}\] Here \(x\) is output, \(i\) is the interest rate, \(\pi\) is inflation, \(\eta=0.6\), \(\sigma=1\), \(\kappa=0.25\), \(\beta=0.95\), \(\phi=1.2\). In this plot, higher interest rates are said to lower inflation. But they lower inflation immediately, on the day of the interest rate shock. Then, as explained above, inflation rises over time. In the standard view, and the empirical estimates from the last post, a higher interest rate has no immediate effect, and then future inflation is lower. See plots in the last post, or this one from Romer and Romer's 2023 summary:Inflation jumping down and then rising in the future is quite different from inflation that does nothing immediately, might even rise for a few months, and then starts gently going down. You might even wonder about the downward jump in inflation. The Phillips curve makes it clear why current inflation is lower than expected future inflation, but why doesn't current inflation stay the same, or even rise, and expected future inflation rise more? That's the "equilibrium selection" issue. All those paths are possible, and you need extra rules to pick a particular one. Fiscal theory points out that the downward jump needs a fiscal tightening, so represents a joint monetary-fiscal policy. But we don't argue about that today. Take the standard new Keynesian model exactly as is, with passive fiscal policy and standard equilibrium selection rules. It predicts that inflation jumps down immediately and then rises over time. It does not predict that inflation slowly declines over time.  This is not a new issue. Larry Ball (1994) first  pointed out that the standard new Keynesian Phillips curve says that output is high when inflation is high relative to expected future inflation, that is when inflation is declining. Standard beliefs  go the other way: output is high when inflation is rising.  The IS curve is a key part of the overall prediction, and output faces a similar problem. I just assumed above that output falls when interest rates rise. In the model it does; output follows a path with the same shape as inflation in my little plot. Output also jumps down and then rises over time. Here too, the (much stronger) empirical evidence says that an interest rate rise does not change output immediately, and output then falls rather than rises over time. The intuition has even clearer economics behind it: Higher real interest rates induce people to consume less today and more tomorrow. Higher real interest rates should go with higher, not lower, future consumption growth. Again, the model only apparently reverses the sign by having output jump down before rising. Key issuesHow can we be here, 40 years later, and the benchmark textbook model so utterly does not replicate standard beliefs about monetary policy? One answer, I believe, is confusing adjustment to equilibrium with equilibrium dynamics. The model generates inflation lower than yesterday (time 0 to time 1) and lower than it otherwise would be (time 1 without shock vs time 1 with shock). Now, all economic models are a bit stylized. It's easy to say that when we add various frictions, "lower than yesterday" or "lower than it would have been" is a good parable for "goes down over time."  If in a simple supply and demand graph we say that an increase in demand raises prices instantly, we naturally understand that as a parable for a drawn out period of price increases once we add appropriate frictions. But dynamic macroeconomics doesn't work that way. We have already added what was supposed to be the central friction, sticky prices. Dynamic economics is supposed to describe the time-path of variables already, with no extra parables. If adjustment to equilibrium takes time, then model that.  The IS and Phillips curve are forward looking, like stock prices. It would make little sense to say "news comes out that the company will never make money, so the stock price should decline gradually over a few years." It should jump down now. Inflation and output behave that way in the standard model. A second confusion, I think, is between sticky prices and sticky inflation. The new-Keynesian model posits, and a huge empirical literature examines, sticky prices. But that is not the same thing as sticky inflation. Prices can be arbitrarily sticky and inflation, the first derivative of prices, can still jump. In the Calvo model, imagine that only a tiny fraction of firms can change prices at each instant. But when they do, they will change prices a lot, and the overall price level will start increasing right away. In the continuous-time version of the model, prices are continuous (sticky), but inflation jumps at the moment of the shock.  The standard story wants sticky inflation. Many authors explain the new-Keynesian model with sentences like "the Fed raises interest rates. Prices are sticky, so inflation can't go up right away and real interest rates are higher." This is wrong. Inflation can rise right away. In the standard new-Keynesian model it does so with \(\eta=1\), for any amount of price stickiness. Inflation rises immediately with a persistent monetary policy shock. Just get it out of your heads. The standard model does not produce the standard story. The obvious response is, let's add ingredients to the standard model and see if we can modify the response function to look something like the common beliefs and VAR estimates. Let's go. Adaptive expectations We can reproduce standard beliefs about monetary policy with thoroughly adaptive expectations, in the 1970s ISLM form. I think this is a large part of what  most policy makers and commenters have in mind.  Modify the above model to leave out the dynamic part of the intertemporal substitution equation, to just say in rather ad hoc way that higher real interest rates lower output, and specify that the expected inflation that drives the real rate and that drives pricing decisions is mechanically equal to previous inflation, \(E_t \pi_{t+1} = \pi_{t-1}\). We get \[ \begin{align} x_t &= -\sigma (i_t - \pi_{t-1}) \\ \pi_t & = \pi_{t-1} + \kappa x_t .\end{align}\] We can solve this sytsem analytically to \[\pi_t = (1+\sigma\kappa)\pi_{t-1} - \sigma\kappa i_t.\]Here's what happens if the Fed permanently raises the interest rate. Higher interest rates send future inflation down. (\(\kappa=0.25,\ \sigma=1.