On the shock-absorbing properties of a banking union: Europe compared with the United States.

Belke, Ansgar ; Gros, Daniel

INTRODUCTION AND MOTIVATION

The euro area started as a pure 'monetary union'. It is now in the process of also becoming a 'banking union' (BU). The leaders of the European Union (EU) have argued that even this step is not enough. In September 2012, close to the peak of the euro crisis, a joint report by the four Presidents of the EU (the Presidents of the European Commission, the European Council, the Euro Group and the European Central Bank), entitled 'Genuine Economic and Monetary Union', argued that much more was needed (Belke, 2013, and Begg, 2014). The four Presidents argued in essence that the establishment of a BU should also be seen as a first step towards further integration. According to their report a fiscal union would be the next logical step. Moreover, a fiscal union was held to imply the need for a political union.

There is surprisingly little analytical support, however, for the claim that a BU needs to lead to a fiscal union (Belke, 2013, 2013a). The key argument most often heard is simply the observation that the euro area has only a very limited central budget (at least compared with other monetary unions), and that therefore there are almost no fiscal transfers to smooth asymmetric shocks. By contrast, the United States, which is of a similar size as the euro area, does have a substantial federal fiscal budget. The US experience is thus usually taken as a model of what is needed for a sustainable monetary union.

This study contributes to this debate by illustrating how the 'banking union' of the United States provides a very tangible insurance against local financial shocks without a major involvement of the 'fiscal union', which undoubtedly also exists in the United States.

The trans-Atlantic financial crisis that started in 2007-2008 and led to the Great Recession provides a key episode in assessing the importance of mechanisms to absorb regional shocks. The financial shocks became quickly regional in the euro area after 2009-2010 when the financial systems of some countries almost collapsed, and their sovereigns lost market access (eg, Ireland, Portugal and Greece). It is often overlooked that the origins of the crisis in the United States were also rather concentrated at the regional level. The housing boom was very concentrated in the United States. The increase in housing prices varied enormously from state to state and only a few states (Arizona, Nevada, Florida and California) accounted for most of the sub-prime lending, overbuilding and thus the subsequent economic distress and losses from delinquent mortgages.

However, the United States experienced 'only' a system-wide crisis in 2007-2009. There was no specific crisis involving only those states where the real estate excesses had been most marked (Nevada, Florida and California). The main thrust of this study is that the United States was better equipped to deal with these regional shocks because it is a fully fledged BU.

The euro area officially has a banking union, but most observers would agree that it is incomplete if one starts with the three 'canonical' elements of a BU (IMF, 2013a, b):

i Common supervision. This has been achieved, since the ECB, under the heading of the SSM (Single Supervisory Mechanism), has become the ultimate supervisor for all banks in the euro area, and the direct supervisor of about 130 of the largest banks accounting for about 2/3rd of banking assets.

ii A common mechanism to resolve banks. This has also been achieved with creation of the Single Resolution Mechanism (SRM), which will be able to rely on a common fund (the Single Resolution Fund) after a transition period. The SRM will cover all banks in the euro area (and in those other EU countries wishing to join the SSM).

iii Common deposit insurance. No agreement has been reached on this point. It remains to be seen how important this lacuna will become.

By contrast, the United States has had all three elements in place at least since 1933. (1) The United States thus has been a banking union for 80 years. But one should also not forget that the US monetary union survived almost a century and a half without being a BU.

The central theme of this study is that the consequences of the US BU could be seen during the financial crisis. A simple comparison of the fate of two different members of a large monetary union, after they were hit by a financial crisis, offers a powerful illustration of the importance of an integrated banking system. Ireland and Nevada, in fact, provide an almost ideal test case. These two entities share several important characteristics. For example, they both have similar populations as well as GDP, and they both experienced an exceptionally strong housing boom. But when the boom turned to bust, the US states did not experience any local financial crisis (nor did any state government have to be bailed out).

This contribution finds that the key difference between Nevada and Ireland is that banking problems are taken care of in the United States at the federal level (effectively a banking union), whereas in the euro area, responsibility for banking losses remains national.

The paper is organised as follows. The next section presents some case studies of the stabilisation properties of a banking union. The section after that then analyses the role of 'foreign-owned banks' as a sort of 'private banking union'. The following section analyses the institutions that paid for the shock absorption provided by the official US federal banking-related institutions: the Federal Deposit Insurance Corporation (FDIC) and the Government-Sponsored Enterprises (GSEs). The penultimate section contains some general considerations with respect to a fiscal union and financial shock absorbers. The final section concludes.

THE MACRO-ECONOMIC STABILISATION PROPERTIES OF A BANKING UNION: SOME CASE STUDIES

In this section, we analyse the implications of a banking union for macroeconomic stability by making comparisons between countries/states that have experienced similar local boom/bust cycle in real estate but are part of different federal systems in terms of financial markets. (2)

The comparison pairs will be Ireland-Nevada, Florida-Spain and Latvia-Nevada. The first of these two pairs is part of a large currency area. The Latvia-Nevada comparison is interesting because Latvia was in the euro area during its boom/bust cycle, but its banking system was dominated by banks from Nordic countries. In this sense Latvia was part of a 'Nordic Banking Union'.

The pairing Ireland-Nevada is the one that comes closest to a natural experiment as these two entities are of a very similar size and had a very similar boom and bust in terms of real estate. The key difference is of course that the banks operating in Nevada are so much part of the fully integrated wider US banking system that one cannot really speak of a 'banking system of Nevada'. The analysis will show that this was decisive for the limited impact of the bust for the local economy and local public finances in Nevada and other US states with similar local real estate booms.

Florida can be compared with Spain. Both these entities represent larger, more diversified economies than either those of Nevada or Ireland. Somewhat surprisingly, real estate investments seem to have played a larger role in Spain although it is somewhat larger than Florida.

Another useful comparison is that between Nevada and Latvia or the other Baltic countries. None of the latter were part of the euro area when the crisis struck in 2008/2009. However, they all weathered the crisis more quickly than Ireland or other peripheral euro area countries because they benefited from the fact that their banks were to a large extent owned by larger Nordic banks that were able to absorb the losses that arose when the housing boom collapsed and the Baltic economies experienced a very sharp recession. It is interesting that the only Baltic country that needed a bail-out was Latvia, which was also the only country that still had a significant local bank.

