Lessons from History: How “Transitory” is Inflation?

By Rob Arnott and Omid Shakernia

“Those who cannot remember the past are condemned to repeat it.”

George Santayana

Bad news: History tells us that once the inflation genie is out of the bottle, it can take far longer to return
to normal levels than most people – including most PhD economists at the Fed! – realize. Consider when
Federal Reserve Chair Paul Volker took office in 1979 and pushed the Fed Funds rate to an unprecedented
20%, 5% above the previous peak inflation rate, the equivalent of Powell embracing double-digit rates
today. Even so, it took two years for this extreme policy intervention to cut inflation to half its peak level,
and over six years to bring inflation to 2%. In a meta-analysis of sixty-seven published studies on global
inflation and monetary policy, Havranek and Ruskan [2013] found that across 198 instances of policy rate
hikes of 1% or more, the average lag until a 1% decrease in inflation for developed economies was between
two and four years.

To quantify how long we can expect before the inflation genie can be put back in the bottle, we studied
all historical cases of inflation above 4%, in 14 OECD developed economy countries 1. Our dataset consists
of inflation rates from January 1970 through September 2022 in each country. 2 We studied cases where
a country’s inflation rate first exceeded each level from 4% to 20%, asking how long one had to wait for
the inflation (1) to fall by half, or (2) to fall below a 3% threshold. For each episode, we measure the time
from first crossing a particular threshold until the first time retreating to the target (either halfway back to
zero, or back to a fixed 3% target). We found 52 instances when inflation rose above 4% level in any of
our 14 OECD economies, tumbling to 6 instances exceeding a 20% inflation level. 3

The results show an interesting pattern. The good news is that, when inflation crosses the 4% threshold, it
usually reverses course; in 32 out of 52 cases, inflation never gets to that next threshold of 6%. We refer

1 Countries in this dataset: Australia, Canada, France, Germany, Great Britain, Italy, Japan, Netherlands, New Zealand, Norway, Spain,
Sweden, Switzerland, United States.
2 We excluded countries that were not developed early in our 52-year history. South Korea, Mexico and Turkey were not developed economies

until recently, and were hardly bastions of monetary discipline at times (including recently for the latter two).
3 For seven of the 52 cases, inflation never got back to 2% before inflation lofted past 4% again, so the graph “Years for Inflation to Shrink

by 50%” is based on 45 cases, while the other is based on 52 cases.

to these cases as “cresting inflation.” The bad news is that these odds flip, at 6% and higher inflation. If
inflation later crosses the next threshold, these are “accelerating inflation” episodes, until we reach a
threshold that is not exceeded. At 6% and higher inflation, cresting inflation is the exception, not the rule.
Indeed, once the 8% threshold is surpassed, as happened this year in the US and much of Europe, inflation
marched to the next threshold, and often well beyond, over 70% of the time. 4 This does not mean that
inflation is destined to move to new highs in the months ahead, only that we dismiss that possibility at our
peril.

When Inflation rises to 4%, it’s often a temporary “blip,” caused by a temporary exogenous shock. Over
60% of the time, it reverts to 2% rather than accelerating to 6% or more.
When inflation crosses 8%, it proceeds to higher levels over 80% of the time.

Source: Research Affiliates, based on data from Bloomberg.

4 We do not show this measure for 20% inflation, as we did not bother looking beyond this threshold.
How Fast Does Inflation Subside?

We are not posing the question, “how fast can inflation subside?” A singular focus on price stability can
rein in inflation almost overnight, as history has shown with German hyperinflation in the 1920s,
Hungarian hyperinflation in the 1940s, and Zimbabwe hyperinflation in the 2000s. In each case, central
bank control of the printing press was rescinded, and the failed currency was ditched in favor of something
that was trusted, typically gold or the then-prevailing reserve currency.

