Address by Governor Patrick Honohan to a conference organised by the Central Bank of Iceland, Reykjavík

28 November 2013 Speech

It is not only in Iceland and Ireland that the failure of big banking sectors has had a devastating effect on entire economies, though they do provide the most spectacular examples. What macroprudential lessons can be learnt from recent experience about vulnerability of small countries with big banking sectors?

1. The Issue

In the age of cloud computing, the actual location of a bank is less unambiguous than it was in the past, but sufficient legal clarity still exists to determine the relevance of national jurisdiction. Indeed, location decisions of banks as providers of international banking services on cross-border banking have been considered a matter for the industrial policy of several countries (trying to become financial centres). Especially in such cases, there is a need for national macroprudential policies to ensure that the overall national policy mix adequately copes with the risks presented by banking location.

To be sure, having an undeveloped or repressed banking system—one that is small relative to the size of the economy—is in itself a restraint on growth. Comparing across countries, the positive relationship between banking depth and economic growth seems to hold at least until the bank-to-GDP ratio gets to about 100 per cent or so. But beyond that, there is little evidence of a growth-enhancing function for having a large banking system per se. And rapid growth in a banking system’s credit to the domestic economy has been long-known as a risk factor for financial instability.

I would like to propose three perspectives on the macroprudential risks that are presented. First: lending and the problem of credit bubbles; second: funding and what I will refer to as the “financial shield” phenomenon; third: exchange rate policy and its complex interaction with banking internationalisation.

When it was famously observed that large banks, though international in life, are national in death, what was meant was that it will be national fiscal authorities and the local customers that are the victims in case of a failing international bank.

But a bank in an international banking centre may not have the same experience. It may not lend much to the domestic economy, it may not source its funding from the domestic economy, it may do all of its business in foreign exchange. In short, it may be a bank whose main connection to the economy is the processing and booking of exported banking services. All-in-all, chances are that the fiscal authorities in the host country will not be interested in providing bail-out to the uninsured creditors of such a bank if it were to fail. In death, as in life, it will remain largely offshore. This is the situation that characterises classic offshore banking.

Nevertheless, much of today’s international banking involves deeper interactions between the banks and the national economy, and as such entails macroprudential risks.


Take lending. This is the most familiar source of leakage of international finance into the domestic economy. With such easy access to foreign funding, internationally connected banks have time and again pumped-up the domestic property market or propped-up the finances of an overspending fiscal authority, creating extensive pockets of domestic over-indebtedness whose deleveraging destabilises economic activity. The result is often the insolvency not only of those who borrowed, but of others whose business model is undermined by the subsequent recession induced by the need for deleveraging. This pattern is familiar and it is now uncontroversial to declare that preventing this pattern emerging calls for macroprudential policy measures to stifle demonstrably excessively credit growth.


On the funding side, what I have to say may not be quite as standard. What we can notice from many highly-banked economies is the fact that the banks cannot be readily partitioned into those catering to the domestic market and those operating an exclusively export service business. This happens when local banks start to tap foreign funding to meet a local demand for funds, or to get into international lending themselves. By taking foreign funding onto a balance sheet already strongly funded by the transactions balances and other core liquid assets of the economy, a domestically significant bank implicitly increases the risk for the domestic fund providers. But the foreign fund providers will feel protected by being pooled with domestic depositors. The domestic fund providers can thus be thought of as a sort of domestic financial shield potentially taken hostage by the foreign funders. After all, if the bank fails and if the national fiscal authorities then decide that the domestic funders should be bailed-out, the national authorities will find it difficult to discriminate against the foreign funders. They may therefore prefer to bail them out also in order to avoid damaging the residents. Bear in mind that it is not just the transactions balances of domestic households and firms that may be at stake, but also their payments services. (Banking systems are what give payment finality to domestic economic transactions).

Since the banks and their foreign funders are aware of this shielding property, it serves to lower the bank’s cost of funding overall. In this way, the indirect access of borrowers to foreign funds is increased by having such funds routed through the domestic banking system.

