Monetary Review 3rd Quarter

Transcription

1 Danmarks Nationalbank Monetary Review 3rd Quarter 999 D A N M A R K S N A T I O N A L B A N K 9 9 9

2 Danmarks Nationalbank Monetary Review 3rd Quarter 999

3 The Monetary Review is published by Danmarks Nationalbank and is issued quarterly. Managing Editor: Jens Thomsen Editor: Anders Møller Christensen The Monetary Review can be ordered from: Danmarks Nationalbank, Information Desk, Havnegade 5, DK-093 Copenhagen K. Telephone (direct) or Schultz Grafisk A/S, Copenhagen ISSN

4 Contents Recent Monetary Trends... The Role of Gold in the Monetary System... 9 Ulrik Bie and Astrid Henneberg Pedersen, the Secretariat The historical background to central-bank holdings of gold reserves is described. The present conditions concerning the gold reserves of Danmarks Nationalbank and other central banks are reviewed. The Market for Government Bonds in the Euro Area Lars Krogh Jessen and Anders Matzen, Financial Markets Department The factors behind the different yields on euro-denominated government bonds issued by various euro-area member states are outlined and the consequences for the issuing policy are described. Is Last Autumn's Financial Crisis Over?... 5 Leif Lybecker Eskesen, Financial Markets Department Analysis of a number of indicators of credit, liquidity and market risk shows that the financial crisis of autumn 998 is over, even though the risk indicators have not quite returned to the very low level in the summer of 998. Can Inflation Expectations Be Deduced From the Development in Danish Consumption? Michael Pedersen, Economics Department It is sought to determine inflation expectations on the basis of the growth in private consumption. The results differ considerably from other research using other methods. The explanation is that the restrictive assumptions of the method are by no means fulfilled in Denmark. More Information Available on the Nationalbank's Web site Liselotte Perch Lind, the Secretariat Adjustments at the Nationalbank in the Event of the Introduction of the Euro in Denmark... 83

5 Press release Tables and graphs section VOL. XXXVIII/No. 3

6 Recent Monetary Trends This review covers the period from the middle of May to August 999 INTERNATIONAL BACKGROUND During the summer long-term bond yields in most industrialised countries rose considerably. At the end of August the yield on 0-year US government bonds was close to 6 per cent, cf. Chart, and had been rising since the turn of the year in step with the publication of continued strong key economic indicators for the US economy. This development affected bond yields in Japan and especially in Europe where an increasing number of indicators point to an upswing in the near future. The cyclical situation is also reflected in the foreign-exchange markets. The euro continued to drop against the dollar until mid-july when the exchange rate fell below 02 dollars per 00 euro, cf. Chart 2. This development was reversed by new positive growth prospects for the euro area, and during August the rate for the euro for a period reached almost 08 dollars per 00 euro. It then declined a little to just under 06 dollars per 00 euro at the close of the month. From mid-may until the end of YIELDS ON 0-YEAR GOVERNMENT BONDS IN THE USA, JAPAN AND GERMANY Chart Per cent Germany USA Japan

7 2 US DOLLAR VIS-À-VIS THE EURO Chart 2 Dollars per 00 euro Calculated Quoted Note: Weekly averages. Most recent observation is week 34. August the Japanese yen strengthened by just over 0 per cent against the dollar, which should be viewed in the light of the publication of favourable growth figures for the otherwise fragile Japanese economy. To prevent excessive strengthening of the yen, the Bank of Japan intervened in the foreign-exchange market on a considerable scale. The upswing in the US economy, now in its eighth year, has been driven primarily by strong growth in consumption and investments. US households now spend more than they earn, which should be viewed in the light of their capital gains on stocks and homes. The stock market in particular has risen strongly and shares are generally traded at prices almost 40 times higher than annual yields. The high level of share prices must be taken to reflect expectations of strong growth in business enterprises' future earnings. Another factor underlying the growth in consumption is that the propensity to borrow is supported by strong competition on the loan market, as well as innovations in the financial sector. At the end of June and again at the end of August the Federal Reserve raised its benchmark official interest rate, the fed funds target rate, by 25 basis points, with reference to the risk of inflation. US consumer prices reacted only moderately to the upswing. This prompted discussion of whether the traditional ties between economic growth and inflation have been broken. However, a considerable proportion of the development in the consumer-price index can be attributed to the sub-index for energy products, which in turn is closely

