UCD CENTRE FOR ECONOMIC RESEARCH
WORKING PAPER SERIES
2009

The Irish Credit Bubble
Morgan Kelly, University College Dublin
WP09/32
December 2009

UCD SCHOOL OF ECONOMICS
UNIVERSITY COLLEGE DUBLIN
BELFIELD DUBLIN 4

 The Irish Credit Bubble.
Morgan Kelly.
21st December 2009

1

Introduction.

During the 1990s, rising employment resulting from improved competitiveness caused Ireland
to experience rapid economic growth.1 As Ireland converged to average levels of Western
European income around 2000 it might have been expected that growth would fall to normal
European levels. Instead growth continued at high rates until late 2007, since when it has
turned sharply negative.
The proximate cause of the boom and bust in Ireland since 2000 is well known: construction. Ireland went from getting 4–6 per cent of its national income from house building in the
1990s—the usual level for a developed economy—to 15 per cent at the peak of the bubble in
2006–07, with another 6 per cent coming from other construction. This construction boom
led to an employment boom which drove wages in all sectors of the economy to uncompetitive
levels; and generated the tax revenues that funded substantial rises in government spending.
However, driving the construction boom was a less recognised boom, in bank lending. As
Figure 1 shows, in 1997, Irish bank lending to the non-financial private sector was only 60
per cent of GNP, compared with 80 per cent in most Eurozone economies and the UK.2 The
international credit boom then saw these economies experience a rapid rise in bank lending,
with loans increasing to 100 per cent of GDP on average by 2008.
1

3

See Honohan and Walsh (2002), especially their Figure 5.
Schularick and Taylor (2009) find similar figures across a wider set of industrialized economies. In what
follows we adopt the usual convention of expressing Irish figures as a fraction of GNP, and international
numbers as a fraction of GDP.
3
Data in Figure 1 on bank lending for UK from Bank of England, for Eurozone economies from the
ECB: Aggregated balance sheet of euro area monetary financial institutions, excluding the Eurosystem
http://www.ecb.int/stats/money/aggregates/bsheets/html/outstanding_amounts_index.en.html. Data for
private sector lending before 2003 includes insurance companies, pension funds, and other non-monetary
financial intermediaries. We remove them in 1997 by assuming that they are the same proportion of lending
in each economy as 2008.
2

1

 200

Percentage of GDP

150

100

50

UK

Spain

Portugal

Netherlands

Italy

Ireland

Greece

Germany

France

Finland

Euro area

Belgium

Austria

0

Figure 1: Bank lending to households and non-financial firms as a percentage of GDP for
Eurozone economies and the UK, 1997 and 2008.
These rises were dwarfed, however, by Ireland, where bank lending grew to 200 per cent
of national income by 2008. Irish banks were lending forty per cent more in real terms to
property developers alone in 2008 than they had been lending to everyone in Ireland in 2000,
and seventy-five per cent more as mortgages.
This more than tripling of credit relative to GNP in 11 years created profound distortions
in the Irish economy. The most visible impact was on house prices. We show below that the
rise in Irish house prices has little to do with falling interest rates and rising population, and
is almost completely explained by increased mortgage lending. In 1995 the average first time
buyer took out a mortgage equal to three years’ average earnings, and the average house
(new or secondhand, in Dublin or elsewhere) cost 4 years’ average earnings.
By the bubble peak in late 2006, the average first time buyer mortgage had risen to
8 times average earnings, and the average new house now cost 10 times average earnings,
while the average Dublin secondhand house cost 17 times average earnings. These rises
in mortgages and house prices became mutually reinforcing, with larger mortgages driving
house prices higher; while rising house prices made banks willing to grant larger mortgages.

2

 As the price of new houses rose faster than the cost of building them, investment in
housing rose. Ireland went from completing around 30,000 units in 1995 to 80,000 in 2007:
one new housing unit for every 20 households.
Like any bubble, the rise of Irish property prices contained the seeds of its own collapse.
Property bubbles grow as long as buyers are willing to borrow increasingly large amounts in
the expectation that prices will continue to rise. This process inevitably hits a limit where
borrowers become reluctant to take on what start to appear as impossibly large levels of
debt, and the self-reinforcing spiral of borrowing and prices starts to work in reverse.
In Ireland buyers started to become nervous sooner than most observers imagine: the
number and average size of mortgages approved (to all categories of borrower: first time
buyers, movers, and investors) peaked in the third quarter of 2006. By the middle of 2007 the
Irish construction industry was in clear trouble, with unsold units beginning to accumulate.
This property slowdown was bad news for the Irish banking system which had lent heavily
to builders and developers to finance projects and to make speculative land purchases. Share
prices of Irish banks fell steadily from March 2007, with the crisis coming to a head in late
September 2008 with a run in wholesale markets on the joint-second largest Irish bank
Anglo Irish. After aggressive denials that the banking system faced any difficulties, the
Irish government has been forced to respond with a sequence of increasingly desperate and
expensive improvised measures.
It guaranteed all deposits and senior debt in the six Irish banks in September 2008; was
forced to nationalise Anglo Irish in January 2009; invested €3.5 Bn in preference shares
in the two large retail banks AIB and Bank of Ireland in February 2009; and established
a National Asset Management Agency (NAMA) to buy non-performing development loans
from banks in November 2009.
It is currently proposed that NAMA buy loans with a face value of €70 Bn for €55 Bn.
However, even assuming that the government’s forecasts of interest costs and the recovery
of property prices—which appear to lie in the upper tail of optimism—prove correct, the
situation of the Irish banking system will remain difficult.
The business model of the Irish banks for the last decade was to borrow heavily in
wholesale markets to lend to developers and house buyers. The collapse of this model leaves
them with three, inter-related problems.
1. Large losses on loans to builders and developers.
2. Heavy reliance on wholesale funding.

