Note · Arthur Mota


The Phillips Curve Evolution: From Flattening to Steepening and the Search for Normalization

Understanding the dramatic transformation of the inflation-unemployment relationship across three distinct eras of macroeconomic policy

The Phillips Curve has undergone one of the most remarkable transformations in modern macroeconomic history, evolving from a seemingly reliable policy guide through a period of apparent irrelevance, to a dramatic resurrection during the COVID-19 pandemic. This evolution encompasses not just empirical observations but fundamental questions about the structural relationships governing inflation dynamics, the role of monetary policy credibility, and the changing nature of price-setting behavior in advanced economies.

Part I: The Theoretical Foundation and Mathematical Framework

The New Keynesian Phillips Curve Specification

The modern understanding of the Phillips Curve relationship begins with the New Keynesian Phillips Curve (NKPC), which provides the theoretical foundation for empirical analysis. The hybrid specification commonly used in empirical work takes the form:

πt = γπt-1 + (1 - γ)Etπt+1 + κ(yt - y*t) + ut (1)

where πt represents inflation at time t, Etπt+1 denotes expected future inflation, (yt - y*t) is the output gap measuring resource utilization, and ut captures supply shocks affecting prices.

The parameter κ, known as the slope of the Phillips Curve, has become the focal point of extensive research and policy debate. As Furlanetto and Lepetit (2024) note, “There are two main reasons for this interest in κ. First, κ is a key determinant of the trade-off between inflation and real activity faced by the monetary authority. For example, a large κ implies that inflation can be reduced at little cost in terms of economic activity. Inversely, a small κ entails that large declines in output are necessary to obtain a significant reduction in inflation.”

Structural Foundations and Marginal Cost

The theoretical foundation connecting inflation to economic activity operates through real marginal costs. The microfounded version relates inflation to average real marginal cost mct:

πt = γπt-1 + (1 - γ)Etπt+1 + ψmct + ut (2)

As Furlanetto and Lepetit (2024) explain, “in equation (1), the output gap should be seen as an imperfect proxy for real marginal cost, and we should keep in mind that κ combines information on the degree of pass-through from costs to prices (ψ) together with the elasticity of real marginal cost with respect to the output gap.”

This insight is crucial because ψ represents a structural parameter capturing price stickiness and strategic complementarities, while the elasticity of marginal cost with respect to the output gap is a reduced-form object that can change depending on economic shocks and structural conditions.

The Phillips Multiplier and Policy Trade-offs

Barnichon and Mesters (2021) propose a more complete characterization of monetary policy trade-offs through the Phillips multiplier, defined as:

Ph = [∂∑hj=0πt+j/∂εmt] / [∂∑hj=0(yt+j - y*t+j)/∂εmt] (3)

where h ≥ 0, providing the cumulative response of inflation to a monetary policy shock divided by the cumulative response of the output gap. This measure captures what central banks fundamentally care about: how much inflation declines when policy changes lower the output gap by a given amount.

Part II: The Great Flattening Era (1990-2019)

Empirical Documentation of the Flattening

The flattening that most estimates date to around 1990 drew full attention after the 2008 financial crisis, when many economists spoke of the “death” of the Phillips Curve. The empirical evidence was striking and widespread. As documented by the San Francisco Fed (2019), “in the years since the Great Recession, inflation has been driven primarily by public expectations of future inflation rather than by economic slack. The vanishing relationship between slack and inflation is known in economics as a ‘flattening’ of the Phillips curve.”

Federal Reserve Chair Jerome Powell’s characterization captured the transformation eloquently. As the St. Louis Fed (2020) reported, Powell described the relationship between resource utilization and inflation as having become “weaker and weaker and weaker to the point where it’s a faint heartbeat that you can hear now.”

The flattening was not merely a U.S. phenomenon. Research by Kuttner and Robinson (2010) documented similar patterns in Australia, while Lie (2024) notes that “the slope of the Phillips curve has been widely shown to be flatter in recent years in many economies.”

Quantitative Evidence of the Transformation

Recent empirical work using sophisticated identification strategies has provided quantitative estimates of this flattening. Barnichon and Mesters (2020) found that “the slope of the Phillips curve has been cut in half between the samples 1960-2007 and 1990-2017.” Similarly, Inoue, Rossi and Wang (2022) found that “the slope of the Phillips curve has been divided by a factor of two to three between 1975 and 2007.”

However, the magnitude of this decline appears less dramatic when proper econometric techniques address endogeneity issues. As Furlanetto and Lepetit (2024) conclude from their comprehensive review, “once appropriate econometric techniques are used to address endogeneity issues in the estimation of the Phillips curve, the estimated decline in the slope of the Phillips curve appears much less dramatic than in the unconditional data.”

Theoretical Explanations for Flattening

Credible Monetary Policy and Anchored Expectations

The most widely accepted explanation centers on the Federal Reserve’s success in establishing credible inflation targeting. As then St. Louis Fed President James Bullard said in 2018, “The Fed has been much more mindful about targeting inflation in the last 20 years,” resulting in “lower, more stable inflation in the U.S.” When central bank promises to maintain low inflation are credible, economic agents may discount short-term fluctuations and anchor their expectations to the inflation target.

