Debt, Financial Crisis and Economic Growth
* **
01-12-2012
Willem H. Buiter and Ebrahim Rahbari
Chief Economist
Citigroup
Global Economist
Citigroup
*
Paper presented at the Conference on Monetary Policy and the Challenge of Economic Growth at the South
Africa Reserve Bank, Pretoria, South Africa, November 1-2, 2012.
**
The views and opinions expressed are those of the authors. They do not necessarily represent the views
and opinions of Citigroup or of any other organisation the authors are affiliated with.
1
Abstract
Following the leverage binge in advanced economies (AEs) over the three decades preceding 2008,
debt growth is generally likely to be low in the years ahead. Deleveraging is likely to continue to weigh
heavily on growth in highly indebted economies, and the deleveraging process will be costlier and take
longer unless adequate policies are implemented to support it.
Debt in the non-financial sector of AEs has almost doubled as a share of GDP between 1980 and 2008
– a period during which GDP grew rather briskly. It initially grew more strongly in the private sector, but
only for public debt shot up sharply after the 2007-09 North-Atlantic financial crisis. Since 2008, debt
growth has slowed by a third in real, and by half in nominal, terms. It would have fallen even more
sharply if public debt growth had not more than doubled.
The speed of deleveraging varies widely in different countries and sectors. On average, household and
non-financial corporate debt has fallen, while public debt is still rising. Private (and sometimes public)
deleveraging has generally been faster where GDP and income growth have held up, and is impeded by
weak income growth in countries where deleveraging pressures are intense, such as in Greece, Ireland,
Portugal or Spain. Safe debt is rapidly becoming an oxymoron.
Hangovers from credit booms are serious. Increases in debt can cause systemic crises which generally
tend to be both long-lived and costly. Large increases in debt also make such crises more painful – we
find that the ‘GDP loss’ relative to trend in the aftermath of financial crises is almost twice as large in
countries which had a large pre-crisis increase in debt than in countries that did not. Today, growth is
weakest, on average, in countries with the largest pre-crisis debt increases. But even when debt does
not cause a major crisis, debt reduction through higher saving rates tends to be contractionary because
the poor coordination of deleveraging, saving and investment decisions give rise to Keynes’s ‘paradox of
thrift’.
Deleveraging – shrinking balance sheets – occurs when households, businesses or the public sector
either desire to save more or are forced to do so. Economic actors may want to save more, or may be
forced to save more by a combination of illiquid assets and restricted access to external funding or
because their net worth is perceived to be inadequate. Both net worth and gross debt therefore matter
for saving and deleveraging behaviour.
Some of the costs of deleveraging are likely unavoidable, but policies can help to reduce the avoidable
costs of deleveraging. First among those is access to liquidity. A well-capitalised banking system would
be a good start, but the private provision of liquidity – a public good - in crises is usually highly
inefficient, so central banks will likely retain a key role in liquidity provision for the coming years.
Mechanisms to allow the gross deleveraging, i.e. the ‘netting’ of assets and liabilities, especially among
banks and other financial intermediaries, should be encouraged. Where higher financial surpluses are
required, policies should encourage higher saving rather than lower investment. Extensive debt
restructuring for governments, banks, and in some countries also households, using yet-to-be-created
orderly debt restructuring mechanisms, is both desirable and likely. In the medium-term, the lessons
should be clear. First, to better coordinate saving and investment decisions, while supporting financial
markets with more effective and sustainable fiscal and monetary policies. Second, on the liability side of
any balance sheet: more equity, less debt.
1
1. Introduction
There is a lot more private and public debt today in the advanced economies than has been the norm
during peacetime periods. In 1980, the total non-financial sector (NFS) gross debt in 17 developed
markets amounted to $12.3trn or 168% of the GDP of these countries. In 2011, the total stood at just
over ten times that value ($128.5trn), amounting to 315% of GDP.
Debt is attractive to holders because it offers, or appears to offer, a predictable safe income stream. It is
attractive to issuers because, among other advantages, it provides leverage, in the economic sense of
the word: “…leverage exists whenever an entity is exposed to changes in the value of an asset over
time without having first disbursed cash equal to the value of that asset at the beginning of the
1
period.” Debt has the further advantage to the issuer that, as long as the borrower adheres to the terms
of the debt contract, the creditor has fewer control rights over the use of the borrowed funds than would
be the case with equity-type liabilities.
Debt has grown in most countries and in most sectors. Private debt, both household debt and debt of
businesses grew strongly from the 1980s until quite recently. Public debt grew more modestly until the
North-Atlantic financial crisis that erupted in August 2007 confronted governments with large revenue
losses, as well as the need for fiscal stimuli and banking sector bail-outs. Public debt growth has now
overtaken the growth in private debt. In 1980, 27% of the debt of advanced economies was household
(HH) debt, 47% non-financial corporation (NFC) debt and 26% general government (GG) debt. By 2011,
the share of general government debt had grown to 37% of the total. It is likely to continue rising in the
coming years, with the share of NFC debt strongly down (to 37%) and the share of HH debt only slightly
down (26%).
Now that debt levels are perceived to be excessive in many places, debt and credit growth from here on
is likely to be low in most DMs for the foreseeable future. Real growth in gross debt in DMs since 2008
has been roughly one third lower than it was between 2001 and 2008. Nominal debt growth has roughly
halved from the pre-crisis average.
Even though credit growth has generally fallen, the years since the North-Atlantic financial crisis of
2007-2009 have on average only seen a modest degree of private sector deleveraging – from a peak of
around 205% of GDP in 2009 to maybe 5ppts of GDP less in Q2 2012. Total NFS gross debt continues
to increase, as public debt has generally risen strongly in nominal and real terms. As a share of GDP, it
has gone up by 30ppts in the space of less than four years since 2008.
We expect debt reductions to have a lot further to run in many countries. Deleveraging pressures are
likely to be particularly severe in Cyprus, Ireland, Portugal, and Spain. In most other countries, private
sector credit growth is likely to remain sharply below the growth rates of previous years. In addition, real
GDP growth is likely to be low during this period of deleveraging, due to an increase in desired net
saving and its adverse affect on the level of economic activity due to the paradox of thrift.
Some of the adverse consequences that large-scale debt reduction brings with it are probably
unavoidable. But policy responses should be focused on minimizing the avoidable costs of
deleveraging. First among those should be measures to allow gross deleveraging (shrinking balance
sheets through equal reductions in assets and liabilities, without the need to raise financial net worth by
running financial surpluses/’saving’) to take place in an orderly and coordinated fashion.
Creating institutions or arrangements to help heterogeneous, decentralized, independent and
uncoordinated private and public entities to coordinate the netting of gross financial assets and liabilities
in complex networks of creditors and debtors should help, too. Clearing houses for a much wider range
of financial claims should therefore be considered.
Additionally, debt restructuring will often be needed to bring about timely net deleveraging, that is adding
to the net worth of financially fragile sectors by running financial surpluses or by saving. Where gross
1
See Counterparty Risk Management Group II (2005), P A1. Clearly debt can be used to leverage equity, but many other financial
instruments other than debt can be used to create leverage. This includes initial margin in futures contracts, and embedded leverage
through options. These broader forms of leverage played a role in the North-Atlantic financial crisis, but will not be part of our focus.
1
debt is excessive and net worth inadequate, socially efficient deleveraging will in many cases require
establishing orderly and efficient debt restructuring mechanisms and procedures for banks and
sovereigns (where they generally don’t exist) and improving insolvency and bankruptcy procedures for
households and non-financial corporates.
2. The Great Leveraging
Figure 1. Advanced Economies – Gross Debt by Sector (% of GDP) – Figure 2. Selected Countries – Non-financial Sector Gross Debt (% of
1980-Q2 2012
GDP) – 1980-Q2 2012
350
300
250
600
HH
NFC
Public
Private
NFS
US
UK
Japan
Germany
France
500
400
200
300
150
200
100
100
50
Jun-2012
0
Jun-2012
0
1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010
1980 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010
Note: Advanced Economy gross debt by sector is constructed as gross debt weighted by the time-varying shares of nominal GDP in 17 countries (Australia, Austria, Belgium,
Canada, Finland, France, Germany, Greece, Italy, Japan, Korea, the Netherlands, Portugal, Spain, Sweden, UK and US). NFS is the non-financial sector (sum of HHs, NFCs
and the GG). Gross debt is equal to total financial liabilities for HH and the public sector, and to total financial liabilities less shares and other equities for NFC. Values are on a
non-consolidated basis except for Portugal and Australia.
Source: IMF, OECD, National Sources and Citi Research
Debt has risen over the past few decades, almost everywhere in the advanced economies and
nd
according to most measures (McKinsey Global Institute (2010, 2012), Cecchetti et al (2011), BIS 82
Annual Report (2012), Tang and Upper (2010). Take gross non-financial sector (NFS) debt (the sum of
the gross debt of households, non-financial corporations, and the general government) in advanced
economies.
In a sample of 26 countries, gross NFS debt relative to GDP rose in every single one between 1995 and
2
3
today. For the 17 countries for which data are available since 1980, debt rose substantially in all. For
these 17 countries, the average NFS gross debt-to-GDP ratio, weighted by GDP shares (which is of
course the same as aggregate NFS gross debt as a share of aggregate GDP), almost doubled since
4
1980 (Figure 1), rising by just under 5ppts of GDP each year, on average.
2
The countries are Australia, Austria, Belgium, Canada, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece,
Hungary, Italy, Japan, Lithuania, the Netherlands, Norway, Poland, Portugal, Slovakia, Spain, South Korea, Sweden, UK, and US. In addition
to these countries, we often consider four countries for which data are only available for a shorter sample period: Ireland (from 2001), Latvia
(from 1998), Slovenia (from 2001) and Switzerland (1999 to 2009).
3 These countries are Japan, Italy, UK, Portugal, Spain, Belgium, Greece, France, Finland, Netherlands, US, Korea, Australia, Austria,
Sweden, Germany and Canada.
Let Di be the gross debt of country i (measured in a common currency) and Yi the GDP of country i
N
∑D
4
i
(measured in a common currency), then,
i =1
N
∑Y
N
≡
∑
i =1
Di
Yi
i
i =1
Y
∑Y
i
N
j =1
j
Elsewhere we also use the unweighted (or rather, equally weighted) average debt-to-GDP ratio,
2
1
N
N
Di
∑Y
i =1
i
.
Since 1995, this aggregate debt-to-aggregate GDP ratio still rose by 75ppts of GDP (4.5ppts of
5
GDP/year on average). Over this period, real GDP, measured in constant USD at market exchange
rates, grew by 37% (2.3%pa) in the 17 countries, and nominal GDP (measured in current USD) by 95%
(5.9%pa), so the growth in real and nominal debt levels was even larger than the growth in debt-to-GDP
ratios. In terms of the increase in the gross NFS debt to GDP ratio, the US was squarely in the middle of
the pack, the UK was in the top quartile, while Germany was in the group of countries with the smallest
increases.
The aggregate picture conceals much diversity. First, there is a difference between smaller and larger
countries: in our sample, larger countries on average had smaller proportional increases in their gross
NFS debt to GDP ratio and more of the total debt increase was accounted for by increases in public
debt. Thus, the simple average (not GDP-weighted) increase in the gross NFS debt–to-GDP ratio
across the sample of 26 countries between 1995 and H1 2012, was 94ppts of GDP (5.7ppts of GDP per
year) compared to the GDP-weighted average increase of 5.3 percentage points; it was 89ppts for the
6
17 countries with longer data series – countries that were on average still larger. These data do not
even include some of the small countries with the largest increases in debt, as data for the earlier period
are not available for them. For example, for Ireland and Latvia, the data are only available from 2001
and 1998, respectively, but between these dates and today, their total non-financial debt as a share of
GDP increased by 307ppts (19ppts per year) and 93ppts (5.6ppts), respectively.
Figure 1. Selected Countries – Non-Financial Sector debt/GDP ratio, change 1995-latest
300 %
250
200
150
100
10.5
HH
8.8
NFC
Public
NFS
9.5
7.5 5.8
1.0
15.7 17.2
6.0
8.5 9.5
5.4
13.5
16.1
6.6
50
5.3
5.1 6.0 4.1 8.6
8.4 5.5 10.5 5.3 5.6
7.3
5.0
0
-50
CY
PT
SP
UK
FR
LT
ET
AT
JP
GR
KO
BE
PL
HU
US
DN
IT
NL
GE
AU
NO
EA
SK
FI
CA
CZ
SW
-100
Note: Public is the general government. For the EA change corresponds to 1999-2011. Latest values are for Jun12, except for Italy, the Netherlands, Ireland (all Mar-12), and Cyprus (Dec-11). Numbers above the columns are
average growth rates of the nominal stock of gross debt in local currency between 1995 and the latest observation.
