Sensitivity Analysis of Denmark's International Investment Position

Transcription

1 61 Sensitivity Analysis of Denmark's International Investment Position Thomas Bie, Statistics, and Frank Øland Hansen, Economics INTRODUCTION Denmark's net international investment position, or net external debt, is the difference between Danish residents' assets and liabilities vis-à-vis abroad, stated at market value. The net external debt fell from 42 per cent of annual GDP in 1990 to 17 per cent in mid This is primarily due to the fact that since 1990 Denmark has had a current-account surplus every year except Both external assets and liabilities have more than doubled since 1990, however. At the same time the composition of the net external debt has changed to include more net liabilities as bonds and net assets as shares. Consequently, the fluctuations on the international financial markets have gained greater importance to the development in the net external debt. Statistical elucidation of the international investment position has improved greatly in recent years. Danmarks Nationalbank has begun to publish quarterly statements of Denmark's international investment position to complement the annual statements issued so far. The statements are prepared on a very detailed basis with breakdowns of Danish residents' foreign assets and liabilities by sectors, financial instruments, country of counterpart and currency of issue, which makes it possible to undertake equivalent detailed analyses. In the following, this data basis is applied to distribute the development in the net external debt in the 2nd quarter of 2002 by the period's transactions, and by market price and currency gains/losses. The new information on the composition of assets and liabilities is also used to analyse the sensitivity of the net external debt to fluctuations in currency rates and in share and bond prices. The effect of hedging is not taken into account, however, since sufficient data is unavailable. The method of analysis is Value-at-Risk (VaR), which with a given probability states an upper limit for the expected fluctuation in a given period. The

2 62 CONTRIBUTIONS TO LOWER NET EXTERNAL DEBT FROM THE CURRENT ACCOUNT AND VALUE ADJUSTMENTS Chart 1 Kr. billion Q1 Q Current-account surplus Value adjustments Source: Statistics Denmark and Danmarks Nationalbank. VaR analysis shows that with a probability of 95 per cent value adjustments will not increase the net external debt by more than approximately kr. 50 billion within one quarter. DEVELOPMENT IN THE NET EXTERNAL DEBT Due to the increase in both assets and liabilities, portfolio value adjustments have acquired greater significance to the overall development in the net external debt. In the short term the effect of value adjustments often eclipses the current-account surplus, cf. Chart 1. However, value adjustments of assets and liabilities tend to be equalised over time, so that the balance of payments is decisive to the long-term development in the net external debt. The strong fluctuations in the international financial markets in recent years are reflected in fluctuations around the trend, which is determined by the current-account balance, cf. Chart 2. The external debt has thus increased since 1999 in spite of the current-account surplus, primarily as a result of strongly receding share prices since the beginning of BREAKDOWN OF THE DEVELOPMENT IN THE NET EXTERNAL DEBT The detailed data basis for calculation of the net external debt, cf. Box 1, makes it possible to break down the development from the start to the

3 63 DENMARK'S ASSETS AND LIABILITIES VIS-À-VIS ABROAD AND NET EXTERNAL DEBT AS A PERCENTAGE OF GDP Chart 2 Per cent of GDP 60 Kr. billion 3, , , , , Q1 Q2 Assets (right-hand axis) Liabilities (right-hand axis) 2002 Net debt as a percentage of GDP 0 Source: Denmark's international investment position, 2nd quarter 2002, Nyt no. 93. end of the period, taking into account the development in both exchange rates and stock-market prices: Portfolio, end portfolio, start + net transactions + exchange-rate adjustments + market-price adjustments The total value adjustments are determined on the basis of information about the portfolios at the start and end of the period, as well as the net transactions during the period. Then the total value adjustments are broken down into exchange-rate adjustments and market-price adjustments. The exchange-rate adjustments are calculated on the basis of the development in the most important currencies. 1 The statement of the international investment position includes hedging as derivatives, e.g. forward contracts under loans, deposits, etc. The degree of detail for derivatives does not permit a breakdown of the total value adjustment for loans, deposits, etc. by exchange-rate and market-price effect. Denmark's net external debt increased from kr. 224 billion to kr. 234 billion in the 2nd quarter of This is primarily related to the fact that the quarter's current-account surplus of kr. 7 billion was more than offset by the price development in the financial and currency markets, 1 The calculation of exchange-rate adjustments is based on the portfolios at the start of the period. It is thus not taken into account that both value adjustments and transactions with abroad during the quarter may affect the size of the portfolios.