\)) Inflation eventually spirals away, but central banks don't leave interest rates alone forever. If we add a Taylor rule response \(i_t = \phi \pi_t + u_t\),  so the central bank reacts to the emerging spiral, we get this response to a permanent monetary policy disturbance \(u_t\): The higher interest rate sets off a deflation spiral. But the Fed quickly follows inflation down to stabilize the situation. This is, I think, the conventional story of the 1980s. In terms of ingredients, an apparently minor change of index from \(E_t \pi_{t+1}\) to \(\pi_{t-1}\) is in fact a big change. It means directly that higher output comes with increasing inflation, not decreasing inflation, solving Ball's puzzle.  The change basically changes the sign of output in the Phillips curve. Again, it's not really all in the Phillips curve. This model with rational expectations in the IS equation and adaptive in the Phillips curve produces junk. To get the result you need adaptive expectations everywhere. The adaptive expectations model gets the desired result by changing the basic sign and stability properties of the model. Under rational expectations the model is stable; inflation goes away all on its own under an interest rate peg. With adaptive expectations, the model is unstable. Inflation or deflation spiral away under an interest rate peg or at the zero bound. The Fed's job is like balancing a broom upside down. If you move the bottom (interest rates) one way, the broom zooms off the other way. With rational expectations, the model is stable, like a pendulum.  This is not a small wrinkle designed to modify dynamics. This is major surgery. It is also a robust property: small changes in parameters do not change the dominant eigenvalue of a model from over one to less than one.  A more refined way to capture how Fed officials and pundits think and talk might be called "temporarily fixed expectations." Policy people do talk about the modern Phillips curve; they say inflation depends on inflation expectations and employment. Expectations are not mechanically adaptive. Expectations are a third force, sometimes "anchored," and amenable to manipulation by speeches and dot plots. Crucially, in this analysis, expected inflation does not move when the Fed changes interest rates.  Expectations are then  very slowly adaptive, if inflation is persistent, or if there is a more general loss of faith in "anchoring." In the above new-Keynesian model graph, at the minute the Fed raises the interest rate, expected inflation jumps up to follow the graph's plot of the model's forecast of inflation. As a simple way to capture these beliefs, suppose expectations are fixed or "anchored" at \(\pi^e\). Then my simple model is \[\begin{align}x_t & = -\sigma(i_t - \pi^e) \\ \pi_t & = \pi^e + \kappa x_t\end{align}\]so \[\pi_t = \pi^e - \sigma \kappa (i_t - \pi^e).\] Inflation is expected inflation, and lowered by higher interest rates (last - sign). But those rates need only be higher than the fixed expectations; they do not need to be higher than past rates as they do in the adaptive expectations model. That's why the Fed thinks 3% interest rates with 5% inflation is still "contractionary"--expected inflation remains at 2%, not the 5% of recent adaptive experience.  Also by fixing expectations, I remove the instability of the adaptive expectations model... so long as those expectations stay anchored. The Fed recognizes that eventually higher inflation moves the expectations, and with a belief that is adaptive, they fear that an inflation spiral can still break out.Even this view does not give us any lags, however. The Fed and commenters clearly believe that higher real interest rates today lower output next year, not immediately; and they believe that lower output and employment today drive inflation down in the future, not immediately. They believe something like \[\begin{align}x_{t+1} &= - \sigma(i_t - \pi^e) \\ \pi_{t+1} &= \pi^e + \kappa x_t.\end{align}\] But now we're at the kind of non-economic ad-hockery that the whole 1970s revolution abandoned. And for a reason: Ad hoc models are unstable, regimes are always changing. Moreover, let me remind you of our quest: Is there a simple economic model of monetary policy that generates something like the standard view? At this level of ad-hockery you might as well just write down the coefficients of Romer and Romer's response function and call that the model of how interest rates affect inflation.  Academic economics gave up on mechanical expectations and ad-hoc models in the 1970s. You can't publish a paper with this sort of model. So when I mean a "modern" model, I mean  rational expectations, or at least the consistency condition that the expectations in the model are not fundamentally different from forecasts of the model. (Models with explicit learning or other expectation-formation frictions count too.) It's easy to puff about people aren't rational, and looking out the window lots of people do dumb things. But if we take that view, then the whole project of monetary policy on the proposition that people are fundamentally unable to learn patterns in the economy, that a benevolent Federal Reserve can trick the poor little souls into a better outcome.  And somehow the Fed is the lone super-rational actor who can avoid all those pesky behavioral biases. We are looking for the minimum necessary ingredients to describe the basic signs and function of monetary policy. A bit of irrational or complex expectation formation as icing on the cake, a possible sufficient ingredient to produce quantitatively realistic dynamics, isn't awful. But it would be sad if irrational expectations or other behavior is a necessary ingredient to get the most basic sign and story of monetary policy right. If persistent irrationality is a central necessary ingredient for the basic sign and operation of monetary policy -- if higher interest rates will raise inflation the minute people smarten up; if there is no simple supply and demand, MV=PY sensible economics underlying the basic operation of monetary policy; if it's all a conjuring trick  -- that should really weaken our faith in the whole monetary policy project.  Facts help, and we don't have to get religious about it. During the long zero bound, the same commentators and central bankers kept warning about a deflation spiral, clearly predicted by this model. It never happened. Interest rates below inflation from 2021 to 2023 should have led to an upward inflation spiral. It never happened -- inflation eased all on its own with interest rates below inflation.Getting the desired response to interest rates by making the model unstable isn't tenable whether or not you like the ingredient.  Inflation also surged in the 1970s faster than adaptive expectations came close to predicting, and fell faster in the 1980s.  