(AGENPARL) – mer 01 marzo 2023 Si invia in allegato Monetary policy and the return of inflation il testo dell’intervento che il Governatore Ignazio Visco tiene oggi presso la Frankfurt School of Finance & Management.
Saluti,
Divisione Relazioni con i media
Servizio Comunicazione
Banca d’Italia
Prima di stampare, pensa verde
Think green before you print

Testo Allegato: After at least three decades of moderate consumer price changes in advanced countries,
in�ation has returned to levels that are severely a�ecting the lives of all citizens (�g. 1). In this
note, I will discuss �ve key issues that often take centre stage in the current macroeconomic
debate. The �rst concerns the nature of the current in�ation and its di�erent sources in the
two main advanced economies. The second is the claim that, in the euro area, monetary
policy reacted too late to the acceleration in consumer prices. The third, closely related to
this, focuses on whether or not the ECB made policy mistakes or forecasting errors and,
if so, why these occurred. The fourth considers the supposed ine�ectiveness of monetary
policy and the channels through which it operates. The last issue tackles monetary policy
prospects and asks whether, in the �ght against in�ation, the ECB should prefer to run the
Even though in�ation is currently a�ecting many economies in an apparently similar
manner, its underlying sources are di�erent across countries, especially so if we compare
the United States, where demand factors have been crucial in triggering the acceleration
First, while �scal policies were expansionary everywhere during the acute phase of the
Covid-19 pandemic, in the United States they were especially bold: the public debt-to-GDP
ratio rose by 25 percentage points in 2020-21, to over 130 per cent. In the euro area,
Monetary
return
of
charts*
This note mostly follows the Bishnodat Persaud Lecture that I gave on 11 February 2023 at the University of Warwick,
during the Warwick Economics Summit. I wish to thank Jens Weidmann for the kind invitation to deliver this
presentation at the Frankfurt School of Finance & Management and Rebecca Kelly, Pietro Rizza, Massimo Sbracia and
instead, the increase was limited to 15 percentage points, to slightly less than 100 per
cent, despite a much deeper decline of nominal GDP in 2020 and a slower recovery in
2021. The exceptional support provided to US households is particularly evident when
comparing the dynamics of GDP and disposable income (�g. 2): in 2020, just as the
former recorded its sharpest collapse in real terms in the entire post-World War II period
(-3.4 per cent), real disposable personal income grew by over 6 per cent, the largest
rise since the mid-1980s. In the euro area, instead, household real disposable income
Second, the di�erent dynamics of household disposable income across the two economies
translated into di�erent e�ects on demand. In the United States, GDP returned to its
crisis trend at the end of 2021, but aggregate data hid a large heterogeneity between
sectors: while demand in the services sector was restrained by pandemic-related factors,
the goods sector increasingly showed signs of overheating (�g. 3). In the spring of 2021,
for example, personal consumption expenditure in the durable goods sector was already
more than 30 per cent higher than its pre-crisis level. The fast recovery in US demand, in
a phase in which global supply was still constrained due to the waves of the pandemic,
caused bottlenecks in the international value chains, which drove up the prices of
intermediate goods everywhere. In the euro area, instead, in the third quarter of last year
(the latest for which we can estimate the decomposition of consumption) demand for
Third, the labour market has been much tighter in the United States than in the euro
area. The US unemployment rate still stands at just 3.4 per cent, a value last seen only
in the late 1960s and about half the level of the euro area (6.6 per cent). The di�erence
between the number of vacancies in the US non-farm sector and the number of people
who are unemployed is, today, over 5 million, i.e. there are many more jobs available
than there are people looking for them, while in the euro area the opposite is true, with
the number of unemployed exceeding the number of job vacancies by about 6 million.