[FIGURE 1 OMITTED]

Before going into these pair-wise comparisons it is useful to consider to what extent the boom/bust cycle is different between the United States and the euro area at the aggregate level.

Regional concentration of real estate cycles within a monetary union

The aggregate data on house prices and construction activity (as percentage of GDP) reveals a considerable similarity; the boom was actually somewhat more pronounced in the United States than in the euro area, at least if one looks at aggregate numbers (Figure 1). House prices increased by more in the United States, then fell by more but also recovered earlier, thus ending up at about the same level relative to the euro area if one looks at the period since the start of monetary union (Figure 1, right-hand side).

An even more important indicator of the potential cost of a real-estate cycle is the amount of construction activity undertaken (Figure 1, left-hand side). A large stock of unsellable houses often constitutes the main reason for losses on mortgages. Here again, one finds that the cycle was somewhat more pronounced in the United States than in the euro area since construction spending fell by about 1.3 percentage points of GDP in the United States but only about 1.1 percentage points of GDP (on aggregate) in the euro area.

[FIGURE 2 OMITTED]

How could one then explain that the United States recovered earlier from the bust of the housing bubble and that there were very serious difficulties at the national level in Europe, even in countries like Ireland or Spain, where public finance had been under control?

It is tempting to argue that this lack of regional problems in the United States was because of a more uniform manifestation of the boom in the United States than in the euro area. Within the euro area the average number hides fundamental differences between the peripheral countries Spain and Ireland, where both house prices and construction activities boomed until 2007, and core countries like Germany where both house prices and construction activity were relatively weak (again until 2007/2008).

However, the boom/bust was also very concentrated in the United States. Figure 2 shows the distribution of the losses sustained by the FDIC during the last crisis in each state. It is apparent that the banking problems were highly concentrated in a few states.

This combination of a similar boom/bust pattern in the aggregate variables and a similar degree of concentration at the regional level already suggests that the structure of the financial system and its back-up mechanism must have played a key role in containing regional problems in the United States.

Ireland versus Nevada

Ireland and Nevada share several important characteristics, as reflected in Table 1. They have similar populations (2.7-4.5 million) and similar levels of GDP (US$120-200 billion). Both federal states experienced a strong recession and a very similar increase in unemployment. However, the fall in GDP was much larger in Ireland than in Nevada. As will be argued below, this was because of the fact that the losses arising from the real estate bust in Nevada were to a large extent absorbed by the US federal financial system.

The most important similarity is, however, that they both experienced an exceptionally strong housing boom and bust. The similarity of the boom-bust cycle is shown in a set of figures: Figure 3, Panel A (nominal) GDP increased by a very similar proportion during the boom and then fell. Figure 3, Panel B, shows the evolution of house prices, which increased up to 2007/2008 and then turned downwards. This was the first fall in house prices during peace time for the United States. Figure 3, Panel C, shows construction activity as a percentage of GDP (for Ireland) and a percentage of GSP (Nevada). It is again apparent that the two series follow the same pattern, but construction activity seems to have been much more important to the economy of Ireland than that of Nevada. However, this difference might be because of a difference in definition of the aggregate 'construction' in the national accounts. Figure 3, Panel D, displays the consequences for the real economy in terms of the unemployment rate, which also follows a similar pattern.

However, there is one fundamental difference between Nevada and Ireland: when the boom turned to bust, Nevada did not experience any local financial crisis and the state government did not have to be bailed out. By contrast, the government of Ireland was for some time unable to issue any new debt on the market and had to be supported by a very large loan financed jointly by the IMF and the European rescue fund (the ESM and its precursor the EFSF).

[FIGURE 3 OMITTED]

The key difference between Nevada and Ireland is that banking problems in the United States are taken care of at the federal level (the United States is a BU), whereas in the euro area, responsibility for banking losses was national, and will remain partially national until the SRF is fully operational.

Local banks in Nevada experienced huge losses (just like in Ireland) and many of them became insolvent, but this did not lead to any disruption of the local banking system as these banks were seized by the FDIC, which covered the losses and transferred the operations to other, stronger banks. In 2008/2009, the FDIC thus closed 11 banks headquartered in the state, with assets of over $40 billion, or about 30% of state GDP. The losses for the FDIC in these rescue/restructuring operations amounted to about $4 billion. (3)

Other losses were borne at the federal level when residents of Nevada defaulted in large numbers on their home mortgages. The two federal institutions that re-finance mortgages have lost between them about $8 billion in the state since 2008. (4)

The federal institutions of the US BU thus provided Nevada with a 'shock' absorber of about 8%-9% of GDP, not in the form of loans, but in the form of an (ex-post) transfer because losses of this magnitude were borne at the federal level. Against this transfer one would of course have to set the insurance premia paid by banks in Nevada before the bust. But they are likely to have been of a smaller order of magnitude.

Of course, a lot of the banking business in Nevada was, and still is, conducted by 'foreign' banks, that is, by out-of-state banks, which just took the losses from their Nevada operations on their books and could set them against profits made elsewhere. (5) This is another way in which an integrated banking market can provide insurance against local financial shocks. One might call this a 'private' BU or a truly integrated banking market. It is impossible to estimate the size of this additional shock absorber, but the losses absorbed by out-of-state banks might very well have been at least as large again as the ones borne by the federal institutions. The total write-downs of the large US banks that operate across the entire US were about $440 billion, twice as much as the $220 billion of losses of the three official institutions (FDIC, Fannie and Freddie). If these losses were distributed in a similar way to the losses of the official institutions mentioned so far, one can conclude that the shock absorption capacity of the large union-wide banks is likely to have been worth about 17% of GDP.

Nevada was also one of the states where 'non-conforming' or 'sub-prime' mortgages became particularly widespread. Non-conforming loans are not eligible for insurance and securitisation by the GSEs, but they were widely packaged into Private Real Estate Mortgage Securities, which then were sold to investors worldwide. Some of these sub-prime securities remained on the balance sheets of the large US banks mentioned above. But a large part was bought by other US and foreign investors. These investors thus absorbed another part of the losses generated locally. The scale of this additional risk-sharing is very difficult to estimate precisely. But given that sub-prime issuance was also particularly widespread in Nevada it is likely that this risk-sharing was also substantial.