Money is very simple. It’s a medium of exchange to buy or sell labor, goods and services, either
contemporaneously or intertemporally (its “store of value”). Money cannot serve multiple masters, and
yet central bankers seem eager to promote an array of goals, that they are ill-equipped to address: price
stability, full employment, low cost of servicing government debt, disrupt bear markets, and so forth.
Accordingly, we ask, “when money is asked to serve multiple masters, how long – on average – will a
burst of inflation linger?” 5

In our first test, we estimate the “half-life” of high inflation – the amount of time it takes for inflation to
fall by half, from 4% to 2%, from 6% to 3%, and so forth. We found 45 instances when inflation rose
above 4% level, diminishing to 6 instances exceeding a 20% inflation level. It bears mention that, for
some of the cases of 4% (or higher) inflation, the inflation arrives and then recedes, without hitting the
next threshold; for others, it accelerated further – in some instances, a lot further – before receding. 6 These
are “cresting inflation” and “accelerating inflation,” respectively.

The graph below shows the median number of years it takes for inflation to be reduced by half, from the
first time it hit a given level of inflation, with the shaded band representing the middle three quintiles
(leaving out the top and bottom 20% of outcomes), measuring the passage of time before inflation has
been cut in half. The medians for cresting inflation and for accelerating inflation are, of course, very
different, and are shown as lighter lines on the graph. 7

5 We think the central bank consensus, that 2% inflation is ideal, is nonsense. True price stability – zero inflation on average over time – is
easy to achieve. The Byzantine Empire managed it for over 600 years. That said, 2% inflation is reasonably benign, with the currency losing
“only” 75% of its purchasing power in an average human lifetime.
6 Note that for nearly 1 in 5 of these instances (38 out of 205), the level of inflation has still not been reduced by 50% to this day. Those 38

cases are not counted in the analysis below, and so the half-life estimates below should be considered conservative lower bounds.
7 Because of a limited number of cases, when we bifurcate between accelerating and cresting inflation, we smooth these two lines by

consolidating the data for 4%, 6% and 8% inflation, for the 6% plot point, and do the same for all but the end points (4% and 20% inflation)
of the graph.
Consider the leftmost end of the graph. Once inflation has reached the 4% threshold, the gray band
stretches from 1 year to 10 years. This means that 20% of the time it takes a year or less to revert to 2%
inflation or less (half of the 4% threshold). At the other extreme, 20% of the time, it takes 10 years or more
to get back down to 2%! Many readers will be surprised to learn that the median result is a 2½-year wait
before a modestly elevated 4% inflation falls below 2%. All of which invites the question: when inflation
was already crossing 4% in April of 2021 (2% of which was in the prior three months, which worked out
to 8% annualized!), what were Powell and Yellen thinking, in declaring the inflation to be transitory?
Would a median expectation of 2½ years, to revert back to 2% inflation, have been considered transitory?

The medians for cresting inflation and for accelerating inflation are very different. Cases of cresting
inflation dominate the lower reaches of the distribution, while case of accelerating inflation dominate the
top of the distribution.If 4% inflation never makes it to the next threshold of 6%, then this cresting inflation
recedes quickly, with a median time of just 1 year to revert back to 2% or less. But, if it is accelerating
and proceeds to the next threshold, we could be in trouble. It takes a median of 10 years before we see
2% inflation again, because we cannot know ex ante how high inflation will go. At 6%, there’s again a
bifurcation. If inflation crests and recedes, we wait a median of 15 months to see inflation fall in half, to
revert to a target of 3% (no longer 2%, which takes longer still); if it goes on to 8% or more, we wait a
median of nearly 11 years to revert to 3%.

When US inflation crossed 4%, in April of 2021 for the first time since 2008, this study of history might
have supported a declaration that the inflation should be transitory. But that’s only true if we were certain
that inflation was cresting, and would not accelerate further to 6% or more. If inflation did go higher, the
median wait to return to 2% inflation increases ten-fold. Once inflation crossed 6% in October of 2021, if
our crystal ball showed us that this was the peak, the median expectation would have been that we recede
to 3% inflation in about 15 months, and that if it did rise further (as it did), we wouldn’t expect to see 3%
inflation for a median wait of another decade. Once inflation crossed 8% in March of 2022, if we can be
sure that it won’t exceed 10%, we could reasonably expect inflation to recede to 4% in about two years.
That would still be twice the Fed’s target, which we would presumably see in March of 2024. If it lofts to
new highs, the average wait is once again 10 years to recede to 4%.
 If we know that inflation is cresting, 4% and 6% inflation reverts by half (to 2% and 3%) in about a
year. If we do not know this, 6% inflation reverts to 3% in a median of about seven years. We dare not
be complacent about inflation of 6% or higher; we dare not count on a rapid return to normal.