8 3 PRICE OF BRENT OIL Chart 3 Dollars/barrel Oil price Note: Weekly averages. Most recent observation is week 34. pegged to oil prices. The weak oil price throughout 997 and 998 had a considerable dampening effect on the energy index and thereby on the overall consumer-price index during those years. In the course of 999 the oil price has shown a strongly rising trend, cf. Chart 3, and the dampening effect on US consumer prices has ceased. Against this background higher inflation must be expected in the coming months. Japan's GDP increased by almost 2 per cent from the 4th quarter of 998 to the st quarter of 999. This is the first sign of growth in GDP for more than a year, and the strongest growth during one quarter since the beginning of 996. The growth should be viewed against the background of last autumn's fiscal-policy crisis measures which temporarily increase government consumption and government investments, although private consumption also rises. There are thus indications of a break with the stagnation of many years. There are also signs that the crisis-ridden Southeast Asian economies are rallying. Growth in the euro area has been weak since mid-998, and during the first half of 999 unemployment stagnated at around 0.3 per cent. However, growth prospects are very promising. The outlook for the German business community is again more positive, cf. Chart 4. Other leading indicators also point to stronger growth in the euro area as a whole in the 2nd half of 999. The oil-price increase will probably push up the rate of inflation in the euro area, where consumer prices in July were. per cent above the level in July 998.

9 4 GERMAN MANUFACTURED OUTPUT AND IFO INDEX Chart 4 Per cent, year-on-year 8 Index Manufactured output, Germany IFO index for Germany (right hand axis) Source: Eurostat and the IFO Institute in Germany. Note: Manufactured output is shown as a 3-month average of the annual rate of growth. The most recent observations are June 999 for manufactured output and July 999 for the IFO index. The UK shows signs of renewed growth after a period of falling growth since the beginning of 998. In view of expectations of a future inflation rate just below the UK government's target, on 0 June the Bank of England lowered the repo rate by 25 basis points to 5 per cent. The moderate development in prices was supported by a stronger pound sterling during the st half of 999. The Swedish economy is in an upswing driven by domestic demand. In Norway, the cyclical course has reversed after a number of years with very high growth. On 7 June Norges Bank lowered its official interest rates, the current-account rate and the lending rate by 0.5 per cent. This was Norway's fourth lowering of interest rates since the turn of the year. DEVELOPMENT IN INTEREST AND EXCHANGE RATES IN DENMARK With effect from 7 June Danmarks Nationalbank lowered the rate of interest for certificates of deposit and the lending rate by 5 basis points to 2.85 per cent. The background was the preceding weeks' substantial purchases of foreign exchange against kroner. The krone has been stable vis-à-vis the euro at a slightly stronger level than its central rate of kr per 00 euro. However, the fluctuations in the dollar's rate against the euro had a slight consequential effect on the krone's rate against the euro. As a consequence of these fluctuations Danmarks Na-

10 5 THE YIELD CURVE ON THE DANISH MONEY MARKET Chart 5 Per cent Millennium rollover Maturity in months End-June 999 End-July 999 End-August 999 Note: The rates at 7, 8, 0 and months are calculated by linear interpolation. tionalbank either sold or purchased foreign exchange on certain occasions in July and August, in order to cushion very short-term fluctuations in the krone resulting from very thin markets. With effect from 2 June 999 Danmarks Nationalbank adjusted the monetary-policy instruments, cf. the account in the Monetary Review, 2nd quarter 999. The mortgage-credit institutes now have access to the monetary-policy instruments on an equal footing with the banks. Three out of ten mortgage-credit institutes have chosen to make use of this access. In total approximately 60 banks and mortgage-credit institutes resort to the monetary-policy instruments. The overall ceiling for current-account deposits is kr. 9,930 million. On the money market the yield curve makes an abrupt leap for the maturity extending into the new year, cf. Chart 5. This increase may be the premium on the reduction of an asset's liquidity in the event of an extraordinary reduction in the trading volume. In connection with the monthly publication of its planned monetary-policy operations Danmarks Nationalbank has announced that the system for both purchase and sale of certificates of deposit will be open on 30 December 999 and on 3 January 2000 in consideration of the planning by banks and mortgage-credit institutes of their positions around the millennium rollover. Normally, the liquidity prognosis covers only the following two months, but for the rest of 999 it will run up to and including January 2000.