3

 3. The likelihood of considerable defaults on mortgages.
While NAMA is intended to repair, for now, the damage to the asset side of Irish bank
balance sheets from developer loans, their liability side appears unsustainable. The aggressive
expansion of Irish bank lending was funded mostly in international wholesale markets, where
Irish banks were able to borrow at low rates. From being almost entirely funded by domestic
deposits in 1997, by 2008 over half of Irish bank lending was funded by wholesale borrowers
through bonds and inter-bank borrowing. This well of easy credit has now run dry. In the
words of Bank of England Governor Mervyn King: “But the age of innocence—when banks
lent to each other unsecured for three months or longer at only a slight premium to expected
policy rates—will not quickly, or ever, return.”4 As foreign lenders have become nervous of
Irish banks, their place has increasingly been taken by borrowing from the European Central
Bank and short-term borrowing in the inter-bank market. Payments from NAMA will allow
Irish banks to reduce their borrowing by a trivial amount.
Without continued government guarantees of their borrowing and, more problematically,
continued ECB forbearance, the operations of the Irish banks do not appear viable. Borrowing in wholesale markets at 5.6 per cent5 to fund mortgages yielding 3.5 per cent is not
a sustainable activity, and Irish banks face no choice but to shrink their balance sheets by
repaying debt and returning to their earlier state of being funded mostly by deposits. It
appears likely that the leverage of the Irish economy will return to normal international
levels, with bank lending in the region of 80–100 per cent of GNP, back where it was in the
late 1990s.
While the stock of credit to the Irish economy will contract by at least half, when we
remember that loans equivalent to 80 per cent of GNP are tied up in mortgages (most of
them of recent origin with terms of 35 years or more), the flow of new lending will contract
even more sharply. It follows that property prices that were inflated by bank lending will
drop steeply.
Should lending criteria return to their late 1990s standards, our results indicate that
the prices of new houses and commercial property will return to an equilibrium two thirds
below their peak levels, with larger falls possible for secondhand property. This means that,
supposing residential prices have already fallen 40 per cent from peak, prices still have to fall
by about half from their present levels. Were prices then to grow in line with real incomes,
4

Cited by Pozen (2009, 165)
The annual yield on a 5 year bond issued by AIB on 5 November 2009, 285 basis points over the mid 5
year swap rate. “AIB sells €750m five-year bond after ratings downgrade.” Irish Independent, 6 November
2009.
5

4

 at around 2 per cent per year, it will take about 50 years for real prices to return to their
2006 peaks.
The third problem of the Irish banks is their mortgages. Recent US experience highlights
two factors that rapidly increase the likelihood of mortgage default: falling house prices, and,
most importantly, unemployment. Both have become realities for many Irish borrowers in
the last year. In September 2009, 14.4 per cent of US mortgages were at least one payment
past due or in foreclosure; while in Florida, whose investor fuelled housing bubble closely
resembles the Irish one, the figure is 25%.6 Dealing with the financial and, more importantly,
human cost of widespread mortgage defaults will be the next, and far more challenging, step
of the Irish banking crisis.
By pushing itself close to, and quite possibly beyond, the limits of its fiscal capacity,
the Irish state has succeeded in rescuing Irish banks from their losses on developer loans.
Despite this, these banks remain as zombies entirely reliant on continued Irish government
guarantees and ECB forbearance, and committed solely to reducing their own debts.
While bank capital levels are, probably, adequate for the markedly smaller scale of their
future lending, we will see below that even fairly modest losses on their mortgage portfolios
will be sufficient to wipe out most or all of that capital. Having exhausted its resources in
rescuing the Irish banks from the first wave of developer losses, the Irish state can do nothing
but watch as the second wave of mortgage defaults sweeps in and drowns them.
In other words, it is starting to appear that the Irish banking system is too big to save.
As mortgage losses crystallise, the Irish government’s ill conceived project of insulating
bank bond-holders from any losses on their investments is sliding beyond the means of its
taxpayers.
The mounting losses of its banking system are facing the Irish state with a stark choice.
It can attempt a NAMA II for mortgage losses that will end in a bond market strike or a
sovereign default. Or it can, probably with the assistance of the IMF and EU, organise a
resolution that shares property losses with bank creditors through a partial debt for equity
swap. It is easy for governments everywhere to forget that their states are not wholly
controlled subsidiaries of their banks but separate entities; and a resolution that transfers
bank losses from the Irish taxpayer to bank bond holders will leave Ireland with a low level
of debt that, even after several years of deficits, it can easily afford.
6

Mortgage Bankers Association,
National Delinquency
http://www.mbaa.org/NewsandMedia/PressCenter/71112.htm

5

Survey,

November

19

2009,

 The rest of this paper is as follows. Section 2 outlines the rise in Irish bank lending since
2007, and how this lending was funded by wholesale markets. Section 3 shows how the rise
in Irish house prices is almost entirely explained by increased mortgage lending. Section
4 shows how the rise in house prices generated a large expansion of construction activity.
Section 5 looks an earlier Irish bubble, in agricultural land. Section 6 examines the problems
of bank funding and mortgages; and Section 7 looks at the causes of the Irish bubble.

2

Irish Bank Lending.

Figure 2 shows deposits and lending to the private sector of Irish banks relative to GNP
since the last quarter of 1992.7 It can be seen that Irish banks were more or less completely
deposit funded until 1997, with loans and deposits both around 75% of GNP, and loans
to the non-financial sector about 60% of GNP. By comparison, loans to the non-financial
private sector of UK banks were one third higher at 80% of GDP, the average level that
Schularick and Taylor (2009) find across 12 industrialized economies at that time.
By 2004, UK lending had risen to 95% of GDP, but had been overtaken by Ireland with
lending of 100%. However, as Figure 2 shows, at this stage Irish lending accelerated rapidly.
By the middle of 2008 the international credit bubble saw UK non-financial lending rise
to 104% of GDP, again close to the international average reported by Schularick and Taylor
(2009). In Ireland however, non-financial lending had risen to 200% of GNP, with total
lending equal to 250% of GNP, and rising to 270% if securitised mortgages are added. By
contrast, deposits had risen only to around 125% of GNP. By the first quarter of 2009, a
combination of continued increases in lending (nominal lending peaked in mid-2008 and has
fallen slightly since) and falling GNP meant that non-financial lending had risen to 225%
of GNP and total lending to 290%, rising to 320% of GNP when securitised mortgages are
included.8
It is this more than tripling of bank lending that accounts for the Irish boom since 1997.
Effectively the Irish economy segued from one driven by competitiveness in the 1990s to one
driven by a credit fuelled bubble in the 2000s.
7

Numbers are from Table C3, column 1 and Table C8 of the Irish Central Bank Quarterly Bulletin
The growth of Irish credit does appear sedate however when compared with Iceland,
where lending to firms and households went from 1.8 times GDP in 1999 to 5 times in
2008.
http://www.sedlabanki.is/?pageid=552&itemid=198dfc1c-a027-4abf-b95f-c51fda8bc5f5,
http://www.statice.is/Statistics/National-accounts-and-public-fin
8

6

 Loans
Non−financial loans
Deposits
250

% of GNP

200

150

100

1995

2000

2005

2010

Figure 2: Lending to the private sector and deposits of Irish banks as a percentage of GNP,
1992 to 2009.

