Market risk – currency and interest rate risk Currency risk

Since the 1970s, the Austrian state borrows in foreign currency. Before the entry into the euro-zone, foreign markets had to be tapped due to the limited capacity of the domestic market. Now, Austrian bonds can be sold in the common (euro) currency area and as a small sovereign borrower Austria’s financial transactions have only a limited impact on the European capital market. Therefore the continued use of foreign currency markets is based on arguments related to expected cost effectiveness. When we issue in foreign currency and subsequently swap the resulting exposure into euros, currency risk is replaced by credit risk associated with the cross currency swap. (The financial soundness of swap counterparties is therefore essential.)

Not all currency risk is swapped away. After our swap operations, 12 per cent of the total debt is still denominated in foreign currency, whereas the quota of the Swiss Franc currently is a little bit larger than the quota of the Japanese Yen. Obviously the speculative element of that business strategy is that we expect in the long term that the appreciation of the currency will be less than the savings in interest payments. Internal calculations show that until year end 2003, the accumulated cost savings are around 3 per cent of GDP on a net basis (after deducting appreciation losses). That means that the Austrian federal debt would have been higher by 3 per cent of GDP if debt management had used solely domestic currency instruments.

Although foreign currency financing has been quite successful in the past, we have defined prudent restrictions for carrying out this risky business.

The risk is due to the appreciation of the foreign currency, because it increases the value of the Austrian debt portfolio and makes the interest payments on foreign currency debt more expensive. Whereas we normally impose limitations on foreign currency operations in terms of quotas of the total debt, the underlying risk is controlled in a more advanced way. The objective was to define a single number that would give a quantitative indication of the expected loss due to exchange rate movements. So we decided to adopt the Value at Risk concept for our portfolio¹ by defining our exchange rate risk as the maximum expected loss over the next 12 months at a level of confidence of 95 per cent. This loss limit is linked to GDP, with our foreign currency position currently having a Currency Value at Risk number of not more than 1.2 per cent of GDP.

When this limit is reached, the forex position is not allowed to increase any further. At the same time it is stated that it may not be sensible to close positions with unrealised losses when the increase in the Value at Risk is due to a currency appreciation. In that situation further steps will have to be discussed with the supervisory board. In addition to exchange rate

movements, the Currency Value at Risk numbers are also influenced by volatilities, correlations and the amounts of foreign currency.

Interest rate risk I (Interest Costs at Risk)

The importance of the risk of rising interest costs stems from the fact that variations in interest expenses directly affect the federal budget balance and, as a result, can severely restrict the flexibility of fiscal policy, especially since the Republic of Austria has committed itself to stay within the deficit limit of the growth and stability pact of the European Union.

We have developed a fairly complex model to address the information needs of those responsible for the budget in the ministry of finance. In this model, interest costs are defined in accordance with Austrian accounting rules so that we can calculate the federal budget balance. Our model provides:

● an estimate of future interest costs under different economic scenarios, as well as

● the determination of the amount of interest costs that will not be exceeded with a certain probability (Interest Costs at Risk).

Interest costs are driven by three major factors:

1. The future path of primary budget balances.

2. The refinancing and portfolio management strategy.

3. Rates and prices in financial market.

The first factor is updated once a month in accordance with the projections made by the ministry of finance. The second factor is defined by

Figure 7.1. Currency value at risk

2.8

Jan. 02Feb. 02Mar. 02 Apr. 02

May 02June 02July 02Aug. 02Sep. 02Oct. 02Nov. 02 Dec. 02Jan. 03Feb. 03Mar. 03

Apr. 03

May 03June 03July 03Aug. 03Sep. 03Oct. 03Nov. 03 Dec. 03Jan. 04Feb. 04Mar. 04

Apr. 04

May 04June 04July 04Aug. 04Sep. 04Oct. 04Nov. 04 Dec. 04Jan. 05Feb. 05 2.62.4

2.2 2.01.8 1.61.4 1.21.0 0.80.6 0.40.2 0 Billions of EUR

AFFA and is only changed within the model if there is a fundamental shift in AFFA’s strategy. In order to generate projections of future financial market rates and prices, two financial market scenarios and a range of scenarios for determining statistical risk estimates are run. The two scenarios are the forward scenario and the steady state scenario. Steady state is defined as freezing all rates at their current levels for the entire time horizon of the analysis. Monte Carlo procedures are used to generate 1 000 possible future paths of rates and prices. The results of this set of future paths is used to determine statistically the total amount of interest costs that will not be exceeded in a certain period (a fiscal year) and confidence interval (95 per cent).

These three scenarios are computed at least once a month. For each computed scenario path (forward, steady state, risk) we look eight years into the future. This period was chosen because it represents two full legislative periods. In near future, the risk model will be enhanced to provide more flexibility in setting the refinancing and portfolio management strategy.²

Interest rate risk II: (Value at Risk)

The Value of Risk (VAR) of the entire position is perhaps of less direct concern for decision makers than interest costs at risk because the immediate political consequences are less visible. Nonetheless, from an economic point of view the VAR of the entire portfolio is an important number. An increase in the value of the debt portfolio reflects an increase in the future burden for taxpayers or it may boost the cost of switches or buybacks. Consequently, Value at Risk is calculated at least once a month as a fixed part of our monthly reporting routine.

Figure 7.2. Interest expense 2004-2011

9.0

2005 2006 2007 2008 2009 2010 2011 2012

8.0

7.0

6.0

Steady state scenario Forward scenario Risk scenario Billions of EUR

In order to account for non-linear exposures related to a number of structured bonds, we decided to use a Monte Carlo simulation methodology instead of the simpler analytical methods for computing Value at Risk. We employ the same stochastic paths as was used for estimating interest rate risk. Once a month, the simulator generates a set of 1 000 stochastic paths which are then used for different risk management applications (market and credit risk), giving us a coherent picture of our risk profile.

For the monthly reports we set the time horizon to one year because we are more interested in longer-term effects. This fairly long VAR horizon is another reason why we prefer a simulation approach over an analytical model.

However, this time feature also makes the assumption that the portfolio is not changing between now and the VAR horizon, quite questionable.³

A future enhancement of the model could be to have quick marginal and incremental analyses available, similar to correspondent functions of the long-term credit risk analysis tool (see below).