Unsurprisingly, the annual change of US nominal wages (measured by the employment
cost index) surpassed 4 per cent as early as in the third quarter of 2021, approached 6 per
cent in 2022, and today still stands above 5 per cent, a level that is di�cult to reconcile
with an in�ation target of 2 per cent (�g. 4). In the euro area, on the other hand, in spite
of current requests for sizeable wage increases in some countries where labour markets
are particularly tight, wage growth has so far remained, on average, moderate, at around
Fourth, the energy shock had a very di�erent role on the two sides of the Atlantic. Since
the second half of 2020, oil prices rose gradually in both the United States and the euro
area. The price of the natural gas delivered in the United States increased much more
markedly, rising from around $10 per megawatt hour before the pandemic to a peak of
over $30 last summer, before sliding back below $10 (�g. 5). However, it was the price
of the natural gas delivered in Europe that recorded the most extraordinary dynamics,
dwar�ng even the 1973 oil price increase (which itself had increased four-fold):
from slightly above â?¬10 per megawatt hour in early 2020, it rocketed to â?¬180 before
the war, soaring to a peak of â?¬350 last summer and then falling sharply, hovering
around â?¬50 in the last few days. This extreme volatility of gas prices was also the result
of a â??bullwhip e�ectâ?, which is the response of demand to uncertain supply, consisting
As a consequence of these dynamics, US consumer price in�ation increased from below
2 per cent in February 2021 to a peak of over 9 per cent last June (�g. 6), declining up to
6.4 per cent in January 2023. Core in�ation (i.e. net of energy and food products) took the
lionâ??s share of the rise, with a peak of 6.6 per cent last September and still in January at 5.6
per cent. In the euro area, on the other hand, headline in�ation reached a record high of
10.6 per cent last October, from less than 1 per cent in February 2021, followed by a return
to 8.6 per cent last January. Core in�ation progressively rose, up to 5.3 per cent in January,
re�ecting the usual lags in the pass-through of energy prices. The upsurge in the latter was
responsible for the largest share of the acceleration in consumer prices: considering the
combination of both their direct and indirect e�ects, in 2022 about 60 per cent of headline
in�ation was attributable to higher energy prices. If we also add the e�ects of the rise in food
2. Did
The Governing Council of the ECB began the process of monetary â??normalisationâ? at
the end of 2021, when it judged that the progress in economic recovery and towards
the medium-term in�ation target was su�cient to allow for the start of a step-by-step
reduction in the pace of asset purchases. Why did we not start earlier? And why did we
To answer these questions, it is useful to recall what the in�ation situation was in June
2021, when we were about to conclude the ECB strategy review. While in the United
States headline in�ation was already above 5 per cent and core in�ation was 4.5 per cent,
pushed by the demand factors discussed above, in the euro area, despite already higher
gas prices, headline in�ation was still below 2 per cent and core in�ation was less than
1 per cent. High in�ation, therefore, seemed to be a phenomenon mostly concentrated
in the United States. The main problem the Governing Council was tackling in that period
was still how to increase the dynamics of consumer prices durably and sustain a rapid
The situation began changing in September 2021, when gas prices went from the
already high level of â?¬50 per megawatt hour to about â?¬100 (a â??supply shockâ?). At that
time, however, futures quotes predicted gas prices would remain at around that level
during the winter season and then decline very sharply, to well below â?¬50 by June 2022.
The prediction of declining gas prices implicit in futures contracts remained more or
less unchanged until late December 2021. With such a steep fall in gas prices in sight,
in�ation could not stay at high levels for long and, in fact, was projected to return
As it turned out, instead of declining by more than 50 per cent as expected at the end
of September, gas prices increased by almost 100 per cent, averaging an unprecedented
level of â?¬200 during the summer of 2022. The Russian invasion of Ukraine had transformed
a temporary shock into a persistent one, warranting an acceleration of the monetary
In the early part of 2022, in fact, the process gained speed. However, we managed to
avoid the potentially dangerous cli�-e�ects of too sharp a swing in our stance, not
least in view of the major uncertainty caused by the con�ict in Ukraine. The end of our
purchases was anticipated to the 1st of July and, shortly after, we started raising our key
o�cial interest rates by a signi�cant size, with the aim of frontloading the exit from their
3. Did
When discussing the high level of headline in�ation reached in the euro area, some
commentators have paid less attention to the sudden occurrence of the energy shock,
its size and its persistence, pointing the �nger instead at the delays of the central bank
in initiating its monetary tightening. Critics citing this hypothetical mistake have also
highlighted the large errors in the in�ation projections made by the ECB/Eurosystem
The forecast errors in predicting consumer price changes over the last year were indeed
sizeable and much larger than those observed in the past. Some have even argued
that these large errors call into question the very credibility of the ECB, although other
While the observed size of the errors may understandably cast doubt on the reliability
of the models used for the projections, our analysis for Italy (with similar results for the
analysis conducted by the ECB on the forecasts for the euro area as a whole) indicates
that the e�ects of energy prices â?? the most important exogenous variables in the
forecasting model, whose changes are usually inferred from the market price of futures
contracts â?? explain, directly and indirectly (i.e. via their e�ects on production costs),
70 per cent of the overall error made in forecasting in�ation in 2022. This share rises
to 80 per cent when the e�ects of food prices, the other volatile component of the
These results suggest that, although all models should be (and are) subject to continuous
checks and improvements, the functioning of the economy has not changed dramatically
over the last year. They do draw our attention, however, to the quality of the forecasts
Undoubtedly, the e�ects of global supply bottlenecks were underestimated, although
demand in the euro area did not contribute greatly to them in any case. The key problem,
however, was the generalised underestimation of the recent geopolitical tensions.