All in all one can thus conclude that the overall loss absorption provided by the public institutions (FDIC and the two GSEs) and the private sector (large banks, sub-prime securitisation) must have been substantially larger than the 25% of GDP coming through the FDIC and the GSEs plus the banks (8.5 + 17 = 25).

In Europe there was no official risk-sharing in the sense that the Irish government had to take the responsibility for saving the banks in Ireland. The ESM did provide financing for the Irish government when it lost market access. But the ESM could provide only loans, which have to be repaid with interest. Moreover, as an implicit counterpart to this support the Irish government was asked not to bail in investors holding the bonds of Irish banks.

One consequence of this lack of risk-sharing was that public debt soared in Ireland. As shown in the Figure 4, just before the start of the crisis Ireland had a very low debt/GDP ratio of around 25% of GDP, which was actually very similar to that of Nevada. The debt ratio of Nevada did not increase much, even though the housing cycle was very similar, as illustrated above. Today the debt ratio of Ireland is above 120% of GDP; six times that of Nevada.

In Europe, this 'private' BU channel of risk mutualisation operates only in some cases. It is of paramount importance only for the smaller Baltic EU countries, whose banks are to a large extent in foreign hands. Estonia, Lithuania (and to a lesser extent Latvia) thus benefited from a similar protection against losses provided by the Scandinavian headquarters of their local banks. By contrast, most of the real estate lending in Ireland (and Spain) had mostly been extended by local banks so that most of the losses remained local (without any federal institution to provide insurance). (6)

[FIGURE 4 OMITTED]

The comparison between Nevada and Ireland thus clearly illustrates the shock-absorbing capacity of an integrated banking system and a BU. For Nevada, the BU resulted in a transfer worth over 25%, possibly up to 30% of its income. Nevada is admittedly an extreme example of the housing boom and bust. Nevertheless, this example illustrates the general point that a BU can provide more shock-absorbing capacity than could ever be provided by any 'fiscal capacity' that is currently being contemplated for the euro area.

As a robustness check, we now assess other factors which could have played a role in conditioning the responses of both economies to the eruption of the crisis--the flexibility of labour markets, the fiscal system and the response of monetary policies.

Labour market flexibility--Ireland versus Nevada

While Ireland has a flexible labour market, it may probably be much less flexible than that in Nevada. This suggests that a part of the very adverse economic performance in Ireland after the crisis might have been because of a less flexible labour market than that in Nevada and not to the absence of a banking union.

[FIGURE 5 OMITTED]

Similarly, while Irish workers have a long history of responding to economy difficulties by emigrating to other English-speaking countries, migration between Nevada and other US states may be greater. Thus, one reason that the highly adverse consequences of the collapse of the housing market in Ireland might be because of differences in labour mobility rather than to the absence of a BU. Also on a more general level, it is widely assumed that the US labour market is more flexible than those of euro area countries and that labour mobility is higher in the US than in Europe. However, both presumptions do not seem to apply if one compares Nevada to Ireland as we will show in the following.

Flexibility of the labour market could be best be measured by wage flexibility, or rather how wages in a state react to an asymmetric shock. Since the shock experienced by Nevada was roughly comparable in terms of the increase in unemployment (as shown above) one could take the adjustment of wages, relative to the average of respective union, as an indicator of flexibility. Figure 5 shows the ratio of wages in Nevada to the US average and wages in Ireland relative to the euro area average. It is apparent that there was more movement in Irish wages than those in Nevada. More flexibility in Ireland can be observed during the boom in the sense that Irish wages increased much more than those in Nevada. But even during the bust, Irish wages have declined (relative to the euro area average) more than those in Nevada. Irish wages declined relative to the euro area average by about 15% if one compares 2014 with the peak in 2009. By contrast, wages in Nevada declined by only 8% from their respective peak. Wages in Ireland were thus more flexible than those in Nevada.

[FIGURE 6 OMITTED]

This suggests that it is difficult to ascribe the adverse economic performance in Ireland after the crisis to a less flexible labour market than that of Nevada.

What about a lack of labour mobility? Similarly, one might suspect that migration between Nevada and other US states is likely to be much greater than that to and from Ireland. However, here again, the raw numbers do not confirm this presumption. Figure 6 shows the rate of net immigration as percentage of the resident population for both Ireland and Nevada. Strikingly, the rate of immigration is almost identical and the decline during the crisis is also very similar, resulting in a net emigration of 0.7% per annum 5 years later. The drop in the net migration rate was about 3 percentage points for Ireland from plus 2.3 to minus 0.7% whereas the drop in Nevada (for the time period for which data is available) was smaller, about 2 percentage points from plus 1.7 to minus 0.3%.

These two observations suggest that the Irish labour market was actually more flexible than that of Nevada, and the important difference was the absence of a BU, rather than lower labour mobility or less flexibility in the labour market and wage setting.

Fiscal federalism--Ireland versus Nevada

Like many countries with property booms, rapid growth of tax revenue during the boom from sales tax payments from the construction sector, not to mention wage taxes from the booming economy, led the US government to reduce taxes during the boom, making the economy even more susceptible to the fall in property prices and aggravating the downturn. One might suspect that local governments in US states with property booms did not cut taxes, leading tax revenue to fall by less than in Ireland, leading to smaller fiscal problems. One might therefore wonder if some of what is attributed to the lack of the BU in Europe might be because of differences in tax systems.

It is true that the federal deficit increased by several percentage points of GDP as a reaction to the deep recession of 2008/2009, which implies that on average in most states residents have received more federal fiscal expenditure than they (or rather their residents) have paid in federal taxes. Yet, it is striking that the residents of those states hardest hit by the real estate boom/bust cycle like Arizona or Nevada, which thus recorded the highest increase in unemployment and large falls in GSP, did not receive more net federal transfers than others (Gros, 2016). This empirical pattern reinforces our conclusion that, all in all, it is difficult to trace back what we attribute to the lack of the banking union to differences in tax systems.

If federal transfers (as defined here) are to be shock absorbing one would need to see a strong correlation between the first differences, that is, changes in unemployment rates by state and the change in the federal transfers received by residents of the states over the same period. However, this is not the case. The correlation between the two is rather low if one takes the post crisis data, that is, the change between 2007 and 2010.