Years for Inflation to Shrink by 50%

16.0
Years Before Inflation Falls by Half

8.0

4.0

2.0

1.0
4% 6% 8% 10% 12% 14% 16% 18% 20%
Starting Inflation
P=20% band Overall Median Cresting Inflation Accelerating Inflation

Source: Research Affiliates, based on data from Bloomberg.

For higher levels of inflation – from when we first cross 6% to when we first cross 20%, we see a slight
hump-shaped response: the median half-life peaks at about 7 years when inflation is between 6% to 8%
(today’s level of inflation!), and falls to about 2½ years for an inflation rate of 20%. We caution readers
to not take too much comfort in the lower half-life at the higher levels of inflation, considering that there
are several instances where inflation has never, to this day, been lowered by half; these instances are not
included in our results. The worst quintile outcomes, across all levels of inflation, requires a wait of
anywhere from 8 to 16 years to lower the inflation by half from the first time it crosses an inflation
threshold.
Few would consider it an applause-worthy “win” to bring 20% inflation back down to 10%. Accordingly,
let’s consider another test in which, after we first cross a starting level anywhere from 4% to 20%, we
declare victory as soon as inflation falls below a fixed 3% threshold, which most of the citizenry would
find tolerable. It should come as no surprise that the median time for bringing 4% inflation down – ever
so modestly – to 3% is brief, about 18 months (which is still longer than many people might expect). But,
after inflation hits a less-benign 6%, the median number of years to cut inflation below 3% soars to 7½
years. From inflation levels of 8% to 20%, the median span required to bring inflation below 3% is
surprisingly flat, requiring anywhere from 9 years to 12 years to do so. Even this may be understated, as
a consequence of the handful of cases that are missing from our data set, because we never did return to
3% inflation, to this day.

Given the recent US inflation rate, above 6% for the past year and above 8% for the past seven months,
history tells us that the median number of years to reduce inflation below 3% is 10 years, with an 20th to
80th percentile range of 6 to 19 years. How many economists – let alone pundits and policy “experts” –
have suggested that we may have elevated inflation for six years, let alone the longer outliers?
 Reverting to 3% inflation, which we view as the upper bound for benign sustained inflation, is easy from
4%, hard from 6%, and very hard from 8% or more. Above 8%, reverting to 3% usually takes 6 to 20
years, with a median of over 10 years.

Years for Inflation to Revert Below 3%

32.0
Years Before Inflation falls back to 3%

16.0

8.0

4.0

2.0

1.0

0.5
4% 6% 8% 10% 12% 14% 16% 18% 20%
Starting Inflation
P=20% band Overall Median Cresting Inflation Accelerating Inflaton

Source: Research Affiliates, based on data from Bloomberg.

When Did the Fed Wake Up?

After Powell retired the term “transitory” in November of 2021, the Fed’s December “dot plot” still
showed a shockingly benign expectation for future Fed Funds rates, hence tacitly for future inflation. 8 The
graph below shows the “dot plot” quarterly, from June 2020 to September 2022. Overlaid on the graph is

8 The “dot plot” shows the views of each of the Fed governors, voting and non-voting members, as to where the Fed Funds rate should be at

yearend, at the next two following yearends, and where it should settle eventually in equilibrium. The median is generally seen as the
consensus of Fed governors, as to likely future Fed policy.
the 12-month inflation through the previous month. For 2022, that inflation rate is (literally) off-the-charts,
at 8% to 8½%.

Fed expectations for inflation were hopelessly behind the curve for over a year, until recently.
Fed expectations for the speed of reverting to 2% inflation levels remain dangerously optimistic.

Fed Projections for the Appropriate Fed Funds Rate, at Yearend 2022

Prior 12-Month
CPI Inflation

Source: Federal Reserve Board and Bureau of Labor Statistics

In June 2020, while we were all still reeling from Covid lockdowns, trailing one-year inflation was 0.1%,
and the Fed Funds rate was much the same (in the 0-25 bps range, which we express as 1/8%). The dot
plot shows that all but one Fed governor believed we would still be at 1/8% at the end of 2022! In
September and December of 2020, inflation had recovered to a still-modest 1%, and the expectations for
yearend 2022 hadn’t budged (except for one lone Fed governor who expected that a modest 25-50 basis
point increase would likely be warranted). No one knew when the Covid mess would end, and many
wondered whether the economy could avoid deflation and depression.