The sharp drop in gas supplies from Russia observed since early 2021 was in fact
(probably mistakenly) attributed at �rst to the e�ects of a particularly cold winter in
Russia and subsequently to the political pressure from the Russian government to
accelerate the opening of the Nord Stream 2 gas pipeline. The new shock caused by
the Russian invasion of Ukraine a year ago has, instead, dramatically changed this
picture, triggering a sharp rise in volatility and pushing both current and expected gas
It may indeed be argued that an earlier rise in interest rates might have reduced the
market uncertainty (and might have more e�ectively managed the initial rise in in�ation
expectations). On one hand, this claim is debatable, since it neglects to consider that
the consequences of the uncertainty generated by the Russian invasion also needed to
be properly evaluated and addressed. On the other, even if we suppose it to be true,
I doubt that a few monthsâ?? delay in the actual implementation of the decision to halt
asset purchases and start raising o�cial interest rates would have had substantial
consequences on the evolution of consumer prices in the euro area, which, as I have
4. Is
As is widely agreed, monetary policy a�ects the real economy, and in�ation, with â??long
and variable� lags. While monetary actions and communications tend to a�ect �nancial
marketsâ?? interest rates and asset prices almost immediately, their transmission to the
�nancing conditions of households and businesses and, subsequently, to consumer
prices, tends to be much more gradual as economic agents revise their decisions to
consume and to invest slowly and, in some cases, infrequently. The initial conditions
of the economy â?? including the level of debt, the degree of economic uncertainty, and
many other domestic and global factors â?? also have important consequences on the
distribution over time of the e�ects of a change in the monetary stance on in�ation and
Empirical evidence for the euro area shows that, on average, a change in key rates exerts
its largest impact on GDP growth after about a year and a half and its maximum e�ect on
in�ation after one to two years. Therefore, most of the economic impact of the rate hikes
that have been implemented so far is yet to be felt, suggesting that a degree of prudence
However, the initial e�ects of our policy measures are already discernible. Following
our latest monetary policy decision, the overall increase of o�cial rates since July has
reached 300 basis points and has been fully and smoothly transmitted to market interest
rates. Since the start of the reduction of monetary accommodation at the beginning of
2022, one-year risk-free rates (measured by overnight index swaps) have picked up from
negative levels to 3.6 per cent, while ten-year rates have increased from barely positive
values to 3.0 per cent. In real terms, using the in�ation-linked swap (ILS) as a de�ator,
they currently stand at about 0.7 and 0.5 per cent respectively, from around -4 and -2 per
cent at the end of 2021 (�g. 8). The fact that long-term interest rates started increasing
well before our �rst key rate hike should not come as a surprise. It is indeed proof of
the credibility of our actions and our commitment to guaranteeing price stability, and it
supports my claim as to the possibly limited e�ects of starting the rise in interest rates a
Further signs of the e�ectiveness of the Governing Councilâ??s actions are found in the
dynamics of in�ation expectations, whose levels are an important anchor for wage
dynamics and actual in�ation. In the euro area, short-term in�ation expectations derived
from �nancial market prices are falling sharply. ILS rates indicate that the expected
in�ation rate twelve months ahead now stands at 2.9 per cent, down from a peak of
almost 9 per cent recorded in late August 2022 (�g. 9). The initial signs of a decline in
At the same time, longer-term expectations, net of risk premia, remain at levels consistent
with our 2 per cent price stability target, and tail-risks of excessive in�ation are gradually
dissipating (�g. 10). The anchoring of in�ation expectations is also supported by the
A quanti�cation of the structural drivers of euro area and US in�ation expectations, as
measured by ILSs, obtained by breaking down their daily �uctuations into domestic and
global shocks (�g. 11), con�rms the di�erent role of supply and demand factors on the
two sides of the Atlantic and provides a measure of the e�ectiveness of the monetary
tightening. The results of this analysis â?? which focuses on the e�ects of changes in policy,
demand and supply with respect to historical regularities â?? show that, since the start
of the war in Ukraine, the in�ation rate predicted over a �ve-year horizon increased
mostly in response to supply shocks in the euro area; the much smaller contribution of
