VoxEU Column International Finance

$60 billion and counting: Carry trade-related losses and their effect on CDS spreads in Central and Eastern Europe

Western bank exposures in Eastern Europe are an issue that is increasingly in policymakers’ sights. This column estimates the losses arising to the non-bank sector and government from foreign currency-denominated debt in Central and Eastern Europe. It also estimates the effect that these losses have had on the market-implied assessment of sovereign default risk. Both losses are reflected in wider CDS spreads, but government losses have a bigger impact.

Driven by the desire to catch up with Western Europe in terms of investment and consumption, and deceived by the low nominal interest rates on the Swiss franc, the euro, or the US dollar, households and non-banking sector firms in several Central and Eastern European (CEE) economies have accumulated the equivalent of $250 billion worth of debt denominated in a foreign currency. In Austria, mostly due to its proximity to Switzerland and the significant interest rate differential between the Swiss franc and the euro, total foreign currency exposure is equivalent to well over $100 billion.

What was advertised by both local and international banks to be a bargain has turned out to be a costly lesson for the entrepreneurs and households that took out these loans. For example, over the course of the last 6 months, the Swiss franc has appreciated by 31% against the Polish zloty, while the Hungarian forint has lost 14% against the euro. Given the large financial positions, these sharp currency appreciations have led to enormous aggregate losses for the borrowing nations.

Private households and firms have suffered a large part of the losses, but much of the currency exposure is concentrated in households and firms that are better equipped to bear the risks, thus somewhat mitigating the adverse real effects of these exposures (see Brown et al. 2008, Beer et al. 2008). Yet market participants anticipate that the government will ultimately bail out the private sector vis-à-vis its resident banks, thus also strongly affecting the outlook for public finances and market perceptions of sovereign default.

In a recent study (Auer and Wehrmüller 2009), we quantify the non-bank sector's losses from foreign currency-denominated debt in nine CEE countries and Austria. We then estimate the effect that these losses have had on the market-implied assessment of sovereign default risk, i.e., on CDS spreads.

Forex exposure and losses from the carry trade

Figure 1 presents the cumulated loss in the non-bank sector that can be attributed to the revaluation of foreign currency loans since the beginning of August 2008 in Austria, Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, and Slovakia. The data used in Figure 1 includes information on all foreign currency-denominated loans to non-bank enterprises, households, and the government that were issued by domestic institutions and adds to that the external debt of the government (please see Auer and Wehrmüller 2009 for the precise calculation of the exposures).1

The total losses in these ten economies sum to nearly $60 billion. In relative terms, Hungary and Poland stand out with cumulative losses totalling respectively 18% and 8% of GDP. Total losses for other countries are non-negligible, but do not exceed 5% of GDP.

Figure 1: Estimated revaluation of non-bank and loans and government debt in % of GDP

The effect on CDS spreads

To what extent do markets assume that the government will ultimately pay for the losses of the private sector vis-à-vis resident banks? To what extent does the expected cost of this bailout, as well as the direct losses from government forex loans, affect the market’s assessment of public finances?

Figure 2 presents the relation between foreign currency-related losses/gains for non-bank loans and sovereign CDS spreads for the period 1 August 2008 to 16 March 2009. It includes all eight nations with available CDS spread. On the horizontal axis, the plot displays daily changes in the valuation of the foreign currency-denominated non-bank loan position as a share of the country’s GDP. A loss corresponds to a negative value and a gain corresponds to a positive value. On the vertical axis, the absolute day-to-day changes in the country’s 5-year sovereign CDS spread are displayed in BP.

Figure 2: Forex-loan gains or losses and 5-year CDS spreads

Figure 2 suggests that foreign currency-related losses in the non-bank sector had a large effect on CDS spreads. In Auer and Wehrmüller (2009), we document that this relation is significant in a proper econometric framework. In the latter study, we adopt a fixed effects panel estimation spanning the eight countries and 163 business days since 1 August 2008. In our estimates, we control for country fixed effects and allow for CDS spreads to revert to their means. Since it is likely that movements of the exchange rates are correlated with CDS spreads through channels other than the country’s foreign currency-denominated loan exposure, we also control for the changes of the exchange rates of the examined economies with the carry trade currencies.