The ends of many inflations come with credible changes in regime. There is a lot of work now desperately trying to fix new-Keynesian models by making them more old-Keynesian, putting lagged inflation in the Phillips curve, current income in the IS equation, and so forth. Complex learning and expectation formation stories replace the simplistic adaptive expectations here. As far as I can tell, to the extent they work they largely do so in the same way, by reversing the basic stability of the model. Modifying the new-Keynesian modelThe alternative is to add ingredients to the basic new-Keynesian model, maintaining its insistence on real "micro-founded" economics and forward-looking behavior, and describing explicit dynamics as the evolution of equilibrium quantities.  Christiano Eichenbaum and Evans (2005) is one of the most famous examples. Recall these same authors  created the first most influential VAR that gave the "right" answer to the effects of monetary policy shocks. This paper modifies the standard new-Keynesian model with a specific eye to matching impulse response functions. The want to match all impulse-responses, with a special focus on output. When I started asking my young macro colleagues for a standard model which produces the desired response shape, they still cite CEE first, though it's 20 years later. That's quite an accomplishment. I'll look at it in detail, as the general picture is the same as many other models that achieve the desired result. Here's their bottom line response to a monetary policy shock: (Figure from the 2018 Christiano Eichenbaum and Trabandt Journal of Economic Perspectives summary paper.)  The solid line is the VAR point estimate and gray shading is the 95% confidence band. The solid blue line is the main model. The dashed line is the model with only price stickiness, to emphasize the importance of wage stickiness. The shock happens at time 0. Notice the funds rate line that jumps down at that date. That the other lines do not move at time 0 is a result. I graphed the response to a time 1 shock above. That's the answer, now what's the question? What ingredients did they add above the textbook model to reverse the basic sign and jump problem and to produce  these pretty pictures? Here is a partial list: Habit formation. The utility function is  \(log(c_t - bc_{t-1})\). A capital stock with adjustment costs in investment.  Adjustment costs are proportional to investment growth, \([1-S(i_t/i_{t-1})]i_t\), rather than the usual formulation in which adjustment costs are proportional to the investment to capital ratio \(S(i_t/k_t)i_t\). Variable capital utilization. Capital services \(k_t\) are related to the capital stock \(\bar{k}t\) by \(k_t = u_t \bar{k}_t\). The utilization rate \(u_t\) is set by households facing an upward sloping cost \(a(u_t)\bar{k}_t\).Calvo pricing with indexation: Firms randomly get to reset prices, but firms that aren't allowed to reset prices do automatically raise prices at the rate of inflation.Prices are also fixed for a quarter. Technically, firms must post prices before they see the period's shocks.Sticky wages, also with indexation. Households are monopoly suppliers of labor, and set wages Calvo-style like firms. (Later papers put all households into a union which does the wage setting.) Wages are also indexed; Households that don't get to reoptimize their wage still raise wages following inflation. Firms must borrow working capital to finance their wage bill a quarter in advance, and thus pay a interest on the wage bill. Money in the utility function, and money supply control. Monetary policy is a change in the money growth rate, not a pure interest rate target. Whew! But which of these ingredients are necessary, and which are just sufficient? Knowing the authors, I  strongly suspect that they are all necessary to get the suite of results. They don't add ingredients for show. But they want to match all of the impulse response functions, not just the inflation response. Perhaps a simpler set of ingredients could generate the inflation response while missing some of the others. Let's understand what each of these ingredients is doing, which will help us to see (if) they are necessary and essential to getting the desired result. I see a common theme in habit formation, adjustment costs that scale by investment growth, and indexation. These ingredients each add a derivative; they take a standard relationship between levels of economic variables and change it to one in growth rates. Each of consumption, investment, and inflation is a "jump variable" in standard economics, like stock prices. Consumption (roughly) jumps to the present value of future income. The level of investment is proportional to the stock price in the standard q theory, and jumps when there is new information. Iterating forward the new-Keynesian Phillips curve \(\pi_t = \beta E_t \pi_{t+1} + \kappa x_t\), inflation jumps to the discounted sum of future output gaps, \(\pi_t = E_t \sum_{j=0}^\infty \beta^jx_{t+j}.\)  To produce responses in which output, consumption and investment as well as inflation rise slowly after a shock, we don't want levels of consumption, investment, and inflation to jump this way. Instead we want growth rates to do so. With standard utility, the consumer's linearized first order condition equates expected consumption growth to the interest rate, \( E_t (c_{t+1}/c_t) = \delta + r_t \) Habit, with \(b=1\) gives  \( E_t [(c_{t+1}-c_t)/(c_t-c_{t-1})] = \delta + r_t \). (I left out the strategic terms.) Mixing logs and levels a bit, you can see we put a growth rate in place of a level. (The paper has \(b=0.65\) .) An investment adjustment cost function with \(S(i_t/i_{t-1})\) rather than the standard \(S(i_t/k_t)\) puts a derivative in place of a level. Normally we tell a story that if you want a house painted, doubling the number of painters doesn't get the job done twice as fast because they get in each other's way. But you can double the number of painters overnight if you want to do so. Here the cost is on the increase in number of painters each day.   Indexation results in a Phillips curve with a lagged inflation term, and that gives "sticky inflation."  The Phillips curve of the model (32) and (33) is \[\pi_t = \frac{1}{1+\beta}\pi_{t-1} + \frac{\beta}{1+\beta}E_{t-1}\pi_{t+1} + (\text{constants}) E_{t-1}s_t\]where \(s_t\) are marginal costs (more later). The \(E_{t-1}\) come from the assumption that prices can't react to time \(t\) information. Iterate that forward to (33)\[\pi_t - \pi_{t-1} = (\text{constants}) E_{t-1}\sum_{j=0}^\infty \beta^j s_{t+j}.