Monetary policies--Ireland versus Nevada

Finally, the Federal Reserve has a 'dual' mandate and took quick and effective action to support the real economy after the crisis stuck. By contrast, the ECB has price stability as its overriding policy objective and was therefore not in a position to respond as rapidly to the downturn in economic activity. This may suggest that the poor outcomes in parts of Europe (eg, in Ireland) might at least partially be because of different monetary policy responses and not to the lack of a BU.

Of course, the lack of a banking union was not the only reason why some parts of the euro area experienced poor outcomes. One other reason might have been the bank centric nature of the financial system in Europe in general (Pagano, 2014). Given that bank loans constitute the main source of financing for investment throughout most of the euro area, the losses from the crisis rendered many banks under-capitalized and some insolvent. The national government had in most cases to finance the recapitalisation of 'its' banks, thus establishing a link between the solvency of the sovereign and that of its banks. De Groen and Gros (2015) show that this could have been avoided to a large extent if the BU with its Single Resolution Fund had already existed at the time the Great Financial Crisis broke out.

Moreover, one could argue that the different mandates of the central banks might also have had an influence. The Federal Reserve had a dual mandate that includes stabilizing the economy whereas the mandate of the ECB is only price stability. This difference does not appear strongly in the initial reaction of the ECB, that is, in 2007/2008, when the ECB did roll out a series of, until then, highly unconventional policy tools to keep the market supplied with liquidity. However, the difference in mandate might explain why later, that is, after the 'euro crisis' had broken in 2009/2010, the ECB focussed only on price stability, and, at one point, even increased its policy rate temporarily, whereas the Federal Reserve continued its course to support the nascent recovery in the United States.

These other factors will have contributed to underperformance of the euro area on average, but the case study of Ireland and Nevada suggests that the much higher intra-area diversity in economic performance was mainly attributable to the lack of a BU.

Seen on the whole, thus, we argue that Ireland and Nevada, the economies studied, do in fact differ in terms of whether they belong to a banking union, but less so in terms of their fiscal system, the flexibility of labour markets, whether labour is de facto internationally mobile, and the response of monetary policy. These factors seem to have played a, if at all, minor role in conditioning the responses of these economies to the crisis.

Florida: another example of the US BU in action

Florida and Spain constitute another pair of countries that can be used to illustrate the difference in the impact of a local real estate boom/bust cycle when there is a fully fledged banking union. Both Florida and Spain are much larger and more diversified economies than Ireland or Nevada and their housing cycles were less extreme. Table 2 provides some of the basic data, showing that Florida is about half the size of Spain, both in terms of population and GDP and that initially the impact on GDP was very similar. However, the local labour markets reacted in a very different way.

Also Figure 7 provides an indication of the similarity in the cycle. Nominal GDP increased a bit more in Spain, but the downturn was also longer so that over the entire cycle the increase in GDP was almost exactly the same. A similar pattern can be seen for house prices, which turned earlier in Florida, but then also started recently to recover whereas Spanish house prices still continue to fall. Investment in construction follows exactly the same pattern, but has always been higher in Spain. It is in unemployment that one sees a decisive divergence. Unemployment rose initially in a similar way, but has continued to increase in Spain and has already declined substantially in Florida.

The key difference one has to explain is that the state government of Florida was barely affected by the crisis whereas the government of Spain had to pay a substantial risk premium for issuing new debt. Moreover, when the full scale of the banking problems became apparent, Spain received a loan of 60 billion euro (about 6% of its GDP) to help finance the recapitalisation of its problem banks (mainly the cajas, which had engaged in most of the real estate lending that caused most of the losses).

By contrast, in Florida one can see again the US BU in action. During the period 2008-2012 the FDIC closed over 70 banks headquartered in Florida, with total losses for the FDIC of roughly $14 billion, or 2% of Florida's GSP. Moreover, mortgages originating from Florida and covered by Fannie Mae and Freddie Mac experienced high default rates, leading to losses of the two GSEs of $19 billion since 2008. Federal loss-sharing on mortgages originating in Florida, but insured by federal 'government sponsored entities' thus amounted to another 2.3% of Florida GDP. Total direct loss absorption through the official BU amounted to about 33 billion or 4.3% of GDP.

As argued above, one has to consider the fact that in Florida, as in Nevada, the large US banks operating nationwide have a very large share. Under the maintained assumption that the losses at the large US banks operating nationwide were about twice as much as FDIC + GSEs, it follows that private sector losses borne out of state might be twice as large as those assumed by the FDIC and GSEs; or probably another 8-9% of GDP.

The total loss absorption [ex post) of the private and public pillar of the US banking union for Florida was thus probably more than 12% of GSP. By comparison, Spain did receive a loan from the ESM, worth about 6% of its GDP to help finance the recapitalisation of Spanish mortgage banks (cajas). But this was a loan and has to be repaid with interest.

As in Nevada, another form of loss absorption came through private sector securitisation. In the United States the most risky part of the mortgages, which accounted for about 20% of all originations in Florida, were securitised and sold to capital market investors not only in the United States, but also internationally. Large US banks retained only part of the remaining risk. A further part of the local risk from sub-prime mortgages was thus borne by 'out of state' investors, protecting the economy of Florida, which could rebound earlier as its debt burden was much lighter.

[FIGURE 7 OMITTED]

Nevada versus Latvia

All of the Baltic states experienced strong growth rates in GDP and house prices and double-digit current account deficits until about 2007. This boom turned into a bust very quickly when global financial conditions turned around in 2007/2008. The adjustment was then very sharp, with GDP falling by double-digit percentages as investment in construction virtually came to a standstill and credit being scaled back (Figure 8). None of the Baltic countries was in the euro area when this occurred. This meant that their local banks could not access the various facilities of the ECB and the national central banks had to be rather restrictive, given that they wanted to defend their exchange rate against the euro. Only one of the Baltic countries, Latvia, needed international financial assistance, mainly to deal with the aftermath of the problems at its only large domestic bank.

Although Latvia was not then (2007/2008) in the euro area it still makes sense to compare it to Nevada because the adjustment patterns were similar because of the strong presence of foreign banks. House prices, available only since 2006, fell strongly when the crisis hit in 2008, but recovered already a few years later as in Nevada (Figure 8b). Unemployment first rose even more than in Nevada, but also started to improve after a few years, mimicking the pattern of Nevada (Figure 8d). In terms of construction activity, the upturn had been shorter and sharper in Latvia, but here also the recovery set in quickly in contrast to Nevada where the longer period of elevated construction activity probably led to a more significant housing overhang (Figure 8c).