4 64 DATA BASIS FOR THE QUARTERLY STATEMENTS OF DENMARK'S INTERNATIONAL INVESTMENT POSITION Box 1 From 2002, the external debt is calculated on the basis of a number of new statistics and data-collection systems. This is necessary owing to requirements from the IMF and the ECB, among other things. As regards the level of detail, it has been attempted to ensure the future orientation of the data collected for each area. The most important data sources for the quarterly statements of the external debt are: Danmarks Nationalbank's MFI statistics for the accounts with non-residents of Danish banks, mortgage-credit institutes and other credit institutions. The data also includes information on foreign branches and subsidiaries of Danish banks with regard to lending to Danish companies, etc. Finally, Danmarks Nationalbank's foreign assets and liabilities, primarily the foreign-exchange reserve, are included. Danmarks Nationalbank's statistics relating to Danish investment associations' holdings of foreign securities and non-residents' holdings of Danish investment certificates. The adjusted VP statistics relating to non-residents' holding of Danish bonds, etc. registered by VP Securities Services. The statistics are adjusted for the extent of repurchase agreements with non-residents, as well as Danish residents' holdings in foreign custodian banks. Information about issues of Danish securities abroad. This information is compiled by Danmarks Nationalbank from Danish issuers. Information from Danish custodian banks about resident customers' holdings of foreign securities and non-resident customers' holdings of Danish portfolio shares. Danmarks Nationalbank's annual questionnaire survey, primarily about nonfinancial enterprises' foreign assets and liabilities particularly direct investments, but also securities deposited in foreign custodian banks, etc. Danmarks Nationalbank's payment statistics for transactions with abroad. These statistics are used for quarterly projections of the annual information relating to direct investments and for breakdown of some of the information from the annual questionnaire survey by countries and currencies. Supplementary information from other sources, above all from the central government's accounts, relating to the government's foreign assets. cf. Table 1. The development in exchange rates reduced the net external debt by kr. 8 billion, while the development in stock-market prices increased the net external debt by kr. 25 billion. The decrease in international share prices in the 2nd quarter of 2002 meant that the foreign portfolio shares held by Danish residents lost kr. 55 billion in value. To this should be added an exchange-rate loss on foreign portfolio shares of kr. 22 billion, primarily reflecting a decline by more than 12 per cent in the value of the US dollar vis-à-vis the krone. Likewise, the falling Danish share prices have meant that Danish portfolio shares held by non-residents have lost kr. 24 billion in value, which

5 65 BREAKDOWN OF DEVELOPMENT IN DENMARK'S INTERNATIONAL INVESTMENT POSITION, 2ND QUARTER 2002 Table 1 Kr. billion Start 2nd quarter Transactions Exchange-rate effect Stock-market effect, etc. End 2nd quarter Total net assets Foreign direct investment Equity securities Debt securities Loans, deposits, etc Reserve assets Total assets 2, ,217 - Foreign direct investment Equity securities Debt securities Loans, deposits, etc Reserve assets Total liabilities 2, ,451 - Foreign direct investment Equity securities Debt securities Loans, deposits, etc Reserve assets Note: Ownership interests of below 10 per cent in a company are defined as portfolio shares. Direct investments are ownership interests (shares) of 10 per cent or more in a company. The great majority of the direct investments are in unlisted companies, stated by the companies at book value or intrinsic value. For direct investments, value adjustments have only been made for the largest stock-exchange-listed companies and solely on the basis of the development in stock-exchange prices. Source: Denmark's international investment position, 2nd quarter 2002, Nyt no. 93, and own calculations. 1 Loans and deposits, etc., primarily involving an exchange-rate effect, also include financial derivatives. The level of detail in the statistics does not permit a breakdown of the total value adjustment by exchange-rate and stockexchange-price effect. reduces Denmark's net external debt. It may seem illogical that falling Danish share prices contribute to reducing the net external debt. However, the lower market value results in losses to both residents and nonresidents. The loss to residents reduces domestic assets, while the loss to non-residents reduces their outstanding claims and thereby the net external debt. An extreme case was observed in when nonresidents' ownership of the rapidly rising Nokia share was reflected in an increase in Finland's net external debt. Correspondingly, a decrease in the market value of this one paper is the major single reason that Finland's net external debt was reduced by 75 per cent from 2001 to mid Another important source of value adjustments is non-residents' holdings of Danish bonds. In the 2nd quarter of 2002 the market value declined, in spite of net purchases and rising stock-market prices. This was due to an exchange-rate adjustment of kr. 18 billion as a result of nonresidents owning Danish bonds issued in US dollars for kr. 136 billion in the 2nd quarter of 2002.