\] We have successfully put the change in inflation in place of the level of inflation. The Phillips curve is anchored by real marginal costs, and they are not proportional to output in this model as they are in the textbook model above. That's important too. Instead,\[s_t = (\text{constants}) (r^k_t)^\alpha \left(\frac{W_t}{P_t}R_t\right)^{1-\alpha}\] where \(r^k\) is the return to capital \(W/P\) is the real wage and \(R\) is the nominal interest rate. The latter term crops up from the assumption that firms must borrow the wage bill one period in advance. This is an interesting ingredient. There is a lot of talk that higher interest rates raise costs for firms, and they are reducing output as a result. That might get us around some of the IS curve problems. But that's not how it works here. Here's how I think it works. Higher interest rates raise marginal costs, and thus push up current inflation relative to expected future inflation. The equilibrium-selection rules and the rule against instant price changes (coming up next) tie down current inflation, so the higher interest rates have to push down expected future inflation. CEE disagree (p. 28). Writing of an interest rate decline, so all the signs are opposite of my stories,  ... the interest rate appears in firms' marginal cost. Since the interest rate drops after an expansionary monetary policy shock, the model embeds a force that pushes marginal costs down for a period of time. Indeed, in the estimated benchmark model the effect is strong enough to induce a transient fall in inflation.But pushing marginal costs down lowers current inflation relative to future inflation -- they're looking at the same Phillips curve just above. It looks to me like they're confusing current with expected future inflation. Intuition is hard. There are plenty of Fisherian forces in this model that want lower interest rates to lower inflation. More deeply,  we see here a foundational trouble of the Phillips curve. It was originally a statistical relation between wage inflation and unemployment. It became a (weaker) statistical relation between price inflation and unemployment or the output gap. The new-Keynesian theory wants naturally to describe a relation between marginal costs and price changes, and it takes contortions to make output equal to marginal costs. Phillips curves fit the data terribly. So authors estimating Phillips curves (An early favorite by Tim Cogley and Argia Sbordone) go back, and separate marginal cost from output or employment. As CET write later, they "build features into the model which ensure that firms' marginal costs are nearly acyclical." That helps the fit, but it divorces the Phillips curve shifter variable from the business cycle! Standard doctrine says that for the Fed to lower inflation it must soften the economy and risk unemployment. Doves say don't do it, live with inflation to avoid that cost. Well, if the Phillips curve shifter is "acyclical" you have to throw all that out the window.  This shift also points to the central conundrum of the Phillips curve. Here it describes the adjustment of prices to wages or "costs" more generally. It fundamentally describes a relative price, not a price level. OK, but the phenomenon we want to explain is the common component, how all prices and wage tie together or equivalently the decline in the value of the currency, stripped of relative price movements. The central puzzle of macroeconomics is why the common component, a rise or fall of all prices and wages together, has anything to do with output, and for us how it is controlled by the Fed. Christiano Eichenbaum and Evans write (p.3) that "it is crucial to allow for variable capital utilization." I'll try explain  why in my own words. Without capital adjustment costs, any change in the real return leads to a big investment jump. \(r=f'(k)\) must jump and that takes a lot of extra \(k\). We add adjustment costs to tamp down the investment response. But now when there is any shock, capital can't adjust enough and there is a big rate of return response. So we need something that acts like a big jump in the capital stock to tamp down \(r=f'(k)\) variability, but not a big investment jump. Variable capital utilization acts like the big investment jump without us seeing a big investment jump. And all this is going to be important for inflation too. Remember the Phillips curve; if output jumps then inflation jumps too. Sticky wages are crucial, and indeed CEE report that they can dispense with sticky prices. One reason is that otherwise profits are countercyclical. In a boom, prices go up faster than wages so profits go up. With sticky prices and flexible wages you get the opposite sign. It's interesting that the "textbook" model has not moved this way. Again, we don't often enough write textbooks. Fixing prices and wages during the period of the shock by assuming price setters can't see the shock for a quarter has a direct effect: It stops any price or wage jumps during the quarter of the shock, as in my first graph. That's almost cheating. Note the VAR also has absolutely zero instantaneous inflation response. This too is by assumption. They "orthogonalize" the variables so that all the contemporaneous correlation between monetary policy shocks and inflation or output is considered part of the Fed's "rule" and none of it reflects within-quarter reaction of prices or quantities to the Fed's actions. Step back and admire. Given the project "find elaborations of the standard new-Keynesian model to match VAR impulse response functions" could you have come up with any of this? But back to our task. That's a lot of apparently necessary ingredients. And reading here or CEE's verbal intuition, the logic of this model is nothing like the standard simple intuition, which includes none of the necessary ingredients. Do we really need all of this to produce the basic pattern of monetary policy? As far as we know, we do. And hence, that pattern may not be as robust as it seems. For all of these ingredients are pretty, ... imaginative. Really, we are a long way from the Lucas/Prescott vision that macroeconomic models should be based on well tried and measured microeconomic ingredients that are believably invariant to changes in the policy regime. CEE argue hard for the plausibility of these microeconomic specifications (see especially the later CET Journal of Economic Perspectives article), but they have to try so hard precisely because the standard literature doesn't have any of these ingredients. The "level" rather than "growth rate" foundations of consumption, investment, and pricing decisions pervade microeconomics. Microeconomists worry about labor monopsony, not labor monopoly; firms set wages, households don't. (Christiano Eichenbam and Trabandt (2016) get wage stickiness from a more realistic search and matching model. Curiously, the one big labor union fiction is still the most common, though few private sector workers are unionized.) Firms don't borrow the wage bill a quarter ahead of time.  