Reflections on the exit strategy
Lorenzo Bini Smaghi
Member of the Executive Board of the European Central Bank
Sveriges Riksbank, Stockholm, 21 January 2010
A lot of attention has been devoted in recent policy discussions to the exit strategy from the
expansionary monetary and fiscal policy stance adopted since the outbreak of the crisis,
and to the interaction between the two. Designing an optimal exit strategy should, in
theory, not be too difficult. In an ideal world, fiscal stimuli should be withdrawn before
monetary policy is tightened in order to ensure that public debt remains sustainable and that
monetary policy is not overburdened.
We know however that first-best solutions are often hard to implement, because they are
based on unrealistic assumptions relating in particular to the rationality of expectations and
the ability of policy-makers to rapidly make decisions and to change them if needed.
Furthermore, policy-makers have to take account of the prevailing uncertainty and of the
impact of their decisions on agents’ expectations. For instance, fiscal policy is normally
decided once a year, through an elaborate budget process, but is not implemented until the
following year. This process introduces significant rigidities into the decision-making and
makes it somewhat less likely that fiscal policy can exit before monetary policy, which is
typically a much more flexible instrument.
Today I would like to focus my remarks on the exit from the accommodative stance of
monetary policy. I will not consider the phasing-out of the non-standard measures that
central banks have adopted over the last few months.
The discussions so far have centred mainly on the timing of the exit decision, which should
neither be premature nor tardy. There are counter-effects in both scenarios. If the decision
to exit is made too early, the economic recovery may be put at risk, as higher interest rates
will produce a tightening effect on consumption and investment decisions at the very time
when the pick-up in the economy is still fragile. Furthermore, it might further restrict credit
conditions while the banking system is still restructuring its balance sheet. If, on the other
hand, the decision is taken too late, monetary conditions will remain too lax for too long,
sowing the seeds of the next crisis. In addition, the later the tightening, the sharper it needs
to be, thus producing valuation changes that may reduce banks’ profitability and undermine
their ability to support the economic recovery.
Given the difficulty of the timing, the discussion has turned to the second-best option. In
other words, which of the two risks – being too early or being too late – would pose the
biggest problems for the economy? Several authors, including international organisations,
have suggested that the policy authorities should err on the side of being late. They argue
1
that an early tightening is difficult to reverse and tends to hit the economy early in its
recovery, and may give rise to a double-dip recession. On the other hand, a late tightening
allows more time to bring things back to normal. In addition, the stronger the economy is
likely to be, the more resistant it will be to the shock produced by the tightening of
monetary conditions.
Overall, history shows that policy authorities have largely erred on the side of being too
late than of being too early. Indeed, late exits have been quite numerous; they were either
the result of deliberate policy decisions to boost the economy, which was probably the case
in the 1970s and early 1980s, or of forecast errors made in over-estimating deflationary
risks and under-predicting the subsequent recovery, which might be the case of the decade
just ended.1
The monetary and fiscal policy tightening that took place in the United States in 1936,
which led to the 1937-38 recession, is regarded as a text-book example of too early an exit,
of a mistake not to be repeated. There are not many other examples of early exits. The
slight increase in interest rates by the Bank of Japan in 2000 has been considered by some
as an early exit, but recent analysis has downplayed it. 2
In fact, examining closely the 1936 episode in the US, it might not even be considered as
an early exit. The very restrictive monetary and fiscal policies that were adopted at that
time might deserve to be categorised as a case of ‘brutal’ exit, the effects of which could in
fact have occurred at any time, exit or no exit. Between mid-1936 and early 1937 reserve
requirements for banks were doubled in a series of three steps. The general government
budget contracted by almost 4% of GDP. These measures appear to have been
disproportionate, even with an economy that had started to recover, with inflation picking
up and credit aggregates accelerating. The severe negative effects of these measures do not
mean that no measure should have been taken at all. They should simply not have been so
abrupt and not have taken economic agents by surprise. In other words, the 1936 episode
might be considered as an example of a badly calibrated and badly communicated exit,
rather than a wrongly timed exit.
1
Ben Bernanke, “Monetary Policy and the Housing Bubble”, speech at the Annual Meeting of the American
Economic Association, Atlanta, Georgia, 3 January 2010.
2
Daniel Leigh, “Monetary Policy and the Lost: Decade: Lessons from Japan, IMF Working paper, October
2009.