Mortgages
140

Development

Billions of 2009 Euro

120

100
Total lending to non−financial private sector in 2000 = ...83 Bn

80

60

40

20

1995

2000

2005

Figure 3: Property lending in billions of 2009 Euro, 1993–2009.
7

2010

 Dublin Secondhand
Dublin New
16

Secondhand
New

14

Years' earnings

12

10

8

6

4

1980

1985

1990

1995

2000

2005

2010

Figure 4: Irish house prices relative to average earnings, 1980–2009.
Particularly large rises occurred in property lending, both in mortgages to households and
in loans to builders and developers. As Figure 3 shows, at the start of the credit bubble in
1997 banks were lending €20 Bn in mortgages in 2009 prices, and €10 Bn to developers.9 By
2008, the value of mortgage lending (including the quarter of mortgages that were securitised)
had risen to seven times its 1997 value, while lending to developers was 11 times its 1997
value.
For comparison, during this time, thanks in large part to the construction activity generated by this lending, real GNP rose by 75%. To put these number in further perspective,
the total value of bank lending to the non-financial private sector in 2000 was around €80
Bn in 2009 prices. Irish banks were therefore lending 75% more in mortgages than they had
been lending to everyone in Ireland 8 years earlier, and 40% more to developers.

8

 Commercial
New House

Index: Real price in 1995=100

300

250

200

150

100

1980

1985

1990

1995

2000

2005

2010

Figure 5: SCS-IPD index of commercial property prices compared with price of new houses,
both in 2009 prices. 1995=100.

3

Lending and Property Prices.

This explosion of bank lending led to predictable rises in the prices of Irish houses and
commercial property. Figure 4 shows that in 1995, the average price of a house in Ireland
(new or secondhand, in Dublin or elsewhere) was equal to 4 years’ average earnings in
industry.10 At the peak in late 2006, new house prices nationally had risen to 10 times
earnings, while Dublin secondhand prices had risen to 17 times earnings.
Figure 5 shows that SCS-IPD index of commercial property prices, adjusted for inflation,
showed similar rises to new house prices. The same underlying processes are driving both
markets.11 The peak in commercial prices occurred about a year after the residential peak,
in late 2007.
9

Classified in Table C8 of the Irish Central Bank Quarterly Bulletin as Construction and Real Estate
Activities.
10
Department of Environment data. The ESRI index is similar.
11
A Jorgensen procedure finds a cointegrating relationship between the two series until 2006 significant at
1% of P = 0.97C where P, C are real price of new houses and commercial property with value in 1995 Q1
set to 100.

9

 FTB Mortgage
9

New House Price

Years' earnings

8

7

6

5

4

3

1990

1995

2000

2005

2010

Figure 6: Average first time buyer mortgage and new house prices, relative to average earnings.
The dearth of transactions makes the extent of subsequent falls hard to gauge. The most
plausible estimates for housing are those of the estate agents Sherry-FitzGerald who estimate
nominal prices nationally have fallen from peak nationally by 37% and in Dublin by 42% by
mid 2009.12
What drove the rise in property prices? The proximate cause was bank lending. In a
market with rising prices, borrowers were willing to accept any loan that banks would give
them so, with property supply fixed in the short run, prices moved in proportion to the size
of new loans.
The cleanest data on new lending is the Department of Environment series on mortgage
approvals for first time buyers. Figure 6 plots the size of mortgages to first time buyers and
new house prices, both relative to average earnings. Two things stand out. First, the two
series move closely together.
12

http://www.sherryfitz.ie/aboutus/NewsItem.aspx?ID=515. One anecdotal data point comes from the
Commercial Court which is hearing a large volume of cases brought against property developers by their
creditors. Mr Justice Peter Kelly has stated that he is dealing “on a daily basis” with cases with valuations
of commercial property showing falls in value of 70–80 per cent. Irish Times, 20 October 2009. “Judge says
cases showing falls in property values of up to 80%”

10

 Intercept
7.5092
(4.2788)

Mortgage

Interest

Popn

SER

R2

DW

1.1259∗∗ −0.1149∗∗ −0.0016 0.3217 0.9676 0.5495
(0.2212) (0.0202) (0.0014)

OLS regression of average new house price relative to average earnings on
average first time buyer mortgage relative to average earnings, real mortgage
interest rates, and population. Quarterly data. Andrews HAC standard
errors in parentheses.

Table 1: Mortgage lending and new house prices in Ireland, 1979–2006.

Table 1 shows that mortgages have a strong impact on house prices, with a €1 rise in
mortgages increasing house prices by €1.13. On the other hand, interest rates have a modest
effect. In order to cause the ratio of price to earnings to rise by one (say, from 4 to 5 years’
earnings), it takes a fall of nearly 9 percentage points in mortgage interest rates. In other
words, rising house prices were driven predominantly by increases in the size of mortgages
that banks were willing to give, with interest rates playing a secondary role, and population
none at all.
Looking at cointegration, the first eigenvalue implies a long run relationship between
house prices, mortgages, and real interest rates of
M
P
= 1.33
− 0.17r
Y
Y
where P is the average price of a new house, Y is average industrial earnings, M is average
first time buyer mortgage, and r is the real mortgage lending rate.13
The second notable thing about Figure 6 is that the peak of first time buyer mortgages
(in number as well as value) occurred in the third quarter of 2006, and then declined quite
sharply. Mortgages for movers and residential investment borrowers follow a similar pattern.14 While the total value of mortgages continued to rise after 2006, it did so at a much
slower rate.
Just as in the United States, rapid increases in credit were accompanied by a marked
deterioration in lending standards. Among first time buyers purchasing new houses in 2006,
13

