An indicator of sovereign risk

Certain EU officials have been vocal in their criticism of the sovereign credit default swap market for its supposed contribution to a rise in borrowing costs for some European countries. However, analysis from Hungary’s central bank suggests CDS spreads provide more accurate information on Hungarian sovereign credit risk than bond yields.

The sovereign credit default swap market has come under scrutiny from regulators across Europe in the wake of the Eurozone debt crisis. Some politicians accused speculators of using it to push spreads wider and reap quick profits – a trend they believe exacerbated the problems faced by some European countries.

Click here for a full version of the article, with charts and equations.

So far, however, there have been few studies devoted explicitly to the functioning of the sovereign CDS market. Gathering information on this sector is essential for the Hungarian central bank, Magyar Nemzeti Bank (MNB), to carry out its responsibilities. As a result, the MNB decided to conduct empirical analysis to explore the market for Hungarian sovereign CDS.

The first study, which began in 2007, looked at the size and liquidity of Hungarian sovereign CDS relative to the global credit derivatives market. We were primarily interested in the information on Hungarian credit risk contained in the CDS spreads, and finding out how reliable that information is compared with other credit spread measures.

However, the deepening of the global financial crisis in the third quarter of 2008 led to a substantial widening in Hungarian sovereign CDS spreads. At this point, determining what portion of the move could be attributed to a global decline in risk appetite – which affected all emerging markets – and how much could be ascribed to country-specific factors became an important issue.

Government bonds vs CDS

Information on sovereign credit risk can be obtained from a variety of sources. One can look at the country credit rating assessments from the major rating agencies and monitor any revisions. However, ratings from individual agencies may differ in the short term, and decisions to upgrade or downgrade a certain country often lag behind market perceptions.

Another option is to quantify the portion of the yield on a government bond that investors expect to receive in exchange for assuming the credit risk associated with that country. As such, the credit spread on a euro-denominated Hungarian government bond can be estimated by deducting the risk-free rate from its yield, where the risk-free rate is the yield of a risk-free bond with parameters – that is, maturity, denomination, interest rate and secondary market liquidity – identical to the given bond. Since all parameters of the two bonds other than credit risk are identical, we can assume their yields contain the same amount of risk premiums corresponding to interest rate risk, liquidity risk, exchange rate risk and reinvestment risk. Any difference between the two yields should purely reflect the credit spread.

We used the euro-denominated German government bond yield for the purposes of the research to approximate the risk-free euro rate. However, a simple yield spread over the German government bond is not an entirely accurate estimate of the credit spread. First, even German government bonds are not completely free of credit risk. To address this problem, we deducted the prevailing German CDS spread from the German government bond yield and used this difference as a proxy for the risk-free yield.

Second, some other minor imperfections exist that can distort the yield spread – for example, Hungarian and German government bonds do not have identical parameters, and the Hungarian euro-denominated sovereign bond market is less liquid than the German government bond market. Nevertheless, empirical experience suggests the overall distortion is not usually very significant.

With the development of global credit derivatives markets, an alternative information source on the price of Hungarian sovereign credit risk emerged: the CDS spread. When trading a Hungarian sovereign CDS contract, the counterparties transfer the credit risk of foreign currency bonds issued by the Hungarian government denominated in any accepted currency.

Reliable information regarding the Hungarian sovereign CDS market is scarce. According to a triennial survey on global foreign currency and derivatives published by the Bank for International Settlements in 2007, Hungarian market participants seldom enter into domestic sovereign CDS contracts. Active market participants include global investment banks, hedge funds and other non-resident fund managers, typically motivated by the possibility of taking advantage of short-term changes in the credit risk premium of Hungary.

According to information from credit derivatives brokers, the Hungarian sovereign CDS market achieved an adequate level of liquidity towards the end of 2005 and early 2006, but a significant increase in activity was observed from early 2008. In that year, brokers typically received 30–40 binding price quotes a day from an average of 10 banks, accounting for 1–3% of all sovereign CDS quotes. At the same time, the daily turnover of actual trades was estimated to be at least $20 million.

The Hungarian sovereign CDS market is therefore considerably more liquid than the secondary market for foreign currency-denominated Hungarian government bonds, which are seldom traded and typically held to maturity. We estimated $10 billion–30 billion of Hungarian sovereign CDS contracts were outstanding at the end of 2007.