Researchers have argued that “the low and stable rates of inflation experienced since the mid-1980s came with a decline in persistence and a bigger role for expectations about future inflation.” This created a self-reinforcing mechanism where successful monetary policy made itself less visible in traditional Phillips Curve relationships.

Globalization and Supply Chain Integration

The increasing integration of global supply chains fundamentally altered how domestic economic conditions translated into inflation. Recent research documents that “the relationship between inflation and unemployment has weakened since the 1980s,” partly “due to forces of globalization.”

One line of argument holds that “increased competition from emerging markets had reduced the pricing power of firms in developed economies,” which “in turn had reduced the ability of firms to pass on increased costs to consumers in the form of higher prices.” This “global slack” hypothesis suggested that domestic unemployment was no longer the primary determinant of domestic inflation pressures.

The theoretical impact of globalization on the Phillips Curve slope is complex and potentially ambiguous. As Furlanetto and Lepetit (2024) note, “Guilloux-Nefussi (2020) introduces endogenous entry in the export market and heterogeneous productivity across firms in an open economy model. In her model, only the largest and most productive firms enter the export market and globalization favors concentration through the emergence of big players. This concentration force dominates the pro-competitive force present in previous papers.”

Technological Change and Market Structure

The digital revolution transformed both production processes and labor markets in ways that weakened traditional Phillips Curve relationships. Research documented that “advances in automation led to a polarization of the labor market, with jobs in the middle of the skill distribution being replaced by machines,” which “in turn led to a decrease in the bargaining power of workers.”

Several studies have explored how market concentration affects the Phillips Curve. As noted by Furlanetto and Lepetit (2024), “Such a mechanism is present in several models that depart from the standard monopolistic competition setup used in the New Keynesian model: see Fujiwara and Matsuyama (2022), Wang and Werning (2022), Heise, Karahan and Sahin (2021), and Baqaee, Farhi and Sangani (2022).”

The Rise of Intangible Investment

A crucial but less discussed factor was the growing importance of intangible assets. Lall and Zeng (2020) note that “intangible investment is growing as a share of economic activity,” with “its distinguishing characteristic of generally greater scalability and lower marginal costs than tangible investment contributing to more elastic aggregate supply in recent years, which is consistent with lower inflation and a flattening of the Phillips curve.”

Labor Market Dynamics and Bargaining Power

Labor market trends also contributed to the decline in κ. As documented by Furlanetto and Lepetit (2024), “Lombardi, Riggi and Viviano (2023) show that κ decreases in response to a decline in the bargaining power of workers in a model with two margins of labor adjustment (employment and hours).” Additionally, “Basso and Rachedi (2024) show that the threat of automating labor tasks lowers workers’ bargaining power in the data and reduces the sensitivity of price and wage inflation to unemployment.”

Econometric Challenges and Identification Problems

The apparent flattening of the Phillips Curve was complicated by fundamental econometric challenges. The baseline Phillips curve includes unobserved variables, requiring reformulation in terms of observables:

πt = γπt-1 + (1 - γ)πt+1 + κ(yt - ŷ*t) + ut + (1 - γ)(Etπt+1 - πt+1) + κ(ŷ*t - y*t) (4)

where ŷ*t is an observable proxy for potential output and the last three terms (the cost-push shock, the inflation forecast error and the measurement error in potential output) form the composite error et, which creates multiple sources of endogeneity.

As Furlanetto and Lepetit (2024) explain, “endogeneity issues arising from the correlation between cost-push shocks and real activity are exacerbated by good monetary policy. Under optimal policy, a central bank will seek to fully stabilize inflation and the output gap in response to shocks that do not create a trade-off between those two objectives.”

This creates a fundamental identification problem: successful monetary policy can make the Phillips Curve appear flat even when the underlying structural relationship remains stable. Recent research demonstrates that “if monetary policy is set with the goal of minimizing welfare losses subject to a Phillips curve, a central bank will seek to increase inflation when output is below potential,” making it difficult to identify a Phillips curve empirically even when it always holds.

Part III: The COVID-19 Shock and Dramatic Steepening (2020-2022)

Empirical Evidence of Steepening

The COVID-19 pandemic and its aftermath marked a dramatic reversal in Phillips Curve dynamics. Research by the Chicago Fed (2023) documented that “Phillips curves have steepened in many industrialized countries since the start of the recovery from the Covid-19 pandemic,” suggesting “a lower sacrifice ratio now than before 2020.”

The transformation was particularly striking in the United States. As the Chicago Fed noted, “Before the pandemic, the U.S. Phillips curve had been flattening for over a decade and was considered ‘dead’ by many economists. It appears this pattern has sharply reversed over the first six quarters following 2020.”

Federal Reserve Analysis of the Steepening

Federal Reserve Governor Christopher Waller provided detailed analysis of this transformation in his speech of 31 March 2023, “The Unstable Phillips Curve,” which describes three phases:

Pre-pandemic stability (2000-2019): The Phillips curve was very flat for the 20-plus years before the pandemic, consistent with the story that the Federal Reserve was credible at keeping inflation low.