The one reason why this relatively early recovery was possible in Latvia (despite the fall in GDP of over 25%) was that the banking system of the country was owned to over 60 % by foreign banks. These banks thus absorbed most of the losses that arose when the Latvian housing market turned in 2007/2008.

It is naturally very difficult to pinpoint the origin of losses occurring within large internationally active banks. The available anecdotal evidence suggests that Swedish banks alone made loan losses in the Baltics of about 12-20 billion USD between 2009 and 2012, which would be several times larger than the capital invested in the local subsidiaries and would amount to between 15 and over 20% of the GDP of the three Baltic states together. (7)

Given that other Scandinavian banks also had a significant part of the market in the Baltic states (about one third, on average) it is thus likely that the total loss absorption by foreign banks in the Baltic states was closer to 30% of their GDP.

[FIGURE 8 OMITTED]

The Baltic states thus benefited enormously from the fact that their banking systems consisted essentially of subsidiaries of foreign banks. As loan losses were in many cases larger than the capital invested in these subsidiaries the foreign (mostly Swedish) banks could have walked away from their daughter companies, which would have forced the Baltic governments to sustain them during the crisis. However, the Swedish and other Nordic banks chose to put additional capital into their Baltic subsidiaries because they were counting on the long-term growth potential of the region. (8)

The broad conclusion that emerges is that one of the reasons why Latvia (as the other Baltic states) weathered the crisis more quickly than Ireland or Spain is that it benefited from the fact that its banks were to a large extent owned by larger Nordic banks, which were able to absorb the losses that arose when the housing boom collapsed and the Baltic economies experienced a very sharp recession. It is interesting to note that the only Baltic country that needed a bailout was Latvia, which was also the only country that had a significant local bank.

FOREIGN OWNED BANKS: A SUBSTITUTE FOR BU? THE EXPERIENCE IN THE EU

The case studies presented in the previous section suggest that in the United States the large banks that operate throughout the entire territory provided a very important channel through which local shocks could be better absorbed. The estimates provided above suggest that the shock-absorbing contribution from internationally active banks could have been twice as significant as the one provided by the official 'Banking Union' institutions (the FDIC and the GSEs). This is generally not the case in Europe. Somewhat surprisingly, however, transnationally operating banks have played a more important role outside the euro area than within.

For example, Spain did not have protection from a BU as there was little activity of foreign owned banks in Spain. Moreover, most of the real estate-related lending that later caused most of the losses was done by the local cajas that financed their loan books not with local savings, but by attracting large inflows of foreign capital, mostly in the form of covered bonds or inter-bank loans, neither of which is loss absorbing.

In the case of Ireland some loss absorption occurred because the large UK banks had a substantial exposure to Ireland and thus also absorbed some losses that occurred ther, but the magnitudes are difficult to ascertain.

However, there has been little cross-border integration of the banking sector within the euro area; much more has occurred within the EU with large banks from the old member states taking over most of the banking systems in the new member states. This trend was particularly strong in the small Baltic states where foreign banks had a market share of 80%-90% and absorbed most losses that occurred when the credit and real estate boom in the region ended abruptly in 2008/2009, as documented above (Buch et al, 2013, p. 9). The only exception to the dominance of foreign banks occurred in Latvia, where one significant local bank remained, but its problems almost pushed the government into insolvency.

The experience of the Baltic States shows that integration via equity markets (ownership) can mimic the shock absorbing properties of a BU. Foreign owned banks can absorb losses. However, this mechanism works only if the (until now national) supervisor allows them to maintain exposure. This willingness of the Swedish and other Nordic supervisors to allow their banks to maintain their exposure in the Baltic states and to recapitalise their subsidiaries there was one crucial element that stabilised the financial sector in the region.

Another condition for loss absorption by 'foreign banks' to be stabilising is that the foreign-owned banks must be strong enough to carry substantial losses. This condition was fulfilled because the Swedish and other banks that had large exposures in the Baltic states were able to absorb substantial losses because business in their home base remained solid and given that their home economies were running large current account surpluses, which effectively insulated them from the flight of cross-border capital that started in 2010/2011 when the broader financial crisis became the euro crisis.

The European experience has also shown that a strong presence of foreign banks can lead to a propagation of financial shocks abroad to the domestic economy. This happened during the first stage of the financial crisis when the large banks from the 'older' Member States came under funding stress and started to pull back capital and credit lines from their subsidiaries in Central and Eastern Europe. These banks came from countries like Italy, Austria or Belgium whose fiscal and balance of payments position was less strong than those of the Scandinavian countries whose banks dominated the Baltic banking market. This pullback by the foreign parents contributed to the economic downturn throughout Central and Eastern Europe and threatened to initiate a self-reinforcing spiral of a withdrawal of financial support from the foreign parents, a deeper recession and therefore more local losses, prompting the foreign parents to accelerate their withdrawal. Moreover, each individual parent bank initially acted in isolation, hoping that the economic impact of its withdrawal would be limited since other banks could, at least in principle, take its place in financing the local economy.

It took an international initiative, coordinated by the international financial institutions, to bring the handful of key parent banks from Western Europe together. Under this so-called Vienna Initiative the banks promised not to reduce their exposure to Central and Eastern European countries and the IMF agreed to provide the countries with substantial balance of payments support.

This combination was sufficient to arrest the vicious circle described above since it helped to stabilise the economies in the region, which in turn limited the losses for the parent banks, thus providing also a justification for the banks to continue to provide financing in the region.

This episode illustrates the general economic principle that 'there is no free lunch'. Large cross-border or cross regional banks can mitigate the local impact of local financial shocks, but they also propagate shocks to the overall financial system to all regions in which they play an important role.

To come back to the US example, one could thus argue that the presence of the large US banks throughout the US provided a shock absorbing mechanism for Nevada or Florida, but also a shock propagating mechanism for the States in the North where there had been no real estate boom. The financial crisis, which started when the sub-prime boom burst, led to a tightening of credit availability all throughout the United States, although the boom had been rather concentrated in a few states as documented above.

