January 29, 2022
With inflation remaining stubbornly high for longer than I, and many others, expected, I want to take another stab at the argument of the inflation hawks. As a jumping off point, I will use the argument put forward by Larry Summers and Jason Furman, probably the two most prominent and coherent economists arguing that we have underestimated the risks of persistently high inflation.
There are three main components to the Summers-Furman (SF) argument. (Their arguments are not identical, so I’m being a bit unfair to both in trying to mash them together.) The first is that the Biden administration provided excessive stimulus to the economy with the American Recovery Act (ARA) passed by Congress last February. They argue that demand for goods and services far exceeded the economy’s ability to supply them, leading to a sharp uptick in the rate of inflation.
Second, the SF position is that this jump in inflation has unmoored inflationary expectations. While households and businesses had long come to expect low and stable inflation, the surge in inflation we saw in the last year has changed people’s expectations. Just to be clear, this is more Summers’s concern than Furman’s. Also, he views this as a serious risk, but not a necessary outcome from the current situation.
If expectations become unmoored, inflation will be self-perpetuating. Workers will demand higher wages in the expectation that inflation will remain high. Employers who share this expectation will be prepared to pay higher wages, which they will then pass on in higher prices. This will lead to a wage-price spiral like what we saw in the 1970s.
The third key component of the SF argument is that we continue to operate the economy above its potential, adding to inflationary pressures. To prevent further acceleration of inflation we will need to reduce demand in the economy, either with aggressive interest rate hikes from the Fed or contractionary fiscal policy, or some combination.
I’ll take each of these in turn.
Did the ARA Spur Inflation?
The answer to this one is obviously yes. If we had a less ambitious stimulus package, the current inflation rate would almost certainly be lower. But we did get a lot from the ARA, along with the previous CARES Acts passed by Congress in 2020.
These pandemic packages shielded the bulk of the population from the economic effects of the pandemic and the shutdowns. In fact, large portions of the population saw their economic prospects actually improve during the pandemic. J.P. Morgan reports that most households have more money in their bank accounts now than they did before the pandemic, with the lowest income households being the biggest gainers in percentage terms.
The expansion of the child tax credit cut the child poverty rate in half and would likely lead to long-term gains for children in low- and moderate-income families, if left in place. And, the ARA provided a huge boost to the labor market. The 3.9 percent December unemployment rate is close to full employment. The biggest gainers from this tight labor market are lower paid workers who are seeing rapid wage increases. They feel the freedom to quit their jobs and seek out new ones in unprecedented numbers.
While it is undeniable that there were major benefits from the ARA, the question is what price did we pay in higher inflation. The answer is not entirely clear. Inflation has jumped pretty much everywhere in 2021 as the world economy reopened. The 7.0 percent year-over-year rate in the Consumer Price Index (CPI) is higher than for other countries, but not hugely so. The rise in Canada was 4.8 percent, in Germany 5.3 percent, and in the United Kingdom 5.4 percent.
The fact that inflation rose almost everywhere indicates that the United States would have seen a jump in its inflation rate even without the aggressive stimulus in the ARA. The rise is attributable to supply disruptions associated with the rapid reopening from pandemic shutdowns. The price of a wide range of commodities soared in 2021, although in many cases, they have recently fallen back to pre-pandemic levels.
In addition, we have seen a rise in the price of many manufactured products due to supply chain disruptions. This is also partly a result of the worldwide reopening and a massive shift in consumer demand from services to goods. People who couldn’t or wouldn’t go to restaurants or travel because of the pandemic instead bought cars, appliances, and other goods. The flood in demand was more than our supply chains could deliver. (Furman points out this shift to goods was much larger in the US than elsewhere, which he attributes to the size of the stimulus.) As a result, prices of a wide range of products soared.
While this happened everywhere, there seems to have been less of an impact in other countries. Jason Furman compared the core harmonized CPI in the eurozone with the core harmonized CPI over the last two years. He finds a jump of roughly two and a half percentage points in the average annual inflation rate over the last two years in this core index for the United States, compared to roughly half a percentage point for the eurozone. This suggests that the inflation from reopening in the US was far more than in Europe.