But beyond a certain point – easily reached with the size of modern banking systems – the losses of a failing bank will exceed the fiscal capacity of the host economy. The controversial re-emergence of capital controls, including exchange controls on inward payments is one of the macroprudential tools that have recently been advocated by some as a potential solution to this problem.

Exchange rate policy

Exchange rate policy also comes into play in a variety of ways (though I will not spend too much time on that topic here). Most international banking works in just a handful of major currencies, with the result that any country with a large banking system is either using one of these currencies itself or has to face the challenges of managing exchange rate risk on those transactions in which the international banking system interacts with the domestic economy.

Low interest rates in the international currency tempt borrowers into assuming foreign exchange risks they may have no capacity to manage or absorb. If the authorities choose to peg their currency in order to limit this risk, they may simply be altering the risk profile: increasing the probabilities both of no change in the currency (in the normal course of affairs), and of a large discrete change in the exchange rate should the peg have to be abandoned.

2. Country examples

Good examples of these aspects come from several small countries caught up in the present crisis. Let’s consider Cyprus, Iceland and Ireland. By the mid-2000s their banking systems had grown to a multiple of GDP (Figure 1); each was tested by the current or previous crises.


The Iceland case is one where public sector borrowing was not a major driver of the crisis. Instead the banks, while based in Iceland, suddenly (in the period 2003-7) became heavily involved in financing non-Icelandic ventures of Icelandic-controlled entities, and other foreign borrowers, as well as fuelling a domestic property boom. Claims on corporate borrowers jumped to over 300 per cent of GDP by 2007, while claims on households also jumped, but to the more moderate level of just over 100 per cent of GDP (Annual growth rate of corporate lending reached the startling figure of over 60 per cent at one stage, whereas for household debt it was only a bit less dramatic, peaking at about 30 per cent – IMF Staff report August 2008). The growth in the activity of the Icelandic banks was extraordinary: little more than 100 per cent of GDP in 2004, the consolidated total assets of the three main banks reached about 10 times that by end-2007.

The Icelandic króna has not been a very stable currency over the years. The Iceland banks funded their expansion substantially in foreign currency and much of their domestic lending was indexed to FX or to inflation.

The banks sourced their funding in different ways as the boom progressed and as different sources became reluctant to increase their Icelandic exposure. Because of this segmentation – bonds, syndicated loans, deposits mobilised in retail markets in the UK and Netherlands – the bulk of the funding came in ways which were more easily separable from the domestic depositor than in the other countries examined here.

And indeed, separated they were. Icelandic deposits (and assets) were carved out of the failing banks and moved to new banks, leaving the bulk of the banking system to be liquidated (Figure 2). Because some of the deposits were placed in branches of one of the Icelandic banks (not a subsidiary) the Icelandic deposit guarantee fund would have been liable to pay, but was unable to do so. Following litigated disputes, the EFTA court adjudicated that, because of the extraordinary circumstances, the Government of Iceland did not have to meet this potentially costly charge. Creditor expectations, that non-discrimination clauses in the EEA agreement would be effective in protecting the foreign depositors, were also dashed in this 2013 court adjudication.

While the Icelandic customers of the banks were far from insulated from the collapse, suffering as they did from protracted exchange controls and currency depreciation (to speak only of the problems of depositors), they did much better than the foreign creditors. After the claims of the UK and Netherlands deposit guarantee system are met from the liquidation, it seems there may be relatively little left for other creditors.

* * *

What the Iceland case exemplifies is a situation where the international and domestic funders were sufficiently separable to disable the “financial shield” mechanism (of which I have been speaking) and prevent it from protecting the foreign funders from bail-in at the expense of the domestic economy. This separation somewhat moderated the fiscal and domestic macro-financial impact of the collapse.


That Ireland’s banking system had total assets approaching nine times GDP in 2008 is a headline fact which tends to mislead. Four different categories of bank can be identified in this total figure (Figure 3). First – making up about 44 per cent of the total – were the locally controlled banks which serviced the domestic economy. Second, a further 12 per cent, were subsidiaries and branches of foreign controlled concerns dealing with the domestic economy. Third, accounting for a further 19 per cent, were foreign-owned concerns which provided mainly export services (not much connected to the domestic borrowers or funders) and which failed as a result of the subprime crisis. Fourth, the remaining foreign-owned concerns also using Ireland as an export base and which remain solid.