Very few prices and wages are indexed in the US. Like habits, perhaps these ingredients are simple stand ins for something else, but at some point we need to know what that something else is. That is especially true if one wants to do optimal policy or welfare analysis. Just how much economics must we reinvent to match this one response function? How far are we really from the ad-hoc ISLM equations that Sims (1980) destroyed? Sadly, subsequent literature doesn't help much (more below). Subsequent literature has mostly added ingredients, including heterogeneous agents (big these days), borrowing constraints, additional financial frictions (especially after 2008), zero bound constraints, QE, learning and complex expectations dynamics. (See CET 2018 JEP  for a good verbal survey.) The rewards in our profession go to those who add a new ingredient. It's very hard to publish papers that strip a model down to its basics. Editors don't count that as "new research," but just "exposition" below the prestige of their journals. Though boiling a model down to essentials is maybe more important in the end than adding more bells and whistles. This is about where we are. Despite the pretty response functions, I still score that we don't have a reliable, simple, economic model that produces the standard view of monetary policy. Mankiw and Reis, sticky expectations Mankiw and Reis (2002) expressed the challenge clearly over 20 years ago. In reference to the "standard" New-Keynesian Phillips curve \(\pi_t = \beta E_t \pi_{t+1} + \kappa x_t\) they write a beautiful and succinct paragraph: Ball [1994a] shows that the model yields the surprising result that announced, credible disinflations cause booms rather than recessions. Fuhrer and Moore [1995] argue that it cannot explain why inflation is so persistent. Mankiw [2001] notes that it has trouble explaining why shocks to monetary policy have a delayed and gradual effect on inflation. These problems appear to arise from the same source: although the price level is sticky in this model, the inflation rate can change quickly. By contrast, empirical analyses of the inflation process (e.g., Gordon [1997]) typically give a large role to "inflation inertia."At the cost of repetition, I emphasize the last sentence because it is so overlooked. Sticky prices are not sticky inflation. Ball already said this in 1994: Taylor (1979, 198) and Blanchard (1983, 1986) show that staggering produces inertia in the price level: prices just slowly to a fall in th money supply. ...Disinflation, however, is a change in the growth rate of money not a one-time shock to the level. In informal discussions, analysts often assume that the inertia result carries over from levels to growth rates -- that inflation adjusts slowly to a fall in money growth. As I see it, Mankiw and Reis generalize the Lucas (1972) Phillips curve. For Lucas, roughly, output is related to unexpected inflation\[\pi_t = E_{t-1}\pi_t + \kappa x_t.\] Firms don't see everyone else's prices in the period. Thus, when a firm sees an unexpected rise in prices, it doesn't know if it is a higher relative price or a higher general price level; the firm expands output based on how much it thinks the event might be a relative price increase.  I love this model for many reasons, but one, which seems to have fallen by the wayside, is that it explicitly founds the Phillips curve in firms' confusion about relative  prices vs. the price level, and thus faces up to the problem why should a rise in the price level have any real effects. Mankiw and Reis basically suppose that firms find out the general price level with lags, so output depends on inflation relative to a distributed lag of its expectations. It's clearest for the price level (p. 1300)\[p_t = \lambda\sum_{j=0}^\infty (1-\lambda)^j E_{t-j}(p_t + \alpha x_t).\] The inflation expression is \[\pi_t = \frac{\alpha \lambda}{1-\lambda}x_t + \lambda \sum_{j=0}^\infty (1-\lambda)^j E_{t-1-j}(\pi_t + \alpha \Delta x_t).\](Some of the complication is that you want it to be \(\pi_t = \sum_{j=0}^\infty E_{t-1-j}\pi_t + \kappa x_t\), but output doesn't enter that way.) This seems totally natural and sensible to me. What is a "period" anyway? It makes sense that firms learn heterogeneously whether a price increase is relative or price level. And it obviously solves the central persistence problem with the Lucas (1972) model, that it only produces a one-period output movement. Well, what's a period anyway? (Mankiw and Reis don't sell it this way, and actually don't cite Lucas at all. Curious.)  It's not immediately obvious that this curve solves the Ball puzzle and the declining inflation puzzle, and indeed one must put it in a full model to do so. Mankiw and Reis (2002) mix it with \(m_t + v = p_t + x_t\) and make some stylized analysis, but don't show how to put the idea in models such as I started with or make a plot.  Their less well known follow on paper  Sticky Information in General Equilibrium (2007) is much better for this purpose because they do show you how to put the idea in an explicit new-Keynesian model, like the one I started with.  They also add a Taylor rule, and an interest rate rather than money supply instrument, along with wage stickiness and a few other ingredients,. They show how to solve the model overcoming the problem that there are many lagged expectations as state variables. But here is the response to the monetary policy shock: Response to a Monetary Policy Shock, Mankiw and Reis (2007). Sadly they don't report how interest rates respond to the shock. I presume interest rates went down temporarily. Look: the inflation and output gap plots are about the same. Except for the slight delay going up, these are exactly the responses of the standard NK model. When output is high, inflation is high and declining.  