All values are expressed on a non-consolidated basis except for Australia and Portugal. See Figure for a list of
country labels.
Source: National sources, Eurostat, OECD, and Citi Research
Cyprus, Portugal and Spain were the countries in our sample that had the largest increases in NFS
gross debt to GDP ratios, with NFS gross debt-to-GDP ratios rising by at least 150ppts (or almost
10ppts/year). Ireland and Latvia would likely also have been in this category, if the data had been
available for the entire period. The countries which saw the largest increase in debt often shared certain
characteristics, including being an emerging European country (the Baltic countries, Hungary), being a
financial centre (Cyprus, UK, Ireland) or having had a housing boom (Baltics, Ireland, Spain). Despite
similarities in economic development and structure, some regional differences exist. For example, the
Czech Republic and Slovakia had among the smallest increases in their gross NFS debt ratio (while
For the broader sample of 26 countries, the GDP-weighted increase in gross debt since 1995 was 89% of GDP.
The GDP weighted average increase in real GDP (measured in constant USD) was 39% (2.4%pa), and nominal
GDP grew on average by 100.4% (6.3%pa) since 1995.
5
6
3
Hungary did not), and gross NFS debt ratios in Finland and Sweden also grew only modestly, while the
7
debt ratio increase in Norway was larger.
Gross debt ratios increased, on average, in each one of the household, non-financial corporate and
general government sectors. Of the 89ppts increase in the GDP-share-weighted gross NFS debt ratio
between 1995 and today, fairly little (less than 25ppts) was due to increases in the NFC gross debt ratio.
Households added 23ppts and general governments the rest – just over 43ppts. However, again, small
and large countries differed: in the simple cross-section of countries the contribution of the HH and NFC
sectors to the increase in the total non-financial gross debt ratio was much higher, on average, at 36ppts
of GDP and 40ppts, respectively, while general government debt increased the total non-financial debt
ratio by a mere 19ppts.
2.1 Debt and debt service relative to debt servicing capacity
Relative to disposable income, HH gross debt also increased strongly in most countries, and the relative
ranking of countries is also broadly similar. The level of HH gross debt currently exceeds annual
disposable income in the majority of advanced economies, and is more than twice annual disposable
income In Denmark, the Netherlands, Norway, Ireland, and Switzerland.
Increases in debt service ratios (interest and principal repayment) relative to disposable income for
private sectors (household and non-financial businesses) were more muted, on average, as increases in
indebtedness were at least partly (and in some cases fully or more than 100 percent) offset by
reductions in interest rates (see Figure 2). In Canada, where gross debt has fallen relative to GDP since
1995, private sector debt service ratios have fallen roughly by half since the peak in the early 1990s. In
Germany and Switzerland private sector debt service ratios also fell, and in France ratios rose only
modestly. In many countries, however, private sector debt service ratios also increased very
substantially in the decade leading up to the financial crisis, including in Ireland, Spain, the UK, and the
US, but also Denmark, Italy, Australia and Norway, despite falls in nominal and real interest rates over
this period.
Debt in the banking sector, and in the financial sector generally has increased enormously over the past
few decades (see Figure 3). In fact, in some financial centres, including Ireland and the UK, increases in
debt in the financial sector dwarfed increases elsewhere, and increases in simple, unweighted,
measures of gross debt and total balance sheet size generally suggest much larger increases in debt
than more complex (or esoteric) measures such as risk-weighted asset ratios, net debt/net worth, or
different definitions of leverage based on (non-independently verifiable) risk weights or on net debt
ratios.
7
In Norway public gross debt remained relatively stable over this period, while it fell sharply in Sweden, Finland and
Denmark. The differential between the CEE countries was mostly driven by differential increases in NFC gross debt.
4
Figure 2. Selected Countries – Private sector and Household Debt Service Ratios (%), 1980-2011
Ireland
Spain
35
. United Kingdom
30
23
Private sector
Household sector
Private sector
30
25
25
20
20
15
15
10
21
Private sector
Household sector
19
17
15
13
11
9
7
10
1980
1985
1990
1995
2000
2005
2010
United States
5
1980
1985
1990
1995
2000
2005
2010
Canada
5
1980
1985
1990
1995
2000
2005
2010
1995
2000
2005
2010
1995
2000
2005
2010
1995
2000
2005
2010
Denmark
20
22
20
30
Private sector
Household sector
19
Private sector
Private sector
Household sector
26
17
18
24
16
16
14
12
10
1980
28
18
1985
1990
1995
2000
2005
2010
France
22
15
20
14
18
13
16
12
14
11
12
10
1980
1985
1990
1995
2000
2005
2010
Germany
16
10
1980
1985
1990
Italy
16
21
Private sector
Private sector
15
15
14
14
13
13
12
12
11
11
10
1980
10
1980
19
Private sector
Household sector
17
15
13
11
9
7
1985
1990
1995
2000
2005
2010
Norway
1985
1990
1995
2000
2005
2010
Switzerland
5
1980
1985
1990
Australia
24
35
Private sector
Private sector
Household sector
22
45
40
Private sector
Household sector
30
20
35
18
25
30
16
25
14
20
12
15
20
15
10
1980
1985
1990
1995
2000
2005
2010
10
1980
1985
1990
1995
2000
2005
2010
Note: The debt service ratio is the sum of interest payments and debt repayments, divided by disposable income.
Source: BIS 82nd Annual Report (2012), and Citi Research
5
10
1980
1985
1990
Figure 3. Selected Countries - Financial Corporation debt/GDP ratio (%), 1995-2011 change
% of GDP
913% 813%
550
550
450
450
350
350
250
250
150
150
50
50
-50
-50
UK
IR
CY
NL
DN
FR
FI
PT
SP
GE
EA
IT
SW
KO
GR
BE
AT
CA
AU
NO
ET
LT
PL
JP
US
CZ
HU
SK
Note: Financial Corporations include NCBs. Gross debt is equal to total financial liabilities less shares and other equities from national balance sheet statistics. For the US,
financial corporations’ debt is “credit market instruments”. For the EA, the change is between 1999 and 2011 and between 2001 and 2011 for Ireland. All values are expressed
on a non-consolidated basis except for Australia and Portugal.
Source: OECD, Eurostat, National sources, and Citi Research
In terms of the levels of NFS gross debt, in our sample of 30 countries, Ireland, Cyprus and Japan are
the mostly highly indebted countries, with NFS gross debt in each case amounting to close to or more
than five times GDP (Figure 4). Portugal and Spain also have very high levels of gross NFS debt. The
average level of gross NFS debt across the countries in our sample is three times the level of GDP
(301% of GDP). The US (275% of GDP), but also Italy (304%), find themselves in the middle of the
pack, and countries like Greece (256%), Germany (258%), but also the Baltics and most CEE countries,
are at the lower end of the spectrum. Lithuania is the country with the lowest level of (NFS gross) debt in
our sample at 171% of GDP.
The composition of debt levels across sectors also varies a lot between countries. In many countries,
including Belgium, Ireland, Spain, but also Sweden, the Baltics and the CEE countries, non-financial
corporates account for most of the gross debt. Only in a few cases is public debt the major contributor to
total NFS gross debt, the most notable case being Japan, but also in Greece, Italy and the US. In some
countries, including Ireland, Cyprus, Portugal and Spain, all three non-financial sectors have relatively
high levels of gross debt.
Figure 4. Non-Financial Sector Gross debt/GDP ratio, Latest
600
%
600
%
HH
500
500
400
400
300
300
200
200
100
100
0
NFC
Public
NFS
0
IR
CY
JP
PT
BE
CA
SP
FR
NL
DN
UK SW EA
IT
NO
KO
FI
US AT HU GE GR CH AU SN LV CZ PL
ET SK LT
Note: NFS is the sum of HHs, NFCs and GG. Latest values are for Jun-12, except for Italy, the Netherlands, Ireland (Mar-12), Cyprus, the EA (Dec-11) and Switzerland (Dec09).
Source: National sources, OECD, Eurostat, and Citi Research
6
2.2 The other side of the balance sheet – changes in net debt and net worth
HHs, NFCs and the public sector also have assets that could potentially be sold to reduce debt or
generate income used to service debt. It therefore makes sense to consider these asset holdings when
assessing debt sustainability, even though the potential liquidity, currency or maturity mismatch between
assets and liabilities suggests that netting assets and liabilities may not generally be advisable.
For broad measures of net debt which only reflect liquid asset holdings, such as gross debt minus
holdings of currency and deposits, the picture is often qualitatively and quantitatively quite similar to that
8
for gross debt (Figure 6). Such levels of net debt have generally increased across most countries
across all three sectors, and the relative ranking of countries according to the broad net debt increase is
similar to the case of gross debt. The size of the increase is smaller, as holding of currency and deposits
have generally increased – a simple average of narrow net debt increased by 62ppts of GDP against
93ppts of GDP for gross debt. In relative terms, the UK in particular looks somewhat better once we
allow for currency and deposit accumulation.
Narrow measures of net debt did not increase to the same extent as gross debt, and often fell. For
example, a narrow measure of net debt that accounts for all financial assets (i.e. including equity and
fixed income claims and pension fund assets, but not ownership claims on land and real estate or
unfunded pension entitlements) fell by 37% of GDP between 1995 and 2011 (or 23.1% in GDP-weighted
terms), most of it due to the fact that net debt of NFCs decreased on average (Figure 5), even though
HH net debt also fell. However, the reduction or at least moderation in levels of net debt was not
generally driven by an increase in private saving rates, but rather an increase in asset values, mainly on
stocks, in the 1990s.
Figure 5. Selected Countries – Non-Financial Sector Narrow Financial Figure 6. Selected Countries – Non-Financial Sector Broad Financial
Net Debt, 1995-2011 change
Net Debt, 1995-2011 change
150
%
200
100
%
150
50
100
0
50
-50
0
-100
-50
-150
-200
HH
NFC
Public
HH
NFS
NFC
Public
NFS
-100
-150
GR
SK
PL
LT
CZ
SP
UK
ET
HU
US
IT
AU
JP
GE
PT
CY
AT
FR
CA
FI
NL
NO
DN
SW
BE
PT
SP
CY
HU
FR
AT
ET
LT
IT
UK
NO
PL
DN
US
BE
GR
SK
JP
NL
SW
AU
FI
CZ
GE
CA
-250
Note: Narrow financial net debt is defined as gross debt minus total financial assets.
All values are expressed on a non-consolidated basis except for Australia and
Portugal. Countries missing include Ireland (data start only from 2001), Latvia (1998),
Slovenia (2001) and Switzerland (1999)
Note: Broad financial net debt is defined as gross debt minus holdings of currency and
deposits. All values are expressed on a non-consolidated basis except for Australia
and Portugal. Countries missing include Ireland (data start only from 2001), Latvia
(1998), Slovenia (2001) and Switzerland (1999)
Source: National sources, OECD, and Citi Research
Source: National sources, OECD, and Citi Research
2.3 Non-financial assets are significant
Our discussion above misses some very substantial components of wealth and net worth, notably
ownership claims to real estate or land. Unfortunately, the availability of data on holdings of real estate
and other real assets is quite limited and measurement and definitional issues make cross-country
comparisons tricky. But the available data indicate that these non-financial assets are sometimes of a
similar order of magnitude for households alone as total financial assets for the entire non-financial
sector. For example, in Spain HH non-financial assets were valued at almost 500% of GDP in 2011
(Figure ), and in France at just under 400% of GDP in 2010. In Germany (2009) or the US (2011) on the
8
We call a measure of net debt that only deducts currency and deposits from gross debt ‘broad’ as only a narrow
range of assets is deducted from gross debt. Narrow net debt therefore reflects a broader range of assets.
7
other hand, non-financial assets of HHs were valued just at around 150% of GDP, although for both
countries this excludes land. There is also at least a suspicion that the prices at which Spanish real
assets (and possibly French real assets as well) were valued in these data err on the side of generosity.
As real estate prices have risen, the value of non-financial assets has generally increased over the last
few decades, often supported further by a boom in real estate construction. Changes in the value of
these assets can easily overwhelm other changes on HH and business balance sheets. Many countries
with long and large real estate booms have seen large falls in HH net worth in recent years. In the case
of Spain, HH net worth has fallen by around 100% of GDP since 2007, mostly driven by a reduction in
the value of non-financial assets. However, in those same countries, HH net worth is often still above the
levels seen in the early 2000s. The fall in real estate valuations in Spain has brought HH net worth back
to the levels of around 2004, with large increases in the years prior to 2004. Of course, continuing falls
in house prices in Spain are likely to erode HH net worth in Spain substantially further in the years
ahead.
