Inflation is back. Annual CPI inflation for the US hit a 40-year high of 8.5% in March 2022. This increase continues a pattern that started in May 2021, when annual inflation breached 5% for the first time in 30 years. It has been rising steadily since. While the high level of inflation is certainly a concern, a key policy issue is whether its current surge represents a transitory or a persistent phenomenon. Describing inflation in terms of its permanence (or lack thereof) is not simply an academic characterisation. Until very recently, the Federal Reserve relied on such a strategy to communicate staff views on the (potentially transitory) nature of current inflationary pressures.1 Against this backdrop, conventional wisdom holds that it takes 12–18 months for the effects of monetary policy to percolate. This suggests central banks should act only if they perceive that inflationary pressures are likely to be persistent.
A salient feature of the current inflation episode has been the dynamics of goods inflation. Figure 1 plots annualised goods and services inflation in the US from 1960Q1 to 2021Q4. While both goods and services inflation have increased over this period, the rise in goods inflation was noticeably more pronounced. This can be largely attributed to two factors. First, the disruption of global supply chains and labour shortages witnessed since the onset of the COVID-19 pandemic have disproportionally affected the shipping and delivery of consumer goods around the world (Coibion et al. 2020). Second, since early 2021 such supply-side bottlenecks were met by an increasing demand for goods fuelled by excess savings, generous fiscal policy, and certain services-based expenditures simply not being an option due to lockdown policies (D’Acunto and Weber 2022).
Figure 1 Annual inflation for US goods and services
Note: The shaded areas denote NBER recession dates.
Although the current high inflation episode has brought into focus the behaviour of goods inflation, Figure 1 also presents less appreciated facts. Specifically, since the 1990s, goods and services inflation displayed somewhat different patterns with goods inflation being more volatile and lower (on average) than services inflation.
If we accept the initial premise that monetary policy should act against the persistent component of inflation, a natural issue is how one should interpret the recent behaviour of goods and services inflation in terms of their implications for aggregate inflation over an extended horizon. In our recent work (Eo et al. 2022), we set out to understand trend inflation (i.e. the rate of inflation that would be expected to prevail if there were no temporary factors) from the perspective of an empirical two-sector model of inflation. We construct trend inflation for both goods and services by adopting statistical methods that strip out the transitory (or noise) component from each sector-specific inflation rate.
Results
Figure 2 presents the estimated trends for goods and services inflation alongside the corresponding inflation rates for these two sectors. A striking result is that before the early-1990s, the estimated trend in each sector mimicked the actual sector-specific inflation. However, since the 1990s, while goods inflation has remained volatile, its trend has become quite flat. Accordingly, our model interprets much of the fluctuation in goods inflation as transitory. In contrast, a quick inspection of the estimated trend in the services sector suggests that the opposite has happened with almost all variation in services inflation being attributed to its persistent (rather than transitory) component.
Figure 2 Estimated sector-specific trend inflation
Notes: The solid line is sector-specific annualized quarter-on-quarter inflation rates and the dotted lines represent our posterior median estimate of the sector-specific trend with its associated 67% credible set. All inflation rates are annualised. The shaded areas denote NBER recession dates.
Given that goods and services represent the entire consumption basket, the variance of aggregate trend inflation can be formally decomposed into the variance of each of these two sectors plus the correlation between them. Figure 3 presents such a decomposition. We note two striking features. First, aggregate trend inflation volatility was high during the Great Inflation of the 1970s, but it has declined drastically since the early-1990s. This result is also documented most prominently by Stock and Watson (2007). Second, while both sectors and their comovement used to contribute to the volatility of aggregate trend inflation in the 1970s, this metric has been dominated by the services sector since the early-1990s. Put differently, inflation in the goods sector over the past three decades has contributed very little or nothing to variation in the persistent component of overall inflation.
Figure 3 Variance decomposition of trend inflation
Notes: Trend inflation is in units of annualised quarter-on-quarter inflation. The shaded areas denote NBER recession dates.
How should we view the recent inflation episode?
We turn to understanding the recent high inflation within the context of our model. Figure 4 plots our estimated trend inflation alongside headline inflation. Our point estimate of trend inflation at the end of 2021 (the last data point) is 2.2%. Our sectoral split provides a clear take on why trend inflation has not accelerated (yet) as much as headline inflation. Since the current high inflation episode is largely manifested in the goods sector and our model interprets goods inflation as being predominantly transitory, much of the increase in overall inflation is thus regarded as transitory. Consequently, aggregate trend inflation remains somewhat muted.