The upper band of this estimate is in line with transaction data reported by the New York-based Depository Trust & Clearing Corporation (DTCC)2, which began publishing this information after we had conducted the survey. According to the DTCC, the gross outstanding stock of Hungarian sovereign CDS contracts totalled $33 billion at the end of October 2008 and $57 billion at the end of June 2010 (nearly $5 billion and $3.5 billion net notional, respectively). In comparison, the outstanding amount of Hungarian foreign currency bonds reached $21 billion at the end of June 2010. Therefore, it is evident Hungarian sovereign CDS should be considered a market of key importance from the perspective of pricing Hungarian sovereign credit risk.

The primary market of price discovery

As discussed, the price of Hungary’s sovereign credit risk can be measured by using either the spread of the Hungarian foreign currency government bond yield over the corresponding risk-free rate, or the CDS spread. If the two levels are not identical, we must decide which price provides more accurate information on the changes in the market’s perception of credit risk.

Over the long run, the two spreads have moved together. However, they deviated from one another on several occasions at the end of 2008 and into 2009. While both the CDS spread and the bond yield spread began to soar in October 2008, the difference between them increased significantly as well – at times exceeding 200 basis points.

We ran a co-integration analysis to examine the relationship between the two time series over different periods. Our empirical analysis reconfirmed that while the two time series co-integrate in the long run (that is, they tend to return to one another), they may temporarily deviate due to factors related to the microstructure of these markets. These include the differences in liquidity, the relatively small proportion of participants who are active in both markets and transaction costs, which are capable of persistently preventing market arbitrage forces from coming into effect. Therefore, the Hungarian CDS spread and the foreign currency bond yield spread at times might contain highly conflicting information about changes in credit risk. We need to decide which of the two has the more reliable information content.

To answer this question and determine the primary market for price discovery, we used error correction analysis. Error correction models focus on changes in time series, and are based on the idea that one of two co-integrating time series usually reacts first to events, with the other lagging the changes.

Based on the results, we concluded the CDS market was the primary market for price discovery between 2006 and 2009, as the CDS spread usually reacted first – that is, new information regarding Hungary’s credit risk was quickly reflected in the CDS spread. The foreign currency bond market was not as effective, as yield spreads tended to lag the changes in the CDS spread.

Consequently, the CDS spread, as opposed to the foreign currency bond yield spread, has been a more reliable measure of Hungarian sovereign credit risk. Nevertheless, we cannot conclude the level of government bond yields is unjustified from an economic perspective. Indeed, the fact the foreign currency bond yield spread was considerably higher than the CDS spread between the end of 2008 and 2009 probably reflects a significant increase in the liquidity premium on Hungarian bonds during this period, as the method we applied to estimate the credit spread of foreign currency bonds cannot separate the liquidity premium from the yield spread.

Hungary’s CDS spread increased significantly in the third quarter of 2008. Our next step was to assess the extent to which this change can be attributed to a substantial global decline in risk appetite observed from the second half of 2008 – that is, a general increase in global credit spreads – and the extent to which country-specific factors contributed to this development.

To answer this question, we conducted a comparative analysis of five-year Hungarian sovereign CDS spreads versus CDS levels observed in other countries in the period between January 2008 and April 2010. Besides Hungary, we used the five-year CDS spreads and the credit rating – the average rating of Moody’s Investors Service and Standard & Poor’s – of 14 other central European and other emerging countries. Our goal was to grab, at each point in time across the sample period, the common information on the relationship between the CDS spread and credit rating of these countries, and examine how the specific values of Hungary compare to that data.

Based on the parameters of the daily regressions, we calculated a rating-implied estimate for the five-year Hungarian CDS spread for each day of the sample period. We then calculated the difference between the actual and estimated CDS spreads.

Since the estimated CDS spread contains information about the relationship between international sovereign CDS spreads and the credit ratings of the selected 15 countries, the difference between the actual and estimated spreads can be considered the country-specific part of the Hungarian CDS spread. In other words, DIFH,t is the part of the CDS spread that is not justified by global developments affecting all credit spreads.

While the estimated and actual values of the CDS spread broadly co-integrated during the sample period, they deviated by more than 100bp in October 2008. Considering the average credit rating of Hungary at the time, global developments justified a widening of the Hungarian CDS spread to nearly 500bp until the middle of October, which is more than twice as high as the highest value observed in the past.