This view is complicated somewhat by the issue with new and used vehicles. These two components added 1.5 percentage points to the overall inflation rate in 2021 and 1.9 percentage points to the core rate. (In 2020, the vehicle contribution to inflation was 0.3 percentage points overall and 0.4 percentage points in the core.) This means that if we pulled out vehicles, most of the gap in the rise inflation rates would disappear.
The cause for the huge jump in vehicle prices is primarily a shortage of semiconductors due to a fire in a Japanese factory, not excessive demand. We would likely have seen a sharp jump in vehicle prices even without any major stimulus from the ARA.
Of course, pulling out vehicles does not make for an entirely apples-to-apples comparison. The European Union countries buy cars too, although their weight in the index is likely far lower than in the US. One important item to note is that the CPI uses a gross rather than net measure for used car sales to determine their weight in the index. The difference is the money that is paid to households for selling their used cars.
As a result, the weight of used vehicles is far higher in the CPI than in the personal consumption expenditure deflator (PCE), which uses the net measure. In December of 2021, the CPI weight was 3.42 percent. By contrast, used vehicles had a weight of 1.65 percent in the PCE for November, the most recent month available.
Ordinarily, this difference would not matter much, but when used vehicles are rising at an annual rate of 37.3 percent, it matters. The use of the gross sales weighting added 0.4 percentage point to the overall CPI in 2021 and 0.5 percentage point to the core measure.
The takeaway is that taking out vehicles would pull out some of the difference in the jump in core inflation rates that Furman shows, but not all of it. The rise in vehicle prices added roughly 1.2 percentage points to the average core inflation rate over the last two years. If we say that the weight in the EU index is half as large, then this would explain 0.6 percentage points of the gap, leaving roughly 1.4 percentage points to be explained by other factors. With the core CPI accounting for just under 80 percent of the overall index, this would imply that roughly 1.1 percentage points of the higher inflation rate in the US is attributable to factors that are not common to the EU and the US.
To sum up, it’s clear that we would have seen a jump in the inflation rate in 2021 due to the economy reopening, even without the boost from the ARA. The data from the EU, where most countries did not have a comparable stimulus package, implies that the increase in the inflation rate would have been 1.0 to 2.0 percentage points less if we did not have a major stimulus/recovery package in 2021.
Far more important than the causes of the inflation in 2021 is the question of its impact on expectations of future inflation. If the economy is to see a 1970s wage-price spiral, it would mean that people had come to expect that high inflation will persist, rather than being just a one-time jump. This does not seem to be the case.
The breakeven inflation rates for inflation-indexed bonds and standard Treasury bonds have risen only modestly. As of January 26, the breakeven inflation rate for the 10-year Treasury bonds was 2.4 percent. This is roughly 0.2 percentage points higher than the peaks in 2018, a period when few people were worried about runaway inflation. Given the differences between the CPI and core PCE that the Federal Reserve Board targets, it is also roughly consistent with the Fed’s long-run 2.0 percent average inflation target.
The breakeven rate for 5-year Treasury bonds was 2.78 percent, this compares to 2018 peaks of 2.15 percent. The higher breakeven rate for the 5-year Treasury bonds is primarily due to the high current inflation rate. However, it is very much consistent with the expectation that inflation will quickly be falling back to the rates we had been seeing prior to the pandemic.
These breakeven rates are very important for the argument that expectations of inflation have become unmoored since the financial markets are daily giving us a measure of what investors expect. Financial markets can, of course, be wrong, as was the case with the stock bubble in the 1990s and the housing bubbles in the 2000s, but the question here is what is expected, not what the future will actually hold.
It is hard to believe that the people investing in financial markets hold qualitatively different views about inflation than the people running businesses, who are making decisions on wage increases and prices. There is a huge amount of overlap in these groups, and they are in constant contact. If expectations of inflation had become unmoored, we should be seeing more evidence of this fact in the bond market.