2
In this context, it is worth mentioning the 1994 episode, when the Fed’s 25-basis-point
increase in its funds rate caused turmoil in the bond market. Even with the benefit of
hindsight, that decision can be regarded as neither tardy nor brutal. The problem was that it
took markets by surprise and led to a major reassessment of the yield curve, with
substantial valuation losses on long-term positions.
These experiences confirm that not only is the timing of the exit crucial, but also its
communication to market participants. Indeed, the impact of any withdrawal of monetary
stimulus on financial market participants depends on whether they have fully anticipated it.
If they have, then the decision tends to have less impact on the yield curve, and thus on the
valuation of banks’ assets and on the rate that banks charge to their customers, especially
on fixed rate loans. As confirmed by a growing body of literature, the main mistake relating
to the 1936-37 decisions was that they were badly communicated to economic agents.3
To sum up, a key element of the exit strategy is that the policy authorities must guide
expectations. How can they do this?
Let me first of all consider how it should not be done.
First, it may be better for central banks not to become embroiled in the debate on whether
exiting too early or too late is better or worse, and especially not to give signals that the
latter is less risky. Such an approach could encourage market participants to become less
risk-averse and to expect the period of low interest rates to be extended beyond what is
necessary. This would induce them to take on further risks, in particular in terms of
maturity mismatches – borrowing short-term to buy long-term bonds – which might be
profitable in the short term but could lead to substantial capital losses when the exit takes
place.
Second, any commitment to prolonged periods of low interest rates is risky. If the aim is to
lower long-term rates, it may also encourage market participants to take substantial long
positions in the fixed income market. As a result, the risk of major losses arising at the time
of the exit increases considerably, as the 1994 experience shows. This may induce the
central bank to further delay the exit to avoid penalising banks. This seems to be the
experience of 2002-2004. As a result, interest rates may remain below the desired level for
too long, fuelling a possible bubble.
3
G. Eggertsson and B. Pugsley, “The mistake of 1937: a General Equilibrium Analysis”, CFS Working
paper, November 2006.
3
Third, central banks need to be very careful in using information extracted from market
data when conducting monetary policy. The anchoring of inflation expectations is a very
important benchmark for assessing whether the stance of monetary policy has become too
expansionary and whether there are risks of falling behind the curve. However, the precrisis period and the crisis itself have shown that expectations are not always formed in a
rational way and might themselves be influenced by the behaviour of the monetary
authorities. Quite often markets participants form their expectations by looking at central
banks’ behaviour, on the assumption that the latter have better and more information about
underlying inflationary pressures. Under these circumstances, market-based inflation
forecasts tend to be biased and may react in a perverse way following a tightening of
monetary policy. In particular, inflation expectations may be revised up, rather than down,
after an interest rate increase, especially at the beginning of a tightening cycle. If the
monetary authorities’ assumption that inflation expectations are well anchored proves to be
wrong, the whole yield curve might move upwards and the decision to increase the rate of
interest may result in a much sharper tightening than expected. 4
Another problem with inflation expectations concerns their measurement, especially in
periods of financial turbulence, when the liquidity of the underlying instruments can affect
their signalling content. A further distortion may arise when the central bank is itself a
major buyer and holder of indexed-linked assets, which may distort prices and thus the
information content. Measurement problems arise also with other components of the
analytical framework, such as the output gap. Experience shows that measurement of the
output gap varies over time. For instance, when measured ex post, with the data available
until 2008, the US output gap over the period 2002-2004 turned out to be very small, even
non-existent, while ex ante it was estimated at around 2% of GDP. 5
So what should a central bank do to implement and communicate an appropriate exit
strategy? Let me make a few comments on this.
To ensure that the exit strategy does not come as a complete surprise, it has to be well
communicated. Over recent months, central banks have devoted a lot of time to explaining
their exit strategies, both in terms of content and conditions. At the ECB we have made it
clear that, as in the past, any decision – even one concerning the exit – will be linked to the
4
5
L. Bini Smaghi, “An Ocean Apart: Comparing Transatlantic Responses to the Financial Crisis”, Rome,
September 2009.
L. Bini Smaghi, “Monetary Policy and Asset Prices”, Freiburg, October 2009.
4
underlying economic conditions and in particular to the re-emergence of inflationary
pressures. Market participants should thus form their expectations on future interest rates
on the basis of projected economic developments rather than on specific unconditional
commitments.