The whole point was to produce a model in which high output level would correspond to rising inflation. Relative to the first graph, the main improvement is just a slight hump shape in both inflation and output responses. Describing the same model in "Pervasive Stickiness" (2006), Mankiw and Reis describe the desideratum well: The Acceleration Phenomenon....inflation tends to rise when the economy is booming and falls when economic activity is depressed. This is the central insight of the empirical literature on the Phillips curve. One simple way to illustrate this fact is to correlate the change in inflation, \(\pi_{t+2}-\pi_{t-2}\) with [the level of] output, \(y_t\), detrended with the HP filter. In U.S. quarterly data from 1954-Q3 to 2005-Q3, the correlation is 0.47. That is, the change in inflation is procyclical.Now look again at the graph. As far as I  can see, it's not there. Is this version of sticky inflation a bust, for this purpose? I still think it's a neat idea worth more exploration. But I thought so 20 years ago too.  Mankiw and Reis have a lot of citations but nobody followed them. Why not?  I suspect it's part of a general pattern that lots of great micro sticky price papers are not used because they don't produce an easy aggregate Phillips curve. If you want cites, make sure people can plug it in to Dynare. Mankiw and Reis' curve is pretty simple, but you still have to keep all past expectations around as a state variable. There may be alternative ways of doing that with modern computational technology, putting it in a Markov environment or cutting off the lags, everyone learns the price level after 5 years. Hank models have even bigger state spaces! Some more modelsWhat about within the Fed?  Chung,  Kiley, and  Laforte 2010, "Documentation of the Estimated, Dynamic, Optimization-based (EDO) Model of the U.S. Economy: 2010 Version" is one such model.  (Thanks to Ben Moll, in a lecture slide titled "Effects of interest rate hike in U.S. Fed's own New Keynesian model") They describe it as This paper provides documentation for a large-scale estimated DSGE model of the U.S. economy – the Federal Reserve Board's Estimated, Dynamic, Optimization- based (FRB/EDO) model project. The model can be used to address a wide range of practical policy questions on a routine basis.Here are the central plots for our purpose: The response of interest rates and inflation to a monetary policy shock.  No long and variable lags here. Just as in the simple model, inflation jumps down on the day of the shock and then reverts. As with Mankiw and Reis, there is a tiny hump shape, but that's it. This is nothing like the Romer and Romer plot. Smets and Wouters (2007) "Shocks and Frictions in US Business Cycles: A Bayesian DSGE Approach" is about as famous as Christiano Eichenbaum and Evans as a standard new-Keynesian model that supposedly matches data well. It "contains many shocks and frictions. It features sticky nominal price and wage settings that allow for backward inflation indexation, habit formation in consumption, and investment adjustment costs that create hump-shaped responses... and variable capital utilization and fixed costs in production"Here is their central graph of the response to a monetary policy shockAgain, there is a little hump-shape, but the overall picture is just like the one we started with. Inflation mostly jumps down immediately and then recovers; the interest rate shock leads to future inflation that is higher, not lower than current inflation. There are no lags from higher interest rates to future inflation declines. The major difference, I think, is that Smets and Wouters do not impose the restriction that inflation cannot jump immediately on either their theory or empirical work, and Christiano, Eichenbaum and Evans impose that restriction in both places. This is important. In a new-Keynesian model some combination of state variables must jump on the day of the shock, as it is only saddle-path stable. If inflation can't move right away, that means something else does. Therefore, I think, CEE also preclude inflation jumping the next period. Comparing otherwise similar ingredients, it looks like this is the key ingredient for producing Romer-Romer like responses consistent with the belief in sticky inflation. But perhaps the original model and Smets-Wouters are right! I do not know what happens if you remove the CEE orthogonalization restriction and allow inflation to jump on the day of the shock in the date. That would rescue the new-Keynesian model, but it would destroy the belief in sticky inflation and long and variable lags. Closing thoughtsI'll reiterate the main point. As far as I can tell, there is no simple economic model that produces the standard belief. Now, maybe belief is right and models just have to catch up. It is interesting that there is so little effort going on to do this. As above, the vast outpouring of new-Keynesian modeling has been to add even more ingredients. In part, again, that's the natural pressures of journal publication. But I think it's also an honest feeling that after Christiano Eichenbaun and Evans, this is a solved problem and adding other ingredients is all there is to do. So part of the point of this post (and "Expectations and the neutrality of interest rates") is to argue that this is not a solved problem, and that removing ingredients to find the simplest economic model that can produce standard beliefs is a really important task.  Then, does the model incorporate anything at all of the standard intuition, or is it based on some different mechanism al together? These are first order important and unresolved questions!But for my lay readers, here is as far as I know where we are. If you, like the Fed, hold to standard beliefs that higher interest rates lower future output and inflation with long and variable lags, know there is no simple economic theory behind that belief, and certainly the standard story is not how economic models of the last four decades work. Update:I repeat a response to a comment below, because it is so important. I probably wasn't clear enough that the "problem" of high output with inflation falling rather than rising is a problem of models vs. traditional beliefs, rather than of models vs. facts. The point of the sequence of posts, really, is that the traditional beliefs are likely wrong. Inflation does not fall, following interest rate increases, with dependable, long, and perhaps variable lags. That belief is strong, but neither facts, empirical evidence, or theory supports it. ("Variable" is a great way to scrounge data to make it fit priors.) Indeed many successful disinflations like ends of hyperinflations feature a sigh of relief and output surge on the real side. 