Using a two lag VAR, a Johansen procedure identified cointegration relationships significant at 5%:
(1, −1.31, 0.16), (1, −0.70, 0.20) with eigenvalues 0.20, 0.14. Including observations after the market peak in
2006 caused the relationship to disappear, consistent with the idea that the post-bubble housing market is
qualitatively different from the bubble one, with falling supply and demand for credit, large stocks of unsold
houses, and uncertainty about the magnitude of future price falls.
14
http://www.ibf.ie/pdfs/IBFPwCMortgageMarketProfileQ407.pdf

11

 Completions
400

New House Price/Earnings

350

Index: 1995=100

300

250

200

150

100

1975

1980

1985

1990

1995

2000

2005

2010

Figure 7: Quarterly housing completions and new house prices relative to earnings.
1995=100.
only 24% had loans to value below the traditional maximum of 80% (and this falls to 15%
in Dublin), 64% had ratios above 90%, and 30% were 100% mortgages. Looking at terms
of mortgages, only 17% of first time buyers of new houses took out mortgages of less than
25 years, and 58% took out loans of more than 30 years, while for Dublin the corresponding
figures are 9% and 69%.

4

House Prices and Construction.

Q theory predicts that residential investment should rise as the price of houses rises relative
to their construction cost. Because labour and material costs move with average earnings,
the ratio of new house prices to earnings gives a proxy for Tobin’s average q. Figure 7 shows
how private housing completions and the ratio of new house prices to average earnings rose
together until 2007. Looking at logs of the two series between 1979 and 2006, applying a
Johansen procedure to a VAR with 8 lags and seasonal dummies, we find a significant at
1 per cent cointegrating relationship ln(C) = 1.56 ln(P/Y ) where C is private completions
per quarter, and P/Y is the ratio of new house prices to average earnings. In other words,
12

 Housing
Other

14

% of GNP

12

10

8

6

4

1970

1975

1980

1985

1990

1995

2000

2005

2010

Figure 8: Share of construction in GNP, 1970–2008.
holding earnings constant, a 10 per cent rise in house prices is associated with a 15 per cent
rise in completions.
Figure 8 shows the marked impact of house building on GNP. Until 1997 Ireland, like
any other industrial economy, got from 4 to 6 per cent of its GNP from building houses.
By 2006–07 this had risen to 13%. For comparison, in the other major European housing
boom, in Spain, housing investment peaked at 9% of GDP. Adding other construction (but
not roads), building was accounting for an extraordinary one fifth of Irish national income
at the peak of the bubble in 2007.15
If we look at the increase of Irish GNP between 2000 and 2007, 28 per cent is accounted
for directly by the growth of construction output.
This construction boom created two serious distortions in the Irish economy. First,
as labour demand rose, particularly for less skilled labour, wage rates across the economy
were driven up out of proportion to productivity growth, leading to a fall in international
competitiveness. Between 2000 and 2008 hourly earnings in manufacturing relative to major
15

These data somewhat overstate the contribution of construction to Irish GNP to the extent that they
include imported inputs. However the bulky nature of building materials ensures that most are produced
domestically.

13

 1998 Irish Pounds per acre

8000

6000

4000

2000

2000

1995

1990

1985

1980

1975

1970

1965

1960

1955

1950

1945

1940

1935

1930

1925

1920

1915

1910

1905

1900

Figure 9: Real price of land in the Limerick region, 1901–1996.
trading partners rose 20 per cent.16 The second effect of the building boom was a large rise
in government revenue which funded a large rise in government expenditure which is proving
painful to reverse with the ending of the boom.

5

The previous bubble.

The credit fuelled bubble in property prices was actually the second asset bubble in Ireland in a generation. Figure 4 shows real land prices in the Limerick region estimated by
Roche and McQuinn (2000), expressed in 1997 pounds (multiply by 1.75 to convert into 2009
euros).
After accession to the European Economic Community in 1973, Irish banks began to
lend heavily to farmers to modernise and expand. As a result, the real price of agricultural
land tripled between 1975 and 1977, reaching a peak equivalent to €14,000 per acre in 2009
prices. Real Irish GNP in 1977 was 28 per cent of its 2008 level, so this price is roughly
equivalent to €50,000 per acre in current purchasing power for purely agricultural land. For
comparison, during the recent boom, when agricultural land prices were driven by demand
16

Central Bank of Ireland, Quarterly Bulletin, Table B4.

14

 for potential development, prices peaked in 2006 at an average of €21,000 per acre nationally,
and €35,000 in Dublin, Kildare and Wicklow.17
The bubble quickly burst as farmers ran into difficulties servicing loans: between 1977
and 1980 real price fell by around 75 percent and remained at this level, more or less where
it had started in 1973, until 1995, eighteen years after the peak. It can also be seen that,
after an earlier boom at Independence in 1921, it took 50 years for prices to regain their
peak.

6

The Irish banks after the bubble.

The collapse of the building boom left Irish banks facing large losses to builders and developers. Despite denials by the banks that they faced any difficulties, their share prices started
to slide steadily after March 2007.
This decline accelerated after May 2008 as domestic banking difficulties started to merge
with the deepening international financial crisis. The crisis came to a head on 29 September
with a run in wholesale markets on the most aggressively expansionary of the Irish banks,
Anglo-Irish.
Although the crisis had been building for 18 months, the government and financial regulators appear to have been taken entirely by surprise. At a late night meeting with banks, the
Irish government committed itself to the unusual step of guaranteeing all existing senior debt
of Irish banks (among European economies, only Denmark subsequently did this) as well as
deposits. In addition, as well as guaranteeing the two large retail banks (AIB and Bank
of Ireland) and two smaller mortgage lenders; for reasons that have never become clear the
Irish government agreed to guarantee two specialist property development lenders (AngloIrish Bank and Irish Nationwide Building Society) despite already well known deficiencies
in their corporate governance.
The most likely rationale for the Irish government’s actions is that it still believed that
the liquidity problems of the Irish banks merely reflected market nervousness in the wake of
the Lehmann collapse, and not justified concerns about the solvency of these institutions.
Despite the liability guarantee the shares of Irish banks continued to slide, with AngloIrish bank being nationalised in February 2009, and the government announcing that it
would establish a bad bank called NAMA to buy non-performing development loans from
banks.
17

