Did supply constraints tilt the Phillips Curve?

Ambrogio Cesa-Bianchi, Ed Hall, Marco Pinchetti and Julian Reynolds

The remarkable stability of US inflation dynamics in the pre-Covid era had led many to think that the Phillips Curve had flattened. However, the sharp rise in inflation that followed the Covid-19 pandemic ignited a debate on whether the Phillips Curve had steepened and, in particular, whether its slope depends on some particular macroeconomic conditions. Which are these conditions, though? In this post, we argue that one important candidate that could explain this kind of state-dependency in the slope of the Phillips Curve is global supply chain constraints. We propose a simple framework to account for this state-dependency, and conduct econometric analysis on US data which supports its implications – showing that inflation in the US is more responsive to slack when supply constraints are tighter.

Global supply constraints and the Phillips Curve: a framework

Several recent episodes of global supply chain (GSC) constraints have severely hampered industrial production by limiting the availability of intermediate inputs, leading to a severe increase in delivery times for durable goods. Chart 1 compares two popular measures of supply pressure, the NY Fed GSCPI Index (cyan line, which focuses on supply chain frictions), and the Dallas Fed IGREA (orange line, which captures the global commodity market pressure) – with a Composite Supply Pressure Index (CSPI, purple line, which we construct as a simple average of the two). While all indices show that in recent years global supply pressures have been exceptionally high by historical standards, in this post we focus on the Composite Supply Pressure Index, which summarises pressures arising from both the shipping and the commodity markets.

Chart 1: Measuring global supply chain pressures

To illustrate the idea that supply constraints can affect inflation responsiveness over and above their direct impact on inflation via the cost channel, we exploit a simple AD/AS textbook model which we augment with a global supply constraint. The framework builds on the anecdotal evidence of the 2020 semi-conductor crisis: the main idea is that, after a certain threshold, it can become very expensive to produce an additional unit of output, and so a change in demand must result in additional pressure on prices and a muted response of quantities. This is represented graphically in Chart 2. The downward-sloping curve represents aggregate demand. The upward-sloping curve represents the Phillips Curve, which steepens at the point where the global supply constraint becomes binding. The presence of this kink in the Phillips Curve should be broadly interpreted as a reduced form way to capture the non-linear effects through which global supply chain pressures can affect the shape of the Phillips Curve.

We consider two cases in the panels in Chart 2. In normal times (left panel), when GSCs are working smoothly, the equilibrium lies on the flat portion of the Phillips Curve. In this region, a change in demand leads to a small change in inflation and a large change in output, as per the commonly held view. But when GSCs are under stress (right panel), the steep part of the Phillips Curve shifts inwards. If the shift is large enough, the same change in demand now results in stronger inflationary dynamics and a muted output response.

Chart 2: The effect of demand shocks on inflation when global supply constraints are loose versus tight

In sum, this simple theoretical framework can jointly explain (i) why the response of inflation to demand shocks can be weak when GSCs are not subject to any particular pressure, and (ii) why it can be pronounced when GSC constraints are tight, as in the recovery from the Covid pandemic.

Global supply constraints and the Phillips Curve: what do the data say?

To investigate empirically whether GSC pressures led to a steepening of the Phillips Curve, we need to isolate exogenous shifts in demand. As explained by McLeay and Tenreyro (2020), this task can be challenging. Central banks with a dual mandate (such as the Federal Reserve) typically attempt to offset demand shocks, while partially accommodating cost-push shocks. This results in simultaneity between supply shocks (eg a change in energy prices) and demand shocks (ie the monetary policy response). Hence, the main goal of the econometric strategies which aim to identify the Phillips Curve is controlling for supply factors. In this work, we explore two possible approaches.

First, we exploit US metropolitan-level data to study the effect of fluctuations in local unemployment on local inflation, while using area and year-fixed effects to control for the other confounding factors. This setup can be particularly well equipped to identify demand shocks, as most supply shocks have a national rather than a regional character, and can therefore be captured by time-fixed effects. To evaluate the relevance of GSC pressure for the slope of the Phillips Curve, we extend this specification by introducing an interaction term between unemployment and global supply pressure, as measured by the CSPI. This term captures the additional effect of a tightening in supply chain pressures on the slope of the Phillips Curve; in other words, if the Phillips Curve were linear, the interaction term should not be statistically significant.

We consider two different specifications of this interaction. First, we employ the continuous values of supply chain pressure (CSPI), and interact these values with unemployment (Unemployment x CSPI). We label this specification ‘continuous’. Second, we interact unemployment with a dummy variable which equals one when the CSPI is above the 75th percentile and zero otherwise (Unemployment x Dummy). We label this specification ‘dummy threshold’. We report the estimated coefficients in Chart 3, for core inflation (left bar in each panel) and services inflation (right bar in each panel). In the chart, the cyan bars represent the linear relationship between inflation and economic slack, ie the drop in inflation associated with a given increase in the unemployment rate. The orange bars represent the marginal effect on the relationship between inflation and unemployment of tighter supply chain pressures.

Chart 3: The role of global supply constraints: evidence from US metropolitan areas

Our results suggest that the coefficient on the interaction between supply and activity measures (unemployment) is generally negative, and in line with our theoretical framework. That is, high supply chain pressures are associated with a steeper Phillips Curve. The coefficient for this interaction term is significant when using core or services inflation as the dependent variable, although not for alternative headline inflation specifications.

A second, complementary, approach to investigate the empirical validity of our framework is to exploit changes in demand driven by identified monetary policy shocks. According to our framework, monetary policy shocks should have a larger impact on inflation when supply chain pressures are high. We can test this hypothesis by estimating a state-dependent local projections model on inflation on monetary policy shocks in the US. We use the same monetary policy shocks as Jarociński and Karadi (2020), which are obtained by measuring the change in market-implied expectations around FOMC announcements. Due to data availability for the monetary policy surprise variables, in this exercise we focus exclusively on the pre-Covid period. Chart 4 summarises the peak impacts of a 50 basis points rate hike on inflation. These results imply that monetary policy may be significantly more powerful in periods of heightened GSC pressures, as demand shocks lead to a greater response of inflation in such circumstances.  

Chart 4: Peak impact of 50 basis points unexpected monetary policy tightening on inflation


The state contingency of the slope of the Phillips Curve on the state of GSCs has some important implications. First, while GSC pressures arguably accounted for a large share of the inflationary impulse in 2022, it is likely that their inflationary effect will fall in 2023 as global bottlenecks unwind. Accordingly, it is reasonable to expect that the direct effect of supply chain constraints will moderate. Second, our exercise highlights that the strength of the effects of monetary policy can be state-contingent, and will depend on the extent of supply constraints.

Ambrogio Cesa-Bianchi, Marco Pinchetti and Julian Reynolds work in the Bank’s Global Analysis Division and Ed Hall works in the Bank’s International Surveillance Division.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.