In the US, HH net worth also fell along with house prices by about 100% of GDP in 2007 and 2008, but
has recently stabilized. In countries that have not seen a major housing bust, HH net worth is generally
close to previous peaks, with the exception of Japan, where HH net worth is still down substantially from
the peak in the early 90s and HH net worth has continued falling at a gradual and slowing pace since
then (Figure 8).
Figure 7. Selected Countries – Holdings of Non-financial Assets by Figure 8. Selected Countries – Household Net Worth (% of GDP), 1990Households (% of GDP), latest
2011
500
800
450
700
400
600
350
500
300
250
400
200
300
150
200
100
100
50
US
Japan
Canada
France
Spain
Germany
0
0
1990
SP FR AU EA UK NL KO JP DN CA GE US SN AT HU IT CZ FI
1995
2000
2005
2010
Note: Values correspond to holdings of total non-financial assets by HHs. Non- Note: Net worth is defined as total assets (financial and non-financial) minus total
Financial Assets include fixed assets, inventories, fisheries, and land. For Italy, values financial liabilities.
correspond to total dwellings, while for the UK and Hungary values exclude land, For Source: OECD, IMF, FED, Bank of Spain and Citi Research
the US, Germany, Slovenia, Austria, Denmark, Finland and Spain, values exclude land
and inventories, due to data availability. Latest is end-2011 except for Japan, Canada,
Australia, France, Czech Republic, Korea, Netherlands, and Slovenia (2010), and
Germany, Italy and Hungary (2009).
Source: National Sources and Citi Research
3. The Drivers of the Great Leveraging
There were many drivers of the increase in debt in the last few decades (starting around 1980 in the UK
and the US), including financial sector liberalisation, financial ‘innovation’, a boom in real estate prices
and construction (themselves fed by the growing debt issuance), a fall in lending standards, a global fall
in real interest rates (often associated with the ex-ante saving glut produced by China and other highsaving EMs and oil-producing countries), and the perception of a fall in macroeconomic volatility and of
enduring faster growth – the Great Moderation.
Of course, profligacy of many governments in the run-up to the financial crisis, which was partly fed by a
misidentification of highly cyclical, or at least unsustainable, revenue increases as permanent, played a
role, too.
8
The recent global recession and financial crisis clearly played a major role in the build-up of sovereign
debt through the collapse of certain sources of unsustainable tax revenues the sovereigns had grown
dependent on (especially taxes on real estate and on financial sector earnings). Other sources of public
debt growth were the operation of the automatic fiscal stabilizers during the downturn caused by the
crises, the discretionary measures to provide fiscal stimuli and the bail-outs of banks, other financial
institutions and sometimes non-financial companies deemed too systemically significant or too politically
well-connected to be allowed to fail. In Europe, and especially the EA, this migration of bad and
impaired private sector assets to the public balance sheet continues. That there are limits to this
migration because at some point ‘too big to fail’ gives way ‘to too big to bail’ should have been clear
since the collapse of Iceland’s banking sector in the Fall of 2008, with recent reminders from Greece,
Ireland, Spain, Portugal, Cyprus and Slovenia.
These factors implied that both the supply curve and the demand curve for credit shifted outwards in the
two or three decades leading up to the financial crisis. Some of the drivers, such as the reduction in
global real interest rates and a perception of greater macroeconomic stability, likely affected both private
and public debt accumulation. The rise in real estate prices and the fall in lending standards likely had a
stronger effect on private debt than on public debt. Many of the drivers were also inter-related and often
reinforced each other. For much of this period, there was what seemed to be a virtuous circle where
credit growth boosted demand, which in turn boosted economic growth and asset prices which both
improved superficial mark-to-market measures of balance sheet health and underpinned further
increases in credit and demand.
One arithmetically obvious candidate as a driver of the increases in debt-to-GDP ratios can plead ’not
guilty’: The rise in debt-to-GDP ratios was not generally due to a fall in or weak growth of the
denominator, i.e. a lack of (real) GDP growth. Real (and also nominal) GDP growth was generally
positive in most countries in recent decades, and was often higher in the ‘95-‘08 period than in the one
or two decades prior to that – although, not surprisingly, lower than real and nominal GDP growth in the
‘Golden Age’ for the advanced economies between 1946 and 1973. Indeed, the countries with the
highest rates of nominal or real GDP growth between 1995 and 2008 generally tended to have larger
increases in (NFS gross) debt ratios and the relationship was pretty tight.
Likewise, rising asset prices in recent decades, and rising of real estate prices in particular, contributed
to the credit boom. As asset prices continued to increase, many households and corporations
interpreted these higher asset prices as sustainable. Even more extravagantly, particularly for housing,
many extrapolated the growth in house prices into the indefinite future. Some of the debt was explicitly
linked to asset price increases – mortgages got larger (in absolute terms or relative to income or GDP)
as the price of houses increased. In other cases, the increase in asset values was leveraged by
withdrawing equity from homes to finance consumption and other spending
Financial liberalization, i.e. the deregulation of financial markets both domestically and for cross-border
transactions, spurred what was at the time often referred to as ‘financial innovation’ and ‘financial
engineering’, but today is more often called ‘financial excesses’ and sometimes worse, including
regulatory and tax arbitrage. Deregulation affected a number of areas, including the reduction of credit
and interest rate controls, the reduction of entry barriers into the financial sector and of restrictions on
cross-border capital account transactions, a lowering of prudential regulations and an easing of
supervision in the banking sector and in securities markets. There were also many reductions in reserve
requirements for financial institutions, and reductions in effective capital requirements through
disintermediation out of more tightly regulated financial intermediaries, products and activities into more
loosely regulated ones, like the shadow banking sector.
Macroeconomic factors also played a role, beside the effect that adaptive or extrapolative expectations
likely implied that robust current economic growth and rising asset prices fed into higher expectations of
future growth and further increases in asset prices. Real interest rates were low in many countries, as
nominal interest rates fell by more than inflation. And the so-called ‘Great Moderation’ phase of low
macroeconomic volatility (see Stock and Watson (2002), Bernanke (2004)) may also have contributed to
an increase in credit demand and supply, as both debtors and creditors, supervisors, regulators and
those in charge of financial legislation underestimated the degree of riskiness of economic activity, as
actual volatility fell. In Europe, the introduction of the euro gave an additional boost, through reductions
9
in interest rates in many countries, rapid financial integration and rapid (if with hindsight unsustainable)
economic growth in some of the countries. The fact that from the launch of the euro in 1999 till 2008,
spreads over ten year Bunds of Irish, Portuguese, Spanish and Italian sovereign debt rarely rose above
25bps, and that the same extraordinarily low spread prevailed for the Greek 10-Y sovereign bond from
2001 till 2008 bears testimony to the wholesale loss of common sense in the markets, and the resulting
massive underpricing of differences in EA sovereign risk (see Buiter and Sibert (2006)).
4. Why debt matters today
There are at least two specific and concrete reasons why debt matters in advanced economies today.
The first is that excessive debt can cause systemic crises, and such systemic crises can have very large
and potentially long-lasting effects on actual and potential output, unemployment, and capacity use. The
second reason is that if debt is considered excessive, the process of bringing down debt can be longlasting and painful, even if it does not create a financial crisis or even after the crisis phase has passed.
Coordination problems in the process of debt reduction often substantially increase the private and
social cost of debt reduction, as agents attempt to raise their saving rates in response to the excessive
level of their debt without a matching increase in planned investment (capital expenditure) by either the
agents planning to raise their saving rates or by other agents at home or abroad. This can give rise to
Keynes’s so-called ‘paradox of thrift’.
4.1 What is deleveraging?
Language use is non-uniform when it comes to debt and deleveraging, something which can create
confusion in a discussion of their significance. It pays to be precise.
The flow-of-funds account of a sector defines its financial surplus – the excess of its saving over its
capital formation (capital expenditure or investment in real reproducible capital) as the value of its net
acquisitions of financial assets minus the value of the additional net financial liabilities it incurs over
some period of time. The change in a sector’s net worth (net worth is also called financial wealth, capital
or equity, although all these terms have multiple different meanings as well), is its saving plus the capital
gains, (or minus the capital losses) on its existing assets and liabilities, real and financial.
Gross balance sheet contraction or gross deleveraging for short is a reduction in the size of the
balance sheet (real and financial) without a change in net worth (i.e. net saving plus capital gains for the
entity or sector in question is zero). Gross deleveraging can be the result of capital losses on real and
financial assets and liabilities, which we shall describe as passive gross deleveraging, as well as of
active gross deleveraging, that is, a reduction in the size of the balance sheet through equal value
reductions in stocks of assets and liabilities at given prices. Active gross deleveraging does not require
any change in either the flow of saving or the flow of investment spending by any individual agent or
sector. However, active gross deleveraging does require coordination of gross sales and purchases of
assets or of gross lending and borrowing across agents and sectors. Either asset markets or some other
mechanism must coordinate the planned transactions in each of the assets and liabilities and translate
them into actual sales and purchases.
However, the problems associated with the ‘paradox of thrift’ discussed below need not strike if all that is
required is passive or active gross deleveraging by one, several or all sectors in the economy. Active
gross financial balance sheet contraction or active gross financial deleveraging is a reduction in the
size of the financial balance sheet alone, that is, excluding the physical capital assets, but with the value
of financial assets and liabilities shrinking by the same amount (at current prices). Capital gains are
excluded.
Net wealth accumulation or net deleveraging by a sector means an increase in the net worth of that
sector, either through saving or through capital gains. Active net wealth accumulation or active net
deleveraging by a sector, which ignores capital gains or losses, is therefore just another name for
positive saving by that sector. Although higher saving is good news from the point of view of the future
growth of actual and potential output if a planned increase in saving is matched by an equal planned
increase in investment, the paradox of thrift warns us about coordination failures between those who
would raise their saving and those who would boost their investment. These coordination failures can
result in short-run and medium term negative impacts on output and employment from a poorly
10
coordinated saving boost by one or more sectors. Active net financial wealth accumulation or active
net financial deleveraging means running a financial surplus, that is, saving exceeding investment.
Note that the terms gross and net are not used to denote saving or investment inclusive of capital
depreciation or excluding it. Gross debt is all liabilities. Net debt is liabilities minus assets.
4.2 Debt causes systemic crises
High debt held by some agents or institutions can make them vulnerable to shocks and unanticipated
(by them) changes in their economic environments. It enhances the fragility of these agents and
institutions. High indebtedness of many agents or institutions, especially if the economic-financial
network is characterised by a high degree of complexity, can result in opacity of that network and in
widespread ignorance (throughout the network and among supervisors and regulators) about the
distribution of exposures and counterparty risk, risk chains and clusters across the network. This can
create systemic fragility.
The high debt burdens in the DMs brought with them vulnerabilities that triggered systemic financial
crises recently. The first was a (mainly) private sector financial crisis – especially in the banking and
shadow banking sectors of the North-Atlantic region, that started in August 2007 and lasted until the end
of 2009. The second crisis is the sovereign debt and banking sector crisis that erupted in the euro area
(EA) at the beginning of 2010, and is still ongoing.
Both the North-Atlantic financial crisis and the EA sovereign debt and banking crises have impacted
more severely on output and employment because in many DM national economies (the main
exceptions have been Germany, Italy and Japan), the private sector too has become highly indebted.
A few simple scatter plots can illustrate the role that debt has played in recent poor economic
performance. Figure 9 plots the difference between what the level of real GDP in 2011 would have been
had real GDP continued to grow at its pre-recession (1997-2004) trend growth rate and actual GDP in
2011 (the ‘GDP loss’) against the change in the ratio of NFS gross debt to GDP between 2001 and
2007. The relationship between the recent growth performance and the extent of the prior buildup in
NFS gross debt is strongly negative – for a 10ppts larger increase in the pre-crisis non-financial sector
gross debt-to-GDP ratio, the GDP loss has been 2.2ppts higher, on average, in our sample of 30
industrial countries. The increase in debt alone can ‘explain’ – in a purely statistical sense – almost 40%
of the variation in GDP performance relative to trend.
Figure 9. Selected Countries – GDP loss in 2011 (% vs trend) and Prior Figure 10. Selected Countries – GDP loss in 2011 (% vs trend) and 2007
Increase in Debt
Debt Levels
60
GDP loss in 2011 (% of pre-crisis) trend)
GDP loss in 2011 (% of pre-crisis trend)
60
50
40
30
20
10
0
-10
-40
-10
20
50
80
50
40
30
20
10
0
-10
100 130 160 190 220 250 280 310 340 370 400 430
NFS Gross Debt (as % of GDP), 2007
Change in NFS Gross Debt (as % of GDP), 2001-07
Note: GDP loss is the deviation of real GDP from its pre-recession trend. The pre-recession trend in calculated as the average growth in real GDP between 1997 and 2004.