In order to promote understanding of how monetary policy needs to adapt to changing
economic conditions, and eventually exit from its very accommodative stance, central
banks would do well to explain how the stance itself is affected not only by the level of the
interest rate they control, but also by the underlying economic developments, such as
projected growth and inflation. In this respect, it is interesting to note that over the last six
months, while economic activity and inflation have picked up, interest rates, both in
nominal and real terms, have continued to fall. Spreads on different types of bond, except
those of some governments, have continued to come down. The slope of the yield curve has
slightly increased, in particular in the US (see Annex).
Finally, an exit is more easily communicated, and understood, by market participants if it is
gradual. Any interest rate adjustment – even a slight one – entails some form of
discontinuity, which may affect the whole yield curve. A credible gradual exit, if well
communicated, minimises shocks to interest rate expectations. However, a gradual exit
strategy is more likely to be credible if it is implemented in a timely way. If it gives the
appearance of being late, the first step by the central bank is likely to cause a substantial
revision of interest rate expectations and thus of the yield curve, strengthening the impact
of the tightening.
To sum up, experience suggests that what matters most in any exit strategy is that markets
are well prepared, so that when the decision is ultimately made they are not taken by
surprise and suffer negative repercussions. Communicating an exit strategy is, in the end,
not that different from communicating monetary policy in normal times. Given the state of
uncertainty about the recovery, market participants have to base their expectations, and
their trading decisions, on projected economic fundamentals rather than trying to read
between the lines of policy announcements.
5
Annex
Chart 1: Economic growth reached its through in
2009 Q1 and now displays a slow recovery
Chart 2: ECB interest rates and the EONIA
Real GDP growth
Quarterly growth
5
Year-on-year growth
5
4
4
3
3
2
2
1
1
09 Q3
08 Q4
08 Q1
07 Q2
06 Q3
05 Q4
05 Q1
04 Q2
03 Q3
02 Q4
-5
02 Q1
-5
01 Q2
-4
00 Q3
-4
99 Q4
-3
99 Q1
-2
-3
98 Q2
-2
97 Q3
-1
96 Q4
0
96 Q1
0
-1
6.0
EONIA
5.0
marginal lending rate
4.0
main refinancing rate
3.0
2.0
1.0
deposit rate
0.0
Aug-08 Sep-08 Oct-08 Nov-08 Dec-08 Jan-09 Feb-09 Mar-09 Apr-09 May-09 Jun-09
Jul-09
Aug-09 Sep-09 Oct-09 Nov-09 Dec-09 Jan-10
Source: Eurostat.
Chart 3: Credit spreads decreased rapidly in
the euro area (basis points)
Non-Financial AA
Non-financial BBB
Financial AA
Financial
3000
Financial BBB (rhs)
500
Chart 4: Corporate bond spreads in the euro
area (basis points)
3000
450
Non-financial
Financial AAA
Financial AA
Financial A
Financial BBB
600
2500
500
2000
400
1500
300
1000
200
500
100
Non-Financial AAA
Non-Financial AA
Non-Financial A
Non-Financial BBB
2500
400
2000
350
300
1500
250
1000
200
150
500
100
0
50
0
0
Jun.07
Dec.07
Jun.08
Dec.08
Jun.09
0
Jul.08
-500
Jan.09
Jul.09
Jan.10
Jul.08
Jan.09
Jul.09
Jan.10
Source: Datastream.
Note: The benchmark bond index used is EMU triple A government bond index calculated by BofA Merrill
Lynch.
Dec.09
Source: Thomson Financial Datastream and ECB calculations.
Latest observation: 19-Jan-10.
Chart 5: Corporate bond spreads in the United
States (basis points)
Financial
2000
1800
Financial AAA
Financial A
Financial AA
900
Financial BBB
800
Non-financial AAA
Non-financial A
Short-term
1400
10.0
600
600
300
400
200
200
100
0
Jan.10
Jul.08
5.5
7.0
5.0
6.0
4.5
5.0
4.0
4.0
3.0
3.5
2.0
3.0
1.0
2.5
0.0
0
Jul.09
6.0
8.0
400
800
6.5
9.0
500
Jan.09
Jul.09
Jan.10
Source: Datastream.
Note: The benchmark bond index used is US Treasury 3-5year maturity government bond index calculated
by BofA Merrill Lynch.
small loans to nfc (over 5 years)
large loans to nfc (over 1 and up to 5 years)
large loans to nfc (over 5 years)
loans for house purchase (over 5 and up to 10 years)
loans for house purchase (over 10 years)
7-year government bond yield
small loans to non-financial corporations
large loans to non-financial corporations
loans to households for house purchase
loans to households for consumer credit
three-month Euribor
Non-financial BBB
1200
1000
Long-term
Non-financial AA
700
Jan.09
(percentages per annum)
Non-financial
1600
Jul.08
Chart 6: MFI bank lending rates in the euro area
2.0
2003
2005
2007
2009
2003
2005
2007
2009
Source: ECB and ECB calculations.