In this respect, it’s worth noting that the breakeven rates are not even moving in the right direction for this view. The 5-year breakeven rate peaked at almost 3.2 percent in mid-November and has since then trended downward. The breakeven rate on the 10-year Treasury peaked at over 2.7 percent at roughly the same time. This means that the additional data on inflation over the last two months has not contributed to rising expectations of inflation.
There is an important qualification to this trend in inflation expectations. The Fed has begun to take a notably more aggressive stance towards inflation over this period, with a high probability of multiple interest rates hikes in 2022. This change in stance likely helped to reduce inflationary expectations, but it still implies that financial markets are confident that, with the expected course of Fed actions, inflation will remain contained.
There is one other point worth noting on the expectations issue. In the 1970s, the dollar fell in value against the currencies of our trading partners. In a simple story, where everything else is held equal, we would expect the value of the dollar to fall if our inflation rate exceeds the inflation rate of our trading partners.
The idea is that if inflation is 10 percentage points higher in the US over the next five years than in the eurozone, then we would expect the dollar to fall by roughly 10 percent against the euro to maintain the relative price differentials between goods produced in the US and Europe. The real world is, of course, much more complicated, and there are large divergences in currency values from what would be needed to maintain this sort of purchasing power parity.
However, it is worth noting that the value of the dollar has risen sharply over the last year, rising more than 6.0 percent against the euro. This rise is hardly conclusive, but it does not indicate investors expect inflation in the US to far exceed inflation in the eurozone over the near-term future.
Are We Above Potential GDP?
The final key issue is whether the economy is currently operating above its potential and therefore, likely to be subject to additional inflationary pressures going forward, rather than just facing the risk that the inflation from 2021 is being locked in due to expectations. The basic argument for being above potential is that GDP in the fourth quarter was 3.1 percent above the level in the fourth quarter of 2019, even though we have 2.9 million fewer people working. If we were producing near the economy’s potential in 2019, then we must be above it now.
A closer look shows a more ambiguous picture. While employment is down by 1.8 percent from the fourth quarter of 2019, hours worked have fallen much less. The index of aggregate hours from the establishment survey was down just 0.7 percent from its level in the fourth quarter of 2019. The reason for the difference is an increase in the length of the average workweek from 34.3 hours in the fourth quarter of 2019 to 34.7 hours in the fourth quarter of last year.
It could be argued that this increase in average hours is not sustainable, that people will come to resent working extra hours and stop doing it. However, there is a big factor arguing in the opposite direction. Spencer Hill, in an analysis for Goldman Sachs, calculates that the increase in people working from home is saving 600 million hours a month in commuting time. This translates into an average per worker savings in commuting time of 3.8 hours a month, or 0.9 hours per week. The rise of 0.4 hours in the length of the average workweek comes to less than half of this savings.
There are huge differences in hours and work experiences across the workforce, but it hardly seems implausible that a worker who formerly spent 8 to 10 hours a week commuting would be entirely comfortable putting in 4 to 5 more hours each week working at home than they did previously at the office. The extent to which time saved commuting translates into increased hours of work remains to be seen (as opposed to more leisure), but presumably it is not zero.
We also need to make an adjustment to the hours index for the increase in the number of people who report being self-employed and therefore are not counted in the establishment data. This figure was almost 400,000 higher (combining incorporated and non-incorporated self-employed) in the fourth quarter of 2021 than the fourth quarter of 2019. If we assume the self-employed put in the same number of hours on average as payroll employees, this reduces the drop in total hours over these two years to just 0.4 percent.
With GDP 3.1 percent higher than in the fourth quarter of 2019 and hours worked 0.4 percent lower, the implied increase in productivity is 3.5 percent. That implies annual productivity growth of a bit less than 1.8 percent. This is higher than the 1.5 percent annual rate in the three years prior to the pandemic, but not hugely so.
It is also possible to identify efficiency gains associated with the pandemic that could account for more rapid productivity growth in the last two years. At the top of this list is business travel. The money spent on business travel in the United States fell from $291 billion in 2019 to $131 billion in 2020, and then rebounded slightly to $157 billion in 2021.