11 6 DANISH INTEREST RATES AND DIFFERENTIALS TO GERMANY Chart 6 Per cent January February March April May June July August 3-month money-market interest rate 0-year yield differential to Gemany 0-year government bond yield The 3-month money-market interest rate rose moderately in July and August, cf. Chart 6, and at the end of August the differential to the equivalent euro interest rate was just under 0.5 per cent. The yield on long-term bonds has risen significantly since the spring, resulting in a steeper yield curve. Since the beginning of May the yield on 0-year government bonds has risen by approximately per cent to 5.3 per cent at the close of August. This is in line with an international trend since the 0-year yield differential to Germany widened by only 0 basis points to approximately 0.5 per cent during the same period. At the end of August the international credit rating agency Moody s gave foreign-exchange denominated Danish government loans the top rating of Aaa. However, the higher rating was not unexpected by market participants, so yield differentials were not affected outright. Since the spring the yield on the typical 30-year mortgage-credit bond, 6-per-cent 2029, has risen by more than per cent. The borrowers' use of bonds with varying coupon rates makes it more difficult to monitor the average 30-year yield to maturity on the issues. On the basis of data for gross issues from Nykredit Danmarks Nationalbank has calculated a weighted yield on 30-year bonds for construction. This method is described in Box. The key result is that the weighted yield to maturity has varied somewhat more than the yield to maturity on the individual series as a consequence of the shift to bonds at a low coupon rate in Cf. Danmarks Nationalbank's Web site

12 7 CALCULATION OF THE WEIGHTED YIELD ON A 30-YEAR BOND FOR CONSTRUCTION Box House purchases and new construction are normally mortgage-credit financed and the related bond yield can be measured as the average yield on open series of mortgage-credit bonds. In recent years the mortgage-credit institutes have primarily issued 30-year series with coupon rates of 5, 6 and 7 per cent. The distribution of the issues by coupon rates has varied over time. This must be taken into consideration on evaluation of the development in the construction-bond yield over a long period, since a shift in the coupon rate implies a similar shift in the yield to maturity. The shift in the coupon rate can be captured by e.g. weighting together the yields to maturity in the open series as a synthetic yield. This weighted yield, where the weights for the underlying yields to maturity are the individual coupon rates' shares of gross issues, is shown in the chart below. The weighted yield tends to show greater variation than the individual underlying yields to maturity. This can be attributed to shifts in the coupon rate. For example, the chart shows a good drop in the weighted yield in January 999, despite the only marginal decline in the yields to maturity. This can be attributed to borrowers in January having switched to bonds with a coupon rate of 5 per cent on a considerable scale. WEIGHTED YIELD ON 30-YEAR BOND FOR CONSTRUCTION Per cent Weighted yield 5-per-cent per-cent per-cent 2029 Source: Nykredit. Note: The weighted yield is a weighting of the yields to maturity in the 30-year series. The individual series are included with a weighting equivalent to their share of gross issues in Nykredit's open series. periods of declining yields, and to bonds at a higher coupon rate when the general level of interest rates in the capital market is on the increase. The overall financial conditions, i.e. the long-term interest rate, the effective krone rate and the short-term interest rate taken as one,