Source: OECD, World Bank, National Sources and Citi Research
11
Interestingly, there is very little evidence of a statistical relationship between the GDP loss and the levels
of the gross debt to GDP ratio at the end of 2007 – the beginning of the North Atlantic financial crisis
(Figure 100). Although we do not want to over-emphasise the significance of this simple (possibly
simplistic) statistical exercise, to us the finding that changes in NFS gross debt ratios are significant in
explaining the variation in cross-country experience suggests both that country-specific factors are very
important (so cross-country comparisons of levels of debt cannot tell the whole story) and that some of
the increase in the NFS gross debt ratios in the years of the Great Leveraging was excessive and is
therefore likely to result in some mean reversion towards historical averages. The pre-crisis increase in
debt ratios may also be a better guide to the extent of desired deleveraging than the realised debt
reduction or the level of debt since the beginning of the crisis – as the process of deleveraging is
nowhere complete and in many countries and sectors has not yet started. The reason is that, in an
environment where there has been a widespread increase in the (precautionary) desire to save, the socalled ‘paradox of thrift’ can exert powerful effects and actual saving may well fall short of desired
saving, a point we will discuss below in more detail.
Previous episodes of deleveraging after financial crises have also generally been associated with poorer
9
economic performance. Figure 11 to Figure 15 depict the behavior of several macroeconomic variables
(relative to their pre-crisis trend) in response to financial crises that were associated with deleveraging in
10
86 countries between 1960 and 2006. There were 18 episodes of financial crisis associated with
deleveraging in our sample, starting with Chile in 1981 and ending with the Dominican Republic in
11
2003.
On average in this sample of episodes, the stock of private sector credit as a share of GDP grew by
30ppts in the eight years preceding the financial crisis and fell by around 15-20ppts over the following
eight years (Figure 11). The effect of financial crises on real GDP was fairly dramatic: GDP fell by
around 10ppts relative to the pre-crisis trend, on average, in the first two years and made up very little
ground in subsequent years. Compared to this historical average of financial crises, the GDP
performance of the US, the UK and the euro area to date have actually been broadly similar, with the UK
underperforming the historical average of our 18 episodes moderately. The increase in private sector
credit in the UK prior to the financial crisis much exceeded those in the US or euro area, and also that of
the average in the 18 countries in our past financial crises sample, which may partly account for the
UK’s sub-par economic performance since 2007.
9
See also Michael Saunders (2011), “What’s the Damage? Debt and Growth in Deleveraging”, Global Economic
Outlook and Strategy - Prospects For Economies And Financial Markets In 2012 And Beyond, “What’s the Damage?
Medium-Term Output Dynamics after Financial Crisis”, IMF World Economic Outlook, September 2009. See also
Cecchetti et al (2012).
10
Episodes of deleveraging are identified, following McKinsey (2010), as episodes where the ratio of gross NFS debt
to GDP has fallen for at least three years and by at least 10ppts of GDP, or where the stock of nominal debt declined
by 10ppts or more. Unlike in other parts of this study, NFS debt is defined as the sum of private sector credit and
public sector debt (both provided by the IMF), due to data availability for this longer sample period. Financial crises
are taken from Laeven and Valencia (2008) and episodes of financial crises-cum-deleveraging are the interface of
the two lists. The exercise provided a total of 31 deleveraging episodes, of which 18 were preceded by a financial
crisis. These 18 episodes were: Argentina (2001), Bolivia (1994), Chile (1981), Dominican Rep (2003), Ecuador
(1998), Finland (1991), Indonesia (1997), Japan (1997), Korea (1997), Malaysia (1997), Mexico (1994), Nicaragua
(2000), Norway (1991), Paraguay (1995), Philippines (1997), Sweden (1991), Thailand (1997), and Uruguay (2002).
Please see the appendix for further details.
11
Our filter rules out ongoing and very recent deleveraging episodes by construction. Our filter also excludes
transition economies during the period of the transition (e.g. Russian and Ukraine) because the output developments
in these economies were strongly related to the shift away from central planning rather than to financial crises per se.
12
Figure 11. Selected Countries – Change in Domestic Credit to the Figure 12. Selected Countries – Real GDP Versus Pre-Crisis Trend,
Private Sector (% of GDP), 2007-11
2007-16F
90
80
70
60
50
40
30
20
10
0
-10
12
8
4
0
-4
-8
-12
-16
-20
-24
-28
EMU
UK
US
Financial Crisis
Average
-8 -7 -6 -5 -4 -3 -2 -1
0
1
2
3
4
5
6
7
US
UK
F
EMU
Financial
Crisis
Average
8
-1
0
Distance in Years from First Year of Crisis
1
2
3
4
5
6
7
8
Distance in Years from First Year of Crisis
Note: Accumulated change in domestic credit to the private sector from T-8 to T+8,
where T is the beginning of the banking crisis. For the US, the crisis is dated (T=0) in
2007, for the UK in 2008 and for the EA in 2009. The shaded area covers the interquartile range of previous episodes, which indicates the middle 50 percent of all crises.
Note: The shaded area corresponds to the interquartile range of previous episodes,
which indicates the middle 50 percent of all crises. From T+4 (2012), real GDP
corresponds to Citi Research forecasts.
Source: IMF, World Bank, BEA, Eurostat and Citi Research
Source: IMF and Citi Research
As Figure 13 and Figure 14 show, both private consumption and investment fall sharply in the aftermath
of financial crises with deleveraging. The fall in consumption is similar to the fall in GDP, but the fall in
investment is more than three times as large – an example of the investment accelerator at work. Net
exports on the other hand add substantially to GDP growth, but the contribution is almost entirely due to
import compression, while exports were on average flat in these episodes. This suggests that the gains
in external competitiveness and real exchange rate depreciations experienced by many of the countries
in the sample (those that had a floating exchange rate, devalued a currency peg or abandoned a
currency board) following their financial crises, boosted the trade balance in much the same way as
fiscal austerity would at a constant real exchange rate: by depressing demand and lowering living
standards. The improvement in external competitiveness was often associated with a worsening in the
terms of trade that acted like a tax by lowering household real income (measured in terms of the
consumption bundle).
Figure 13. Selected Countries – Real Private Figure 14. Selected Countries –
Consumption vs Pre-Crisis Trend, 2007-11
Investment vs Pre-Crisis Trend, 2007-11
20
10
0
-10
-20
-30
-40
-50
-60
-70
-80
8
US
4
UK
EMU
0
Financial Crisis
Average
-4
-8
-12
-16
-20
-24
-28
-1
0
1
2
3
4
5
6
7
8
Distance in Years from First Year of Crisis
US
UK
Real Figure 15. Selected Countries – Real Net
Exports vs Pre-Crisis Trend, 2007-11
50
46
42
38
34
30
26
22
18
14
10
6
2
-2
-6
EMU
Financial
Crisis
Average
-1
0
1
2
3
4
5
6
7
8
US
UK
EMU
Financial
Crisis
Average
-1
0
Distance in Years from First Year of Crisis
1
2
3
4
5
Note: The shaded area corresponds to the interquartile range of previous episodes, which indicates the middle 50 percent of all crises.
Source: IMF, World Bank, BEA, Eurostat and Citi Research
Excessive debt not only creates the vulnerabilities that lead to financial crises. It also increases the cost
of financial crises, as Figure 16 and Figure 17 show. In these figures we divide the sample into two
groups, depending on the increase in debt prior to the financial crisis during the four years before the
12
crisis. The average increase in the debt-to-GDP ratio was 18ppts of GDP for the group with the larger
debt increases, while this debt ratio actually fell by 3.5ppts in the other group, on average, in the 4 years
13
before the financial crisis. As we can see in Figure 16, the fall in GDP for the larger-debt increase
12
We define a ‘large’ increase in debt as an increase above the median of all episodes over the three years leading
up to the crisis (between T-4 to T-1), as in IMF (2012).
13
The cutoff increase in NFS debt was 8.8ppts of GDP.
13
6
Distance in Years from First Year of Crisis
7
8
group was almost twice what it was in the other group two years after the crisis. Even worse, it
continued to fall relative to the trend, while the ‘smaller debt increase’ group crept back to trend. The fall
in private sector debt post-crisis on the other hand was much steeper for the large debt increase group,
while private sector credit fell modestly. The falls in investment and consumption were larger and more
persistent, and so were the increases in saving rates for the countries where debt had risen more ahead
of the financial crisis.
For the many countries that had large increases in private debt up until the North Atlantic financial crisis,
the outlook may therefore be even gloomier than the average experience depicted in Figure 17 would
suggest. Out of the 30 countries in our sample, all but 6 (Germany, Netherlands, Canada, Japan,
Slovakia and Czech Republic) had increases in NFS gross debt in the three years leading up to the
crisis that would have put them into the ‘larger debt increase’ group of the financial crisis sample.
Figure 16. Real GDP Versus Pre-Crisis Trend After Banking Crises, Figure 17. Change in Domestic Credit to the Private Sector (% of GDP)
1980-2011
after Banking Crises, 1980-2011
2
0
-2
-4
-6
-8
-10
-12
-14
-16
-18
40
35
Large debt increase
30
Low debt increase
25
Large debt increase
20
Low debt increase
15
10
5
0
-1
0
1
2
3
4
5
6
Distance in Years from First Year of Episode
-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8
Distance in Years from First Year of Episode
7
Note: “Large debt increase” group includes countries with above-median increases in Note: “Large debt increase” group includes countries with above-median increases in
gross debt in the three years leading up to the crisis.
gross debt in the three years leading up to the crisis.
Source: IMF, World Bank, and Citi Research
Source: IMF, and Citi Research
4.3 Even ‘orderly’ deleveraging is likely to be costly - coordination failures and the ‘paradox of
thrift’
Systemic crises are particularly painful, as they often combine impairments to credit availability with an
additional desire to increase saving. But debt reduction can impose heavy costs even outside episodes
that feature a weak banking system and widespread restrictions on credit availability.
Increased planned saving implies lower spending on goods and services, and lower net income from the
production of these goods and services for somebody unless that shortfall in demand is somehow
replaced by increased spending elsewhere. However, the main reason that heavy economic, social and
human costs are often associated with deleveraging by the public and private sectors is the fundamental
coordination problem faced by decentralised capitalist market economies with large financial sectors
and significant financial intermediation. This coordination problem can arise from an increased desire to
save, no matter whether this was driven by liquidity or solvency concerns. This coordination problem
has preoccupied macroeconomists since Keynes, and probably before Keynes also.
In a closed system (like the world economy) it has to be the case that system-wide aggregate saving
has to equal system-wide aggregate investment. Even though this relationship holds identically ex-post,
that is, for realized saving and investment flows and for actual purchases and sales of financial
instruments, it need not hold ex-ante, for planned investment and saving and for planned financial asset
accumulation and decumulation. It can therefore be viewed, ex-ante, as a coordination constraint.
Failure for it to hold ex-ante can result in the revenge of the ‘paradox of thrift’.
Unlike in a subsistence peasant economy, where a decision to save (that is, a decision to abstain from
consumption of current goods and services), constitutes ipso facto an identical decision to invest (to add
to the stock of real reproducible capital – by adding the grain that is not consumed to the stock of grain
to be used in sowing for the next harvest), in a decentralised financially developed economy,
14
households save (abstain from consumption) but invest very little in the form of capital expenditure.
Instead household saving flows into a range of financial instruments. Likewise, firms do most of the
capital expenditure, but when they cut their investment, they don’t raise corporate consumption demand
by the same amount. Instead they either retain profits or distribute their disposable income to
households and other beneficial owners. This physical, institutional and legal separation of the saving
and investment decisions places a big coordination burden on the financial markets linking households
and NFCs, and, in a more complex world, households, NFCs, financial institutions, the government and
the rest of the world. Much of the time, financial markets do a reasonable job of performing the task of
balancing saving and investment at levels of employment close to full employment. But as the years
since 2007 remind us, there can be spectacular (financial) market failures, sometimes aided and abetted
by labour market failures and by policy failures.
In the Keynesian textbook model, the paradox of thrift described a situation where a planned increase in
saving by households (that is, a planned or ex-ante reduction in household consumption demand at a
given level of household disposable income) weakens output and employment to the point that actual,
realized or ex-post saving instead of rising as planned, rises less, stays constant or even falls because
lower consumption demand lowers production and thus household disposable income. We can see
variants of these destructive feedback loops at work throughout the periphery of the euro area, in the UK
and in core EA countries like the Netherlands, where the realization in 2011 by households that they had
excessive gross (mortgage debt) and illiquid assets that were falling in value contributed to a major
slowdown in private consumption demand and a recession.