Business travel is, in effect, an expense of doing business, like the steel used to construct an office building or the energy used to power a factory. If companies can produce the same output with less business travel, it is effectively an increase in productivity, just as if they needed less steel to put up a building or less energy to operate their factories.
While it may be the case that businesses are actually less productive as a result of the decline in business travel, it is reasonable to believe that this is not the case. As it stands, we are producing a slightly higher level of GDP with much less business travel. (It’s possible there will be some long-term effect where we see smaller productivity gains in future years because of less person-to-person contact, but we’ll have to hold off in assessing that one.)
Anyhow, it might go against our economist instincts to think that companies would needlessly waste money sending their employees flying around the country and the world, but it is at least possible that this is the case. If we treat the reduction in travel as eliminating waste, then we went from spending 1.5 percent of GDP on business travel to 0.8 percent of GDP in 2021, implying a gain in productivity from this pandemic induced innovation of 0.7 percentage points, equivalent to 0.4 percentage points of annual growth over the last two years. This change alone could fill the gap between the implied productivity growth of the last two years and the average rate for the three years prior to the pandemic.
While there are big question marks here on the extent to which longer hours can be maintained and the impact of reduced business travel on longer term productivity, it is at least plausible that these factors can allow for the increase in output we have seen over the last two years without overburdening the economy.
It is also worth mentioning that we will likely see further gains in labor supply if we can bring the pandemic under control. In December, 1.1 million people reported that they were not working or looking for work because of the pandemic. They were either sick with COVID-19, caring for a family member who was sick, or worried about catching the disease. If the pandemic is brought under control, most of these people will presumably come back into the labor market. We also see many people unable to work for part of the month, because they had COVID-19 or had to care for family members who were sick.
If we can bring the pandemic under control in the months ahead, it should lead to a substantial expansion in the labor supply. It will also be good for people’s health and lives.
The Path Forward: Back to Normal
While it is difficult to know when the supply chain problems will ease, it is not hard to identify their impact. The Bureau of Labor Statistics reports that the price index for final demand for transportation and warehousing services rose 16.6 percent last year. This is a cost that gets factored into the price of just about everything.
One example that shows the impact clearly is apparel. The index for apparel prices in the CPI rose 5.8 percent last year. The overwhelming majority of our apparel is imported, most of it from developing countries like China and Bangladesh. The index for the price of imports of apparel rose by 1.5 percent over the last year.
Presumably, most of the 4.3 percentage gap between the CPI index and the import price index is explained by higher transportation costs. If we can expect that at some point in 2022 that these supply chain problems will be overcome, then not only will apparel prices stop rising, but much of their increase in the last year will be reversed.
That should not sound far-fetched. We already have a great example where we saw exactly this sort of story. Television prices fell 6.5 percent from August to December after rising 10.2 percent from March to August. Prior to their run-up in the spring and summer, the index for television prices had been consistently falling for decades.
It seems plausible that there will be many other items, most notably new and used vehicles, where we will see a similar price trajectory to what we saw with televisions. The price increases of the last year may not be fully reversed, but it is more likely that the price of many of these items will be going down in 2022 rather than rising further.
Clearly, much hinges on getting this story right. No one would want to see the sort of wage-price spiral that we saw in the 1970s. But, we also don’t want to see anything like the surge in unemployment from the 1981–82 recession, especially if it’s intended to combat an inflation problem that does not exist.
 Jason Furman and Larry Summers gave me very useful comments on an earlier draft of this post.
 The Bureau of Labor Statistics article on the harmonized index just gives an aggregate weight for transportation. It appears that the harmonized index applies the gross measure for the United States, although it is not clear from the article which weight is used since the transportation index doesn’t match up with the category in the CPI news release.
 Goldman Sachs, 2022. “U.S. Economics Analyst: Productivity Gains Will Outlast the Pandemic,” January 16,2022.
 To make this entirely comparable, we would need to also take account of increased spending on Zoom and other services that were needed as a result of the increase in people working from home.