Costs and benefits of government borrowing in foreign currency: is it a major source of risk for EU member states outside the Euro?

* The author would like to thank Igor Ljubaj and two anonymous reviewers for providing useful comments during the preparation of this paper. The views and opinions expressed in this article are those of the author and do not necessarily reflect the official policy or position of the Croatian National Bank.

¹ Kaminsky and Reinhart (1999) empirically investigated the link between banking and currency crises. They found that banking and currency crises often coincide, with the banking typically preceding the currency crisis. The currency crisis in turn aggravates the banking crisis, creating a vicious circle with an adverse impact on the real economy. However, although banking crises usually start before currency crises, they are not necessarily the immediate cause of currency crises. In fact, two crises are sometimes manifestations of the same root cause – excessive credit growth backed by favourable access to international financial markets.

² Frankel and Rose (1996) found that the likelihood of a currency crisis is high when international reserves are modest, the share of FDIs in total external debt is low, the real exchange rate is overvalued, credit growth is strong and interest rates in the advanced economies are rising. Most of these variables are identified as critical also by Sachs, Tornell and Velasco (1996).

³ This risk is often referred to as foreign currency-induced credit risk because the materialization of currency risk on the side of the borrower leads to the materialization of credit risk for the creditor.

⁴ A deterioration in investor confidence may concern a single country or a group of countries sharing similar characteristics. A notable example of a broad-based confidence shock was the so-called "taper tantrum" of May 2013. As the Federal Reserve announced that it was considering winding down its program of unconventional monetary policy measures, a sudden reassessment of global risk premiums took place, which had a negative effect on several emerging market economies that were vulnerable at the time due to elevated macroeconomic imbalances (Sahay et al., 2014).

⁵ For instance, in November 2016 the Mexican peso depreciated sharply against the dollar following Donald Trump's victory in US presidential elections, as investors feared that the new President would keep his campaign promise to revise the North American Free Trade Agreement (NAFTA) agreement (Banco de Mexico, 2017).

⁶ There is a large body of literature dealing with the vulnerability of emerging market countries to self-fulfilling panics (Obstfeld, 1986; Obstfeld, 1994; Eichengreen, Rose and Wyplosz, 1995; Flood and Marion, 1996; Cole and Kehoe, 2000).

⁷ For a review of unconventional policy measures of the ECB, see Constancio (2012), Gros, Alcidi and Giovanni (2012) and Micossi (2015).

⁸ In June 2020, the ECB decided to scale up the total size of the program from EUR 750 billion to EUR 1,350 billion.

⁹ The capacity of the central bank to support government debt in foreign currency is limited by the size of its foreign exchange reserves. On the other hand, the capacity to intervene in domestic currency funding markets is not directly determined by the level of foreign exchange reserves. At the peak of the COVID-19 crisis, several emerging market central banks pursued unconventional monetary policies to support their economies, including the purchase of government debt securities denominated in domestic currency (IMF, 2020a). However, for such programs to be credible, the central bank should have comfortable foreign exchange reserves to assure financial markets that the newly created money will not lead to a currency devaluation.

¹⁰ The first thing the central bank will do in such a context is reduce the reserve requirement to release banks' own foreign currency liquidity buffers. If these buffers prove insufficient, the central bank may decide to sell a portion of its foreign exchange reserves to banks through foreign exchange interventions or swap transaction. Obviously, the ability of the central bank to supply foreign currency to banks in this way is constrained again by the available stock of foreign exchange reserves (Chang and Velasco, 1998).

¹¹ However, in the case of a severe banking crisis, even large foreign exchange reserves may prove insufficient to contain the turmoil in the banking system and support the currency at the same time. An example of this is the Latvian crisis from 2008, when a dramatic outflow of foreign currency deposits from a major bank led to a rapid depletion of central bank reserves, which undermined the credibility of the currency peg. This in turn prompted the Latvian authorities to request international financial assistance (Purfield and Rosenberg, 2010; IMF, 2009).