Auffällig allgemein: Die Antworten von Bundesbildungsministerin Stark-Watzinger auf eine Große Anfrage sind so, dass die Opposition von "Hinhaltetaktik" spricht. Ein paar interessante Nuancen enthalten sie aber doch.






Foto: StockSnap / pixabay - cco.






DIE VERHANDLUNGEN zwischen Bund und Ländern liefen zuletzt wieder, zudem hatte sich Bundesbildungsministerin Bettina Stark-Watzinger (FDP) im Herbst mehrfach demonstrativ zur Fortsetzung des
Digitalpakts bekannt. Doch über die Einzelheiten, wie ein neues Bund-Länder-Programm aussehen könnte, hält sich das BMBF weiter bedeckt. 



 



Auch eine Große Anfrage der CDU-/CSU-Bundestagsopposition zu den Regierungsplänen "hinsichtlich eines Digitalpaktes 2.0" hat daran nichts ändern können. Die meisten Antworten
der Bundesregierung fallen auffällig allgemein aus. Gleich an zwei Stellen heißt es: Ja, man beabsichtige weiterhin eine Digitalpakt-Fortsetzung, aber: unter Berücksichtigung der "gegebenen
haushalterischen Rahmenbedingungen".



 



Was für sich betrachtet trivial erscheint, da das Parlament und nicht die Regierung der Haushaltsgesetzgeber ist. Doch die zweifache Betonung, verbunden mit der Weigerung, Zahlen zum geplanten
Umfang zu nennen, wird nicht geeignet sein, die Zweifel auf Seiten der Kultusminister zu zerstreuen: "Die Ausgestaltung eines Digitalpakts 2.0 ist Gegenstand laufender Verhandlungen", lautet
die Auskunft auf die Frage, in welcher Höhe ist die Bundesregierung Finanzmittel für einen Digitalpakt 2.0 zur Verfügung zu stellen. Genau wie zuvor schon auf die Frage, welche
Bedingungen die Bundesregierung an die Länder stelle. 



 



"Seit Monaten weicht die Ampel bei Fragen nach dem Digitalpakt 2.0 aus und verweist auf laufende Verhandlungen", kommentiert der bildungspolitische Sprecher der Unionsfraktion, Thomas Jarzombek,
die Antwort der Bundesregierung. Diese "Hinhaltetaktik" sei nicht weiter hinnehmbar. "Länder und Kommunen brauchen Verlässlichkeit, mit welchem bildungspolitischen Engagement des Bundes sie
in den nächsten zwei Jahren noch rechnen können." 



 



Die Bundesregierung sieht keine Förderlücke



 



Ein paar interessante Nuancen enthält die von Stark-Watzinger unterzeichnete Regierungsantwort dann aber doch. Etwa dass beim Digitalpakt 2.0 als Lektion aus der ersten Digitalpakt-Runde
bürokratische Antragsverfahren "keine Hemmschwelle mehr sein" dürften. Dass die Bundesregierung beim Digitalpakt 2.0 verstärkt auf Daten und wissenschaftliche Begleitung setzen wolle: "Es muss
stärker gemessen werden, was in den Klassenzimmern tatsächlich funktioniert und was nicht."



 



Und dass man im BMBF anders als in Opposition oder Kultusministerkonferenz der Meinung ist, dass die zweite Runde sehr wohl nahtlos an die erste anschließt: Die Mittel aus dem Digitalpakt Schule
könnten laut Verwaltungsvereinbarung von 2019 bis Ende 2025 abgerufen werden, heißt es in der Regierungsantwort. Den Ländern sei frühzeitig und zuletzt bei der KMK-Sitzung im Oktober 2023
kommuniziert worden, "dass Mittel eines Digitalpakts 2.0 daher frühestens im Jahr 2025 zur Verfügung stehen werden". Während die CDU-/CSU-Opposition in ihrer Anfrage hervorhebt, dass der
"Digitalpakt Schule (Laufzeit 2019-2024)... zum 16. Mai 2024 nach aktuellem Planungsstand der Regierungskoalition ersatzlos" auslaufe. Ab dann stünden für neue Digitalisierungsvorhaben
eben keine Gelder zur Verfügung.



 



Zudem verweist die Bundesregierung in ihrer Antwort selbst auf die von Ende Juni stammende, jüngste Erhebung zur Verwendung der fünf Bundesmilliarden aus dem Basis-Digitalpakt (in der
Corona-Zeit kamen für Sonderprogramme nochmal 1,5 Milliarden hinzu). Die ergab, dass schon damals 90 Prozent der Mittel gebunden waren. Anderthalb Jahre, bevor das BMBF frühestens die
Fortsetzung starten will. Die Bundesregierung fügte in ihrer Antwort hinzu: "Beim Digitalpakt geht es nun mit erhöhtem Tempo auf die Zielgraden (sic)."



 



Und wie geht es jetzt weiter mit den Verhandlungen? Die Antwort aus dem Bildungsministerium: Bei der KMK-Sitzung im Oktober sei ein  gemeinsames Papier mit Grundsatzpositionen von BMBF und
Ländern über den Digitalpakt 2.0 zur Kenntnis genommen worden. "Auf dieser Grundlage werden die Gespräche des BMBF mit den Ländern fortgeführt", "innerhalb eines gemeinsam abgestimmten Zeitplans
mit regelmäßigen Sitzungen einer eigens eingerichteten 'Verhandlungsgruppe' mit Vertreterinnen und Vertretern der Länder und des BMBF".



 



"Ohne Kompass und Konzept"?



 



Was in den Ausführungen der Bundesregierung noch fehlt: Zuletzt hatte Stark-Watzingers Haus sogar einen ersten Entwurf für eine Bund-Länder-Vereinbarung vorgelegt, wobei auch dort eine Leerstelle
beim geplanten Finanzvolumen stand – dafür aber an anderer Stelle eine konkrete Zahl: 50 Prozent. So hoch stellt sich das BMBF den Länderanteil beim Digitalpakt 2.0 vor. In der ersten Runde
mussten die Länder nur zehn Prozent drauflegen. Das Gerangel geht also weiter. 