13 8 FISCAL POLICY AND FINANCIAL CONDITIONS Chart 7 Percentage GDP effect since Fiscal effect Financial index Source: Danish Ministry of Finance and own calculations. Note.: The fiscal effect includes the effects of the reduced tax deductibility of interest payments under the Whitsun package of economic measures. continue to be relaxed, even though the expansionary effect has diminished in step with the buoyancy of the long-term yield, and since mid-july the strengthening of the effective krone rate. The effects of the financial conditions can be summarised in an index by weighting together the long-term yield, the effective exchange rate and the short-term interest rate with their impact on GDP, cf. Chart 7. An increase in the financial index indicates more relaxed financial conditions. A fiscal index, i.e. an index of the impact of fiscal policy on activity, can likewise be set out on the basis of the official fiscal effects of the fiscal policy in individual years. An increase in the index indicates a more relaxed fiscal policy. Given Denmark's monetary-policy objective to maintain a stable krone rate vis-à-vis the euro, the development in the financial index is almost entirely determined exogenously. An increase in the financial index must therefore be countered by a tightening of fiscal policy in order to neutralise the relaxed financial conditions. Due to the reduced tax deductibility of interest payments under the Whitsun package of economic measures, among other factors, the overall fiscal policy is estimated to have had a moderately contractive effect in the years However, the fiscal policy did not offset the relaxation of financial conditions in the period after 995 during which the Danish See e.g. Niels Lynggård Hansen, Monetary Conditions Indices, Monetary Review, May 997.

14 9 economy was characterised by strong domestic activity, with pressure on capacity and high wage increases. DOMESTIC ACTIVITY AND WAGES Domestic demand increased by 2 per cent in the st quarter of 999, according to the latest national accounts. Business investments accounted for the strongest growth. Private consumption showed a moderate increase, while government consumption fell slightly. At the same time, exports increased. A relatively large proportion of demand was met by depletion of stocks. Both imports and output (GDP) showed a slight decline in the st quarter of 999, cf. Chart 8. The considerable depletion of stocks may indicate rising output in the 2nd quarter, although the compilation of stockbuilding is usually subject to great uncertainty. The current indicators of activity in the 2nd quarter paint a rather mixed picture. On the one hand, the manufacturing industry reports higher sales and greater optimism, and unemployment declined further during the 2nd quarter, after its flattening out around the turn of the year. On the other hand, a proportion of the decrease in unemployment may be due to contraction of the labour force. Employment in the manufacturing industry and the entire private sector compiled on the basis of payments to ATP (the Danish Labour Market Sup- OUTPUT AND DEMAND Chart 8 Index 993 = Per cent Contribution of stockbuilding to growth (right) Imports Exports GDP Domestic use (excluding stocks) Source: Statistics Denmark. Note: The contribution of stockbuilding to growth indicates the change in stockbuilding as a proportion of GDP in the preceding quarter.

15 0 HOURLY WAGE COSTS IN DENMARK AND THE EURO AREA Chart 9 Per cent, year-on-year Denmark Euro area Source: The Danish Employers' Confederation (Denmark) and Eurostat (the euro area). plementary Pension Scheme) thus fell a little in the 2nd quarter from a high level in the st quarter. The tendency of lower car sales continued into the summer months, while the index of retail sales subsided a little. Both factors indicate that the dampening of consumption continues. Furthermore, consumer confidence continues to be low, although it has been rising since the turn of the year. The unemployment rate in July was 5.5 per cent. The decline in unemployment since 994 has entailed high wage inflation, which may cast doubt on the sustainability of this development in the longer term. Despite the braking in the 2nd quarter wage inflation in Denmark is still far higher than seen in the euro area, cf. Chart 9, which contributes to the persistent deterioration in Denmark's competitiveness. A precondition for maintaining the current level of unemployment on a permanent basis is thus a more flexible labour market. Viewed in isolation, the higher long-term interest rate will dampen private domestic demand. Concurrently the higher growth abroad will stimulate the demand for exports. Both factors indicate an improvement in the balance of payments. However, the extent to which domestic demand will be dampened and will thereby contribute to sustaining this course is uncertain. All in all, there is only little "export capacity" available on the labour market. It is therefore important to maintain the economic-policy objective of keeping domestic demand stable.