Initial steepening (2021): The Phillips curve suddenly looked relatively steep. That is, the labor market became extremely tight as the U.S. economy emerged from pandemic lockdowns.

Vertical relationship (2022): “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)” (PCE inflation in 2022, as of March 2023).

International Evidence

The steepening was not confined to the United States. IMF research by Gudmundsson, Jackson, and Portillo (2024) provided “evidence that Phillips curves shifted up and steepened across advanced economies,” with quantitative estimates showing that “across specifications the slope of the curve has increased by 170–270 percent relative to pre-Covid.”

The Czech National Bank documented similar patterns: “The latest empirical studies conclude that the Phillips curve began to steepen after the pandemic, even when using simpler specifications that had shown a ‘flatter’ Phillips curve in the previous period. This is evident in the US and the EU, but also in 28 other OECD economies, including the Czech Republic, Hungary, Poland and Slovakia.”

Three Competing Explanations for Steepening

1. Unanchored Inflation Expectations

The first explanation suggests that inflation expectations became temporarily unanchored during the pandemic. Waller’s analysis showed that “shorter-run inflation expectations followed observed inflation up and down in recent years, with a notable increase in 2021 before reversing course in 2022 and into early 2023,” but “inflation expectations over a longer horizon remained within the range of values seen in the years before the pandemic.”

The mechanism works through Phillips Curve shifts. If inflation expectations increase, the short-run Phillips curve shifts upward, generating what appears to be a vertical relationship when observed over short time periods.

2. Increased Frequency of Price Changes

A more encouraging explanation for policymakers focuses on structural changes in price-setting behavior. The slope of the Phillips curve is determined by the frequency of price changes—as Waller (2023) put it, “the more often prices change, the steeper is the Phillips curve.”

Early evidence is consistent with this mechanism. Preliminary work using U.S. Consumer Price Index micro-data examining about 90,000 individual prices each month shows that “monthly price change frequency increased from about 10 percent of the sample in 2019 to around 20 percent by the end of 2021 and held around that level in 2022.”

This increased price flexibility had several drivers:

Digital Transformation: Analysis of granular price quote data indicated that “increased digitalization may have raised price flexibility,” while “de-globalization may have made inflation more responsive to domestic economic conditions again.”

Supply Chain Disruptions: The pandemic exposed the fragility of just-in-time supply chains, forcing many firms to adopt more flexible pricing strategies to manage volatile input costs.

3. Sectoral Dynamics and Supply Constraints

The pandemic revealed important sectoral differences in Phillips Curve behavior. IMF research showed that “differences in the inflation response across sectors imply the relative price of goods has been pro-cyclical this time around rather than a-cyclical as during previous cycles.”

A two-sector New Keynesian model demonstrates that “unbalanced recoveries that run against a supply constraint in the goods sector” can endogenously generate Phillips curves that “shift up and steepen,” with the “shift up and steepening most prominent in the goods sector.”

Theoretical Mechanisms: Menu Costs and Rational Inattention

The pandemic period provided natural experiments for testing theories about state-dependent pricing. In models with menu costs, “changing prices is costly because it requires acquiring information about costs, price setting by competitors, and many other factors. Therefore, most firms will not change their price in response to a small increase in costs. In contrast, a large share of firms immediately increase their prices in response to a large shock (like the post-COVID inflation surge).”

Similarly, in rational inattention models, “agents ‘rationally ignore’ certain data and react sluggishly in response to small shocks. In contrast, households and firms change their behavior (including their prices) much more quickly in response to large shocks.”

Weber et al. (2024) provided extensive cross-country evidence showing that “both consumers and firms are substantially more informed (and thus able to process information) about the state of the economy in periods when inflation is high than in periods when inflation is low.”

Part IV: The Search for Normalization (2023-2024)

Mixed International Evidence

As economies moved further from the acute phase of the pandemic, evidence began emerging about whether Phillips Curve relationships were returning to pre-pandemic patterns. The evidence varies significantly across countries and methodologies.

Australia: No Steepening Found

Research on Australia provides a contrasting case. Mallick (2024) found that “after controlling for cost-push shocks and equilibrium unemployment, we obtain a modest slope of the Phillips curve and, more importantly, we do not find any steepening of the Phillips curve in the post-COVID period.”

This study suggests that “without correct identification, the post-COVID Phillips curve would be steeper as documented for other countries including the United States,” highlighting the importance of proper econometric identification in drawing conclusions about structural changes.

European Evidence: Modest Steepening

Research on European economies found more modest changes. Analysis indicated “some steepening in the UK, Spain, Italy and the euro area as a whole, but at magnitudes that are too small to explain the entire surge in inflation in 2021–22, suggesting an important role for outward shifts in the Phillips curve.”

United States: Ongoing Debate

Recent Federal Reserve research continues to explore the persistence of steepening effects. Smith, Timmermann, and Wright (2024) find, in the FEDS Note “Nonlinear Phillips Curves,” that the Phillips curve has flattened since about 2000 in normal labor markets but remains steep in tight ones, in both the United States and the European Union.