 



CDU-Bildungspolitiker Jarzombek sagt, die Ampelregierung agiere "ohne Kompass und Konzept". Dabei sei die inhaltliche wie strukturelle Weiterentwicklung des Digitalpakt Schule so wichtig.
"Während der Fokus bisher auf digitaler Infrastruktur und Hardware-Ausstattung lag, muss es jetzt auch um digitale Lernmedien und innovative Unterrichtskonzepte gehen." Die Ständige
Wissenschaftliche Kommission der KMK gebe da wichtige Hinweise, zum Beispiel zum Potenzial adaptiver Lernsysteme. 



 



Die Bundesregierung teilt derweil in ihrer Antwort mit, strategisches Zielbild der Digitalpakt-Fortsetzung sei "die Steigerung der Leistungsfähigkeit der kommunalen Bildungsinfrastruktur durch
ein flächendeckendes digital unterstütztes Lehren und Lernen und dadurch Leistungssteigerungen der Schülerinnen und Schüler". Bund und Länder hätten durch den Digitalpakt Schule eine positive
Bewegung hin zu Digitalisierung in der schulischen Bildung bewirkt. "Dieser Aufbruch wird gefestigt werden."




Kostenfreien Newsletter abonnieren




In eigener Sache: Blog-Finanzierung














Wie dieser Blog finanziell aufgestellt ist, warum er Ihre Unterstützung braucht – und welche Artikel im November am meisten gelesen
wurden. 



 



Mehr lesen...

Email



Share

Tweet

Share


Comment










Image



Young girls in Niger. Erick Kaglan / World Bank






















Nearly every country around the world is grappling with more than one crisis: the still-simmering pandemic and continued vulnerability to future health emergencies; historic spikes in food insecurity, exacerbated by supply shortages arising from the war in Ukraine; fragility, conflict, and violence; and the steadily rising tide of climate change's assaults on the environment.

Neutralizing even one of these crises can be confounding and perilous. Some countries, unfortunately, face them all at once, fighting on multiple fronts. That usually keeps them from attending to the longer-term task of giving people the knowledge, skills, access to health care, and opportunities they need to live out their full productive potential. Investing in resilient, shock-responsive systems is critical to protect human-capital gains and improve resilience to future shocks.

Niger is an example of a country that faces many complex and interconnected challenges. Shocks and crises are increasingly frequent and overlapping in Niger, disrupting efforts to sustain broad-based growth, build human capital, and reduce poverty. Regional instability has led to the displacement of families and the closure of schools, threatening social stability and increasing insecurity; that, in turn, complicated Niger's efforts to respond to the COVID-19 pandemic and worsened the food insecurity that is now affecting more than 4.4 million of the country's people. Climate shocks have triggered localized flooding, while steady rises in temperatures threaten the more than 80 percent of Niger's citizens who depend on agriculture for their nourishment and livelihoods.

The government of Niger is determined not to lose any ground in its steady climb to protect and invest in all its citizens by pressing ahead with programs and reforms that are having transformational impact on people's lives. A great example of this is the Wadata Talaka safety-net program, a partnership between Niger and the World Bank that focuses on poverty reduction, resilience building, and women's empowerment. The program provides monthly cash transfers to extremely poor households to smooth their consumption expenditures and improve their ability to cope with shocks. It also provides "economic inclusion" support—life and micro-entrepreneurship skills training, coaching, and support to village savings groups—and helps poor children get essential mental stimulation in their early years. Such programs can respond quickly to help poor and vulnerable families prepare for, cope with, and adapt to shocks such as the COVID-19 pandemic: As the virus spread, the program expanded to four hundred thousand households to protect them from the pandemic's adverse economic consequences. The program is well-placed to assist poor households with rising food insecurity and climate shocks.

A successful response will need to include supporting women and innovation. Because women are the primary beneficiaries of Wadata Talaka, the program is an important vehicle for their empowerment. Evaluations of the economic inclusion program show that in the eighteen months since it began, it improved household consumption and food security. The total income of women beneficiaries has increased (by 60 to 100 percent, much of it from non-farm businesses), and there is strong evidence of gains in their mental health and social wellbeing.

To develop such systems reaching the poorest and most vulnerable, countries will need strong social registries and good enrollment, delivery, and payment systems, often leveraging technology. The government of Niger is fully committed to these efforts. For example, responding to climate change, Wadata Talaka was the first program of its kind in West Africa to use satellite data to quickly anticipate drought hotspots and provide emergency funds more quickly than usual (three months ahead of the traditional response) to help people before they entered the lean season. Research is currently underway to measure the impact of that speed.

At a time when countries are forced to contend with the ebb and flow of shocks like climate change, pandemics, conflict, or food price increases, investments in social protection systems are more critical than ever. Niger's programs serve as an example of just how impactful such adaptive systems can be.

This blog was initially published on Atlantic Council site.









Topics

Climate Change
Environment
Fragility Conflict and Violence
Inequality and Shared Prosperity
Jobs & Development
Poverty
Social Protection









Countries

Niger









Regions

Africa
















Authors















twitter.com/intent/follow?screen_name=@MamtaMurthi


Mamta Murthi
Vice President for Human Development
More Blogs By Mamta






































Mahamadou Ouhoumoudou
Prime Minister of Niger
More Blogs By Mahamadou









































0











Join the Conversation