In 2021, the inflation picture changed rapidly. Home prices had already risen 10% in the prior year, 2020,
which was tied for the biggest one-year jump in 15 years. 9 By mid-2021, inflation was already 5%, and
was officially deemed to be “transitory,” even though the BLS measure for “shelter” inflation was miles
behind home price appreciation and rental rates. 10 As history tells us, transitory inflation was a plausible
outcome, but not the median expectation, as long as inflation would go no higher. Which it did. By
yearend, it was 6.6%, five times the year-earlier level. Shockingly, the dot plot still showed that every
single Fed governor believed that the appropriate yearend-2022 Fed Funds rates should be at least 500
basis points lower than the latest 12-month inflation report. Not a single Fed governor expected that Fed
Funds should be even 1% above the yearend-2020 expectations.

2022 is a very different picture. By mid-September, the governors were forecasting a consensus year-end
Fed Funds rate of 4.38%, and were on-track to create a full yield-curve inversion (3-month T-bill rates
higher than 10-year T-bond yields) by yearend. 11 But, inflation remains stubbornly above 8%, and – based
on our analysis (also in Arnott and Harvey [2022]) – likely to remains so through yearend. Cam Harvey’s
pioneering 1986 work on the yield curve suggests that an inversion from 3 months to 10 years, if it lasts
more than a couple of weeks, has a 100% batting average since World War II in predicting recessions,
with no false positives.

Conclusion

“To lose one parent may be regarded as a misfortune; to lose both looks like carelessness.” Oscar Wilde

It was a mistake (Wilde’s “misfortune”) to declare inflation transitory, when it was rising fast, and when
history tells us that, even at 4%, it’s often not. It inflicted serious damage (Wilde’s “carelessness”) to
continue a too-easy policy and the transitory messaging – which, in different terminology, continues to
this day – even as inflation lofted past 6%, then 8%. A cursory glance at history would have told them that

9 2013 saw an identical 10% jump, as a rebound from the GFC housing rout.
10 History, once again, is a useful guide. We find that owners equivalent rent (OER) and rental of primary residence (RPR) are both smoothed
and lagged; they play catch-up on the upside or on the downside, with an average lag of one to two years, and residual catch-up for much
longer. We are seeing that play out in 2022, with OER and RPR soaring, even as home prices and rents are beginning to moderate. See
Arnott and Harvey [2022].
11 The yield curve has gone to full inversion for a few days in October; the November/December rate hikes should make this a more serious

inversion, if current expectations are correct.

“transitory” is possible, but hardly a sensible central expectation, and that their messaging and their policy
response should reflect the relatively high empirical risk of accelerating inflation. Is it possible that
inflation will recede to 4% and then to 2% in the coming year or two? Of course it’s possible. The range
of historical outcomes is vast. But, history says that a fast retreat to 4%, let alone 2%, is unlikely.

We believe that our policy leaders, for both fiscal and monetary policy, should embrace a variant of the
Hippocratic oath, “first, do no harm.” Our fiscal and monetary policy has done far more harm than good
in recent years. We believe that, as George Box famously observed, “all models are wrong, some models
are useful.” Our leaders seem to believe that their economic models are reality, and seem eager to dismiss
reality that does not fit the models. We believe that scientific method demands that we welcome well-
reasoned contrary opinions and alternative perspectives, and always seek to test our ideas against the
principle of “falsifiability.” Indeed, a correct application of scientific method requires us to test our
models, not with an eye to prove them correct, but with a goal of breaking our models and finding their
weaknesses. This is the norm in the hard sciences, but not in the soft sciences (if they can be called
sciences!).

We perceive a resistance to alternative views, in both fiscal and monetary circles, a desire to create an
echo-chamber of similar views, and a reluctance to learn from past mistakes. We are fools if we allow our
hopes for a rapid dissipation of inflation to become our central expectations.
References

Arnott, R. and C. Harvey. 2022. “No Excuses: Plan Now for Recession.” Research Affiliates.

Havranek, T. and M. Rusnak. 2013. “Transmission Lags of Monetary Policy: A Meta-Analysis”,

International Journal of Central Banking.