16 THE BALANCE OF PAYMENTS Preliminary current-account statistics show a surplus of just over kr. 6 billion for the first five months of the year. This is an improvement of no less than kr. 5 billion on the same months of 998. The improvement in the trade balance and the balance of payments in 999, cf. Chart 0, may reflect a relative increase in demand abroad, including the euro area, compared to Denmark. However, the improvement from 998 to 999 is obviously also influenced by specific factors. Firstly, the balance of payments was very poor in the first half of 998 as a consequence of the labour-market dispute. Secondly, the depletion of stocks in the st quarter of 999 contributed to a temporary weakening of imports. Thirdly, possible improvement in the interest item and other transfers may reflect a periodical deviation, and may thus be offset by poorer statistics for the rest of 999. For the 2-month period from June 998 to May 999 the current account shows a deficit of kr. billion. For the same period there is a deficit of kr. 3.5 billion on current net payments, according to the Nationalbank's payments statistics. For the first time for a long period net payments are thus poorer than the official balance of payments, although the payments do show some improvement. Current net payments are related to the time of payment, not the time of transaction as in the case of the official EXPORTS, IMPORTS AND CURRENT ACCOUNT Chart 0 Kr. billion 29 Kr. billion Current account 2-month sum (right-hand axis) Exports Imports Source: Statistics Denmark. Note.: Seasonally-adjusted exports and imports of goods, excluding ships and aircrafts. Most recent observation is May 999.

17 2 DEVELOPMENT IN CONSUMER PRICES AND NET RETAIL PRICES Table Consumer-price index Index of net retail prices Energy Imports Total Foodstuffs Domestic prices Rent Public services Other factors Weights HICP CPI Year-on-year growth, per cent st qtr nd qtr rd qtr th qtr st qtr nd qtr Note: Weighting basis as of September 996. The index of net retail prices is the consumer-price index adjusted for indirect taxes, duties and subsidies for general price reductions. "Other factors" is a measure of domestic market-determined inflation. "Other factors" normally increases faster than the index of net retail prices due to an overweight of services, for which the price development is typically stronger than for other commodities. At the same time, the demand for services viewed in a more long-term perspective will typically increase faster than the demand for other products. HICP is the Harmonised Index of Consumer Prices. balance of payments, so a certain divergence is natural, even though Statistics Denmark uses elements of the payments statistics in its compilation of the balance of payments. DEVELOPMENT IN PRICES Growth in consumer prices has increased during 999. In July, the consumer-price index was 2.4 per cent higher than in July 998. The rise in inflation must be viewed in the light of the higher oil price and stronger growth rates for food prices and prices for public services. Furthermore, inflation is technically pushed up by the bringing forward of the biannual inclusion of the rent item. The inflation rate was pushed up by 0.2 percentage points in the months from February to April, and will be lifted again in August-October. However, the tendency towards a higher price-increase rate is also seen in "Other factors", which reflects domestic wages and profits and thus the underlying development in prices,