6 66 BREAKDOWN BY CURRENCY, END OF 2ND QUARTER 2002 Table 2 Kr. billion Assets Liabilities Net assets Euro US dollars Japanese yen Pounds sterling Swiss francs Norwegian kroner Swedish kronor Other foreign currencies Danish kroner 404 1,520-1,116 Total 2,217 2, Note: Direct investments are included in the calculations on the liabilities side at approximately kr. 400 billion in kroner, and on the assets side at approximately kr. 90 billion in dollars, since the country of investment has been used as approximation for currency. Source: Denmark's international investment position, 2nd quarter 2002, Nyt no. 93, and own calculations. SENSITIVITY OF THE NET EXTERNAL DEBT TO EXCHANGE RATES AND SHARE PRICES The isolated financial consequences of fluctuating exchange rates and stock-market prices can be assessed using the detailed statement of the international investment position. Here the effects of currency fluctuations and the development in the stock markets are examined. The calculations do not include the effect of derivatives, for which detailed information is not available, and consequently the analysis does not provide a full picture. Since derivatives are used more to hedge exposures than to incur risks, the calculations can be seen as high-end estimates. Danish residents hold more assets in euro than in any other currency, cf. Table 2. The assets in euro exceed the liabilities by almost kr. 500 billion, so that exchange-rate fluctuations vis-à-vis the euro would result in large losses or gains. In practice the Danish fixed-exchange-rate policy entails that fluctuations are very small, and value adjustments due to the rate of the euro have been insignificant. Danish residents also hold considerable assets and liabilities in US dollars, but since they are on more or less the same scale the net exposure is modest. If the rate of the dollar falls by 10 per cent, Denmark's net external debt will increase by less than kr. 2 billion. The largest net exposures are in pounds sterling and Swedish kronor, both of which fluctuate a good deal vis-à-vis the Danish krone. A 10-per-cent decline in one of these currencies would result in a loss of kr. 9 billion or kr. 7 billion respectively. The remaining foreignexchange exposure is relatively modest, and fluctuations in individual exchange rates will not lead to major value adjustments of the overall net external debt.

7 67 HOLDINGS OF FOREIGN PORTFOLIO SHARES BY OWNER Chart 3 Kr. billion Q Social security funds MFI Insurance and pension Investment associations, etc. Non-financial enterprises Households Note: Ownership interests below 10 per cent in a company are defined as portfolio shares. Source: Denmark's international investment position, 2nd quarter 2002, Nyt no. 93. The Danish holdings of foreign shares have increased significantly since the early 1990s, cf. Chart 3. Especially insurance and pension companies, investment associations, etc. and social security funds have invested large proportions of their assets in foreign stock-exchange-listed portfolio HOLDINGS OF FOREIGN PORTFOLIO SHARES, BREAKDOWN BY COUNTRY Chart 4 Rest of world 15 per cent Japan 6 per cent USA 31 per cent Sweden 8 per cent UK 15 per cent Germany 4 per cent Rest of euro area 21 per cent Source: Denmark's international investment position, 2nd quarter 2002, Nyt no. 93, and own calculations.

8 68 shares. For pension companies this development reflects the fact that the share ceiling has been raised several times, from 25 per cent in 1990 to a maximum of 70 per cent at present. After the falling share prices in 2001 and the first half of 2002 the market value of Danish residents' foreign portfolio shares was kr. 342 billion at the end of the 2nd quarter of Almost a third was in US shares, cf. Chart 4. A drop by 10 per cent in the US stock market would thus result in a direct Danish wealth loss of almost kr. 11 billion. Since there tends to be considerable co-variation in the development on the international stock markets, simultaneous losses will often be suffered in other stock markets. Danish shares also tend to shadow the fluctuations abroad, which reduces the total value adjustment of the net external debt, regardless of the size of the losses incurred by individual investors. VALUE-AT-RISK To gain a full overview of how fluctuations in the financial markets affect the net external debt, the probability of fluctuations in the factors affecting the net external debt should be taken into account. Value-at-Risk (VaR), a tool used in financing theory, is an expression of the risk connected to a given portfolio. With a chosen probability, VaR states the greatest loss to be expected in a future period under normal market conditions. For portfolio managers VaR is a useful management tool which e.g. takes into account the co-variation in the price development of the assets. Danmarks Nationalbank thus applies VaR in connection with overall portfolio decisions and the ongoing monitoring of Danmarks Nationalbank's market risk. 1 VaR can also be applied to analyses of the net external debt. Since the latter is a compilation of all residents' assets and liabilities vis-à-vis abroad, the calculation of VaR for the net external debt solely serves analytical purposes. A VaR analysis of the net external debt can be used to estimate the size of a "normal" fluctuation in the net external debt as a result of value adjustment. The VaR estimate depends on the chosen time horizon for which the risk is assessed, since a longer time horizon results in a higher VaR. For the calculation of VaR for the net external debt a time horizon of three months has been selected, to match the statistical statements. In addition, the estimate depends on a number of technical assumptions, including whether it is based on daily, monthly or quarterly observations, and the estimation period applied for the risk factors used. The distribution and 1 See Morten Malle Høyer, Value-at-Risk as a Measure of Danmarks Nationalbank's Market Risk, Danmarks Nationalbank, Monetary Review, 3rd Quarter 2002.

