Whether post-pandemic inflation can be explained by demand-side versus supply-side factors is critical for setting monetary policy. Nevertheless, there is widespread disagreement about the underlying sources of recent inflation. This column argues that part of the disagreement is due to differences in (i) the variables being studied (inflation or interest rates), (ii) the countries analysed, and (iii) whether the focus is on domestic or global shocks. It finds that demand shocks were the primary driver of recent increases in interest rates, while supply shocks were the primary driver of post-pandemic inflation. However, there are important heterogeneities across countries, and in the contribution of domestic versus global shocks.
How much of the post-pandemic inflation was driven by demand versus supply? This question has triggered extensive debate over the last few years – a debate that may last longer than the inflation behind it. This column explains how some of the seemingly discordant arguments can be mutually consistent.
This debate is not just of academic curiosity, but critically important for setting monetary policy. Monetary policy should respond more forcefully to demand-driven shocks, but can, in certain circumstances, look through some supply-driven shocks. 1 If central banks initially overestimated the role of supply shocks in driving post-pandemic inflation, this could have delayed the necessary tightening of monetary policy and even contributed to subsequent inflation.
Recent research attempting to parse out these relationships yields seemingly discordant conclusions. 2 In a recent study, Bernanke and Blanchard (2024) highlight the role of price shocks (primarily for energy and food) in driving post-pandemic inflation. Many other studies focus on the role of supply shocks, a conclusion supported by President Lagarde (2024) in her opening comments at the latest ECB forum in Sintra.
At the same event, however, Giannone and Primiceri (2024) argue the opposite: that demand shocks were more important than supply in driving the post-pandemic inflation in both the euro area and the US. Also at the same forum, Forbes et al. (2024) show that global shocks were the primary drivers of interest rates for a group of five advanced economies since 2020, and although supply shocks played a larger role than in any other historical episode since 1970, demand shocks were still more important on average.
Our analysis – and a new extension reported below – shows that some of these seeming disagreements simply reflect differences in what is being compared: apples versus oranges, or even apples versus broccoli. There are important differences in the methodology, data, and samples in the studies cited above (and much larger literature on these topics), but our analysis provides a simple example of how the relative importance of demand and supply can nearly flip based on whether the analysis focuses on the drivers of inflation (for which supply is more important) versus interest rates (for which demand is more important). The relative importance of shocks can also change based on what economy is being examined, with a much greater role for supply shocks in the euro area than the US. In addition, it is critical to consider whether global or domestic forces drive the underlying shocks, with a greater role for global shocks in the euro area than the other economies in our sample. We will examine each of these in turn.
What variable is being explained?
Any debate on the role of demand versus supply (or other disturbances) needs to clarify exactly what variable is being explained. A majority of the studies in the monetary policy literature decompose the variation in inflation, but some others decompose the variation in interest rates, growth, or other variables.Although many of these variables move together over the business cycle – there are periods when they do not – especially if monetary policy looks through certain types of shocks.
To show the importance of making these distinctions in the context of inflation and interest rates, we use the Factor Augmented VAR (FAVAR) model presented in Forbes et al. (2024). The model decomposes the drivers of the variances of inflation and interest rates into four global shocks (oil prices, supply, demand, and monetary policy) and three domestic shocks (supply, demand, and monetary policy). Figure 1 shows these decompositions over the post-pandemic period of 2020-2023 averaged across five advanced economies (Canada, the euro area, Japan, the UK, and the US) – referred to as the G5.
Demand and supply shocks play different roles based on the variable under consideration. Demand shocks (both global and domestic) contributed 44% to the variation in interest rates over 2020-23, as compared to only 29% for supply shocks (both global and domestic, as well as oil prices). These shares are basically reversed for the variation in inflation, for which demand shocks contributed 28%, and supply shocks 49%. If monetary policy shocks are included as demand shocks, the differential is even larger for interest rates (with the contribution of demand now jumping to about 2½ times that of supply), while the overall contribution of demand to inflation is roughly equal to that of supply.
Figure 1 Average variance decomposition of interest rates and inflation in G5 economies
Source: Based on data and model presented in Forbes et al. (2024).
Notes: Forecast error variance decompositions (in percent) of interest rates and CPI inflation over the 40-month forecasting horizon, based on country-specific FAVAR models that consist of four global variables (inflation, output growth, policy rates, and oil prices) and three domestic variables (inflation, output growth, and policy rates). Results are averages across G5 economies. Interest rates are shadow interest rates, as estimated in Krippner (2013). These are the shortest maturity rate based on the shadow yield curve and are essentially the policy interest rate in ‘non-lower’ bound or unconventional monetary policy environments.