http://www.knightfrank.ie/documents/Farms%20Market%20-%2009.pdf

15

 The idea of a bad bank is simple. Suppose that a bank has €10 Bn of capital and has
loans of €100 Bn, including €20 Bn of development loans. Suppose that they suffer 50%
losses on these development loans, which wipes out their capital and leaves them insolvent.
If the government buys the €20 Bn of non-performing loans at their face value, the banks
are recapitalized and can continue lending at their previous level, although the tax payer
faces a significant loss, while the bank’s shareholders and bond-holders lose nothing.
This appears to be the arrangement originally envisaged by the Irish government. To
mitigate adverse selection problems, every borrower of the the banks associated with NAMA
with more than €5 million in property loans (performing or otherwise) would have all their
loans, including non-property loans, transferred to NAMA which would then function as
their bank. NAMA’s portfolio would be approximately one third development land, one
third unsold development projects, and the remaining third being so-called associated loans,
the non-property loans of transferred borrowers.
Two complications intervened. First, the borrowers being transferred to NAMA included
many of the largest and most profitable firms in Ireland. Predictably, a large UK bank
offered to repay the loans of these firms to NAMA banks if they transferred their business
to it. As a result, the Irish government has apparently agreed that Irish banks will be
given a veto on the transfer of borrowers to NAMA. This means the the adverse selection
problem reappears, and the portfolio of NAMA, in particular the associated loans that were
supposed to generate most of its cash flow, will be worse than the government’s business
plan predicted.18
The second complication came from Europe. First, the European commission dictated
that the Irish government could pay only 70 per cent of the face value of non-performing
loans. In terms of our example, this would mean that the bank receives €14 Bn for its €20
Bn of bad loans, leaving it with only €4 Bn of capital. To maintain a minimum capital ratio
of 8% of loans, the bank must either raise another €4 Bn of capital in the market, or halve
its loan portfolio to €50 Bn. For the two main Irish banks, the 30 per cent NAMA haircut
on transferred loans would reduce their book value of their equity by approximately half.
For AIB, transferring €24 Bn of loans to NAMA entails a haircut of €7.2 Bn. Given that
they have already written down these loans by €2.2 Bn, their common equity will fall by €5
Bn or 49 per cent.
18

http://www.ibf.ie/pdfs/IBFPwCMortgageMarketProfileQ407.pdf

16

 For Bank of Ireland, transferring €15 Bn of loans to NAMA entails a haircut of €4.6 Bn.
Given that they have already written down these loans by €0.8 Bn, their common equity
will fall by €3.8 Bn or 57 per cent. 19
However, more recently the ECB has required the Irish government to value each asset
being transferred individually at market rates, rather than paying 70 per cent of their face
value, sight unseen. It therefore seems likely that the amounts that banks receive, particularly for development land, could be considerably less than the 70 per cent first envisaged.
To understand if NAMA will work, we must first ask what problems the Irish banks face.
There are three, inter-related ones.
First, Irish banks have suffered large losses on developer loans. Secondly, to fund sharply
increased lending, Irish banks borrowed heavily in wholesale markets. Thirdly, losses on
other loans, in particular mortgages, are likely to be substantial.
The Irish government has, through NAMA, addressed the first problem of developer loans
and, in doing so, has pushed the Irish state close to, and quite possibly beyond, the limits of
its fiscal capacity. However, NAMA does not address the other two problems: on the liability
side of heavy wholesale debt, and on the asset side of further possible losses on other loans.
We address these problems in turn.

6.1

Bank liabilities.

The first problem that the Irish banks face is that while NAMA remedies many of the current
(if not future) problems with the asset side of their balance sheet, the liability side remains
difficult. As Figure 2 shows, the extraordinary expansion of Irish bank lending after 1997 was
funded mostly in wholesale markets, both through inter-bank borrowing and bonds. Like
other financial institutions, Irish banks were able to take advantage of the great moderation
before mid 2008 to borrow wholesale at almost central bank rates, and to lend at low rates
in the Irish property market.
Figure 10 shows the evolution of the non-capital liabilities of Irish banks.20 It can be
seen that in 1999, public deposits were much the largest liability. By December 2007, bonds,
were almost as large as public deposits, while inter-bank deposits were considerably larger.
19

Figures based on Supplementary Documentation National Asset Management Agency 16 September
2009 www.nama.ie/Publications/2009/Supplementary_Documentation.pdf and Hank Calenti Irish Banks–
Vegetative State–Reduce to Underperform. RBC Capital Markets, 13 August 2009.
20
Figures are total liabilities of Irish banks to Irish and foreign residents from Central Bank Quarterly
Bulletin Table C3 minus liability totals from Table C6 for credit institutions with mostly foreign operations.
The inter-bank deposit figure excludes lending of Irish banks to each other by subtracting loans to Irish
resident financial intermediaries. 1999 figures have been multiplied by 1.4 to express them in 2009 prices.

17

 1999−Jan

2007−Dec

2009−Jun

250

Billion Euro

200

150

100

50

Other

Inter−bank

ECB

Bonds

Deposits

Other

ECB

Inter−bank

Deposits

Bonds

Other

Inter−bank

ECB

Bonds

Deposits

0

Figure 10: Non-capital liabilities of the Irish banks.
From early 2008, the Irish banks have faced a sustained loss of bond funding, which have
been replaced by inter-bank borrowing, and borrowing from the European Central Bank.
There has also been a marked decline in “Other liabilities” corresponding to swap contracts.
The positions of AIB and Bank of Ireland are particularly worrying. AIB had outstanding
bonds in August 2009 of €24 Bn (plus €5 Bn subordinated debt) and ECB borrowings of €34
Bn. Bank of Ireland had bonds of €45 Bn (plus €8 Bn subordinated) and ECB borrowing
of €17 Bn. By comparison, even if valuations go as they hope, AIB will receive only €17 Bn
from NAMA, while Bank of Ireland will receive €11 Bn.
The problem of the size of bank debt is compounded by its short maturity. In 2010 AIB
has €9.8 Bn of bonds maturing, Bank of Ireland €12.2 Bn, and the nationalised Anglo Irish
€9.2 Bn. In addition the Irish state will have to borrow about €20 Bn, and roll-over €7.4
Bn in maturing debt. Borrowing such sums, on top of issuing €55 Bn of NAMA debt to the
banks, is likely to prove challenging, even if markets do not continue to grow more nervous
about the sovereign and quasi-sovereign debt of weaker Euro-zone states.
Difficulties of funding Irish bank activities are likely to be aggravated by the announced
withdrawal of ECB support. During the international financial crisis Irish banks were able
18