18 3 cf. Table. Since the turn of the year this item has increased by just over 3 per cent year-on-year. The residual factor normally increases by 0.5 percentage points more than the index of net retail prices, due to a relatively high content of services. The current differential is percentage point, thus indicating very strong inflationary pressure. In July the annual growth rate measured by the Harmonised Index of Consumer Prices (HICP) was 2.0 per cent in Denmark, against. per cent in the euro area. Denmark continues to meet the Maastricht Treaty's inflation criterion for participation in EMU. This criterion stipulates a 2-month average of the HICP index compared to the average for the 2 preceding months. In the period from August 998 to July 999 the price index on average was.4 per cent higher than in the period from August 997 to July 998. A candidate country's inflation rate must not exceed the equivalently calculated inflation rate in the three EU member states with the lowest inflation by more than.5 per cent. At present, this implies a value of.8 per cent. However, there is a tendency for Denmark to move closer to the threshold value since the other EU member states have typically not experienced the same inflation updrift from rising wages. IMF REPORT ON DENMARK On 26 August 999 the IMF published its latest report on Denmark. Previously, the reports after the article IV consultations, which in Denmark's case are conducted every second year, were confidential, but as from 999 it has been decided on a trial basis to publish the IMF's analysis and recommendations. The IMF report is available on the IMF's Web site at: INTERNATIONAL PROPOSALS FOR A CAPITAL ADEQUACY FRAMEWORK The framework of the financial system is being discussed in several international fora. A principal item on the agenda for these discussions is the existing capital adequacy rules, which are a cornerstone in the regulation of financial institutions. In June 999 the Basel Committee thus submitted a proposal for amendment of the requirements which were recommended in 988. The new recommendations are expected to be issued some time after the expiry of the consultation period in March The Basel Committee, whose secretariat is located at the Bank of International Settlements (BIS) in Basel, was established in 975 for the purpose of strengthening the stability of the international financial system. The Committee consists of the G0 countries (Belgium, Canada, France, Germany, Italy, Japan, the Netherlands, Sweden, and the UK, the USA) as well Switzerland and Luxembourg.

19 4 THE BASEL COMMITTEE'S PROPOSALS FOR A NEW CAPITAL ADEQUACY FRAMEWORK Box 2 In June 999 the Basel Committee submitted the consultation document "A new capital adequacy framework" to all interested parties with a deadline for response by end-march The Committee proposes a new capital adequacy framework consisting of three pillars: minimum capital requirements, a supervisory review process and effective use of market discipline. Minimum capital requirements are a cornerstone in the existing regulation of financial institutions and will continue to be so. However, the method of calculating the solvency ratio must be brought up-to-date in order to ensure that the banks' solvency ratio continues to reflect the actual risk. The existing 988 Accord with subsequent amendments and additions has a number of weaknesses. The credit risk weights are too coarsely meshed and new risk-management instruments, such as credit derivatives and netting, are not taken into sufficient consideration, while risks other than credit and market risks are not included explicitly. 2 The Basel Committee's new proposals seek to eliminate these weaknesses. The minimum capital requirements constitute a three-step approach, whereby the first step is a review of the current standardised approach. The proposals include compilation of credit risk weights for individual loans on the basis of the borrowers' rating by external rating agencies. The next step is to allow the banks to compile the weights on the basis of internal ratings of their own customers. Finally, the possibility of using risk models as for market risk is being investigated. There is still considerable work to be done in this area, so the third step should be viewed in a longer-term perspective. In all three steps hedging of risk is more widely addressed. As mentioned, explicit capital adequacy is proposed for other risks than credit and market risks. The supervisory review process implies supervision by the relevant authorities of financial institutions and the "tools" used in this connection. Improved access for the supervisory authorities to impose capital requirements above the minimum requirement of 8 per cent (differentiated capital requirements) is proposed, as well as early supervisory intervention. Supervisory authorities must focus particularly on the banks' internal risk management processes. The previous Accord contained no explicit recommendations concerning the supervisory process. Market discipline is the stabilising effect of borrowers and other market participants refraining from trading with unsound financial institutions. This encourages the banks to operate a sound and efficient business. Market discipline requires transparent markets, or in other words that market participants have access to credible, timely information, so that they can assess the (credit)worthiness of financial institutions themselves. The Basel Committee intends to publish a report later with a more detailed description of the requirements for public disclosure of accounts, etc. 2 The consultation document is available on BIS' Web site at: For example the interest-rate risk on the banking book and the operational risk. The previous recommendations of the Basel Committee, which to a great extent served as the basis for the EU's capital adequacy framework, have contributed to strengthening the stability of the financial