What economy is being analysed?
Any debate on the role of demand versus supply in driving the variation in any variable should clarify exactly what entity is being analysed. The results above are averages for the G5, but the drivers could vary meaningfully across economies based on their characteristics and vulnerability to specific shocks.
Figure 2 shows this heterogeneity across the economies used to construct the G-5 averages in Figure 1. It uses the same FAVAR decomposition as above. The role of supply and demand shocks varies across these five advanced economies – particularly for the euro area.
More specifically, for the US, 51% of the variation in policy interest rates is explained by demand and only 25% by supply; in the euro area the roles are almost reversed, with only 28% of the variation explained by demand and 42% by supply. The decompositions in the other economies are closer to that in the US, with demand estimated to play a much larger role than supply.
Figure 2 Variance decomposition of interest rates and inflation by G5 economy
Source: Based on data and model presented in Forbes et al. (2024).
Notes: Forecast error variance decompositions (in percent) of shadow interest rates and CPI inflation over the 40-month forecasting horizon, based on country-specific FAVAR models that consist of seven variables listed above.
In the inflation decompositions, supply shocks play a greater role than demand (excluding monetary policy shocks) for each economy, but they continue to play the largest role for the euro area (albeit only slightly more than in Japan and the UK). More specifically, in the euro area, supply shocks drove a majority of the variation in inflation over 2020-23 (53%), while demand shocks contributed only 22%. In contrast, supply shocks explain only slightly more of the inflation variation than demand shocks in the US (42% versus 38%, respectively).
These differences are intuitive. The euro area was more reliant on oil, natural gas, and food from Russia and Ukraine, while the US is not only a net exporter of both food and energy, but also had a greater boost to demand from multiple large fiscal packages during this period. The decompositions of inflation in the other G5 economies were between that of the euro area and the US, albeit with a greater role of supply shocks (around 50%) as compared to demand shocks (26-28%), and substantial heterogeneity across countries (such as the much larger role of domestic supply in Japan and the UK).
How much of the shocks originated domestically versus globally?
Our findings also shed light on an important issue that is often not discussed in the debate on demand versus supply, and often not even decomposed in most models: the role of global versus domestic shocks. Figure 3 repeats the analysis from Figure 2, but now highlights the differences between the global and domestic shocks. The global shocks explained a majority of the variation in both interest rates and inflation on average, and in most of the individual economies over 2020-23, particularly in the euro area. The only exceptions were Japan (where domestic shocks explained a majority of the variation in both variables) and the UK (where domestic shocks accounted for a majority of the variation in inflation).
Figure 3 Global and domestic shocks: Variance decomposition of interest rates and inflation by economy
Source: Based on data and model presented in Forbes et al. (2024).
Notes: Forecast error variance decompositions (in percent) of shadow interest rates and CPI inflation over the 40-month forecasting horizon, based on country-specific FAVAR models that consist of seven variables listed above. Global shocks include shocks to supply, demand, monetary policy and oil prices, while domestic shocks include shocks to supply, demand and monetary policy.
The pronounced role of global shocks is striking, but also reflects a continuation of a longer-term trend; since the early 1970s, global shocks have played a more and more prominent role in explaining the variation in interest rates, inflation, and economic activity. While central banks will continue to set monetary policy to meet domestic mandates, this will increasingly be in response to global events. A global shock, however, does not necessarily imply a supply shock, and Figures 1 and 2 suggest it more often has a larger global demand component.
Just like any model, the framework and shock decompositions used in this analysis are simplifications of the real world. They do not capture the non-linearities that are likely important. Like some VAR models, they have large standard errors, and results can vary based on the number of variables, types of shocks, and restrictions to identify the shocks.
Nonetheless, a simple application of our model suggests that the following statements are all consistent for the G5 economies we study:
- Demand shocks were the primary driver of post-pandemic increases in interest rates on average.
- Supply shocks were the primary drivers of post-pandemic inflation on average.
- Supply shocks played a greater role in both the post-pandemic inflation and interest rates in the euro area than in the other economies.
- Global shocks explain a majority of the variation in the post-pandemic period for both interest rates and inflation on average and in Canada, the euro area, and the US.
The greater role for demand shocks in explaining the variation in interest rates (on average), and supply shocks in driving the variation in inflation, can be interpreted as consistent with the idea that monetary policy should respond more forcefully to the positive demand shocks behind inflation, but can have a more muted response to supply shocks.
Clarifying these points will hopefully provide common ground for debating the many other unresolved questions about post-pandemic inflation and corresponding policy responses.
Source : VOXeu