¹² There are, indeed, some alternative sources of revenue for the central bank other than foreign exchange reserves. In particular, central banks typically charge commercial banks a positive interest rate on repo loans granted through regular monetary operations. In addition, if the central bank implemented a government bond purchase program, it could earn interest on bonds accumulated during the implementation of the program. This interest revenue, however, would at some stage be paid back to the state budget as part of the distribution of central bank profits.

¹³ As an example, if the government issues a bond internationally with a coupon rate of 4%, and the central bank makes 1.5% on the reserve assets it acquired after converting the bond placement proceeds into domestic currency, the effective nominal interest rate on the bond will be roughly 2.5% (assuming that all central bank profits are allocated to the state budget).

¹⁴ Most studies that deal with currency crises identify inadequate reserves as one of the best early warning indicators of currency crises (Kaminsky, Lizondo and Reinhart, 1998; Kruger, Osakwe and Page, 1998; Vlaar, 2000; Abiad, 2003; Babecky, et al., 2012).

¹⁵ An event that clearly confirms the importance of the maturity composition of government debt is the 1994 Mexican crisis, which was triggered by a run on short-term US dollar-denominated government liabilities called Tesobonos (Sachs, Tornell and Velasco, 1996).

¹⁶ As the recent crises in Iceland, Spain and Ireland have shown, even if fiscal indicators are generally sound, debt sustainability may soon come into question if severe losses in systemically important banks force the government to implement costly recapitalization programs in order to avoid the collapse of the banking system.

¹⁷ At the end of 2018, the Danish government had virtually no foreign currency liabilities (Danmarks Nationalbank, 2019). Although about a quarter of Sweden's government debt is denominated in foreign currency, the level of foreign currency exposure is much lower as the national debt office has purchased large amounts of foreign currency in recent years to reduce the exposure to currency risk (Swedish National Debt Office, 2019).

¹⁸ While Bulgaria borrows almost exclusively in euros, a small part (less than 5%) of Croatia's government debt is denominated in US dollars. However, given that the Croatian government uses EUR/USD currency swaps to hedge the exposure to the US dollar, its public finances are sensitive only to fluctuations in the EUR/HRK exchange rate (MoF, 2017).

¹⁹ In addition to investing in foreign currency-denominated debt securities issued domestically, local banks with a preference for foreign currency assets may also purchase eurobonds issued by the government in international financial markets. This is a common practice in Croatia.

²⁰ Specifically, in Bulgaria and Croatia the shares of foreign currency debt in total debt amount to 81% and 72% respectively, while the shares of debt held by non-residents are much lower, at 44% and 33%. By contrast, in Poland and the Czech Republic, which are much less euroized, the shares of foreign currency debt are actually lower than the shares of debt held by non-residents (28% and 11%, compared to 44% and 41%), suggesting that foreign investors hold non-negligible stocks of domestic currency debt issued by these two countries.

²¹ The Czech Republic is a special case, as it experienced an increase in both international reserves and non-FDI liabilities from 2007 to 2018. This was partly driven by the Czech National Bank's decision of November 2013 to introduce an exchange rate floor with the aim of stimulating recovery and fending off deflationary pressures. This decision triggered large speculative non-FDI flows to the Czech Republic, as investors expected the koruna to strengthen against the euro once the floor has been removed. In 2017 alone, the central bank bought as much as EUR 42.5 billion (22% of GDP) in the foreign exchange market to defend the floor (Czech National Bank, 2018).

²² If economic ties between two countries are weak, the correlation of their business cycles is likely to be low, so monetary policy tailored to the needs of one of them will not necessarily be appropriate for the other. For example, if the anchor country is in a more mature phase of the economic expansion than the follower country, it may choose to raise interest rates to prevent its economy from overheating, and this in turn could depress growth in the follower country, which would prefer monetary policy to remain unchanged.