It is key to recognise that the ‘paradox of thrift’ is not restricted to the consequences of fiscal austerity
implemented by governments that are trying to reduce their debt burdens or deficits. It applies to the
adverse feedback loops created by any economic agent, or sector, whose individually rational defensive
actions when faced with an unsustainable debt and deficit configuration (or with any other reason for
boosting his individual saving) creates negative income or demand externalities for other agents in the
economy by cutting his consumption and thus the effective demand for output and actual output –
externalities that are not effectively captured by the price signals, quantity signals or other information
conveyed by these actions. Indeed, the original paradox of thrift does not involve fiscal austerity at all.
Instead, it analysed the consequences of a ‘spontaneous’ decision by the private sector to raise the
household saving rate.
Financial markets and financial asset prices and yields are supposed to coordinate the spending and
saving plans of producers, consumers and other economic agents. Unfortunately, they do so least
effectively when it is most needed. Allocation over time and the pooling, sharing, pricing and trading of
risk are the areas of economics where both markets and governments are weakest. The incompleteness
of markets (due to private and asymmetric information and costly contract enforcement, poor
governance of private and public enterprises, the inability of governments to commit their successors
and indeed often even themselves for any length of time, a pervasive lack of trust in people and
institutions and a scarcity of all key ingredients of social capital) is a major obstacle to the efficient
allocation of resources over time and across states of nature.
5. How to reduce debt over time? - mostly for sovereigns, but with lessons for
other sectors
To guide our discussion of the various ways to bring down sovereign debt (or indeed the debt of any
14
economic agent), an accounting identity is useful:
∆d = (r − g )d − s
= (i − π − g ) d − s
(1)
Here d is the net debt-to-GDP ratio, i is the one-period (strictly the instantaneous) nominal interest rate,
g is the growth rate of real GDP, r is the ex-post (actual or realized) one-period real interest rate, π is the
actual rate of inflation and s is the primary (non-interest) surplus as a share of GDP. To get from the first
14
The identities hold only in continuous time. For discrete periods, slightly messier expressions exist.
15
identity in equation (1) to the second, we use the fact that the ex-post real interest rate equals the
nominal interest rate minus the actual rate of inflation (r = i – π) . Equation (1) says that the change in
the net debt-to-GDP ratio is given by the primary surplus (as % of GDP) and a ‘snowball’ factor that
depends on the difference between the real interest rate and the growth rate of real GDP.
Now let
~
r
be the ex-ante or expected real interest rate. The nominal interest rate equals the ex-ante
real interest rate plus the expected rate of inflation, π , that is,
e
(
i=~
r + π e . It follows that:
)
∆d ≡ ~
r +π e −π − g d − s
(2)
From equations (1) and (2) we can see that there are five distinct ways to deleverage (strictly to engage
in net deleveraging), that is, to reduce d. When we list these five modalities, we are strictly keeping all
else constant, even if this may make no sense in practice because there are other economic
relationships linking the variables in equations (1) and (2). The five ways to deleverage are:
I.
Practice fiscal austerity (increase s by increasing the numerator of s, tax revenues minus noninterest public spending in the case of the public sector).
This approach – fiscal pain through cuts in public spending or tax increases – is painful and
unpopular. This is partly because, even holding constant the level of economic activity (GDP and
employment), public spending cuts deprive the beneficiaries of public spending of some of the
benefits they receive, whether in cash or in kind, and because tax increases reduce disposable
income or wealth. In addition, since the real world is Keynesian, at least in the short run, fiscal
tightening almost always depresses economic activity. The expansionary contractionary fiscal policy
paradigm of Giavazzi and Pagano (1996) and Alesina and Ardagna (2010) is a theoretical curiosum.
The announcement effects, today, of a credible commitment to future fiscal austerity may be
expansionary (because it lowers long-term interest rates), but when the pre-announced fiscal
tightening occurs, it will almost surely depress aggregate demand and economic activity. There is also
no empirical evidence of a Keynesian ‘Laffer curve’ where tax increases or cuts in public spending
reduce economic activity to such an extent that the tax base shrinks to the point that the deficit
increases despite the fiscal tightening (see Cottarelli and Jaramillo(2012)), although excessive and
misdirected fiscal zeal can do lasting damage to potential output, by depressing capital formation and
through hysteresis in the unemployment rate.
II.
Reduce the effective nominal interest rate on the public debt, i.
This can be done (a) by influencing the market equilibrium interest rate (say through QE or other
large-scale asset purchases of sovereign debt or private debt), (b) by ensuring the funding of the
sovereign by the private sector (typically in the primary markets) at a cost below the market
equilibrium interest rate, that is, through financial repression, or (c) by getting access to sovereign
funding at below-market interest rates from external official entities, as Greece, Portugal and Ireland
do through their access to the concessional and conditional funding of the IMF, the Greek Loan
Facility, the EFSF and soon also the ESM. Holding constant the actual inflation rate, π, this is
equivalent to lowering the ex-post real interest rate, r. In the post-World War II sovereign debt
deleveraging in the US and the UK, as well as in many other countries, this has been an important
mechanism for deleveraging (see Sheets (2012, 2011)). Reinhart and Sbranica (2011) found that
between 1945 and 1980 financial repression, working through a reduction in the real rate of interest
on public debt, was a major contributor to the reduction in public debt seen in many countries.
III. Pursue policies that raise the actual rate of inflation, π.
From equation (1), this will work provided these policies don’t raise the nominal interest rate, i, don’t
lower the growth rate of real GDP, g, and don’t raise the primary deficit, -s too much. The most
obvious problem is raising inflation without raising the nominal interest rate. Consider equation (2). If
~
the equilibrium or ex-ante real interest rate r is not affected by the inflation-raising policy (this is
sometimes referred to as the Fisher hypothesis), then higher actual inflation lowers the debt to GDP
e
ratio only if and to the extent to which it is unanticipated (if π rises by more than π ). If anticipated
inflation rises as much as actual inflation, the nominal interest rate will rise one-for-one with the
16
expected and actual inflation rate and there is no deleveraging. Financial repression can come to the
rescue here too, of course. If the authorities stop the nominal interest rate on the public debt from
~
rising with expected inflation, there is a de-facto reduction in the ex-ante real interest rate r and
deleveraging will occur regardless of whether the inflation is anticipated or not.
Unanticipated inflation (or anticipated inflation combined with financial repression that keeps nominal
yields from rising in line with anticipated inflation) can always be used to inflate away the real burden
of servicing a given outstanding stock of (public) debt that is denominated in domestic currency (but
not of course, inflation-linked debt or foreign-currency-denominated debt).
Temporary inflation can solve a fiscal unsustainability problem when the proximate cause of the fiscal
unsustainability is a very large stock of debt and when the real value of the flow (primary) deficit does
not present a material problem. Italy fits that category. If the general government debt burden is more
modest but the (primary) general government deficit is large – which was the situation in Ireland and
Spain in early 2008 before bad private assets began their migration to the public sector balance sheet
– a short sharp burst of inflation cannot solve the fiscal unsustainability problem by itself. If both the
public debt burden and the public sector primary deficit are large in real terms and as a share of GDP,
as is the case in the US and in Japan, inflation can only provide relief on the stock component of the
fiscal unsustainability conundrum. The bulk of the real flow primary deficit will have to be eliminated
some other way.
IV.
Raise the growth rate of real GDP, g.
This, of course, is everyone’s favourite deleveraging option because it is effectively painless,
especially if it means raising output by reducing economic slack and involuntarily idleness of
resources rather than by raising potential output along with actual output, which will in general require
sacrificing valued leisure and/or private or public consumption to boost capital expenditure. Raising
the level and/or growth rate of real GDP increases the real resources available for public debt service
without the need for fiscal austerity – cuts in public spending or tax increases. Some of the writings of
the ‘growth instead of austerity’ school make it look as though the governments of the EA member
states, the UK and other countries engaged in fiscal austerity either don’t recognize that fiscal
austerity hurts output and employment in the short and medium run or somehow forgot to push the
‘growth button’. The problem with this view is that unlike fiscal austerity, which is a policy (or rather a
set of two broad categories of policies: public spending cuts and tax increases), growth is not a policy.
Growth is an outcome that a country enjoys if it has (1) the right policies, (2), the right institutions and
culture, (3) the right initial conditions, (4) the right external environment, (5) a bit of luck and (6)
affordable funding for the sovereign and other systemically important institutions.
V.
Write down the debt or mutualise it.
The final deleveraging option is default (restructuring) or mutualisation – effectively making the debt
(or part of it) jointly and severally guaranteed by a wider community.
Debt default or debt restructuring takes two canonical forms from an economic perspective. The first
is equitisation: part or all of the debt is turned into equity. This option is rarely applied even in part to
sovereign debt, although it is common in the financial sector and the corporate non-financial sector.
The second is a write-down. Repudiation is a 100% write down. From an economic perspective, what
matters is the net present discounted value (NPV) of a write-down relative to the value of servicing the
debt in full according to the letter of the debt contract. Whether the restructuring is voluntary or
coercive and the details of the restructuring (maturity extension, lower interest payments, write down
of face value or notional value of the debt) is of interest to lawyers, credit rating agencies, the ISDA
Determinations Committees and politicians who don’t understand the difference between face value
and NPV and/or hope that their voters don’t understand the difference either, but is of secondary
economic significance.
6. When is Deleveraging Most Harmful?
The most damaging forms of deleveraging, from the point of view of their short-to-medium run impact on
aggregate demand, output and employment, as well as possible long-run or even permanent effects on
17
potential output, occur when the ex-ante desire to increase saving rises sharply and when the
coordination of decisions on saving, investment and on sales and purchases of financial instruments are
poorly coordinated by markets and governments. These circumstances generally are more likely to
arise,
if the state is among the sectors that need to deleverage. As discussed before, the government
is often tasked with stabilizing the economy when the non-financial private sector deleverages. If,
however, the government is preoccupied with its own debt burden, it is often constrained in its
ability to support the private economy. It is also less effective as a focal point for coordinating
private sector decisions. Furthermore, as noted earlier, the state is also usually the ultimate source
of financial support for the banking sector. Weak banking sectors can exacerbate the harm done by
NFS or government deleveraging. Finally, debt restructuring for sovereigns, while far from rare in a
historical context, is often done inefficiently, as timely and orderly debt restructuring is often
impeded by the lack of clear (contractual and/or statutory) procedures and by partisan political
considerations;
if the banking sector is in poor shape. Weak banking sectors strengthen precautionary saving
motives of households and non-financial corporates, and often lead to liquidity hoarding behaviour
by banks themselves. What is more, as discussed above, the risks of disorderly and contagious
bank deleveraging and bank runs (encouraged by the ‘sequential service constraint’ on bank
deposits when it is feared available reserves are insufficient to meet likely deposit withdrawals) are
larger than for other sectors, not only owning to the banking sector’s higher leverage, but also
because of the lack of clear and efficient procedures for bank debt restructuring in many countries –
even though both excessive banking sector leverage and a lack of orderly resolution regimes for
banks could be solved through collective action;
if more/larger sectors are attempting to deleverage at the same time. Coordination becomes
more complex and finding a sector that is willing to reduce its financial surplus, while others are
attempting to raise theirs, more difficult;
if the objective is to increase net wealth/reduce net debt rather than to bring down gross
balance sheet size or levels of debt, i.e. if there is a desire to increase active net financial
deleveraging (a larger planned sectoral financial surplus) or to increase active net deleveraging (a
higher planned sectoral saving rate). The capital adequacy ratio of an agent/sector can be raised
and its leverage ratio reduced without this requiring either active net deleveraging (‘saving’) or
active net financial deleveraging: the agent or sector does not have to raise its saving or reduce its
investment. All that is required is that assets and liabilities be reduced by the same amount. This is
true even if there are no capital gains or losses. The coordination problem is not eliminated –
distressed asset sales to pay off maturing debt can set in motion damaging feedback loops
between lack of market liquidity and lack of funding liquidity – but is in principle simpler than when
saving and investment decisions have to be coordinated as well.