 to benefit from the ECB’s policy of quantitative easing by borrowing against lower quality
assets at 1 per cent interest. Now that the French and German economies have technically
left recession, the ECB has announced that special lending facilities will start to be curtailed.
To the extent that the ECB feels it necessary to re-establish a reputation for political independence, harsh treatment of one of the smaller, peripheral Euro-zone economies could be
seen as a useful way to send a signal to international markets and larger Euro zone economies
that it still takes the stability pact conditions seriously.
In summary then, the business model of Irish banks for the last decade of borrowing
heavily in wholesale markets to lend in Irish property markets appears defunct. Irish banks
now face wholesale lending rates of 5.6% to fund mortgages paying 3.5% interest.
Even domestic deposits are proving problematic as the Irish economy shrinks. Deposits
from Irish residents have fallen by 5.5 per cent in the year to September 2009, probably as
a result of dis-saving by households and firms. Banks may also lose deposits as people vote
with their wallets on the politically unpopular bank bailout, and transfer their savings to
institutions not involved with NAMA and guaranteed by more credit-worthy sovereigns.
In these circumstances, the Irish banks must shrink their balance sheets, by reducing
lending and repaying debt. It seems likely that Irish credit levels will return to average
international levels in the region of 100% of national income. However, because €110 Bn of
their lending (equal to 80 per cent of GNP) is tied up in recently issued mortgages of 35
years or longer, it seems unlikely that there will be very much new lending to any sector of
the Irish economy for the foreseeable future.
The large mortgage book of Irish banks creates another problem for them. The interest
rate on most of these mortgages is set as a fixed markup over ECB rates, so the Irish banks
have a limited capacity to recapitalize themselves by widening their lending spreads, as US
banks did in the early 1990s after the Savings and Loan Crisis (Koo, 2008, 10).
This sharp contraction in the supply of credit is likely to coincide with a sharp fall in
demand. Irish firms and households have run up unusually large debts to fund purchases of
property which has fallen sharply in value. Just like Japan in the 1990s (Koo, 2008, 1–37),
their overwhelming priority will be to reduce borrowings to sustainable levels.
As expanding credit caused property values to rise, so contracting credit will cause them
to fall. If a return to reliance on the retail funding used until 1997 leads to a return to the
lending standards of the mid-1990s, then Figures 6 and 5 would predict a fall in new house
prices and commercial property prices of about two thirds from peak levels. There is a risk,
moreover, that a more or less complete cessation of lending for two or three years if banks

19

 are forced rapidly to reduce wholesale borrowings, could result in even larger falls than this
in the medium term.
The return of the Irish economy to normal levels of bank lending and property prices is
to be welcomed. However, it does make it inevitable that the Irish state will suffer large
losses on bad loans taken over by NAMA.
The Irish government is assuming that property prices have fallen by 40 per cent from
peak, and that if prices rise by 10% in the next decade (in other words to 2004 levels), that
NAMA can break even. However, what we have seen is that the only way that property
prices can return to 2004 levels is for bank lending to grow at 2004 levels. Given the state
of Irish bank balance sheets, even after NAMA, this is an impossibility.

6.2

Mortgages.

As well as problems of excessive debt on the liability side, Irish banks face the prospect of
further loans losses, particularly on mortgages. As we saw in Figure 3, mortgage lending rose
to €150 Bn, or over 100 per cent of late 2009 GNP, about €40 Bn of which was securitized.
In the US in September 2009, over 14 per cent of mortgages were at least one payment
overdue or in foreclosure; with 25 per cent in Florida. Looking at a very large sample of US
mortgages, Foote et al. (2009) identify two factors that overwhelmingly drive default: falls
in house prices and unemployment. Among prime mortgages, they find that a 10 per cent
fall in house prices increases the risk of a mortgage being in default for 90 days by 82 per
cent; while a one percentage point increase in unemployment increases the risk by 22 per cent
(Table 5). The level of strategic default is low even in states with non-recourse lending, and
most homeowners go to considerable lengths to continue paying their mortgages even when
faced with severe negative equity, viewing the prospect of losing their homes as terrifying.21
A notable aspect of US mortgage delinquency is the relatively high rate of self-cure:
delinquent mortgages that start performing again. Adelino, Gerardi and Willen (2009) find
that about 30% of US mortgages that are two months delinquent have started performing
again within twelve months. This reflects the relative ease of finding a new job in the US. The
loss of competitiveness of the Irish economy during the bubble, and the fact that many of the
unemployed from the building industry have little formal education, suggests that transitions
21

White (2009) argues that the levels of default are irrationally low for homeowners, reflecting their shame
and fear at defaulting on loans, and suggests that a higher level of default would be a socially efficient means
of giving more bargaining power to households in negotiating with banks over sharing the losses from falling
house prices.

20

 from unemployment to employment will be rarer, and the rate of self-cure correspondingly
lower.
The recent falls in Irish house prices and swift rise in Irish unemployment to 12.5 per
cent (compared with 10 per cent in the US) suggests that US rates of default are a strong
possibility here.22 In addition to unemployment, Ireland has a uniquely large source of
default risk in emigration. Emigration is one circumstance where strategic default becomes
compellingly easy, and the possibility of walking away on a deeply underwater property loan
is likely to join unemployment as a compelling reason to make a new life overseas.
Default rates for homeowners in the US start to rise sharply as the loan to value ratio
exceeds 120 per cent.23 How close is the average Irish borrower to this cutoff? If we look at
the average first time buyer who took out a 35 year, 90 per cent mortgage at 3.5% interest
three years ago at the peak of the market, it would take a fall in prices of 29 per cent to
leave their outstanding loan balance (which has now been paid down by 5 per cent) twenty
per cent larger than the value of their home. For the average mover, taking out a 30 year,
70 per cent loan to value mortgage at that time, it would take a fall of 45 per cent. Given
that house price falls around 40 per cent have already occurred, it can be seen that many
first time buyers who bought around the market peak are well within the danger zone, with
movers getting close to it.
In the US, banks currently recover around 57 per cent of their outstanding loan in a
foreclosure (Foote et al., 2009). In Ireland, the difficulties of evicting people from their
family homes may make the losses of Irish banks from non-performing mortgages somewhat
larger.
For AIB, it would take a loss of 16 per cent on its mortgage book of €33 Bn to wipe out
its post-NAMA book capital of €5 Bn. However, it would only a loss of 5 per cent on Bank
of Ireland’s much larger mortgage book of €59 Bn to wipe out its post-NAMA book capital
of €3 Bn.
The difficulties of the Irish banks do not end with property loans. Loans to businesses
account for about one quarter of the loans of the two large Irish banks. Again, substantial
losses are possible here. With construction accounting for one fifth of national income in
2006–07, many Irish firms were heavily or entirely reliant on supplying the construction
sector and have had their market effectively disappear.
22