demand shocks rose progressively over the course of 2022, re�ecting improved business
cycle conditions. Results also document strong spillover e�ects of US monetary policy
on euro area in�ation expectations since the second half of 2022 and con�rm the initial
signs that the ECBâ??s monetary policy tightening is having the desired soothing e�ects
on the economy. In the United States, on the other hand, in�ation expectations had
been steadily sustained by domestic demand until the end of last October, and have
since been reined in by contractionary monetary policy e�ects. The credibility that central
5. Should
There is no question that the tightening of the euro area monetary stance must continue
to ensure that a temporary increase in in�ation caused by a supply shock does not
become a more persistent phenomenon sustained by demand factors. This said, in early
February the Governing Council of the ECB assessed that the risks to the in�ation outlook
have become more balanced, while reiterating that uncertainty remains very high. In this
context it has raised its key rates by 50 basis points and has announced the intention to
raise them by a further 50 basis points in March. In any case, the pace of any further rate
hike will continue to be decided on the basis of incoming data and their impact on the
It will remain essential to continue balancing the risk of a too-gradual recalibration
doing too little
), which could cause in�ation to become entrenched in expectations and
in wage
setting processes, with that of an excessive tightening (
doing too much
), which
would result in signi�cant repercussions for economic activity, �nancial stability and,
ultimately, medium-term price developments. In line with our symmetrical price stability
On the one hand, doing too little would come at a cost for the economy if this led to
the need for a stronger and more prolonged restriction of monetary policy. On the other
hand, however, the costs linked to the opposite risk may be relevant if â??doing too muchâ?
were to determine an undershooting of the target and possibly even lead to serious
debt-de�ation phenomena, triggering nonlinear perilous ampli�cations. In the face of
both of these risks, the central bank decisions should continue to be characterised by
wisdom and balance and be guided by careful quantitative evaluations of incoming data.
When it comes to reducing in�ation, a recession is not always inevitable. Communicating
a strong commitment to bringing in�ation down to target in a speedy manner is
fundamental, but doing so by minimising the costs for the real economy is not any less
A cautious approach is also advisable due to a series of other considerations. The �rst
is related to the high level of economic uncertainty and the Brainard principle, which
states that when the central bank is uncertain about the e�ects of its actions, it should
move conservatively. An exception to this principle is the case of uncertainty around the
persistence of in�ation: when persistence is high, in fact, a strong monetary reaction may
be required to avoid high in�ation becoming entrenched in agentsâ?? mind-sets. While this
possibility should be carefully monitored, data on market- and survey-based in�ation
expectations â?? including their recent decline at short horizons and their decreasing pro�le â??
and the marked deceleration of prices on a 3-month annualised basis (�g. 12), may call
into question the persistence of in�ation at high levels in the euro area, reinforcing the
The second concerns the â??long and variable lagsâ? of the monetary transmission process
that I have already mentioned. Credit dynamics provide signals that are especially
relevant in this respect. On a 3-month (annualised) basis, the growth of loans to �rms
in the euro area was negative in January (-1.3 per cent) from an almost double-digit
expansion in October (9.8 per cent). Although this deceleration is the natural (and desired)
consequence of monetary normalisation, both its size and speed require caution on the
The potential risks to �nancial stability also require a good dose of caution. The
unprecedented coordinated rises in o�cial rates around the world may create spillover
e�ects that are di�cult to quantify but may not be negligible. Financial instability risks are
particularly relevant in the Economic and Monetary Union, whose incomplete architecture
â?? especially its decentralised �scal policy and the delays in completing the banking and
Finally, in the current uncertain environment, models and forecasts should necessarily
be taken
, in particular when determining the â??terminalâ? level of key
interest rates. Quantitative assessments are still useful, but their insights have to be
assessed together with the information that will gradually become available on in�ation
The monetary policy tightening started by the ECB in December 2021 has been crucial