7. How much deleveraging has taken place?
The leverage party has mostly stopped. Growth in debt and credit have fallen in most developed
markets, sometimes precipitously. In the period 1995-2006, gross non-financial sector debt grew by
9.3%pa in nominal terms, on average, but nominal NFS debt growth fell to 3.8%pa between 2008 and
15
Q2 2012. The fall in real credit growth is somewhat smaller on average than the drop in nominal credit
growth rates, as inflation rates have also fallen in many countries recently (relative to pre-2006 growth
rates), but real credit growth still fell in the post-2008 period relative to the pre-crisis trend in all but 4
countries in the sample (Belgium, Canada, Czech Republic and Japan). With very few (and small)
exceptions, the most recent data do not indicate any pickup in the rate of NFS credit growth
The pace of deleveraging, in what follows mostly measured by the change in the stock of debt relative to
GDP, has been very uneven across countries in recent years. Substantial deleveraging has taken place
in some countries and sectors. Ten countries (Italy, Poland, Netherlands, Czech Republic, Slovakia,
Belgium, Finland, France, Japan and Cyprus) have not seen any decrease at all in the NFS gross debt
15
In both cases these values are GDP-weighted growth rates in local currency.
18
to GDP ratio by Q2 2012. In many countries, gross debt as a share of GDP has increased further since
2008, mostly through increases in the public debt ratio, while the private debt ratio has fallen more often
(Figure 18). Thus, in 14 out of 28 countries the latest data indicate a reduction in gross debt-to-GDP
ratios of NFCs relative to 2008 and in 11 countries for HHs, while only three countries had decreases in
public debt – one of which is Greece as a result of its debt restructuring. Public debt ratios peaked only
very recently in some countries and were in fact still rising in almost half of our sample (in 13 out of 28
countries).
Figure 18. Selected Countries – NFS gross debt/GDP ratio, Change 2008 – 2012Q2 (% of GDP)
100 %
80
100 %
HH
NFC
Public
NFS
137%*
80
60
60
40
40
20
20
0
0
-20
-20
-40
-40
ET SW NO GR LV KO HU US AU IT GE EA SN PL SP
AT DN LT NL UK CZ SK BE CA FI PT FR JP CY IR
Note: Total non-financial sector gross debt equals the sum of households (HH), non-financial corporations (NFC) and general government (public) gross debt. All values are
expressed on a non-consolidated basis except for Australia and Portugal. For Italy, the Netherlands, and Ireland latest data correspond to 2012Q1, while for the EA and Cyprus
correspond to 2011*In Ireland, HH gross debt to GDP ratio declined by 1ppts, NFC debt/GDP increased by 74ppts, while GG gross debt/GDP increased by 64ppts.
Source: OECD, Eurostat, National Sources and Citi Research
However, the most recent data suggest that private debt ratios at least have peaked in most countries –
in all but two (Belgium and Portugal) for NFCs, and in all but four (Belgium, Canada, Slovakia and
Czech Republic) for households, even though in many cases the peaks were very recent. Across
countries, gross deleveraging in recent years seems to have been a ‘Nordic’ phenomenon. In the Baltic
and Scandinavian countries gross debt ratios have fallen strongly from their respective recent (post2006) peaks (Figure 19
). In many other countries, including the UK, Ireland, Portugal or France, NFS gross debt ratios have not
fallen at all. In some countries, including the US and Spain, the aggregate amount of deleveraging has
been rather small, but as noted, more substantial private sector deleveraging has been met with
increases in public debt.
Figure 19. Selected Countries – Non-Financial Sector gross debt/GDP ratio, peak – 2012Q2
20
20
10
10
0
0
-10
-10
-20
-20
-30
-30
-40
-40
HH
-50
NFC
Public
NFS
-50
NO LV SW HU ET GR LT SN JP KO IT DN PL SP
US CZ EA AT BE FR GE NL CA PT AU IR UK FI SK
Note: Peak corresponds to maximum since 2006 for NFS gross debt. All values are expressed on a non-consolidated basis except for Australia and Portugal. For Italy, the
Netherlands, and Ireland latest data correspond to 2012Q1, while for the EA and Cyprus correspond to 2011
Source: OECD, Eurostat, National Sources and Citi Research
19
Where it occurred, deleveraging seems to have been driven by differential economic growth or default
rather than variations in credit growth, i.e. the countries with the largest debt reductions were not
generally the ones with the largest reductions in (nominal or real) credit growth. As noted above,
nominal and real credit growth has fallen quite substantially in many countries, and particularly so in
highly leveraged economies. Growth rates in debt-to-GDP ratios have also fallen quite substantially. But
the fact that real and nominal GDP growth have been very weak in recent years has made the job of
deleveraging much harder – the average yearly rate of nominal GDP growth between 2008 and 2011
was a whopping 5.5ppts lower than for 2000-08, while nominal debt growth fell by 4.7ppts.
8. How much more deleveraging is to come?
Likely and desirable levels of sectoral debt are likely to be lower than prior to 2007. Economic theory,
however, provides little guidance on optimal levels of debt and leverage. In the absence of
fundamentally-based criteria for debt sustainability, focal points can be useful benchmarks:
For public debt, the Maastricht Treaty of the EU provides one such focal point with a threshold of
60% for general government gross debt-to-GDP ratios
Cecchetti et al (2011) find that on average debt is associated with lower GDP growth when gross
debt-to-GDP ratios exceed 85% for the public sector (close to the Reinhart and Rogoff (2009)
threshold of 90% of GDP), 90% for the non-financial corporate sector and 85% of GDP for HHs
(even though the threshold was not statistically significant in the case of HHs).
Debt levels experienced during a period for which there is general agreement that financial excesses
were absent may also provide useful benchmarks.
8.1 Households are likely to require plenty of additional deleveraging
Figure 20 highlights the difference between HH gross debt levels (relative to GDP and disposable
income) today and in 2001. On (an unweighted) average, HHs would need to reduce their gross debt by
around 30ppts of GDP to get back to 2001 levels, not a small order given that debt only fell by around
1.5ppts of GDP on average in the two and a half years since the end of 2009. While the magnitudes
differ, the picture painted by ratios of HH gross debt relative to HH disposable income is very similar.
Figure 20. Selected Countries – HH Change in Gross Indebtedness required to return to 2001
levels
120
% of GDP
Gross Debt/GDP
Gross Debt/Disposable Income
100
80
60
40
20
0
FI
GR
PT
SP
AU
JP
PL
UK
ET
HU
SK
CZ
CY
LT
LV
SN
IR
IT
NL
NO
SW
CA
US
AT
BE
DN
FR
GE
-20
Note: HH refers to households. Values correspond to the difference between HH gross debt divided by GDP or
disposable income at the latest available date and in 2001.
Source: OECD, Eurostat, National Sources, and Citi Research
If we distinguish the countries in our sample according to the pressures for households to deleverage,
we observe the following regularities.
20
First, there are countries with unambiguously large and likely long-lived deleveraging pressures for HHs.
This group includes countries that have had very large increases in gross debt over the past decade,
substantial increases in most measures of net debt and recent (or likely future) substantial reductions in
net worth (financial and non-financial). Levels of gross and net debt are also often relatively high in
these countries, and the degree of deleveraging achieved in recent years has been modest. Among the
countries in this group are Cyprus, Greece, Ireland, Spain, and Portugal, where HH gross debt relative
to GDP are 58ppts of GDP, 40ppts, 63ppts, 32ppts and 23ppts, respectively, higher than in 2001 and
where house prices have fallen in recent years (generally leading to falls in HH net worth, including non16
financial assets where data are available), even though in Portugal by rather little to date. The latest
available data indicate that HH nominal gross debt is falling at an annual rate of around 2.5% in Spain,
4-4.5% in Portugal and Ireland, and 7% in Greece but is still increasing in Cyprus. Assuming that
deleveraging continues at this pace, bringing gross debt back to their 2001 levels would under our
17
assumptions for nominal GDP growth take around or above another decade in these five countries.
Second, there are countries with more moderate, but still deleveraging pressures, at least in the
medium-term. This group includes countries with relatively large increases in HH gross debt (and
usually high levels of gross debt) but where HH net worth (often reflecting a combination of financial and
non-financial wealth) has remained relatively robust. This group of countries includes Australia, Canada,
Denmark, Korea, Norway, Sweden, the Netherlands, and the UK. In our view, the large increases (and
often high levels) of gross indebtedness and the fact that house prices are often high in these countries
suggests that at least some gross deleveraging will likely be necessary in coming years in these
countries. In some, such as Denmark, house prices have already fallen substantially, which has led to
some pressures to reduce debt. In most other countries (except the UK), deleveraging has not yet
started in earnest.
The group of countries with moderate deleveraging pressures on HHs also includes countries such as
Belgium, France (where increases in gross debt have been sizable but smaller, and where HH net worth
has also held up) and Italy (where HH gross debt is low, but has also risen quite a bit in the past decade,
and where HH net worth has suffered) – where deleveraging has also not yet started. This group also
includes the US (where HH deleveraging has gone quite a long way, but HH net worth has fallen), and a
number of Eastern European countries (where HH debt and house prices have fallen after sizable prior
increases, but where levels of debt remain low).
In the group of countries with moderate deleveraging pressures the time horizon over which the
deleveraging will play out is difficult to pin down – precisely, because there is no significant deleveraging
process in place by now, with some exceptions
Third, there is a small and select circle of countries with very modest increases in HH gross debt and no
need for net deleveraging, either. This select circle includes Austria, Germany and Japan, and maybe
Switzerland. Germany and Japan are in fact the only countries among the 30 countries in our sample,
where gross debt-to-GDP ratios for HHs at the latest available date (usually Q2 2012) were below those
in 2001.
8.2 Non-financial corporations
For non-financial corporations, gross debt is on average (unweighted) 26ppts of GDP higher than it was
in 2001, but with greater differences between the countries – in Ireland, gross debt levels of NFCs are
165ppts of GDP higher! In a number of countries, NFC gross debt has fallen relative to GDP since 2001
(in at least 5 countries: the Netherlands, Sweden, Japan, Poland, and the Czech Republic). Accounting
16
Generally, in countries with very long housing booms, even after recent falls in house prices, house price levels are
still up from the levels of the later ‘90s or early ‘00s, with the level of non-financial asset holdings up by more still, as
supply has increased.
17
Our assumption for nominal GDP growth is based on the average Citi forecasts (IMF for Cyprus) for 2012-2016 for
real GDP and CPI inflation. Under these assumptions, returning HH gross debt-to-GDP to its 2001 levels would take
6 years in Portugal, 7 years in Greece, 11 years in Ireland and 13 years in Spain. These assumptions would not
suggest any deleveraging in Cyprus but assuming that HH gross debt-to-GDP falls by 3ppts/year, it would take 16
years there.
21
for currency and deposit accumulation leaves the picture broadly unchanged, but considering narrower
measures of net debt (i.e. netting out broader sets of assets) suggest lower increases in indebtedness.
Generally, the list of countries where prior debt increases (and therefore future deleveraging needs)
were high is rather similar to that for HHs: Gross NFC debt increased strongly in Cyprus, Ireland,
Portugal, and, Spain, but also in Belgium and the UK and a number of countries in Emerging Europe. In
few of these countries have levels of gross debt to GDP come down substantially from the peak.
However, even in these countries NFC financial net worth is often still substantially higher than it was in
2001, including in Portugal, Spain, and Ireland. This truly highlights the roles of gross debt and access
to liquidity in creating acute deleveraging pressures.
In Greece, the levels of gross debt and broad measures of net debt are low in the cross-country
comparison, and the increases have been below the cross-country average. But equity has fallen even
more steeply, so that debt-to-equity ratios have seen among the largest rises in our sample (Figure 21).
In the US, NFC gross debt-to-GDP ratios have recently come down and have registered below average
increases in the previous decade.
However, we consider the argument that DM corporates generally have a strong balance sheet position
to be overplayed. First, Figure 21 highlights that debt to equity ratios, while very volatile, have risen, not
fallen, in many countries since 2001. Second, corporate profit margins are high and are unlikely to rise
further in many countries, and could well fall over time, putting pressure on measures of corporate
leverage that are based on earnings or profits. Third, in an environment where public sectors are under
pressure to consolidate finances, cash-rich corporate balance sheets may offer an opportune source of
revenue for fiscal consolidation. A number of countries, including Spain and France, have already raised
corporate taxes in recent years against a year-long trend of falling corporate tax rates. Of course, the
opposite holds in the few countries that currently have lots of fiscal space – Sweden just decided to
lower its corporate tax rate to stimulate a slowing economy.
Taking the Cecchetti et al-suggested value of 88% of GDP as a benchmark, many countries (23) could
raise potential GDP growth by achieving a lower level of corporate debt. Only in Australia, the US,
Germany, Greece, Slovakia, Poland and Lithuania, did NFCs remain below that threshold, with Irish
NFCs at 206ppts above the benchmark.
