Our review of three academic papers. The first paper (What Does New Micro Price Evidence Tell Us About Inflation Dynamics and Monetary Policy Transmission? by ECB economists Luca Dedola and coauthors) shows that firms adjust prices more frequently during periods of high inflation, altering the transmission of monetary policy. The second paper (“The Long-Run Phillips Curve is … a Curve” by Guido Ascari (Uni Pavia) and coauthors) finds evidence of a non-linear and negatively sloped long-run Phillips curve, contrary to the traditional view of a vertical curve. In other words, the paper challenges the notion that inflation has no long-term impact on real economic activity, suggesting instead that higher inflation can lead to significant output losses. Finally, the third paper (“Bond Market Views of the Fed” by Luigi Bocola (Stanford) and coauthors) argues that markets began to view the Fed as more tolerant of inflationary fluctuations after the pandemic, in contrast to the preceding two decades. This change in perception, alongside regime shifts in the Fed’s monetary policy rule, can account for approximately half of the observed inflation increase in 2021, according to the paper.
Paper #1: Dedola et al. (2024)
“What Does New Micro Price Evidence Tell Us About Inflation Dynamics and Monetary Policy Transmission?” by ECB economists Luca Dedola, Lukas Henkel, Christian Höynck, Chiara Osbat, and Sergio Santoro examines the implications of micro-level price data on inflation dynamics and the effectiveness of monetary policy. Using granular price data, the authors show that firms adjust prices more frequently during periods of high inflation, leading to faster inflation pass-through and altering the transmission of monetary policy.
What the paper does and main results
The authors use new evidence from the Price-Setting Microdata Analysis (PRISMA) network to explore how state-dependent pricing affects the speed and magnitude of inflation in response to economic shocks. The dataset comprises 135 million individual price quotes across 11 Eurozone countries from 2010 to 2019, covering approximately 60% of the Harmonised Index of Consumer Prices (HICP) basket. This microdata allows the authors to analyze how often and by how much prices change, shedding light on the underlying mechanisms of inflation. The authors contrast two key price-setting behaviors: state-dependent models, where firms adjust prices based on economic conditions, and time-dependent models, where price changes occur at regular intervals regardless of economic shocks.
The frequency of price adjustments increases significantly in response to large aggregate shocks, leading to a non-linear relationship between the frequency of price changes and inflation. During periods of high inflation, firms are more likely to adjust prices, resulting in a faster and potentially more volatile inflationary process. Figure 1 shows that firms adjusted prices more frequently during the high-inflation episode of 2022-2023. More generally, price adjustment is “lumpy,” meaning that price changes occur infrequently but tend to be large when they do happen. This behavior is also more pronounced during periods of elevated inflation.
Figure 1. Higher inflation is associated with more frequent price adjustment.
The increased frequency of price changes during inflationary periods suggests that monetary policy actions, such as interest rate adjustments, may have quicker but potentially less persistent effects on inflation. The authors’ model simulations support this idea, showing that the pass-through of monetary policy shocks to inflation is more immediate in a high-inflation environment, but the inflationary effects dissipate more quickly compared to periods of low inflation.
A critique of the paper
The paper offers valuable insights into the dynamics of inflation by leveraging detailed microdata, which provides a more granular understanding of price-setting behavior. The focus on state-dependent pricing models is particularly relevant given the current economic environment characterized by frequent and significant economic shocks. This type of micro evidence is an important input to inform our economic theories, some of which (see “The Inflation Accelerator” by Blanco et al. (2024)) have recently started to focus on the inherent nonlinearities arising from endogenous changes in the price-setting behavior. The analysis could benefit from a more detailed sectoral breakdown, exploring how price-setting behavior varies across industries with different cost structures and competitive dynamics.
Policy implications
Central banks may need to act more decisively to address initial increases in inflation because of a nonlinear acceleration mechanism via changes in price setting frequency. While the authors show that interest rate decisions feed into prices more quickly when inflation is high, more frequent price adjustments also reduce nominal rigidities such that central banks have less impact on real rates. This matches the authors’ findings that the effects of monetary policy could be less persistent in high-inflation environments, necessitating more frequent adjustments to interest rates or other policy tools to maintain price stability. This challenges the conventional wisdom of gradualism in monetary policy and suggests that central banks should be prepared for more dynamic and responsive policy frameworks.
Paper #2: Ascari et al. (2024)
“The Long-Run Phillips Curve is … a Curve” by Guido Ascari (De Nederlandsche Bank), Paolo Bonomolo (De Nederlandsche Bank), and Qazi Haque (The University of Adelaide) examines the long-run relationship between inflation and output in U.S. data. Contrary to the traditional view of a vertical long-run Phillips curve, the authors find evidence of a non-linear and negatively sloped Phillips curve. Specifically, they identify a threshold level of trend inflation around 4%, above which potential output decreases as inflation rises. Their findings challenge the notion that inflation has no long-term impact on real economic activity, suggesting instead that higher inflation can lead to significant output losses.
What the paper does and main results
The authors use a piecewise linear Bayesian vector autoregression (BVAR) model to investigate the relationship between inflation and potential output in the long run. They develop an empirical framework that generalizes the standard BVAR with stochastic trends, allowing for non-linearities in the long-run relationship. Their analysis focuses on U.S. data from 1960 to 2008, excluding the post-2008 period to avoid complications from the zero lower bound on interest rates.