to respond to the risks stemming from high and rising in�ation. Even before exerting
its e�ects on aggregate demand, it has operated by contributing to containing
in�ation expectations, avoiding an increase to excessively high levels. This anchoring
of expectations has therefore laid the foundations for preventing the occurrence of
Low and stable in�ation expectations, however, do not completely eliminate the risk
of â??second-order e�ectsâ?. To this end, labour and business in all euro-area countries
must continue to behave responsibly. It has to be fully understood that the energy
shock is like a tax on the euro area economy, which unfortunately cannot be returned to
sender and cannot be circumvented through a fruitless race between wages and prices
nor through an excessive and permanent increase in public debt. Wage negotiations,
therefore, cannot go back in time to when they were purely backward-looking. Making
up for the loss of purchasing power must, rather, rely on achieving sustained productivity
growth, although targeted and temporary �scal measures to alleviate the burden on
Today, gas prices in Europe are returning to more moderate levels and their volatility
is declining, setting the stage for the convergence of in�ation, over the medium term,
to the ECBâ??s target of 2 per cent. In this context, the pricing strategies of businesses
will play a central role. In particular, we will have to closely monitor whether, after the
pass-through of the higher energy costs observed in 2022, �rms will allow �nal prices to
re�ect the most recent declines, which would imply a less intense tightening of �nancial
In this phase, monetary policy should not be left to work alone. The contribution of all
policies, including perhaps some new versions of old-fashioned income policy recipes,
the ECB target. The return
to price stability will greatly bene�t from �scal policies and negotiations between workers
Figure 1
(monthly data; annual percentage changes)
1970
1980
1990
2000
2010
202
UK
EZ
IT
Source: Eurostat, Istat, UK Office for National Statistics and US Bureau of Labor Statistics.
Note: EZ denotes the euro area (changing composition after 1999 and weighted average of the 11 countries partecipating
Figure 2
Disposable income and GDP
(annual data; percentage changes)
disposabl
e
income
GDP
disposabl
e
income
GDP
Source: Eurostat and US Bureau of Economic Analysis.
Figure 3
Demand in the goods and services sectors
(monthly and quarterly data; indices: Jan. 2020 / 2019 Q4 = 100)
Goods sector
Services sector
202
0
202
202
0
202
Source: US Bureau of Economic Analysis and estimates based on Eurostat data.
Note: dashed lines show pre-pandemic trends.
Figure 4
Nominal wage growth
(quarterly data; annual percentage changes)
United States
2007
2012
2017
202
Source: ECB and US Bureau of Labor Statistics.
Figure 5
Natural gas prices
(daily data)
United States (dollars)
2021
202
Source: Renitiv.
Note: Title Transfer Facility (TTF) quotations for European gas and Henry Hub for US gas.
Figure 6
Headline and core ination
(monthly data; annual percentage changes)
headline
core
2020
2021
2022
headline
core
2020
2021
2022
Source: Eurostat and US Bureau of Labor Statistics.
Figure 7
ECB/Eurosystem projections errors for euro area headline ination
(percentage points)
1 quarter ahead projection errors
4 quarters ahead projection errors
2007
2011
2015
2019
202
2007
2011
2015
2019
202
Source: Bank of Italy and ECB.
Note: dashed lines denote an interval around zero of plus/minus two standard deviations of projection errors realized in
2003-2020; latest observation: 2022 Q4.
Figure 8
Real interest rates in the euro area
Term structure, spot rates
Term structure, 1-year forward rates
202
mid
202
27 February 2023
2021
mid
2022
27 February 2023
Source: based on Bloomberg and Renitiv data.
Note: nominal OIS interest rates deated by the corresponding ination
linked swap rates.
Figure 9
(daily data; per cent)
ear
10
year
2021
2022
Source: Bloomberg.
Note: 1-year and 10-year ination-linked swap rates.
Figure 10
Ination tail risks in the euro area
(daily data; per cent)
2021
202
Source: based on Bloomberg data.
Note: probabilities inferred from ination options; 0 (1) is the probability of ination being smaller than 0 (1) on
average in the next 5 years;�� 3 (4) is the probability of ination being larger than 3 (4) on average in the next 5 years;
50-days moving averages.
Figure 11
Drivers of changes in ination expectations
(daily; percentage changes)
euro area
United States






























































the US”, mimeo, Bank of Italy, Rome.
Note: 5-year ination swap rates; changes with respect to 3 January 2022.
Figure 12
(monthly data; 3-month annualised percentage changes)
202
UK
EZ
Source: Eurostat, UK Office for National Statistics and US Bureau of Labor Statistics.
Note: EZ denotes the euro area.