Our main point estimates suggest that if, on a given day, the losses occurring to the government that arise from appreciating foreign currencies amount to 1% of GDP, the country’s CDS spread on government bonds increases by 0.1998%, or 20 basis points. We also find that losses that occur to the non-bank sector affect sovereign CDS spread by substantially less than do direct losses of the government. The point estimate is that if losses of the private non-bank sector amount to 1% of GDP, the country’s CDS spread on government bonds increases by 11 basis points. This coefficient suggests that losses in the non-bank sector, on average, pass through to the public sector’s financial position with a rate of 11/20 or 55%.

To give a better impression of how our predictions compare to actual movements of CDS spreads, Table 1 presents the CDS spreads in the countries examined in this study on 1 August 2008, on 16 August 2009, and the difference between those two dates. Column 4 shows what part of the increase of CDS spreads can be explained by the losses on forex loans.

Table 1. Changes in CDS spreads that can be explained by forex losses

  CDS spreads (BP) Difference
  (1) (2) (3) (4)
  01.08.2008                             16.03.2009                                Tot. Act. Change =(2)-(1)        Part of (3)Expl. by Forex Losses
Austria 11 195 184 27
Croatia 82 507 425 37
Czech Republic 37 249 212 27
Hungary 114 523 409 265
Lithuania 138 770 632 10
Poland 51 309 258 123
Slovakia 34 162 128 8

Notes: For calculation see Auer and Wehrmüller 2009.

As can be seen by comparing the last two columns of Table 1, foreign currency-related losses can explain a large part of the skyrocketing CDS spreads for Hungary and Poland. The losses can, however, only explain a small fraction of the similar increase in CDS spreads for other nations such as the Czech Republic. This pattern suggests that financial market participants have not taken into account that the forex positions of CEE economies are very heterogeneous, but have lumped these economies together in the same risk category on a rather ad-hoc basis.2

It is also noteworthy that CDS spreads have not only trended upwards, but have also differentiated, i.e., the spread differentials between nations such as the Czech Republic and Hungary have widened somewhat. This differentiation points towards a generally rising price of insuring against government default risk in CEE, rather than a genuine upward shift in the likelihood of default (see Remolona et al. 2007).

Authors' note: An extended version of this column is available here.

Disclaimer: Views expressed in this article do not necessarily reflect those of the Swiss National Bank


Auer, Raphael, Martin Brown, Andreas Fischer, and Marcel Peter, 2009. Will the crisis wipe out small carry traders in Central and Eastern Europe?, VoxEU.org, Published on 29 January 2009.
Auer, Raphael and Simon Wehrmüller, 2009. Carry Trade-Related Losses and their Effect on CDS Spreads in Central and Eastern Europe, mimeo, Swiss National Bank.
Beer, Christian, Steven Ongena, and Marcel Peter, 2008. Borrowing in Foreign Currency: Austrian Households as Carry Traders, Swiss National Bank Working Paper 2008-19.
Brown, Martin, Steven Ongena, and Pinar Yesin, 2008. Foreign Currency Borrowing by Small Firms, working paper presented at the SNB-CEPR Conference: “Foreign Currency Related Risk Taking by Financial Institutions, Firms and Households”, 22/23 September 2008.
Brown, Martin, Marcel Peter, and Simon Wehrmüller, 2009. Swiss Franc Lending in Europe, Policy Paper, Swiss National Bank.
Remolona, Eli M. Michela Scatigna, and Eliza Wu 2007. The Dynamic Pricing of Sovereign Risk in Emerging Markets: Fundamentals and Risk Aversion, 20th Australasian Finance & Banking Conference 2007 Paper, Journal of Fixed Income, Vol. Spring, 2008.

1 This definition does NOT include private sector loans that are obtained directly from a foreign creditor. We exclude these loans on the basis that each country’s government is much less likely to bail out investors owing money to a foreign rather than a local bank; therefore, we assume that the direct exposure to foreign creditors will not affect public sector finances by much.
2 Lithuania constitutes a special case since the country has until now been able to defend its currency peg to the euro, hence eliminating the losses on euro-denominated loans. The recent increase of the country’s CDS spread probably reflects the likelihood of a devaluation and the associated losses in the future.

210 Reads