WHO PAYS FOR THE SHOCK ABSORBERS? FISCAL UNION AND FINANCIAL SHOCK ABSORBERS

One key issue for any shock-absorber mechanism is whether the mechanism is self-financing or needs public funding. This issue has played a key role in the political debate, both in the United States and in the EU. In the United States a key political slogan has been that 'the industry' should pay for its own mistake, and that 'tax payer's money' should not be used to bail out banks.

The aim of protecting the tax payer has also been a key consideration for the construction of the Single Resolution Mechanism and its Single Resolution Fund, which will be financed by contributions from the industry. The size of the SRF could be kept relatively small because another piece of EU legislation, namely, the BRRD (Bank Resolution and Recovery Directive), established tough rules on the 'bail in' of creditors before a bank can receive financial support from the SRF. Ex ante there is thus a clear intention to make the key pillar of the banking union in the euro zone self-financing, obviating the need for financial support from the budgets of Member States. An additional key element of the political economy of the design of the BU in the euro area was to avoid the prospective of transfers across member states. This reinforces the importance of the question whether a banking union can be constructed in such a way that it does not need fiscal support if one averages over several financial cycles. It is of course too early to say whether the European BU will be self-financing.

In the United States there are two official shock-absorber mechanisms, FDIC and securitisation by the GSEs, which have been operating for long enough to measure whether, ex post, over several financial cycles, the system did finance itself, that is, whether the costs that had to be sustained during crisis times were on average borne by the industry (the contributions during the non-crisis times).

The best known element of the US BU is the Federal Deposit Insurance Corporation with its associated fund to finance bank rescues and make depositors whole. Historically, the losses of the FDIC have come in two waves: the savings and loans crisis of the 1990s and the 'subprime' crisis of the last decade. The losses the FDIC had to sustain after 2008 were larger than the fund it had accumulated during the previous boom years. At the start of the crisis the FDIC had slightly over 50 billion USD at its disposal, equivalent to about 1.2% of insured deposits. However, already about 1 year into the crisis the available funding ran out. The FDIC fund became negative by about 21 billion USD in 2010.

The FDIC had thus to be supported by a large line of credit from the Treasury. But its fund was replenished quickly because the FDIC was able for force banks to pre-pay assessments several years in advance, bringing the fund quickly back into the black. However, it will take another decade or so before the FDIC will again reach its target level of 1.25% of insured deposits. (9) The FDIC's funding will thus be reconstituted by contributions from industry. This part of the US BU thus needed liquidity support from the Federal institutions during the biggest financial crisis in living memory. But in the end, the FDIC remained solvent.

A second key element of the US BU is the securitisation of 'conforming' mortgages by Fannie May and Freddy Mac. During the early years of the crisis, the GSEs made large losses, especially in some states as documented above. These losses were larger than the capital they had. The government thus had to step in and refinance them. However, the losses incurred during 2008-2012 are now in the process of being made up as the insurance premium has increased and delinquency rates are falling. This part of the US BU is thus likely to become self-financing very soon as well.

Both elements of the US BU are thus in the end self-financing. The insurance against regional shocks they provided will have been financed entirely by banks, depositors and mortgage holders.

FISCAL UNION AND FINANCIAL SHOCK ABSORBERS

A key finding of this study is that a common backup system for banks combined with the overall integration of national financial systems greatly increases the ability of financial markets to reduce the negative spillovers among members of a monetary union resulting from national or regional financial crisis. This finding is confirmed in more macroeconomic terms by the empirical literature that measures the channels of stabilisation of regional income in existing monetary unions like the United States and Germany. For example, Asdrubali et al (1996) find that, in the United States, around 40% of shocks to per capita gross state product is smoothed by capital markets and around 25% by credit markets. This implies that about two thirds of shocks to state income are absorbed by financial markets. Similarly, Athanasoulis and van Wincoop (2001) find that around 70% of the shocks in the United States are smoothed through private and public risk-sharing mechanisms: financial markets play the biggest role, allowing around 60% of the total smoothing, while the Federal fiscal policy covers the other 10%. More recently, Hepp and von Hagen (2013) find that for Germany, in the pre-unification period, most of the smoothing was provided by the Federal tax-transfer and grant system (55%), while for the post-unification period, factor income flows have become the most important channel, contributing about 51 % of total income smoothing.

The introduction of the euro and the associated common payments infrastructure (the TARGET system) reduced barriers to financial integration and credit flows via the wholesale interbank market and boosted financial integration as measured by the size of cross-border flows and stocks. This was expected to facilitate risk-sharing among investors. However, the euro crisis has shown that sometimes larger cross-border financial flows and stock can actually be at the origin of a crisis (Mink and de Haan, 2014). The main reason for this is that banks have been the primary financial intermediaries in the European Union and in the euro area. Most euro area member countries' financial systems are heavily bank-centred and stock and bond markets provide a relatively modest share of the financing to the private sector in most countries. Total bank assets account for 283% of GDP in the EU, compared with about 65% of GDP in the United States (Furceri and Zdzienicka, 2013).

There is no robust evidence in the literature that a financial system dominated by banks rather than the market and by debt instead of equity has increased the capacity of the economy for risk-sharing. Kalemli-Ozcan et al. (2010) and Demyank et al (2007) find evidence that increased cross-banking integration has fostered ex-post the optimality of the currency union by improving cross-country risk sharing. By contrast, Furceri and Zdzienicka (2013) finds that 'the decrease in private credit smoothing after the creation of the EMU reflects the fact that credit flows have become less counter-cyclical'. By contrast, as is the main focus of this paper, the United States shows a model how a high degree of banking integration can absorb shocks. The main reason for this that in the US banking integration has not taken the form of cross-border credit, but de facto cross-border equity as a few large banks are operating nation-wide.

CONCLUDING REMARKS

The existing BU in the United States has been very successful in managing the local real estate booms and busts that the United States has experienced. A careful examination of the cases of Nevada and Florida compared with Ireland and Spain, respectively showed that these financial shock absorbers have a higher shock-absorbing capacity than could ever be provided by any 'fiscal capacity' for the euro area. The macro-economic literature confirms this in the sense that it finds that, in the United States, the shock absorption provided by financial markets is much larger than that provided by the fiscal system (see, for instance, Begg, 2014).