Although the first three categories all got into trouble and had to be rescued, only the first category was implicated in the fiscal cost to the Irish Government, as it did not guarantee the other categories. The shareholders of the second category and foreign fiscal authorities of the parent entities of the third category bailed-out creditors of those entities. The financial shield mechanism insulated only those foreigners who were creditors of the nationally controlled banks. Lesson: the presence of strong foreign parents can reduce or eliminate the financial shield problem for the national authorities.

* * *

The large Irish banking system was structured in such a way as to partially—but only partially—ringfence the international part from the domestic. The part that was not ringfenced was still big enough to cause great damage. The wider destabilisation of the Irish economy, including the boom and bust in property prices which has left a large section of the household sector under stress of over-indebtedness, was created by the first two categories of bank (and especially their ability to source foreign funding), but not by the third or fourth category.


Cyprus (Figure 4) is a case where the foreign depositors were embedded in the domestic system, but when it failed, the shield I have been speaking of did not in the end protect them. Instead in each of the two main banks, creditors suffered equally, whether domestic or foreign. The rate of growth in the scale of the Cyprus banking system was not as rapid as in Iceland or even Ireland, but total deposits had reached 3½ times GDP by the end. As in Ireland, exchange rate movements were not an issue insofar as the euro is the national currency.

For Cyprus, the classification “resident” is sometimes challenged, but taking it at face value we find that loans to residents in Cyprus at 74% of the total, exceeded residents’ share of deposits at 62%. The difference meant that residents borrowed more than €10 billion more than they deposited with the banks – about 60 per cent of GDP. Thus, while the banking system in Cyprus has been considerably exposed to other countries, notably Greece (both restructured Government bonds and problem loans reportedly associated in particular with Laiki Bank), it is neither a situation where foreign funds have been employed in net terms to fund the domestic fiscal authorities, nor to fund a domestic property bubble on the scale that occurred in Ireland. Instead, the scale of foreign funding enabled the main banks to become exposed to Greek official debt and to make loans of doubtful quality to Cypriot, Greek and other borrowers, which resulted in losses that promised to demand recapitalisation on a scale unaffordable by the Cypriot State. While the mechanism adopted for bailing-in the bank creditors (and certainly some of the earlier proposed mechanisms) in order to boost the capital position of the banks can be criticised—and certainly underlines how welcome is the European legislation on bank resolution, now in advanced stage of negotiation—there are also lessons for macroprudential regulation.

* * *

The fact that the most affected banks in Cyprus have also been the main retail and payments banks for the national economy has resulted in broad economic disruption, including the introduction of exchange and payments controls, and haircuts on transactions balances of companies. Here the “domestic financial shield” mechanism has been ineffective in protecting the foreign depositors, but was itself broken in the collapse.

3. Concluding remarks

By avoiding the melange of export banking with the domestic banking sector, some of the macrostability risks of having a large banking sector doing international business can be reduced. Fiscal resources will not be stressed and the domestic economy will not be destabilised by an export banking sector if its funding and lending is largely focused on the rest of the world.

Without such separation, extremely prudent and active management is needed to avoid trouble in systems with large banking sectors. One can even imagine devising macroprudential policies to be put in place to prevent expansion of banks’ balance sheets beyond a certain scale, putting the onus on authorities to determine what is the optimal size and structure of the country’s financial sector – not an enviable task.

If large banking systems with access to foreign finance become involved in providing credit to the local economy they have the capacity to rapidly destabilise it. Furthermore, it will be difficult or impossible for the national fiscal authorities to deal with the failure of such a bank, if it is providing systemically important payment or transactions balance services to the domestic economy.

Of course, these considerations are directed to the situation of a small country with a large banking system. More generally, the interpenetration of banking systems has been seen as an important contributor to limiting uncompetitive practices in terms of price and improving provision of services. The degree of segmentation of international finance along national borders that has been seen in the course of this crisis is surely an over-reaction which should, and will in time, be reversed.