The Role of Nonlinearities

Recent research has increasingly focused on nonlinear aspects of the Phillips Curve. The Czech National Bank notes that “the cyclical relationship between wage inflation and unemployment changes over time,” with “this relationship being weaker in a low-inflation environment and stronger when inflation is higher.”

This suggests threshold effects where Phillips Curves may exhibit regime switching. As documented, “discussions about the alleged flattening of the Phillips curve lasted practically until the end of the pandemic and the subsequent onset of the global energy crisis in 2021. With rising energy prices and increasing inflation, more and more studies are now concluding that the Phillips curve is back and has ‘reawakened’.”

Policy Implications of Normalization

The potential normalization of Phillips Curve relationships has important implications for monetary policy conduct. If the curve has indeed steepened permanently, it suggests that central banks may be able to achieve disinflation with less economic disruption than previously thought.

As Governor Waller noted, “the higher-frequency-of-price-changes story is a more encouraging guide for central bankers” because “a steep Phillips curve means inflation can be brought down quickly with relatively little pain in terms of higher unemployment.”

However, the uncertainty surrounding the true nature of recent changes calls for policy humility. Central banks must be prepared for multiple scenarios while maintaining the credibility that helped flatten the curve in the first place.

Part V: Advanced Econometric Approaches and New Evidence

Addressing Identification Challenges

Recent methodological advances have addressed many of the identification problems that plagued earlier Phillips Curve research. As Mavroeidis, Plagborg-Moller and Stock (2014) concluded, “New identification approaches and new datasets are needed to reach an empirical consensus.” Researchers have responded with several innovative approaches.

Conditioning on Demand Shocks

Barnichon and Mesters (2020) developed an approach “using sequences of independently identified structural demand shocks as instruments.” This method addresses endogeneity by focusing on variation in economic activity that is clearly driven by demand rather than supply factors.

The approach requires that identified demand shocks satisfy both exogeneity and relevance conditions. As Furlanetto and Lepetit (2024) explain, “the exogeneity condition is satisfied if the sequence of demand shocks is orthogonal to all the elements of the error term,” while “the relevance condition holds if and only if the impulse responses of lagged inflation, future inflation, and the output gap to the identified structural shocks are not linear functions of one another.”

Regional Data Approaches

A second major advance involves using regional data to estimate Phillips Curves. As documented by Furlanetto and Lepetit (2024), “The motivation for using regional data is two-fold. First, there is potentially more variability in the regional data than in the aggregate data, which facilitates inference. Second, changes in the aggregate environment can be captured by time-fixed effects.”

The key insight is that “the deviation of the regional output gap from the aggregate output gap is independent of monetary policy” because “the policy rate drops out of” the regional equation. This eliminates the endogeneity bias from monetary policy responses.

Studies using this approach include significant contributions by Hazell et al. (2022), who provide “evidence from U.S. states” on the slope of the Phillips curve, and various Federal Reserve researchers analyzing metropolitan-level data.

Firm-Level Data

Gagliardone et al. (2023) represent a third methodological advance, using “detailed micro-data for the Belgian manufacturing sector at a quarterly frequency from 1999 to 2019” and relying on “a bottom-up approach to estimating the slope of the Phillips curve.”

This approach provides unique insights because researchers “observe costs, prices, and quantities of production at the firm-product level,” allowing them to “estimate dynamic pass-through regressions relating prices to measures of marginal costs at the firm level.”

Synthesis of Recent Slope Estimates

The table below synthesizes recent estimates of Phillips Curve slopes from studies using advanced identification methods:

Study Sample Period Dependent Variable Forcing Variable Slope Estimate
Barnichon and Mesters (2020) 1990-2017 Core PCE inflation (quarterly change) Unemployment gap -0.24
Inoue, Rossi and Wang (2022) 1974-2007 Core PCE inflation (quarterly change) CBO unemployment gap -0.6 (1974) to -0.25 (2007)
Del Negro et al. (2020) 1989-2019 Core PCE inflation (quarterly change) Unemployment rate ≈0
Hazell et al. (2022) 1991-2018 Headline CPI (annual change) Discounted sum of current and future unemployment rates -0.055
McLeay and Tenreyro (2019) 1990-2017 Core CPI (semiannual change) Unemployment rate -0.38

As Furlanetto and Lepetit (2024) note, these estimates “are not directly comparable” because they depend on factors such as “the horizon of the inflation expectations included in the regression” and “the persistence of the forcing variable.”

Time-Varying Parameter Models

Recent advances include sophisticated time-varying parameter models. Research asks “Is the slope of the Phillips curve time-varying?” using “unobserved components models” to track how the relationship evolves over time.

These approaches are particularly relevant for understanding “the role of individual shocks and policy feedback coefficients in determining the optimal targeting window as the slope of the Phillips curve varies.”

Part VI: Nonlinear Dynamics and State Dependence

Theoretical Foundations of Nonlinearity

Recent research has moved beyond linear Phillips Curve specifications to explore fundamental nonlinearities in the inflation-unemployment relationship. As the Czech National Bank documents, “A complementary perspective is offered by non-linear models, which allow for a framework of two inflation regimes or two slopes of the Phillips curve.”