20 5 system. However, amendment of the rules is required in view of the latest financial product innovations and sophisticated new riskmanagement techniques. The Basel Committee's proposed new recommendations are described in further detail in Box 2. The new proposals of the Basel Committee are based on a wish to continue to ensure the stability of the financial system and to provide a level playing field for the participants. The proposed model of different capital-requirement approaches provides an opportunity for large internationally active banks to apply more sophisticated risk-management techniques in a regulatory context, while small niche and regional banks are not subject to complicated capital requirements with appurtenant administrative burdens. Similar work is taking place under EU auspices. In cooperation with the Banking Advisory Committee, which consists of representatives of EU supervisory authorities, the European Commission is working on a proposal for revision of the capital adequacy framework (the solvency directive and the own funds directive). The Commission intends to submit the proposal for consultation during the autumn of 999. CONCENTRATION TRENDS IN THE EUROPEAN BANKING SECTOR Since 985 the number of credit institutions in the EU has fallen significantly from approximately 3,000 in 985 to approximately 9,300 in 997. This decline conceals considerable variations in levels and development trends among the EU member states. In 997 Germany had around 3,600 credit institutions, or twice as many as France, in second place. There were 90 credit institutions in the Netherlands in 997. Sweden, Finland and Denmark account for the greatest percentage decline in the number of credit institutions in the period, while in a number of large countries the reduction is less significant. These differences are reflected in the statistics for the concentration of credit institutions in the EU, cf. Chart. The degree of concentration is relatively low in the large countries, but high in a number of small countries, including the Nordic countries. The concentration in the Nordic banking sector was seen as early as at the end of the 980s and at the beginning of the 990s. In some countries the development was a consequence of a serious financial crisis in connection with the recession around the beginning of the 990s. The high concentration in Portugal and Greece is the result of a series of privatisations. The reduction of the number of credit institutions in the larger European countries led to only a small increase in the degree of concentration up to 997, due to such factors as that mergers were primarily of small savings banks and cooperative banks. The most recent

21 6 CONCENTRATION OF CREDIT INSTITUTIONS IN EU MEMBER STATES IN 997 Chart Per cent Sweden Netherlands Finland Portugal Denmark Greece Belgium Austria Spain Ireland France UK Italy Luxembourg Germany Source: ECB "Possible effects of EMU on the EU banking systems in the medium to long term". Note: The concentration is compiled as the 5 largest credit institutions' share of the total assets of credit institutions as of 3 December 997. mergers in e.g. Spain, France and Italy, involving a number of large credit institutions, will lead to a significant increase in the degree of concentration in these countries. So far, the concentration trends in the EU are primarily a result of mergers and acquisitions among domestic banks. From 995 up to the beginning of 998 there were approximately 400 mergers and acquisitions between credit institutions from the same EU member state, and only around 90 cross-border mergers and acquisitions involving a bank from an EU member state. The cross-border operations were primarily between neighbouring countries such as the Nordic or the Benelux countries. The relatively few and geographically limited cross-border mergers and acquisitions may indicate that they are subject to certain barriers. Cultural, regulatory and supervisory differences between countries may hinder mergers and acquisitions, or at least impede them. Furthermore, more politically motivated barriers may exist since many countries prefer that the largest banks on the domestic market are in resident ownership. However, the relatively limited number of cross-border mergers should also be viewed in the light of the relatively low degree of concentration in several large EU member states. Domestic consolida- For example Merita Bank - Nordbanken (Sweden, Finland), ING Bank NV Banque Bruxelles Lambert (the Netherlands, Belgium), Den Danske Bank Östgöta Enskilda Bank (Denmark, Sweden), Den Danske Bank - Fokus Bank (Denmark, Norway).

22 7 tion may thus be a requirement for cross-border mergers and acquisitions to be considered a relevant strategic opportunity. The importance of EMU to transparency and competition in the banking sector can be expected to encourage the concentration trends. After consolidation at national level the number of cross-border mergers can be expected to increase, despite national differences. At the same time, the development of new technology will provide new opportunities for cross-border activities without the establishment of a physical presence in the relevant market (e.g. Internet banking). It is evaluated that the structure of banking within the EU will become more homogenous and that the degree of concentration in the European banking sector overall will increase. The effects of EMU are described in further detail in Jens Thomsen, Karsten Biltoft and Jesper Ulriksen Thuesen, "The Euro and the New Perspectives", Monetary Review, st quarter 999.