9 69 CALCULATION OF VALUE-AT-RISK FOR THE NET EXTERNAL DEBT Box 2 Denmark's net external debt is influenced by a large number of risk factors. This survey has been limited to the effects of changes in selected exchange rates, share prices and interest rates. The currencies are stated in Table 2. The share indices used are Standard & Poor's 500 for the USA, Euro Stoxx 50 for the euro area, Nikkei 225 for Japan and corresponding indices for the UK, Norway and Sweden. For Denmark the KFX index is used. For the bond market 10-year government bonds have been used, and for the money market 3-month money-market interest rates. The same geographical breakdown applies as for shares. For loans and deposits interest-rate adjustments do not result in any value adjustments. The importance of the individual risk factors depends on their variation and covariation and the size of the net assets affected. The distribution of net assets is determined on the basis of a detailed breakdown of the international investment position at the end of the 2nd quarter of The calculation of Value-at-Risk using the so-called analytical method requires that the distribution of value adjustments is determined. It is assumed that changes in risk factors have a normal distribution, so that portfolio value adjustments will also be normally distributed. The actual fluctuations in e.g. share prices tend to have fatter tails than the normal distribution. The normal-distribution assumption will then imply that VaR is understated. The standard deviation of the portfolio can be determined by σ 2 = ω T Σ ω, where σ is the standard deviation, ω is a vector stating how many billion kroner are affected by each risk factor, ω T is ω transposed, and Σ is the covariance matrix stating the risk factors' variance and covariance. Σ is determined on the basis of daily observations for January 2001 September 2002 to reflect the current risk scenario. The portfolio's standard deviation can thus be stated at kr. 4.1 billion with a time horizon of one day. Since we want to calculate VaR with a probability of 95 per cent, and the risk factors are assumed to have a normal distribution, the standard deviation must be multiplied by The VaR of the external debt with a time horizon of 3 months (65 trading days) is thus: VaR 95 = 1.65 x kr. 4.1 billion x 65 1/2 = kr. 54 billion. co-variation of the risk factors are here estimated on the basis of daily observations for the period from January 2001 to September In the calculation the selected risk factors are fluctuations in key currencies, share indices, 10-year government bonds and 3-month money-market instruments, cf. Box 2. The use of aggregate risk factors reflects the level of detail in the statement of the international investment position, which is a simplification in relation to basing the calculations on information about the individual securities. With a time horizon of 3 months and a probability of 95 per cent, VaR for the net external debt can be calculated at kr. 54 billion at the end of the 2nd quarter of In other words, with a probability of 95 per cent value adjustments are not likely to increase the debt by more than kr. 54 billion in the subsequent quarter.