The theoretical basis for these nonlinearities comes from several sources:

State-Dependent Pricing Models

In menu cost models, “the slope of the Phillips curve becomes state-dependent” because “changing prices is costly” and “most firms will not change their price in response to a small increase in costs” but will “immediately increase their prices in response to a large shock.”

This creates a fundamental asymmetry: research demonstrates that “expansionary shocks are more inflationary because the probability of price increases is higher than the probability of price decreases.”

Labor Market Nonlinearities

Benigno and Eggertsson (2023, 2024) show that “when unemployment is high, increased demand reduces unemployment with minimal inflation, whereas when unemployment is low, the economy is operating at full capacity and any increase in demand leads to significant inflation.”

This suggests that the Phillips Curve should be flatter at high unemployment levels and steeper when unemployment is low—exactly the pattern observed during the pandemic recovery.

Capacity Constraints and Sectoral Dynamics

Boehm and Pandalai-Nayar (2022) use “pre-pandemic industry-level data” and document that “industries’ supply curves are convex.” They “develop a model to show that the convexity could be due to capacity constraints in production.”

Guerrieri et al. (2021) show that “a combination of a shift in the composition of demand and convexities in sectoral supply curves can give rise to aggregate cost-push shocks.” The logic is that when demand shifts between sectors with different supply elasticities, “in the aggregate, prices increase and quantities decline: the shock ends up looking like a cost-push shock.”

Empirical Evidence of Nonlinearities

Pre-Pandemic Evidence

Furlanetto and Lepetit (2024) document that “unconditional aggregate data indicates a modest degree of steepening of the price Phillips curve at low levels of the unemployment rate if the data extends back to the 1960s. There is, however, more robust evidence of nonlinearities in the wage Phillips curve.”

Regional data provides stronger evidence: “Babb and Detmeister (2017) and Hooper, Mishkin and Sufi (2020) find robust evidence of a steeper price Phillips curve at low levels of unemployment in U.S. MSA-level data.”

Pandemic-Era Nonlinearities

The pandemic provided compelling evidence of state dependence. Cerrato and Gitti (2022) use “MSA-level data and compare pre-COVID estimates with those obtained on a sample ranging from March 2021 to August 2022: they find that the slope of the Phillips curve has tripled between the former and latter sample.”

Asymmetric Shock Transmission

Recent research documents fundamental asymmetries in how shocks affect inflation. Studies show that “asymmetry in the transmission of shocks” means “expansionary shocks are more inflationary because the probability of price increases is higher than the probability of price decreases.”

This asymmetry helps explain why Phillips Curves may appear steep during inflationary episodes but flat during disinflationary ones, contributing to the apparent instability of estimated relationships.

Volatility and Curve Steepness

Another source of nonlinearity comes from volatility effects. Research suggests that “increased volatility, driven by significant shocks that cause greater fluctuations in inflation in both directions” may contribute “to the observed steepening of the Phillips curve.”

The pandemic period was characterized by unprecedented economic volatility, which may have contributed to the apparent steepening of Phillips Curve relationships through this mechanism.

Part VII: Sectoral Analysis and Disaggregated Evidence

Goods vs. Services Phillips Curves

One of the most important insights from recent research concerns sectoral differences in Phillips Curve behavior. IMF research documents that “differences in the inflation response across sectors imply the relative price of goods has been pro-cyclical this time around rather than a-cyclical as during previous cycles.”

The pandemic created an unusual pattern of demand shifts. Analysis shows that “the shift up and steepening was most prominent in the goods sector, consistent with the price of goods relative to services becoming much more sensitive to the recovery.”

Two-Sector Model Insights

Theoretical analysis using a two-sector new-Keynesian model shows that “when we introduce unbalanced recoveries that run against a supply constraint in the goods sector,” three patterns emerge endogenously: steepening, shifting, and relative goods price pro-cyclicality.

The model generates quantitatively significant effects: “When recoveries are unbalanced and constrained, the model generates an increase in the slope of the Phillips curve of between 50 and 60 percent, depending on the representation of the Phillips curve (dynamic versus static).”

Regional Heterogeneity

U.S. State-Level Evidence

The comprehensive analysis by Hazell et al. (2022) using U.S. state data provides crucial insights. Their two-region model shows that it is the slope of the regional Phillips curve for nontradeable goods that is informative about the aggregate slope.

Their findings suggest relatively modest slopes when properly identified, with careful attention to the distinction between tradeable and non-tradeable sectors.

Metropolitan Area Analysis

Studies using Metropolitan Statistical Area (MSA) data provide additional granular evidence. Fitzgerald et al. (2024) estimate “a price Phillips curve with the unemployment rate as the forcing variable using metropolitan statistical areas (MSA) data and find that the slope of the Phillips curve has modestly declined after 2000, although that decline is not statistically significant.”