At that point, however, it became apparent that investors’ perception about the credit risk of Hungary was significantly more unfavourable than their assessment of the credit risk of other emerging countries. Indeed, the Hungarian sovereign CDS spread surpassed the regression line between the credit rating and five-year CDS spreads of emerging countries by 150bp on October 16, 2008 – a historical high.

In conclusion, the significant decline in risk appetite triggered by the financial crisis, which affected all emerging markets, was particularly devastating for Hungary, as the perception of sovereign credit risk was significantly poorer than its prevailing average credit rating.

In addition to fundamental reasons, technical factors also contributed to the increase of the Hungarian CDS spread from below 200bp up to 600bp. Due to the significant decline in the liquidity of the government bond market, many investors were unable to reduce their Hungarian credit risk exposures to the extent they had wished to. This created a sudden excess demand for Hungarian sovereign CDS contracts. As these investors were seeking protection on their Hungarian government bond portfolios, their behaviour triggered an abrupt and substantial increase in the CDS spread, a part of which can therefore be considered overshooting.

The emergency 300bp increase in the policy rate (from 8.5% to 11.5%) by the Monetary Council of the MNB on October 22, 2008, followed by the announcement of the International Monetary Fund credit facility agreement on October 26, largely contributed to contain the profound loss of confidence in Hungarian investments. In the days that followed, the difference between the actual and the estimated Hungarian CDS spreads dropped to nearly zero from the previously observed historical high.

Country-specific factors played a significant role in the development of the Hungarian credit spread once again in March 2009 (see figure 2). Reflecting a substantial weakening of the euro/forint exchange rate, the Hungarian CDS spread exceeded the level warranted by global developments and Hungary’s prevailing credit rating by nearly 100bp. However, the difference between the estimated and actual Hungarian CDS spread fell below 50bp by the third quarter of that year and remained there until the end of the period examined (April 2010). The level of the actual CDS spread also fell substantially, by about 200bp.

Based on these facts, we can conclude Hungary’s relative credit risk position improved together with global risk appetite between March 2009 and April 2010, and the relative disadvantage of the country vis-à-vis other emerging countries significantly diminished.

Conclusion

The importance of the Hungarian sovereign CDS market has grown steadily over the past decade, associated with a marked improvement in liquidity. The sector has been dominated by foreign investors hedging credit risk on their Hungarian government bond portfolios or speculating on Hungarian credit risk. This latter approach may be motivated by the fact CDS offer investors an easier and more flexible way to manage their credit risk positions than the less liquid bond market, where they may face difficulty closing out their positions.

More recently, attention has focused on the role of CDS markets during the Eurozone debt crisis – and in particular, whether speculative activity contributed to a rise in borrowing costs for some European governments. In response, some supervisors have proposed new regulations for the CDS market, including restrictions on naked short selling of sovereign CDS contracts.

The analysis conducted by the MNB had a different focus, so has not been able to directly address these issues. However, some of our results are interesting in this context. While there is undoubtedly a need to adopt uniform and comprehensive regulations on CDS markets, assessing the role of these markets and the potential to take credit risk positions solely through CDS contracts is a subtle issue.

Our results suggest the Hungarian sovereign CDS spread is currently the best indicator of Hungarian credit risk. Nonetheless, the Hungarian CDS market is not immune to overshooting in the short term, as illustrated in October 2008.

In other words, although we believe the CDS spread to be the best available measure of Hungarian credit risk, it does not mean it is a reliable indicator at each point in time taking fundamentals into account.

As such, a large and lasting overshooting leading to unduly negative expectations and an unjustified sell-off of Hungarian assets cannot be ruled out.

On the other hand, during some episodes of market turbulence, it was very beneficial that investors were able to manage their exposure to Hungarian credit risk in the CDS markets and were not forced to sell more of their Hungarian assets, which would have led to a further deterioration in liquidity.

Moreover, our results show the CDS spread reflected the global and country-specific shocks influencing Hungarian credit risk over longer periods in the past, and as such is a very useful indicator of general risk perception and risk appetite towards Hungary.

Lóránt Varga is a senior economist at Magyar Nemzeti Bank, the Hungarian central bank. The views expressed in this article are those of the author and do not necessarily reflect the official view of the Magyar Nemzeti Bank