Moreover, while the current high inflation inevitably evokes comparison to the Great Inflation of the 1970s (e.g. Ha et al. 2022), our empirical exercise offers other perspectives on how to view comparisons relative to the 1970s. Trend inflation estimates for the 1970s still sit at a much higher level than our current estimates. This suggests that inflation expectations remain anchored, as Cascaldi-Garcia et al. (2022) argue. Moreover, our model suggests more broad-based inflation during the Great Inflation of the 1970s. Figure 3 shows that the comovement term between the goods and services sectors used to contribute to the variation of aggregate trend inflation. This comovement term now plays a negligible role in our analysis of what drives variation in aggregate trend inflation at the end of 2021. Consequently, the current high inflation episode, at least at this point, does not seem to echo the persistent broad-based inflation across sectors observed during the 1970s (see also Borio et al. 2022).
It is, however, important to describe some nuances associated with our findings by offering two specific points. First, the degree of uncertainty associated with our trend inflation estimate of 2.2% at the end of 2021 is much larger than usual.2 Figure 4 reveals that the 67% credible interval for the estimate of trend inflation at the end of 2021 is between 1% and 4%, which entertains the possibility that aggregate trend inflation may indeed be much higher relative to the Federal Reserve’s longer-term average inflation target of 2%. The credible interval also suggests that the risk to our trend inflation estimate of 2.2% is very much skewed to the upside. Therefore, from a risk management perspective, one should not necessarily unequivocally accept that the current high inflation is transitory.
Figure 4 Estimated aggregate trend inflation
Notes: The solid line denotes annualized quarter-on-quarter Personal consumption expenditures inflation together. The dotted lines denote our posterior median estimate of aggregate trend inflation with the associated 67% credible interval. All inflation rates are annualized. The shaded areas denote NBER recession dates.
Second, according to our model, even if the point estimate for trend inflation remains at a level that the monetary authority may be comfortable with, trend inflation has been gradually rising and is associated with moderately higher levels of trend inflation not only in the goods but also (and arguably more importantly) in the services sector. The latter may be consistent with the idea that wages have been increasing, and if so, we know from our work that movement in services inflation feeds into trend inflation. Considering the years between the Great Recession and the COVID-19 pandemic saw the US, and much of the world, experience consistently lower-than-target inflation, some may welcome recent increases in trend services inflation as it could shift trend inflation into a range where monetary policymakers would like it to be. That said, our work also indicates that if services inflation continues to increase, it is likely that trend inflation will increase beyond its current modest estimate to levels that are perhaps more concerning.
Conclusion
We estimate a model of trend inflation of the goods and services sectors to understand how both sectors contribute to the persistent component of inflation. Our main finding is that, since the early-1990s, goods inflation has played a negligible role in driving trend inflation. This contrasts with the Great Inflation of the 1970s, when goods inflation was a larger contributor to variation in aggregate inflation. Our results were sustained throughout the COVID-19 pandemic. Therefore, the estimated trend inflation is at a relatively modest 2.2% at the end of 2021 and we believe that the high goods inflation that we are currently experiencing is likely to be transitory. The latter assessment is largely grounded on much of the recent high inflation (still) being concentrated in the traditionally high-frequency goods sector. Nonetheless, we should note that the model finds greater-than-usual uncertainty associated with the estimate of trend inflation during the COVID-19 pandemic recovery. The latter point seems important to reiterate since the risk of trend inflation is skewed towards the upside, which remains a relevant consideration from a risk management perspective.
Authors' note: The views expressed are our own and do not necessarily reflect those of the Bank of Canada.
References
Borio, C, P Disyatat, D Xia and E Zakrajšek (2022), “Looking Under the Hood: The Two Faces of Inflation”, VoxEU.org, 24 January.
Coibion, O, Y Gorodnichenko and M Weber (2020), “Labor Markets During the COVID-19 Crisis: A Preliminary View”, NBER Working Paper No. 27017.
D’Acunto, F and M Weber (2022), “Rising Inflation is Worrisome. But Not for the Reasons You Think”, VoxEU.org, 04 January.
Cascaldi-Garcia D, F Loria and D López-Salido (2022), “Is Trend Inflation at Risk of Becoming Unanchored? The Role of Inflation Expectations”, FEDS Notes, 31 March.
Eo, Y, L Uzeda and B Wong (2022), “Understanding Trend Inflation through the Lens of the Goods and Services Sectors”, CAMA Working Papers 2022-28.
Ha J, M Ayhan Kose and F Ohnsorge (2022), “Today’s Inflation and the Great Inflation of the 1970s: Similarities and Differences”, VoxEU.org, 30 March.
Stock, J H and M W Watson (2007), “Why Has U.S. Inflation Become Harder to Forecast?”, Journal of Money, Credit and Banking 39(s1): 3-33.
Endnotes
1 See, for example, the FOMC statement on 22 September 2021 at https://www.federalreserve.gov/newsevents/pressreleases/monetary20210922a.htm
2 Our current trend inflation estimate and its uncertainty are largely consistent with those reported in Cascaldi-Garcia et al. (2022).