10 70 VALUE-AT-RISK FOR DENMARK'S NET EXTERNAL DEBT, END OF 2ND QUARTER 2002 Table 3 Kr. billion VaR at 95 per cent VaR at 99 per cent Exchange-rate risk Share-price risk Interest-rate risk for bonds Interest-rate risk for money-market paper Diversification gain Total Note: The calculation of VaR with a time horizon of one quarter is based on the daily fluctuations in the risk factors in the period January 2001 September Source: Denmark's international investment position, 2nd quarter 2002, Nyt no. 93, and EcoWin and own calculations. With the current risk scenario, price fluctuations for portfolio shares are the major source of value adjustments. If all other risk factors are disregarded, VaR can be calculated at kr. 51 billion on the basis of share-price risk, cf. Table 3. It can therefore be expected that the developments in the international share market in each quarter have a greater impact on the net external debt than the current account. Bond prices fluctuate far less than share prices, but due to the large portfolios, particularly on the liabilities side, bond prices are also a major source of value adjustment. The sum of the contributions to VaR from the individual risk factors exceeds the portfolio's total VaR. The reason is that the risk factors do not always pull in the same direction, and thereby do not always result in losses/gains at the same time. This diversification gain is significant and can be calculated at kr. 31 billion for VaR at 95 per cent. In connection with the compilation of the net external debt the diversification gain reflects the fact that the risk on the net external debt as value adjustments is exaggerated if single factors are viewed in isolation. The value adjustments of the net external debt amounted to kr. -11 billion in the 1st quarter of 2002 and kr. -17 billion in the 2nd quarter of Compared with the calculated VaR of kr. 54 billion the realised value adjustments were modest. This was due to moderate price changes in the 1st quarter, while large net losses on portfolio shares in the 2nd quarter were partly offset by gains on e.g. direct investments. On the other hand, the net loss on portfolio shares of kr. 53 billion in the 2nd quarter of 2002, as a result of a decline in all international share indices, is higher than the calculated VaR with a probability of 95 per cent for portfolio-share-price risk. For portfolio shares, value adjustments on this scale can be expected to occur a few times in a decade. Total value adjustments exceeding the calculated VaR for the overall net external debt are also observed from time to time. For instance, positive value adjustments, primarily resulting from increasing foreign

11 share prices, reduced the external debt by kr. 114 billion in For comparison, the previously calculated VaR based on daily observations projected to a time horizon of one year was kr. 108 billion. We have used daily observations over a relatively short period to reflect the current risk scenario. When daily observations are used to calculate VaR with a time horizon of one quarter, it is assumed that the development tomorrow is independent of the development today. This is a theoretical assumption, which in some situations is unrealistic. For instance, the krone stays close to its central rate vis-à-vis the euro, so that fluctuations are equalised over time. If the risk scenario is determined on the basis of observations with a frequency corresponding to the time horizon for which the uncertainty is to be assessed, in this case quarterly observations, this problem is avoided. If quarterly observations are used, a longer estimation period is also required in order to have a sufficient number of observations. A longer historical period may be used to determine a VaR which is less influenced by the recent unrest on e.g. the stock market. This would give a less true picture of the current market risk, but may be used to assess value adjustments in relation to a more average market risk. If quarterly observations for the period 1st quarter nd quarter 2002 are used, we find that with a probability of 95 per cent value adjustments will not increase the debt by more than kr. 41 billion. As was expected, the more average market risk is lower than the current market risk, but the altered calculation basis does not significantly change the picture of the expected fluctuations in the net external debt. It is primarily the share-price risk that is lower if a longer estimation period is used. The negative value adjustments of the net external debt seen in the last few quarters are not unusual either in relation to the lower average market risk. There is no point in calculating VaR for the net external debt by sector. To calculate the total risk of the individual sectors, the exposure to other residents must be included, and all types of hedging taken into account. The external-debt statements do include statements of financial derivatives, but firstly they only include the market value vis-à-vis nonresidents, and secondly the statistics are not sufficiently detailed to be used as a basis for a sector analysis. It is therefore only possible to make qualitative statements about the sectors contributing to the net external debt VaR. Share-price risk naturally relates to the sectors which own foreign shares, mainly insurance and pensions, as well as investment associations. Non-residents also bear a risk in relation to Danish shares. Interest-rate risk is primarily attributable to government and mortgage-credit bonds. Exchange-rate risk is not attributable to particular sectors, as residents to a large extent hedge and 71

12 72 take on exchange-rate risks via Danish banks, which in turn hedge their exposures abroad. CONCLUSION The statement of Denmark's international investment position is now made on a more detailed basis than before. The more detailed information will gradually be presented in the coming quarterly statements of the international investment position. In this article the detailed data basis is used to break down the net external debt and its development over a given period of time. Value-at- Risk is used to illustrate how fluctuations in share prices, exchange rates and interest together can influence the net external debt. In the longer term the effects of the upward and downward fluctuations in the net external debt will offset each other, but the VaR analysis based on historical data shows that quarterly fluctuations in the net external debt of approximately kr. 50 billion as a result of developments in the financial markets may be expected from time to time.