There are several channels through which regional financial shocks are absorbed at the Federal level in the United States. The FDIC is the most visible one, but the system of securitisation of mortgages, especially the so-called GSEs contributes as well. Moreover, the large banks, which operate nation-wide, dominate the banking sector. They are able to absorb local losses in their overall results. By contrast, in Europe large banks operating in different member countries are still perceived as foreign banks outside their home country. Integration via international groups has so far been limited in the euro area but has been very important for the new Member States. The prevalent form of financial market integration across borders within the euro area is debt, which does not act as a shock absorber in the case of systemic shocks. By contrast there has been much more cross-border equity outside the euro area through large-scale foreign ownership of banks in Central and Eastern Europe.

If the really important and costly shocks are national financial boom-bust cycles, followed by a financial crisis, the question arises: What arrangement provides the best protection against these shocks? The United States experience seems to provide a clear answer: the shock-absorbing power of explicit federal transfers is rather small, but the US BU provides important support in the case of large shocks to the local financial system. This has one simple implication: To insure its stability, the euro area needs a strong banking union, but not a fiscal union. (10) The usual argument that the former needs to be followed by the latter should thus be turned on its head: an area with a well-functioning banking union has much less need for fiscal shock absorbers and does not need a fiscal union (see also Belke, 2013). From the latter, it follows that there is also no need for a political union.

In a BU, excessive spending by individual member states might lead to difficulties for the state concerned, but it should no longer destabilise the entire system. This implies that political responsibility for fiscal policy can remain at the national level. Technically speaking one can thus argue that a BU reduces the negative external effects of excessive deficits and debts. The BU thus represents a key element to make the original Maastricht view with its 'no bail out clause' viable in reality.

Two elements of the US BU that do not exist, at least not yet, in the euro area are widespread securitisation and the existence of large banks that operate throughout the entire area. These two characteristics of the US financial system allow it to absorb regional shocks. But these two characteristics also have their own drawbacks. Large banks are often more prone to generate systemic risk, and it has been shown (ASC, 2014) that most of the growth in the banking sector over the last decade has come from the largest banks. The drawbacks of widespread securitisation also became apparent during the 'sub-prime' crisis when it was shown that the originating banks were subject to serious conflicts of interests as they earned fees from originating mortgages irrespective of the quality of borrower and his/her ability to service the loans. A system that deals more easily with regional crises might thus have other drawbacks. The challenge for Europe will be to build a system that breaks the 'diabolical' feed-back loop between weak banks and their sovereign but one that is not dominated by a handful of very large banks which are then not only too large to fail, but also too large to be saved. (11)

Acknowledgements

The authors are grateful to Matthias Busse for valuable research assistance and to an anonymous referee for valuable comments. They also gratefully acknowledge comments received by participants at the International Finance and Banking (FIBA) Conference, 26-27 March 2015, Bucharest, Romania, where this paper was presented as a keynote lecture.

REFERENCES

Advisory Scientific Committee (ASC). 2014: Is Europe Overbanked?, Reports of the Advisory Scientific Committee of the ESRB, No. 4/June 2014, https://www.esrb.europa.eu/pub/pdf/asc/Reports _ASC_4_1406.pdf.

Asdrubali, P, Sorensen, BE and Yosha, 0. 1996: Channels of interstate risk sharing: United States 1963-1990. The Quarterly Journal of Economics. MIT Press 111(4): 1081-1110, November.

Athanasoulis, S and van Wincoop, E. 2001: Risk sharing within the United States: What do financial markets and fiscal federalism accomplish? Review of Economics and Statistics 83(4): 688-698, November.

Begg, I. 2014: Genuine economic and monetary union. In: Durlauf, SN and Blume, LE (eds). The new palgrave dictionary of economics, Online Edition.

Belke, A. 2013: Towards a genuine economic and monetary union--comments on a roadmap. Politics and Governance 1(1): 48-65.

Belke, A. 2013a: Debt Mutualisation in the Ongoing Eurozone Crisis--A Tale of the 'North' and the 'South', The New Palgrave Dictionary of Economics, Online Edition. In: Steven ND and Lawrence EB (eds). Palgrave Macmillan: Houndmills, Basingstoke.

Belke, A, Dobrzanska, A, Gros, D and Smaga, P. (forthcoming): (When) should a non-euro country join the banking union? Journal of Economic Asymmetries, in press.

Buch, CM, Korner, T and Weigert, B. 2013: Towards Deeper Financial Integration in Europe: What the Banking Union Can Contribute, Working Paper 02/2013, German Council of Economic Advisors, Wiesbaden, August.

De Groen, WP and Gros, D. 2015: Estimating the Bridge Financing Needs of the Single Resolution Fund: How expensive is it to resolve a bank?, CEPS Special Report No. 122, CEPS, Brussels, November, www.ceps.eu/publications/ estimating-bridge-financing-needs-single-resolutionfund-how-expensive-it-resolve-bank. Also published on the European Parliament's website, www.europarl.europa.eu/RegData/etudes/IDAN/2015/542687/IPOLJDA(2015)542687_EN.pdf.

Demyank, Y, Ostergaard, C and Sorensen, B. 2007: Risk Sharing and Portfolio Allocation in EMU, European Commission Working Paper, Brussels.

Furceri, D and Zdzienicka, A. 2013: The Euro Area Crisis: Need for a Supranational Fiscal Risk Sharing Mechanism?, IMF Working Paper WP/13/198, International Monetary Fund, Washington DC, September.

Gros, D. 2012: Banking Union: Ireland vs. Nevada, an illustration of the importance of an integrated banking system', CEPS Commentary, Centre for European Policy Studies, Brussels, October, http://www.ceps.eu/book/banking-union-ireland-vs-nevada-illustration-importance-integratedbanking-system.

Gros, D. 2016: A Fiscal Capacity for the Euro Area? What to Learn from the US Experience, Centre for European Policy Studies, Brussels, February, mimeo.

Gros, D and Thygesen, N. 1995: European Monetary Integration. Longman: London.

Hepp, R and von Hagen, J. 2013: Interstate risk-sharing in Germany: 1970-2006. Oxford Economic Papers 65(1): 1-24, January.

IMF. 2013a: A Banking Union for the Euro Area, IMF Staff Discussion Note 13/01, International Monetary Fund, Washington DC, February.

IMF. 2013b: Toward a Fiscal Union for the Euro Area, IMF Staff Discussion Note 13/09, International Monetary Fund, Washington DC, September.