Figure 21. Selected Countries – Change in leverage required to return to 2001 levels
200%
120
165%
70
20
-30
-80
Gross Debt/GDP
Gross Debt/Equity
FI
GR
PT
SP
AU
JP
PL
UK
ET
HU
SK
CZ
CY
LT
LV
SN
IR
CA
IT
NL
NO
SW
US
AT
BE
DN
FR
GE
-130
Note: Values correspond to difference between NFC gross debt/GDP and gross debt/equity at the latest available
date and 2001.
Source: OECD, Eurostat, National Sources, and Citi Research
22
8.3 Public Debt
Figure 22 presents the levels of general government gross liabilities to GDP for our sample of countries
18
in Q2 2012. Of the 30 countries in the sample, 21 were above the 60% of GDP benchmark. Among
those that were below 60% of GDP, four (Estonia, Czech Republic, Lithuania, Latvia) are EMs. 14 of the
30 countries were above the 90% of GDP benchmark too, by only a small margin for Spain and
Germany (both at 92% of GDP) but with Japan carrying general government liabilities of a staggering
240% of GDP. The US and the UK have both recently crossed even the higher threshold. These
numbers are not what one would expect for safe haven countries, a designation used not just for the US
but also for the UK these past two years.
Figure 22. Selected Countries – General Government Liabilities (% of GDP), Q2 2012
250
200
150
90% of GDP
100
60% of GDP
50
0
US AT BE DN FR GE IT NL NO SW CA FI GR PT SP AU JP PL UK ET HU SK CZ CY LT KO LV CH SN IR EA
Note: For Italy, the Netherlands, and Ireland latest data correspond to 2012Q1, while for Cyprus and the EA it
correspond to 2011
Source: OECD, Eurostat, National Sources, and Citi Research
What is more, as indicated above, public debt levels and ratios to GDP are still rising in many countries,
including the US, Japan and the UK.
Now that the illusion of the existence of risk-free sovereign debt is broken, probably beyond repair, we
doubt that private investors will continue to finance or refinance such levels of government debt even in
the safe havens, for very much longer at anything near current levels of yields, without much friendly or
not-so-friendly encouragement (aka financial repression) by the relevant national authorities (usually a
combination of the national/federal treasuries, central banks and regulators/supervisors). Financial
repression and/or continued activity by the central bank as the buyer of last resort may for a while still
maintain the appearance of easy ‘market access’ for many of these sovereigns. High levels of private
saving and limited capital mobility make the job of financial repression somewhat easier, but many
fiscally weak euro area countries do not have either. For private investors to stay or to return voluntarily,
sustainably and with confidence, a long and painful period of gradual public debt reduction through fiscal
pain is likely to be needed in many countries. In countries where government debt is above 90% of GDP
and still rising, the period of fiscal pain needed to bring debt down to sustainable levels is likely to cover
most of the rest of this decade, unless sovereign debt restructuring is resorted to. In a number of
countries, including Greece, Ireland, and Portugal, and potentially Spain, Italy, Cyprus, and Slovenia,
18
We use the general government gross liabilities as presented in Flow of Funds Accounts rather than the more
commonly used measures for General Government Debt for the following reasons. First, it often paints a more
accurate and timely picture of government indebtedness. Conventional measures of general government debt, such
as those under the EU’s Excessive Deficit Procedure Definition, exclude some items, such as accounts payable,
which later transition into recognized items under eve the EDP measures, but with a lag. Second, these data are
available quarterly for most countries, while the general government debt data can often only be obtained on an
annual basis.
23
sovereign debt restructuring is most likely necessary to restore solvency of the sovereign. In others, a
short period of swift and, one hopes, orderly sovereign debt restructuring may be a benign alternative to
years of fiscal pain.
9. Conclusion: what lies ahead and what is to be done?
There remains far too much debt in the balance sheets of most advanced economies. Reducing this
debt burden to more tolerable levels will take many years unless recourse is had to debt restructuring on
a much greater scale than currently contemplated. Higher real growth is neither a policy nor a realistic
expectation as a means to deliver painless deleveraging in the excessively indebted advanced
economies. As regards the growth of potential output, most of these economies are at or close to the
technology frontier and have unfavourable demographics. More open immigration policies could ease
the demographic crunch. Even under favourable circumstances, in the absence of excessive leverage,
the growth rates of potential output would be modest (see Gordon (2012)). Clearly, many of the most
afflicted economies in the EA have deeply distorted and dysfunctional labour markets, closed-shop
professions, badly managed and poorly regulated utilities, excessive state ownership of productive
resources and a host of other man-made supply-side distortions whose removal could lift potential
output significantly. Unfortunately, the reforms are slow in coming and will often require considerable
time to be implemented. And even when they have been implemented, the benefit in the form of higher
actual output still requires demand to come from somewhere. Animal spirits alone are unlikely to do the
job with acceptable speed.
In addition, net deleveraging by sovereigns and banks (in most countries), by households in many
countries and by the non-financial corporate sector in some countries, means that these sectors attempt
to run financial surpluses without matching increases in planned financial deficits by other sectors,
except possibly the central banks. The result is that the paradox of thrift strikes and activity is well below
its potential level.
An inflationary solution to the excessive leverage problem is all but impossible in the euro area, highly
unlikely in Japan, unlikely in the US and quite unlikely in the UK. The reason for this is the much
increased independence of central banks in the advanced economies and their commitment to price
stability. Financial repression will play a modest role in the deleveraging process of the DMs. This will
occur partly through central bank purchases of sovereign debt in the primary markets (except for the
ECB which cannot engage in primary market purchases of sovereign debt because of Article 123 of the
19
Treaty) at yields below those prevailing in the secondary markets. Sovereign and private debtors can
also benefit from purchases in the secondary markets that drive down yields there - sovereign bond
markets are inefficient and the supply of and demand for sovereign debt influences its yield. Banks and
other regulated financial intermediaries will be cajoled by the national authorities to hold more sovereign
debt than they would choose to hold voluntarily at yields lower than what they would accept voluntarily,
with financial repression sometimes masquerading as prudential probity, as in the case of the LCR and
the NSFR. In the absence of at least moderately high inflation (say 5 percent or more per annum),
financial repression only has a modest effect on real bond yields, however.
Private and public austerity will continue to be important mechanisms for deleveraging in the years to
come. In the euro area, so will mutualisation of sovereign debt and restructuring of sovereign debt and
bank debt. Restructuring of household debt (especially mortgage debt) would be desirable in many
countries with excessive gross household debt (e.g. the Netherlands, Denmark, Ireland and Spain) but
is for political reasons unlikely on a large scale.
Debt restructuring, for sovereigns in the periphery and for banks in both periphery and core is, in our
view, inevitable during the next two or three years. This is likely to start with another sovereign debt
restructuring for Greece, regardless of whether it exits the euro (as we expect to happen during 2013 or
2014). Portugal, with its inexorably rising sovereign debt burden, poor growth prospects and growing
19
It would of course be possible for the ECB to engage in de-facto primary market purchases of sovereign debt by
arranging ‘back-to-back’ purchases of sovereign debt in the primary markets at below-fair yields by commercial banks
that then sell on that sovereign debt ‘in the secondary market’ to the ECB at the same (favourable to the sovereign)
price.
24
austerity fatigue will likely have to restructure its sovereign debt, most likely when its current troika
programme terminates, in the second half of 2014. Unless Ireland achieves ample retroactive
mutualisation of public debt incurred as a result of its banking sector bail-outs since 2008, it too will have
to restructure its sovereign debt.
In Spain, the consolidated sovereign and banking sector (allowing for likely rapidly rising residential
mortgage losses and a deep and long recession) is most likely insolvent, in our view, so the operational
question is what combination of debt mutualisation through the EA sovereigns or the Eurosystem, bank
debt restructuring and sovereign debt restructuring will be used. Cyprus will require bank debt
restructuring unless the bad assets of the bank are transferred to the sovereign, in which case Cyprus
will require sovereign debt restructuring. Slovenia faces a similar conundrum. Finally Italy, despite its
strategic-sovereign-default-inviting combination of a very large public debt and a primary general
government surplus (actual and structural), is certainly able to service its sovereign debt in full (following
accession to a programme that grants it access to OMT support). One key risk in Italy is that the next
elections (no later than April 2013) could produce an anti-euro, nostalgia-for-the-lira, let’s-restructuresovereign-debt-held-by-banks-and-foreigners, populist coalition government.
In addition to, and where possible instead of reducing the size of gross liabilities of sovereigns, banks,
and in many cases also households and non-financial corporations through haircuts or write-downs, a
change in the composition of these liabilities away from debt-type instruments and towards more equitytype instruments is highly desirable. In the case of banks, we would hope that bailing in unsecured
creditors would not take the form of haircuts but of a mandatory partial or complete conversion of
unsecured debt into equity. For households, the equitisation of existing mortgages, when a nonperforming household has negative equity, plus a much greater future issuance of equity-type mortgage
products would make sense.
Islamic or joint-equity-type mortgages have much better risk sharing properties than conventional
western repayment or interest-only mortgages, whose inflexible debt contract features are most
inappropriate for households – typically entities with very limited financial flexibility whose main asset,
their human capital, cannot be sold or used as security. With a stylised Islamic mortgage, the seller of a
residential home sells it to the bank. When the bank deals with a would-be buyer, the mortgageequivalent contract consists of two parts. The first is a contract between the bank and the buyer to
establish a joint ownership. The buyer commits to buy, typically in a sequence of purchases over time,
the share of the bank (which could be 100% initially). At the same time, the bank leases its share to the
buyer – effectively a rental contract for the share of the property not (yet) owned by the buyer. As the
buyer over time purchases additional fractions of the bank’s equity, the stream of rental payments from
the buyer to the bank diminishes. If the household cannot make these rental or lease payments, it can
be evicted, like any tenant who does not pay the rent.
Finally, sovereigns should not only incur fewer liabilities, their liabilities should be more equity-like. Real
GDP growth warrants or a long-term floating rate instrument where the ‘interest rate’ is some constant
plus the growth rate of nominal GDP, are examples. When the government’s ability to service its debt is
lower, its debt service is likely to be lower also. There are practical problems: inflation and real GDP data
can be manipulated by unscrupulous governments. One would hope that an agency like Eurostat in the
EU would be able to prevent the opportunistic deliberate manipulation of macroeconomic price and
quantity data in the future.
The sequence of crises the advanced economies have inflicted on themselves and on the rest of the
world since 2007 is by no means over. Entire new chapters remain to be written. Mr. Micawber’s recipe
20
for happiness deserves to be on the wall in every financial kitchen.
20
Mr Micawber’s principle states: "Annual income twenty pounds, annual expenditure nineteen pounds nineteen and
six, result happiness. Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery."
From Charles Dickens’s novel, David Copperfield, 1850.
25
References
Alesina, A. and S. Ardagna (2010) “Large changes in fiscal policy taxes versus spending” in J.R. Brown
(ed.), Tax policy and the economy, NBER and University of Chicago Press.
Arrow, Kenneth (1953), "The Role of Securities in the Optimal Allocation of Risk-Bearing", Econometrie (in 1963, RES).
Arrow, Kenneth and Gerard Debreu (1954), "Existence of an Equilibrium for a Competitive Economy",
Econometrica.
Ayub, Muhammad (2007), Understanding Islamic Finance, Wiley Finance, Jhn Wiley and Sons, Ltd,
Bernanke, Ben S.(2004). "The Great Moderation". Remarks made at the meetings of the Eastern
Economic Association, Washington, DC February 20, 2004,
http://www.federalreserve.gov/BOARDDOCS/SPEECHES/2004/20040220/default.htm
Bhutta (2012), “Mortgage debt and household deleveraging: accounting for the decline in mortgage debt
using consumer credit record data”, Federal Reserve Board Finance and Economics Discussion Series,
2012–14, http://www.federalreserve.gov/pubs/feds/2012/201214/201214pap.pdf
Buiter, Willem H., Giancarlo Corsetti and Nouriel Roubini (1992), "Excessive deficits: sense and
nonsense in the Treaty of Maastricht", Economic Policy, 1993(1), Reprinted in The Political Economy of
Monetary Union, Edited by Paul de Grauwe, pp. 297-331, Edward Elgar Publishing Ltd, Cheltenham,
2001
Buiter, Willem H. (2012), "The Role of Central Banks in Financial Stability: How Has it Changed?", paper
presented at the Fourteenth International Banking Conference organised by the Federal Reserve Bank
of Chicago and the European Central Bank in Chicago, Ill., on November 10-11, 2011, CEPR Discussion
Paper No. 8780, January.