Unemployment shows no sign of levelling off: the Department of Enterprise redundancy series has been
steady at 6,000 per month since the summer: http://www.entemp.ie/employment/redundancy/statistics.htm
23
First
American
CoreLogic
Negative
Equity
Report
2009
Q3,
Figure
4
http://www.calculatedriskblog.com/2009/11/negative-equity-report-for-q3.html

21

 The property related borrowings of the owners of smaller Irish companies are a particular
concern. By self-selection, the entrepreneurs who establish and run small and medium
enterprises are more motivated than average to make money, and the surest way to make
money in Ireland during the bubble decade was to borrow heavily to invest in land and
property. The heavy personal debts of these business owners are now an impediment to their
companies’ survival and may lead to large job losses as owners are forced into bankruptcy
over losses in property speculation. The destruction of the Irish entrepreneurial class may
prove one of the most enduring and costly consequences of the property bubble.

7

Explaining the Irish Credit Bubble.

Credit booms occur sufficiently often and with sufficiently alarming results to have generated
a large economic literature. Financial accelerator models emphasise how, in a model of asymmetric information about project quality, rising prosperity increases the value of collateral,
which stimulates lending, which further increases the value of collateral (Kiyotaki and Moore,
1997, Matsuyama, 2007). Rajan (1994) emphasises herding among banks, Dell’Ariccia and Marquez
(2006) show how reduced adverse selection during booms causes lending standards to be relaxed, while Berger and Udell (2004) examine the loss of institutional memory among loan
officers during booms. In the context of the Asian crisis Corsetti, Pesenti and Roubini (1999)
argue that the existence of state guarantees of bank liabilities, explicit or implicit, encouraged moral hazard where foreign lenders extended credit to domestic banks for sub-optimal
projects.
Notable recent empirical studies of international bubbles include Mendoza and Terrones
(2008) who show how rapid TFP growth and relaxed supervision precede credit booms in
developed economies, while developing economy booms are associated with capital inflow.
Among studies of the US sub-prime crisis, Dell’Ariccia, Igan and Laeven (2008) find that
areas with the largest credit expansions experienced the largest declines in credit standards,
while Mian and Sufi (2008) show that income and mortgage credit growth are negatively
correlated across zip codes during the sub-prime boom of 2002-2005, consistent with the
idea that an increased supply of credit rather than improved fundamentals drove lending.
All of these factors were present for Irish banks, and their impact was magnified by
failures of regulation by the Central Bank and government. The rapid expansion of credit
in the Irish economy and the consequent rise in property prices and construction activity
represent systematic failures of control at all levels of the Irish economy.
22

 At the first level, bank management lost awareness of the riskiness of their portfolios,
extrapolating past rises in incomes and property prices to assume that prices could only go
on rising or, at worst, stabilise. The financial accelerator was amplified in Ireland by the
extreme narrowness of markets, especially for commercial property, so expansions of credit
by a single bank could have a large impact on prices.
The mis-management of Irish financial institutions was amplified by the presence of a
genuinely rogue bank, Anglo Irish. Through aggressive property lending, this had gone
from an insignificant merchant bank in the 1990s to the joint-second largest bank by 2007.
The two large retail banks, AIB and Bank of Ireland came under sustained pressure from
analysts to match the profits and growth of Anglo Irish, and responded with ultimately
baleful consequences.
Since the seventeenth century, financial innovation has consisted in banks finding new
ways to lose money. However, while US, UK and European banks lost money in exotic
derivatives, Irish banks lost money the old fashioned way, by making bad property loans.
It is well known that banks get carried away during booms, which is why institutions
called central banks exist to curb their enthusiasm. The Irish Central Bank has not had a
history of independence from government and, after joining the Euro-zone, contented itself
with gathering statistics and issuing currency, and made no effort to control the obvious credit
bubble engulfing the Irish economy. While there was no explicit relaxation of regulation as
occurred in the Nordic economies prior to their credit boom in the late 1980s (Englund,
1999), the mind set of Central Bank that Ireland was a small region of a larger economy
where prices and interest rates were set exogenously so policy could have no impact, did
lead to the effective disappearance of the active regulation that had been maintained when
Ireland had an independent currency.
While the Irish Central Bank lost power to set interest rates, we have seen above that
interest rates actually had a limited impact on property prices: the decisive effect was
increased size of loan. Had the Central Bank restricted mortgages to traditional levels of
80% loan to value, and three to four times income, house prices would have risen in line with
income and many of the distortions of the past decade would have been avoided.
The failings were even graver with respect to development loans where, in many cases,
collateral was dispensed with, and banks lent against so-called “personal guarantees” that
the bank would have recourse to the borrower’s personal assets in the event of default.
Large developers were able to borrow hundreds of millions without posting collateral on the
strength of their believed equity in other highly leveraged projects they had undertaken.