Panel Evidence with Estimated Breaks

Smith, Timmermann and Wright (2023) estimate price and wage Phillips curves on U.S. “sectoral, state, or MSA-level data” with “Bayesian panel methods with breakpoints that allows them to estimate the timing of Phillips curve breaks instead of imposing them.”

Their analysis reveals important heterogeneity: “In the sectoral data, they uncover a marked flattening of the price Phillips curve between a pre-2001 and a post-2001 sample. However, the state-level data indicates that the slope of the wage Phillips curve has been relatively stable over time while the MSA-level data shows a similar result for the slope of the price Phillips curve.”

Industry-Level Dynamics

Recent firm-level research has provided unprecedented insight into micro-foundations. Gagliardone et al. (2023) use “detailed micro-data for the Belgian manufacturing sector” and find “a relatively high slope for the aggregate marginal cost-based Phillips curve in the range of 0.05 to 0.06.”

Importantly, they find that “the elasticity of marginal cost with respect to the output gap is low, which helps reconcile their results with the low slope estimates usually found in the literature estimating unemployment or output gap-based Phillips curves.”

This distinction between marginal cost-based and output gap-based Phillips curves helps explain much of the apparent inconsistency in the literature.

Part VIII: Price-Setting Behavior and Micro Evidence

Frequency of Price Changes

One of the most compelling pieces of evidence for structural change comes from detailed analysis of price-setting behavior. Montag and Villar Vallenas (2023) “uses U.S. Consumer Price Index micro-data to examine about 90,000 individual prices each month and determines if there was a change and, if so, was the change an increase or decrease.”

Their findings are striking: “The monthly price change frequency (the black line) increased from about 10 percent of the sample in 2019 to around 20 percent by the end of 2021 and held around that level in 2022. The increase in price change frequency is primarily driven by firms upwardly adjusting their prices (the red dashed line) during this period.”

Calvo vs. Menu Cost Models

The distinction between time-dependent (Calvo) and state-dependent (menu cost) pricing models becomes crucial for understanding recent dynamics. In the standard New Keynesian model with time-dependent pricing, “an increase in the frequency of price adjustment at the micro level mechanically leads to a higher value of κ.”

However, “the frequency of price changes is an exogenous parameter in the standard New Keynesian model with time-dependent pricing,” but “it changes with economic conditions in models with menu costs, thus implying that the slope of the Phillips curve becomes state-dependent.”

Information Processing and Attention

Models of rational inattention provide additional insight into price-setting behavior. In these models, “households and firms have a limited amount of attention that they can dedicate to processing information. It is costly and time consuming to constantly adjust behavior in response to shocks.”

Weber et al. (2024) provide “extensive cross-country evidence showing that both consumers and firms are substantially more informed (and thus able to process information) about the state of the economy in periods when inflation is high than in periods when inflation is low.”

This research suggests that the apparent steepening of Phillips Curves during high inflation periods may reflect increased information processing rather than fundamental structural changes.

Digital Pricing and Technological Change

The role of digitalization in changing price-setting behavior has received increasing attention. Research using “granular price quote data from the UK indicates that increased digitalization may have raised price flexibility.”

The pandemic accelerated digital adoption across many sectors, potentially making it easier and cheaper for firms to adjust prices frequently. This technological change could represent a permanent shift toward more flexible pricing, implying persistently steeper Phillips Curves.

Part IX: Global Perspectives and Cross-Country Evidence

European Union Analysis

The European experience provides important comparative evidence. Czech National Bank research documents that recent steepening is “evident in the US and the EU, but also in 28 other OECD economies, including the Czech Republic, Hungary, Poland and Slovakia.”

Central European countries show particularly interesting patterns: “recent studies point to an increasing importance of the backward-looking component of inflation, especially as inflation persistence increases, particularly in Central and Eastern European countries.”

Developed vs. Emerging Economy Differences

Research on emerging markets reveals different patterns. Studies examining “price adjustment patterns in Armenia from January 2007 to December 2021” found that “prices in Armenia change more frequently than in advanced economies, with an average price spell of 2.4 months.”

This suggests that the flattening observed in advanced economies during the 2010s may have been less pronounced in emerging markets, where “high inflation volatility and frequent external shocks” maintain more responsive pricing behavior.

Japanese Experience

Japan provides a particularly interesting case study given its long experience with low inflation. Recent research examining “Japanese parameters, such as a low output gap response to the real interest rate, discounted IS curve, alternative inflation persistences” provides insights into how Phillips Curve relationships function in persistently low inflation environments.

Cross-Country Synthesis

Comprehensive international analysis by the Chicago Fed documented that Phillips curve steepening “is apparent in figure 1, which shows the Phillips curves of the United States, United Kingdom, and France during two time periods: the seven years (or 28 quarters) before the pandemic (2013:Q1–2019:Q4) and the six quarters of the recovery for which we have data (2021:Q1–2022:Q2).”

The evidence shows that “These three countries’ Phillips curves have steepened significantly during the recovery, in stark contrast with their flat Phillips curves of the pre-pandemic period.”