Latest observation: November 2009 for MIR rates and December 2009 for market rates.
Latest observation: 19-Jan-10.
6
Chart 7: Bank lending rates on loans to non-financial
corporations in the euro area (percentages per annum)
Short-term
Chart 8: Bank lending rates on loans to households
in the euro area (percentages per annum)
Short-term
Long-term
small loans
large loans
three-month Euribor
small loans (over 1 and up to 5 years)
small loans (over 5 years)
large loans (over 1 and up to 5 years)
large loans (over 5 years)
7-year government bond yield
7.0
10.0
7.0
6.0
5.0
6.0
4.0
Long-term
house purchase (over 5 and up to 10 years)
house purchase (over 10 years)
consumer credit (over 1 and up to 5 years)
consumer credit (over 5 years)
house purchase (over 1 and up to 5 years)
7-year government bond yield
house purchase
consumer credit
three-month Euribor
5.0
9.0
10.0
8.0
9.0
7.0
8.0
6.0
7.0
5.0
3.0
6.0
4.0
4.0
2.0
5.0
3.0
4.0
2.0
3.0
1.0
3.0
1.0
2.0
0.0
2003
2005
2007
2003
2009
2005
2007
Jun-07
4.31
3.33
3.20
3.41
5.40
Real long-tem MFI lending rates
Real cost of market-based debt
Real cost of quoted equity
Oct-09
3.10
1.81
1.60
2.51
6.39
Nov-09
3.07
1.81
1.65
2.36
6.28
2005
2007
2003
2009
Dec-09
Jan-10
2.22
6.00
2.24
5.92
Euro area
United States
7
7
30-Jun-09
30-Sep-09
15-Jan-10
6
5
6
5
2
4
4
1
1
3
3
2
2
1
1
0
30-Jun-09
30-Sep-09
15-Jan-10
0
0
1
2
3
4
5
6
7
8
9 10
0 1 2 3 4
5 6 7 8 9 10
Ja
n
A -08
pr
Ju 08
O l-08
ct
Ja -08
n
A -09
pr
Ju -09
l
O -09
ct
Ja -09
n10
20
10
2
20
09
5
20
08
3
5
20
07
3
20
06
6
20
05
4
6
20
04
4
20
03
7
20
02
8
20
01
2009
Chart 10: Nominal implied forward curves for the
euro area and the United States (percent per annum)
7
20
00
2007
Latest observation: November 2009 for MIR rates and December 2009 for market rates.
8
19
99
2005
Source: ECB and ECB calculations.
Chart 9: Real overall costs of financing (percentages)
Real short-term MFI lending rates
2003
2009
Source: ECB and ECB calculations.
Latest observation: November 2009 for MIR rates and December 2009 for market rates.
Overall cost of financing
2.0
0.0
Sources: ECB, calculations, Thomson Financial Datastream, Merrill Lynch and Consensus Economics
Forecast. Last observation: MFI lending rates: November 2009; Cost of equity and cost of market-based debt:
6 January 2010.
Source: EuroMTS, ECB, FED. Notes: euro area based on triple A government bonds. The underlying
curve of the United States excludes on-the-run bonds.
Chart 11: Break-even inflation rates in the euro
area and the United States (percent per annum)
Euro area
United States
5-year BEIR (seas. adj.)
10-year BEIR (seas. adj.)
5-year forward 5 years ahead BEIR (seas. adj.)
5-year BEIR
10-year BEIR
5-year forward 5 years ahead BEIR
4.00
4.00
3.00
3.00
2.00
2.00
1.00
1.00
0.00
0.00
-1.00
-1.00
-2.00
Jul-08
Oct-08
Jan-09
Apr-09
Jul-09
Oct-09
Jan-10
-2.00
Jul-08
Oct-08
Jan-09
Apr-09
Jul-09
Oct-09
Jan-10
Source: Reuters, ECB calculations, and R. Gürkaynak, B. Sack, and J. Wright (2008), "The TIPS Yield
Curve and Inflation Compensation", Federal Reserve Working Paper.
Latest observation: 19-Jan-10 for the euro area, 15-Jan-10 for the US.
7