Ingves, S. 2010: The Crisis in the Baltic--the Riksbank's Measures, Assessments and Lessons Learned, Speech by Mr Stefan Ingves, Governor of the Sveriges Riskbank, to the Riksdag Committee on Finance, Stockholm, 2 February, http://www.bis.org/review/rl00203b.pdf.

Kalemli-Ozcan, S, Papaioannou, E and Peydro, JL. 2010: What lies beneath the euro's effect on financial integration? Currency risk, legal harmonization, or trade? Journal of International Economics 81(1): 75-88.

Mink, M and de Haan, J. 2014: Spillovers from Systemic Bank Defaults, CESifo Working Paper 4792, Munich, May.

Pagano, M. 2014: Is Europe Overbanked?, Reports of the Advisory Scientific Committee, European Systemic Risk Board (ESRB), European System of Financial Supervision, No. 4, June, Frankfurt/Main.

Van Rompuy, H in close collaboration withBarroso, JM, Juncker, J-C and Draghi, M. 2012: Towards a Genuine Economic and Monetary Union, http://www.consilium.europa.eu/uedocs/cms_data/ docs/pressdata/en/ec/134069.pdf.

(1) In that year a common mechanism and fund for both deposit insurance and resolution was created in the form of the FDIC (Federal Deposit Insurance Corporation). The creation of the FDIC came after most of the 50 different state-based deposit insurance schemes went bankrupt as a countrywide banking crisis led to the failure of hundreds of banks.

(2) The starting point for this section is Gros (2012)

(3) The initial loss estimates of the FDIC were later revised downwards to $2.4 billion. As some of the assets, which the FDIC had to value at crisis prices, later recovered partially. However, the initial estimate constitutes the more important figure because it shows the amount of risk the FDIC was prepared to assume at the height of the crisis. During a financial crisis the perception of risk by the market and the ability to bear risk is more important than the exact amount of the losses that materialise once the crisis is over. The loss estimates of Fannie and Freddy were not revised as they represent just the sum of mortgages that did not perform. At first sight it appears that the loss rate for the FDIC was about 10 %, not much higher than the 8 % of bail-in-able debt instruments that EU banks are supposed to hold under the regulations. This would seem to suggest that the likelihood that the SRM could face large losses should be minor. However, Washington Mutual, which had its headquarters in the state, represents a large part of the balance sheet of the intervened banks. However, given that there was no loss for the FDIC in this operation (WAMU was sold for 1 dollar) the loss rate on the other banks was much higher, about 30%.

(4) Fannie Mae and Freddy Mac have taken the unusual step of indicating their credit losses for those states hardest hit by the crisis (including Nevada, Florida, California, for example).

(5) The experience of Washington Mutual (WaMu) constitutes a somewhat special case. The biggest bank to have failed in US history, a mortgage specialist, WaMu had its headquarters in Nevada and some small operations there. However, its failure did not lead to any local losses as WaMu was seized by the FDIC and its banking operations were sold for a very low sum to another large US bank (JP Morgan Chase) --but without any loss for the FDIC. Such an 'overnight' operation would have been impossible in Europe where no euro area-wide institution would have carried through a cross-border takeover of this size. Moreover, WaMu received about $80 billion in low-cost financing from the US Federal Home Loan Bank. Irish banks received massive amounts of low-cost emergency liquidity assistance from the European Central Bank, but the Central Bank of Ireland had to guarantee these loans, which was not the case for the State of Nevada or for any bank in Nevada.

(6) It appears, however, that the larger UK banks, like RBS, also had substantial operations in Ireland, where they had to write off of about 8 billion t. Unfortunately, it is not possible to establish what proportion of the write off resulted in actual losses and what part of any losses was incurred in the Republic of Ireland and what part in Northern Ireland.

(7) See 'Swedish banks can handle Baltic losses of 20 billion dollars', http://www.baltic-course. com/eng/finances/?doc=14707zz; see also 'Riksbank sees 2010 Baltic bank losses at USD 3.7 bln', http://www.baltic-course.com/eng/finances/?doc=23185.

(8) See Ingves (2010) and 'SEB banka has not yet recovered what it lost during financial crisis', http://www.baltic-course.com/eng/finances/?doc=88286.

(9) For more detailed information see 'FDIC Statistics at a glance', http://www.fdic.gov/bank/ statistical/stats/2012mar/fdic.html.

(10) For an early discussion of fiscal and political union see Gros and Thygesen (1995). The view that a fiscal and political union is needed is expressed at the political level by the report of the four EU Presidents on Genuine Economic and Monetary Union. See van Rompuy et al. (2012).

(11) Belke et al. (forthcoming) analyse the benefits and costs of a non-euro country opting-in to the banking union.

ANSGAR BELKE [1,2,3] & DANIEL GROS [2]

[1] University of Duisburg-Essen, UniversitatsstraEe 12, 45117 Essen, Germany. E-mail: [email protected]

[2] Centre for European Policy Studies, Congresplaats 1, Brussels 1000, Belgium. E-mail: [email protected]

[3] Institute for the Study of Labor, Schaumburg-Lippe-Strasse 5-9, 53113 Bonn, Germany.

This symposium paper is based on a keynote address given at the X111th edition of the International Finance and Banking (FIB A) Conference organized by the Faculty of Finance of the Bucharest School of Economic Studies which was held on March 26-27, 2015 in Bucharest, Romania.

Table 1: Ireland and Nevada compared

                                                 Nevada   Ireland

Population (in million, 2011)                     2.7       4.5
GDP (in billion ($) 2011)                         120       200
Change in GDP(2007-2010)                         -5.3%    -17.6%
Average net migration rate since 'bust' (2008)   0.32%     0.09%
  as percentage of total population
Unemployment rate (2011)                         13.5%     14.4%

Source: Eurostat and BEA, US Census Bureau

Table 2: Comparison Spain-Florida

Comparison Spain and Florida              Spain      Florida

Population (in million, 2011)              46.1        19.1
Nominal GDP (in billion ([euro]), 2011)   1,063      542 (770
                                                   billion USD)
Change in nominal GDP (2007-2011)          1.0%       -0.9%
Unemployment rate (2011)                  21.7%       10.5%
Change in unemployment rate (2007-2011)   13.4pp      6.5pp

Source: Own calculations