Buiter, Willem H. and Ebrahim Rahbari (2012), "Looking into the deep pockets of the ECB", Citi
Economics, Global Economics View, 27 February 2012.
Buiter, Willem H. and Anne C. Sibert (2006), "How the Eurosystem's open-market operations weaken
financial market discipline (and what to do about it)," in Fiscal Policy and the Road to the Euro, National
Bank of Poland, Warsaw, Oct. 2006, pp. 29-58.
Cecchetti Stephen G, M S Mohanty and Fabrizio Zampolli (2011), “The real effects of debt”, September
2011, http://www.bis.org/publ/othp16.pdf
Cottarelli, Carlo and Laura Jaramillo (2012), “Walking Hand in Hand: Fiscal Policy and Growth in
Advanced Economies”, IMF Working Paper, WP/12/137, May.
Dynan, K (2012), “Is a household debt overhang holding back consumption?”, Brookings Institution,
http://www.brookings.edu/~/media/Files/Programs/ES/BPEA/2012_spring_bpea_papers/2012_spring_B
PEA_dynan.pdf
Galbraith, John Kenneth (1954), The Great Crash, 1929, Houghton Mifflin Harcourt.
Gambacorta, Leonardo, Boris Hofmann and Gert Peersman (2012), “The Effectiveness of
Unconventional Monetary Policy at the Zero Lower Bound: A Cross-Country Analysis”, BIS Working
Papers, No 384, August.
Garber, Peter (2000), Famous First Bubbles: The Fundamentals of Early Manias, MIT Press.
Gerhardt, Maria (2009), Consumer Bankruptcy Regimes and Credit Default in the US and Europe: A
comparative study”, CEPS Working Document No. 318/July 2009, http://aei.pitt.edu/11336/1/1887.pdf
26
Giavazzi F., and M. Pagano (1996), Non-Keynesian Effects of Fiscal Policy Changes: International
Evidence and the Swedish Experience, Swedish Economic Policy Review, vol. 3, n.1, Spring, 67-112.
Gordon, Robert J. (2012), “Is U.S. Economic Growth Over? Faltering Innovation Confronts the Six
Headwinds”, NBER Working Paper 18315, August. http://www.nber.org/papers/w18315.
IMF (2009), “What’s the Damage? Medium-Term Output Dynamics after Financial Crises”, IMF World
Economic Outlook, September 2009.
IMF (2012), “Dealing with Household Debt”, Chapter 3, World Economic Outlook, Spring 2012,
http://www.imf.org/external/pubs/ft/weo/2012/01/pdf/c3.pdf
th
Kindleberger, Charles (2000), Manias, Panics and Crashes, 4 edition, Wiley.
Kingsley, G. Thomas, Robin Smith and David Price (2009), “The Impacts of Foreclosures on Families
and Communities”, The Urban Institute, May,
http://www.urban.org/UploadedPDF/411909_impact_of_forclosures.pdf
Mack Adrienne and Martinez-Garcia Enrique (2011), “A Cross-Country Quarterly Database of Real
House Prices: A Methodological Note”, Federal Reserve Bank of Dallas, Globalization and Monetary
Policy Institute, Working Paper No. 99
Modigliani, Franco and Merton Miller (1958). "The Cost of Capital, Corporation Finance and the Theory
of Investment". American Economic Review 48 (3): 261–297.
Modigliani, Franco and Merton Miller (1963). "Corporate income taxes and the cost of capital: a
correction". American Economic Review 53 (3): 433–443.
Reinhart, Carmen M. and M. Belen Sbrancia (2011), "The Liquidation of Government Debt," NBER
Working Papers 16893, National Bureau of Economic Research
Reinhart, Carmen M. and Kenneth S. Rogoff (2009), This Time is Different: Eight Centuries of Financial
Folly, Princeton University Press
Shleifer, Andrei and Robert Vishny (1997), "The Limits of Arbitrage", Journal of Finance, Vol. 52, No. 1. ,
March, pp. 35-55
Shiller, Robert (2000), Irrational Exuberance, Princeton University Press.
Sheets, Nathan (2012), “Fiscal Deleveraging, Financial Repression, and Central Bank Independence—
Lessons from the U.S. Experience after World War II”, Citi Economics, Empirical and Thematic
Perspectives, 14 May 2012.
Sheets, Nathan (2011), “U.S. Household Saving and Deleveraging – What’s Next?”, Citi Economics,
Empirical and Thematic Perspectives, 20 December 2012.
Sherman, Matthew (2009), “A Short History of Financial Deregulation in the United States”, Center for
Economic and Policy Research, http://www.openthegovernment.org/sites/default/files/otg/deregtimeline-2009-07.pdf
Stock, James H. and Watson, Mark W., “Has the Business Cycle Changed and Why?”, NBER
Macroeconomics Annual 2002, Volume 17 (2003), MIT Press
Townsend, R.M., 1979. Optimal contracts and competitive markets with costly state verification. Journal
of Economic Theory 22, 265–293
27
Appendix
Figure 23. Country Labels
Country
Australia
Austria
Belgium
Canada
Cyprus
Czech Republic
Denmark
Estonia
Finland
France
Germany
Greece
Hungary
Ireland
Italy
Japan
Korea
Latvia
Lithuania
Netherlands
Norway
Poland
Portugal
Slovakia
Slovenia
Spain
Sweden
Switzerland
UK
US
Euro Area
Abbreviation
AU
AT
BE
CA
CY
CZ
DN
ET
FI
FR
GE
GR
HU
IR
IT
JP
KO
LV
LT
NL
NO
PL
PT
SK
SN
SP
SW
CH
UK
US
EA
Source: Citi Research
Nonfinancial Sector Debt
1) Financial Accounts by Sector:
The time series constructed are taken either from national balance sheet statistics (flow of funds) from
the OECD or national sources (usually national central banks) at an annual and quarterly frequency. The
30 countries included in the sample are: Australia, Austria, Belgium, Canada, Cyprus, Czech Republic,
Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Japan, South Korea,
Latvia, Lithuania, the Netherlands, Norway, Poland, Portugal, Slovakia, Slovenia, Spain, Sweden,
Switzerland, the UK and the US. These countries accounted for 62.3% of world GDP at market
exchange rates in 2011.
On average annual data start around 1995, but the data go back to 1950 (US), 1980 (Spain, Canada,
Korea, Japan) and 1990 (Germany, Netherlands, Hungary) for some countries, and generally end in
21
2011. Annual data going back to 1980 were extended/backdated using data from Cecchetti, Mohanty,
and Zampolli (2011) for Austria (for which data from the original source start in 1995), Belgium (1994),
France (1994), Germany (1992), Italy (1995), the Netherlands (19990), Sweden (1995), Finland (1995),
Greece (1995), Portugal (1995), Australia (1990), and the UK (1987).
Quarterly data start as early as 1952 (US), 1964 (Japan), 1975 (Korea), 1980 (Spain), 1987 (UK), 1990
(Canada), 1991 (Germany), and 1993 (Belgium). For all other countries, quarterly data start after 1995.
21
By extrapolating the time series, for which the level of the new series and growth rates of the reference series are
reflected in the final time series.
28
The quarterly data generally end in 2012:Q2. Data for Cyprus and Switzerland are not available
quarterly and these countries are therefore not included in any analysis that relies on quarterly data. For
Ireland the data start in 2001, for Switzerland in 1999, for Slovenia in 2001, and for Latvia in 1998,
hence these countries are not included in the comparisons of debt levels between 1995 and today.
The sectors covered are (i) households and non-profit institutions serving households, (ii) nonfinancial
corporations, and (iii) general government. Following Cecchetti et. al (2011), debt is defined as the
following: gross liabilities for households and general government, and total liabilities less shares and
other equities for nonfinancial corporations. For US nonfinancial corporations, “credit market
22
instruments” is used as a measure of gross debt.
OECD countries missing in the general analysis include: Chile (data start in 2005), Iceland (no data for
the household sector available), Israel (data only for 2010), Luxembourg (data start in 2006), Mexico
(data from 1997), New Zeeland (no data for the household and non-financial sector), and Turkey (no
data for the household and non-financial sector).
For household disposable income (net), we use OECD and Eurostat data that define it as the sum of
household final consumption expenditure and saving (minus the change in net equity of households in
pension funds). These values are equivalent to the sum of wages and salaries, mixed income, net
property income, net current transfers and social benefits other than social transfers in kind, less taxes
on income and wealth and social security contributions paid by employees, the self-employed and the
unemployed. Due to data availability, values for the UK, Portugal and Spain are taken from respective
national sources. For the UK and Portugal, values are for gross disposable income, which do not
discount the change in net equity of households in pension funds
2) Other data used in the analysis
Domestic credit to the private sector (IMF): total domestic credit provided by domestic banks to resident
private sectors of the economy (e.g. other financial corporations -insurance companies, pension funds,
and the like-, nonfinancial corporations, and households). Domestic banks include all deposit-issuing
financial institutions operating within the country. They include domestic banks and domestic branches
of foreign banks.
Deleveraging Episodes
Identification
Based on our data for NFS gross debt across 86 countries for the period 1960-2006 (constructed as the
sum of private sector credit and public sector debt, both provided by the IMF), we identify deleveraging
episodes, following Mckinsey (2010), episodes where either the ratio of total debt to GDP declined for at
least three consecutive years and fell by 10ppts of GDP or more OR an episode in which the total stock
of nominal debt declined by 10ppts or more.
23
These exercise provided 31 deleveraging episodes, of which 18 were preceded by a financial crisis.
22
Credit market instruments include the following financial liabilities for nonfinancial corporations: i) commercial
papers, ii) municipal securities, iii) corporate bonds, iv) total loans and v) mortgages.
23
Financial crisis list episodes are from Laeven and Valencia (2008).
29
Figure 24. Deleveraging Episodes – Overview Table
Argentina
Bolivia
Chile
Dominican Republic
Ecuador
Finland
Indonesia
Japan
Korea
Malaysia
Mexico
Nicaragua
Norway
Paraguay
Philippines
Sweden
Thailand
Uruguay
Deleveraging
start
end
2003
2009
1998
2008
1986
1994
2004
2008
2000
2008
1993
1999
1998
2008
2000
2003
1998
1999
1998
2008
1995
1999
2002
2008
1994
1996
1999
2005
2004
2007
1993
2000
1998
2001
2003
2007
NFS Debt (% of GDP)
start
end
150.2
72.6
125.2
72.2
227.7
89.1
60.3
46.2
112.6
47.4
135.1
99.0
125.8
59.8
361.3
354.7
82.8
91.0
194.6
143.2
86.0
72.0
255.3
113.8
106.1
95.3
69.6
55.6
102.0
76.7
189.8
107.0
206.2
154.1
142.5
86.4
Domestic Credit (% GDP)
start
end
10.8
13.5
64.1
34.7
62.7
48.1
23.4
20.9
29.9
26.3
80.8
53.3
53.2
26.6
219.3
187.6
68.6
74.7
158.5
100.3
29.2
20.4
19.6
37.6
54.7
58.1
30.4
17.6
32.2
28.9
111.7
42.3
155.9
96.9
43.2
23.4
Public Debt (% of GDP)
start
end
139.4
59.0
61.2
37.5
165.0
41.0
36.9
25.3
82.6
21.1
54.2
45.7
72.5
33.2
142.1
167.2
14.3
16.3
36.1
42.8
56.8
51.6
235.7
76.2
51.4
37.2
39.2
38.0
69.7
47.8
78.2
64.7
50.3
57.2
99.3
63.0
Financial Crisis
start
2001
1994
1981
2003
1998
1991
1997
1997
1997
1997
1994
2000
1991
1995
1997
1991
1997
2002
Note: The list corresponds to deleveraging episodes that were preceded by a financial crisis
Source: IMF and Citi Research
Estimating Macroeconomics Responses
We estimate responses in macroeconomic variables following a deleveraging episode for real GDP,
private consumption, gross capital formation, net exports, the stock of domestic credit to the private
sector (from IMF, see above), and public debt.
Responses were approximated by estimating deviations from the pre-recession (pre-deleveraging) trend
after the episode, following IMF (2009). This approach consists of comparing the medium-term level of
the variable to the level it would have reached following the pre-crisis (pre-deleveraging) trend, with the
medium term defined as seven years after the crisis.
First, we estimate a linear trend through the actual (output) series during a seven-year pre-crisis period
that ends three years before the onset of the crisis (e.g. between t-10 and t-3, t being the year of the
crisis). This trend is then applied to values from t onwards to construct a (output) series trend (e.g. GDPt
= GDPt-1*(1+trend), with GDPt = GDP trend at t). The (output) series is then subtracted from the
(output) series trend.
30