23

 In summary, the activities of the Irish banks remained extremely simple by international
standards and could easily have been regulated had the will to do so been present.
Given the weak independence of the Irish Central Bank, the will to control banks derived
ultimately from government. As Johnson (2009) argues, the expansion of banking activities
since 1990 has seen the effective capture of governments by the financial industry, as politicians in industrialized economies came to view finance as the root of national prosperity.
Two factors in Ireland aggravated this tendency. The first is the small size of the country
which ensures that politicians and financiers are inevitably well known to each other: it is
easy to become “too connected to fail.”
Secondly, the unusual magnitude of the Irish credit bubble made the apparent bounty
of the banks’ activities appear larger here than elsewhere. In particular the rise in employment stemming from the construction boom—particularly among low skilled workers in rural
areas without other forms of employment—generated a natural alliance of interests among
politicians, developers and banks.24
The Irish government was therefore poorly equipped to understand the crisis when it
finally occurred in late September 2008. Instead of recognising the borrowing difficulties of
Irish banks as the result of well grounded market apprehension about their solvency, the
Irish government responded to the crisis as if it were a temporary problem of liquidity in the
aftermath of the Lehman collapse.
Once committed to guarantee all senior debt as well as deposits of all six Irish banks, the
Irish government found itself in the position of not being able to change direction to share
losses with bank bond holders without being forced to admit that it had made a mistake in
its initial guarantee.
However, the question remains of why, given that Ireland’s bankers were probably no
more reckless, its regulators no more spineless, and its politicians no more clueless than
their counterparts elsewhere, how did Ireland come to have a far larger credit boom than
other wealthy economies, with the exception of Iceland? The most likely reason is that
Ireland’s credit boom was preceded by a decade of real, competitiveness driven growth. Irish
lenders, borrowers, and regulators became accustomed to economic growth of 6–7 per cent
that disguised the magnitude of the credit bubble that followed it.
24

This is not to deny that Irish banks may have exerted more direct means to control the political process
but, until Irish members of parliament have to declare their liabilities as well as their assets, it is not possible
to establish how many leading politicians received large, interest free loans from two banks.

24

 8

Conclusions.

In the last decade, the Irish economy has experienced an unusually large credit bubble.
Lending as a fraction of GNP increased from 60% in 1997, to over 200% in 2008, twice the
level of other industrialized economies.
In the aftermath of this bubble, the Irish banking system faces three inter-related problems. The first is that it has made large losses on loans to property developers. The second
is that it has large wholesale liabilities to international bond holders and, increasingly, to
the European Central Bank. The final problem is that it faces likely further large losses on
mortgages and business loans.
The Irish government’s policy response has been solely to address the first problem of
losses on developer loans by establishing a state institution to buy these loans. However, by
ignoring the second problem of the large wholesale liabilities of the Irish banks, this project
will inevitably end in expensive failure.
As Irish banks are forced to repay this wholesale borrowing and to shrink their balance
sheets to normal international levels, the sharply diminished supply of credit will lead inevitably to continued sharp falls in property prices. These falls in property prices will result
in severe losses for the Irish taxpayer on the ill-conceived NAMA project.
Despite having pushed the Irish state close to, and quite possibly beyond, the limits of
its fiscal capacity with the NAMA scheme, the Irish banks remain as zombies whose only
priority is to reduce their debt, and who face complete destruction from mortgage losses.
The issue therefore is not whether the Irish bank bailout will restore the Irish banks so
that they can function as independent commercial entities: it cannot. Rather it is whether
the Irish government’s commitments to bank bond holders when added to its existing spending commitments, will overwhelm the fiscal capacity of the Irish state, forcing outside entities
such as the IMF and EU to intervene and impose a resolution on the Irish banking system.

References
Adelino, Manuel, Kristopher Gerardi and Paul S. Willen. 2009. Why don’t lenders renegotiate
more home mortgages? redefaults, self-cures, and securitization. Working Paper 2009-17
Federal Reserve Bank of Atlanta. 20
Berger, Allen N. and Gregory F. Udell. 2004. “The institutional memory hypothesis and the
procyclicality of bank lending behavior.” Journal of Financial Intermediation 13(4):458–
25

 495. 22
Corsetti, Giancarlo, Paolo Pesenti and Nouriel Roubini. 1999. “Paper tigers?: A model of
the Asian crisis.” European Economic Review 43(7):1211–1236. 22
Dell’Ariccia, Giovanni, Deniz Igan and Luc Laeven. 2008. Credit Booms and Lending Standards: Evidence From The Subprime Mortgage Market. Discussion Paper 6683 C.E.P.R.
22
Dell’Ariccia, Giovanni and Robert Marquez. 2006. “Lending Booms and Lending Standards.”
Journal of Finance 61(5):2511–2546. 22
Englund, Peter. 1999. “The Swedish Banking Crisis: Roots and Consequences.” Oxford
Review of Economic Policy 15(3):80–97. 23
Foote, Christopher, Kristopher Gerardi, Lorenz Goette and Paul Willen. 2009. Reducing
Foreclosures: No Easy Answers. Working Paper 15063 National Bureau of Economic Research. 20, 21
Honohan, Patrick and Brendan Walsh. 2002. “Catching Up with the Leaders: The Irish
Hare.” Brookings Papers on Economic Activity 33(2002-1):1–78. 1
Johnson, Simon. 2009. “The Quiet Coup.” The Atlantic . May. 24
Kiyotaki, Nobuhiro and John Moore. 1997. “Credit Cycles.” Journal of Political Economy
105(2):211–48. 22
Koo, Richard. 2008. The Holy Grail of Macroeconomics: Lessons from Japan’s Great Recession. New York: Wiley. 19
Matsuyama, Kiminori. 2007. “Credit Traps and Credit Cycles.” American Economic Review
97(1):503–516. 22
Mendoza, Enrique G. and Marco E. Terrones. 2008. An Anatomy Of Credit Booms: Evidence From Macro Aggregates And Micro Data. Working Paper 14049 National Bureau
of Economic Research. 22
Mian, Atif and Amir Sufi. 2008. The Consequences of Mortgage Credit Expansion: Evidence
from the 2007 Mortgage Default Crisis. Working Paper 13936 National Bureau of Economic
Research. 22
26

 Pozen, Robert. 2009. Too Big to Save? How to Fix the US Financial System. New York:
Wiley. 4
Rajan, Raghuram G. 1994. “Why Bank Credit Policies Fluctuate: A Theory and Some
Evidence.” The Quarterly Journal of Economics 109(2):399–441. 22
Roche, Maurice J. and Kieran McQuinn. 2000. Speculation in agricultural land. Economics,
Finance and Accounting Department Working Paper Series n1010700 Department of Economics, Finance and Accounting, National University of Ireland - Maynooth. 14
Schularick, Moritz and Alan M. Taylor. 2009. Credit Booms Gone Bust: Monetary Policy,
Leverage Cycles and Financial Crises, 1870–2008. NBER Working Papers 15512 National
Bureau of Economic Research, Inc. 1, 6
White, Brent T. 2009. Underwater and Not Walking Away: Shame, Fear, and the Social
Management of the Housing Crisis. Arizona Legal Studies Discussion Paper 09-35 University of Arizona. 20

27