Below a threshold of 4% for trend inflation, potential output remains unaffected but above this threshold, every percentage point increase in trend inflation is associated with about a 1% decrease in potential output per year. This suggests that the long-run Phillips Curve is not vertical but instead slopes downward when inflation exceeds the threshold. To support these findings, the authors estimate a New Keynesian (NK) model that incorporates time-varying trend inflation. The structural model confirms the existence of a negatively sloped long-run Phillips Curve, consistent with the empirical results from the BVAR model. The NK model quantifies the output losses associated with higher trend inflation, estimating that an inflation target of 4% could result in an output loss of about 0.6% per year compared to a 2% target. Figure 2 compares the long-run Phillips curve estimates of the VAR (blue) and the NK model (black). The slope estimates are very similar, with tighter confidence intervals for the NK model due to some modelling assumptions.
Figure 2. Estimate of the long-run Phillips curve from the BVAR (blue) and NK model (black). Dashed lines are 90% confidence intervals.
A critique of the paper
The paper makes a significant contribution by challenging the traditional view of the long-run Phillips Curve as vertical. The use of a piecewise linear BVAR model is innovative, allowing the authors to capture non-linearities that standard models might miss. It is encouraging to see that an estimated New Keynesian model delivers very similar results.
The estimation of the slope of the long-run Phillips curve is plagued by the same issues as estimates for the short-run curve. Empirical estimates generally suffer from very wide confidence intervals, as can be seen for the BVAR model. This is because of the short time series available for estimation. The nonlinearity in the slope might also be time-varying, though estimating this variation is even more challenging.
The short sample also means that the results depend on a single period of increased trend inflation (1970s/1980s). The estimates must therefore be taken with a grain of salt since they can represent period-specific factors, and it is unclear if the real effects of the monetary contraction during that time affect the results. In this regard, it would be encouraging to use the recent inflation episode to extend the sample.
Policy implications
The evidence of a negatively sloped long-run Phillips Curve suggests that central banks should be cautious about allowing inflation to rise above certain thresholds. While moderate inflation might not harm potential output, sustained inflation above the 4% threshold could lead to significant long-term output losses. This challenges the conventional wisdom that inflation is primarily a short-run concern and underscores the importance of maintaining price stability for long-term economic growth.
Central banks need to consider the non-linear effects of inflation on output when setting inflation targets. The results suggest that a higher inflation target, such as 4% (but also 3%), could have substantial costs in terms of lost output, which might outweigh the benefits of avoiding the zero lower bound on interest rates. Policymakers should carefully weigh these trade-offs when designing monetary policy frameworks, particularly in the context of ongoing debates about raising inflation targets in response to long-run supply side pressures and the risk of falling into ZLB territory.
Paper #3: Bocola et al. (2024)
“Bond Market Views of the Fed” by Luigi Bocola (Stanford), Alessandro Dovis (UPenn), Kasper Jørgensen (ECB), and Rishabh Kirpalani (University of Wisconsin-Madison) studies shifts in financial markets’ perceptions of the Federal Reserve’s monetary policy, especially in the post-2020 period. The authors argue that markets began to view the Fed as more tolerant of inflationary fluctuations after the pandemic, in contrast to the preceding two decades. This change in perception, alongside regime shifts in the Fed’s monetary policy rule, can account for approximately half of the observed inflation increase in 2021.
What the paper does and main results
The paper uses daily data on nominal Treasury yields and inflation-protected bonds (TIPS) to identify changes in financial market expectations about the Federal Reserve’s response to inflation. Based on market-implied expectations of future nominal interest rates and inflation, the authors observe that bond markets displayed a stable relationship between expected future inflation and nominal interest rates from 2000 until 2020, consistent with the Fed’s inflation-targeting policy. However, between 2020 and 2022, this relationship weakened considerably. Figure 3 shows that the correlation between expected future inflation and expected future nominal rates has flattened since 2020. The left panel shows the scatter plot for 2017-2019 and 2020-2022. The right panel estimates the coefficient of future expected nominal rates on inflation expectations, showing a significant drop from around 1.5 to below 1 since 2020.
Figure 3. Relation between expected future inflation and expected future interest rates. The left panel plots the relation for two subsamples. The right panel estimates the coefficient (rates on inflation) for six subsamples.
To quantify the macroeconomic effects of this change, the authors incorporate high-frequency financial data into a standard New Keynesian model that includes regime shifts in the Fed’s policy rule. The model’s key insight is that financial markets began to expect the Fed to place less emphasis on stabilizing inflation post-2020, aligning with the introduction of the Fed’s flexible average inflation targeting (FAIT) framework. Under this framework, the Fed allowed for inflation to temporarily exceed its 2% target to make up for prior periods of below-target inflation. A counterfactual analysis shows that inflation in 2021 would have peaked at around 5% rather than the observed 9% if the Fed had maintained its pre-2020 policy regime. Therefore, shifts in the perceived monetary policy framework, as reflected in bond market data, played a central role in amplifying inflationary pressures during the pandemic recovery.
A critique of the paper
The paper usefully combines market data on inflation and interest rate expectations with the canonical New Keynesian model but it is unclear that market-based expectations are the relevant metric to feed into our macro models. Making the connection between market data and New Keynesian model requires the authors to take a stance that these expectations are accurately capturing aggregate expectations. However, firms and households may have different expectations with different implications for the perceived policy regime. The strength of the exercise therefore mostly lies in measuring market perceptions of policy regime changes as opposed to studying the macroeconomic implications of these perceptions.