Part X: Monetary Policy Implications and Framework Evolution

Sacrifice Ratio Dynamics

The evolution of Phillips Curve slopes has fundamental implications for the sacrifice ratio—the amount of economic output that must be sacrificed to achieve a given reduction in inflation. Recent steepening suggests “a lower sacrifice ratio now than before 2020,” meaning that inflation can be brought down at a smaller cost in economic activity.

This represents a dramatic change from the post-2008 period when flat Phillips Curves implied high sacrifice ratios and prolonged disinflation processes.

Average Inflation Targeting

The changing slope of the Phillips Curve has important implications for optimal monetary policy frameworks. Research on “the optimal targeting window for a central bank conducting an average inflation targeting” framework shows how curve flattening affects policy design.

Analysis reveals that “a flatter slope also implies an increasing role of inflation variation in determining household welfare,” suggesting that “the central bank might want to increase their focus on inflation stabilizations, sacrificing the output-gap stabilizations.”

When the Phillips Curve is steep, as appears to be the case post-pandemic, central banks may have more flexibility in their policy frameworks.

Communication and Expectations Management

The role of central bank credibility remains crucial. Waller’s analysis emphasizes that when a central bank’s inflation promises are credible, “the Phillips curve should be relatively flat.” The recent steepening may partly reflect temporary questioning of central bank resolve during the initial stages of the inflation surge.

However, “if unanchored inflation expectations were driving what looks like a shifting Phillips curve, this would be extremely problematic for central bankers because it could require dramatic actions by the FOMC to lower inflation expectations to shift the Phillips curve back down.”

Forward Guidance and Policy Rules

The evolution of Phillips Curve relationships affects optimal policy rules. Writing in March 2020, with inflation below target, Del Negro, Lenza, Primiceri and Tambalotti argued in a Brookings paper that “systematic [policies] like average inflation targeting could be especially effective in bringing inflation back to 2 percent in the current environment.”

The key insight is that “A flat Phillips Curve requires the monetary authority to work harder to stabilize inflation: Unemployment needs to get lower to bring inflation back to target after a recession.”

Part XI: Looking Forward: Scenarios and Implications

Scenario Analysis for Phillips Curve Evolution

Looking ahead, several scenarios emerge for the future evolution of Phillips Curve relationships:

Scenario 1: Return to Pre-Pandemic Flatness

In this scenario, the steepening observed during 2021-2022 represents a temporary response to pandemic-specific shocks. As supply chains normalize and inflation expectations re-anchor, the Phillips Curve would return to its pre-pandemic flat configuration.

Implications: This would suggest that the 2010s pattern of weak inflation responses to labor market tightness would resume, requiring more aggressive monetary policy to achieve inflation targets.

Scenario 2: Permanent Steepening

This scenario assumes that structural changes—increased digitalization, reduced globalization, or changed price-setting behavior—have permanently steepened the Phillips Curve.

Implications: Central banks would face more favorable trade-offs, with the ability to achieve disinflation with less economic disruption.

Scenario 3: State-Dependent Nonlinearity

This scenario suggests that Phillips Curves are inherently nonlinear, appearing flat in low-inflation environments but steep when inflation is elevated.

Implications: Policy frameworks would need to account for regime-dependent relationships, with different strategies for different inflation environments.

Climate Change and Supply Shocks

Future Phillips Curve dynamics will likely be influenced by climate-related supply shocks. Energy transition and climate-related disruptions may create new sources of inflation volatility.

These supply-side factors could interact with Phillips Curve relationships in complex ways, potentially creating more frequent episodes of stagflationary pressures.

Demographic Transitions

Demographic changes represent another important factor. Research by Ambrocio (2023) examines “Demographic Aging and the New Keynesian Phillips Curve,” suggesting that aging populations may affect both labor market dynamics and inflation trends.

Technological Evolution

The continued advancement of artificial intelligence, automation, and digital platforms will likely continue to affect price-setting behavior. The question is whether these changes will enhance price flexibility (steepening Phillips Curves) or reduce pricing power through increased competition (flattening curves).

Geopolitical Fragmentation

The trajectory toward economic nationalism and supply chain reshoring may strengthen Phillips Curve relationships by reducing the role of global slack in domestic inflation determination. Research already documents that “higher digitalization and lower trade intensity are shown to be associated with steeper Phillips curves.”

Part XII: Methodological Lessons and Future Research

Advances in Identification

The Phillips Curve literature has made substantial progress in addressing identification challenges. As Furlanetto and Lepetit (2024) conclude, “While these approaches can and should still be refined, they have brought new robust results, notably that traditional approaches to estimating the Phillips curve using lagged macroeconomic variables as instruments tend to underestimate its slope.”

Remaining Challenges

Despite methodological advances, important challenges remain:

Weak Instruments Problem

Even sophisticated identification strategies face limitations: “truly exogenous monetary policy shocks are small and constitute weak instruments, thus leading to large weak-IV robust confidence intervals for the parameters.”

Model Uncertainty

Regional approaches face issues when “the underlying inflation dynamics in the data are heterogeneous across regions (different γi,t and κi,t by region): indeed, in that case the error term is correlated with the regressor.”

Structural Break Dating

Many studies impose break dates rather than estimating them endogenously. More sophisticated approaches like those by Smith, Timmermann and Wright (2023) that “estimate the timing of Phillips curve breaks instead of imposing them” provide more reliable evidence.

Future Research Directions

Several areas warrant continued investigation:

Real-Time Analysis

Understanding how Phillips Curve relationships evolve in real-time rather than only in retrospective analysis would enhance policy relevance.

High-Frequency Data

The use of “about 90,000 individual prices each month” represents the frontier of micro-price analysis, but expansion to higher frequency and broader coverage could provide additional insights.

Cross-Country Comparative Studies

Systematic comparison across countries with different institutional arrangements could help identify the structural factors most important for Phillips Curve stability.

Integration with Other Structural Relationships

Future work should better integrate Phillips Curve analysis with other key macroeconomic relationships, including the IS curve and monetary policy reaction functions.

Conclusion: The Enduring Relevance of the Phillips Curve

The Phillips Curve’s journey from apparent death to dramatic resurrection during the COVID-19 pandemic illustrates the dynamic nature of macroeconomic relationships. Rather than viewing these changes as evidence of theoretical failure, they represent the normal evolution of economic structures and relationships in response to changing technological, institutional, and policy environments.

Key Lessons

Structural vs. Reduced-Form Relationships: The experience demonstrates the importance of understanding the structural foundations of macroeconomic relationships rather than relying solely on reduced-form empirical correlations. As the literature shows, “the overwhelming majority of empirical explorations of the Phillips curve is based on a relationship between inflation and a measure of real economic activity,” but “in structural models, microfounded versions of the Phillips curve relate inflation to average real marginal cost.”

Policy Endogeneity: The fundamental insight that “if monetary policy is set with the goal of minimizing welfare losses subject to a Phillips curve, a central bank will seek to increase inflation when output is below potential,” making it “difficult to empirically identify a Phillips curve even when it always holds,” has important implications for both research and policy.

State Dependence: The evidence for nonlinear, state-dependent Phillips Curves suggests that policy frameworks should account for regime-dependent relationships rather than assuming constant parameters.

Technological and Structural Change: The role of “increased digitalization” and changing “trade intensity” in affecting Phillips Curve slopes highlights the importance of considering how technological and structural changes affect macroeconomic relationships.

Implications for Monetary Policy

The evolution of Phillips Curve relationships has several implications for monetary policy conduct:

Framework Flexibility: Central banks need frameworks robust to changes in underlying economic relationships, including the possibility that key parameters may be time-varying or state-dependent.

Credibility Maintenance: The central role of expectations in Phillips Curve dynamics reinforces the crucial importance of maintaining central bank credibility, as when a central bank’s inflation promises are credible, “the Phillips curve should be relatively flat.”

Humility and Adaptation: As the San Francisco Fed research suggests, “A failure to maintain inflation expectations around the target could greatly undermine the Federal Reserve’s ability to achieve stable prices—part of its dual mandate—in the future.”

The Future of Phillips Curve Research

Looking ahead, Furlanetto and Lepetit (2024) note that “At a time when inflation is once again a major policy issue, these methodological advances should form a useful springboard for further research on inflation dynamics.”

The research agenda should focus on:

  1. Understanding when and why Phillips Curve relationships change
  2. Developing robust policy frameworks that can adapt to evolving relationships
  3. Integrating micro-foundations with macro-level analysis
  4. Incorporating technological and structural change into theoretical models

Final Reflections

The Phillips Curve remains a useful framework for understanding inflation dynamics, but its application requires careful attention to changing economic structures and institutional arrangements. The pandemic experience has demonstrated that structural economic relationships can reassert themselves quickly when conditions change, providing important lessons for both economic theory and policy practice.

Rather than asking whether the Phillips Curve is “alive” or “dead,” we should focus on understanding how and when it operates, what factors influence its strength and stability, and how policy frameworks can best adapt to its evolution. The relationship between inflation and economic activity, while fundamentally rooted in economic logic, will continue to manifest differently across different economic environments and institutional arrangements.

The recent experience serves as a powerful reminder that in macroeconomics, relationships that appear stable can change quickly when underlying conditions shift. This underscores the importance of maintaining both theoretical understanding and policy flexibility to navigate an ever-changing economic landscape, while recognizing that the fundamental trade-offs captured by the Phillips Curve remain central to monetary policy analysis and implementation.

The story of the Phillips Curve’s evolution is far from over. As economies continue to evolve in response to technological change, demographic shifts, climate challenges, and geopolitical developments, the Phillips Curve will likely continue to adapt. The key for both researchers and policymakers is to remain vigilant to these changes while maintaining the theoretical and empirical tools necessary to understand and respond to them effectively.


This analysis draws from extensive recent research in monetary economics, including contributions from the Federal Reserve System, International Monetary Fund, Czech National Bank, and leading academic institutions. The citations provided throughout reflect the current state of knowledge in this rapidly evolving field.

Sources

The studies and speeches quoted above, with the documents they come from:


Personal page. Views are my own and do not represent my employer. Nothing here is investment advice.