23

24 9 The Role of Gold in the Monetary System Ulrik Bie and Astrid Henneberg Pedersen, the Secretariat SUMMARY The great period of gold in the monetary system lasted from the 870s to the outbreak of World War I. During this period a global fixed-exchange-rate system was established, based on a fixed definition of each currency vis-à-vis gold, together with unequivocal rules for gold convertibility and gold coverage. As gold was a scarce resource, its use as a direct means of transaction, i.e. as coins, was limited. Instead, banknotes gained ground, so that the convertibility of banknotes into gold came to play a leading role in the system and the gold standard evolved into a so-called gold exchange standard. During the inter-war years some countries sought to stretch their gold reserves by introducing a gold bullion standard, whereby only amounts equivalent to whole gold bars could be converted into gold. For individual citizens gold thus played a limited role, but it was still the foundation of the monetary system. At the beginning of the 930s more and more countries had to abandon the gold standard. After World War II the Bretton Woods system was established. It was based on an implicit pegging to gold, i.e. currencies were pegged either to the dollar or to gold directly. In general, only central banks had access to convertibility into gold. The Bretton Woods system collapsed in 97 as gold had outlived its central role in the international monetary system. Today, gold is still part of most central banks' foreign-exchange reserve, although gold's share of the total foreign-exchange reserves has declined. INTRODUCTION OF THE GOLD STANDARD At the beginning of the 8th century Britain defined its currency in relation to a certain amount of gold, i.e grains, corresponding to grammes, per pound. This marked the de facto introduction in Britain of a gold standard which lasted albeit interrupted for certain periods until the outbreak of World War I in 94.

25 20 In contrast to Great Britain, throughout the 9th century most other countries in Europe defined their currencies in relation to silver, although some had a bimetallic standard whereby the value of the currency was defined in relation to both gold and silver. The result was an official fixed parity between gold and silver. In the most important bimetallic country, France, this parity was 5½:. The amount of silver required for one franc was thus 5½ times higher than the amount of gold required. The value in the market fluctuated around this parity until the 870s, cf. Chart. The role of gold was determined by the limited amount available worldwide, but in the 850s considerable gold strikes were made in Australia and California, cf. Chart. For a short period this led to a weakening of the price of gold. In the 860s gold production fell, while silver production rose and the value of silver thus fell relative to gold. This caused concern in the silver-standard countries, among other things because citizens could buy silver at lower prices in the market and have it minted. At the same time the amount of gold in Europe rose because the USA financed a substantial share of its civil-war expenditure (86-65) by lifting the convertibility into gold and selling its gold reserves. The stage was thus set for gold to play a new, more prominent role in the international monetary system. The growing importance of gold in the international monetary system became clear in 865 on the establishment of the Latin Currency Union consisting of France and a number of other European countries. The GOLD PRODUCTION AND THE COMPARATIVE VALUE OF GOLD AND SILVER Chart Tonnes 700 The comparative value of gold and silver Gold production (left-hand axis) Price of gold in relation to price of silver (right-hand axis) Source: Wilcke (930).

26 2 Latin Currency Union was officially on a bimetallic standard, but gold played a far more prominent role than silver, which was widely perceived as a step towards a gold standard. The official parity between silver and gold continued to be 5½:. Two of the world's most important currencies the French franc and the pound sterling were thus both based on gold. The German states had traditionally adhered to a silver standard. After France was defeated in the Franco-Prussian war in 87 Germany received five billion gold francs as war indemnity. This amount of gold provided for a rapid transition to a gold standard. It constituted the basis for a new pan-german currency which was proposed in December 87. As a result of Germany's transition to the gold standard, it was likely that large amounts of silver would be released on the free silver market once gold had replaced silver coins in circulation. An increase of this magnitude in the supply of silver to the market would cause the price of silver to plummet, and the silver-standard countries, whose units of payment were defined in terms of a fixed amount of silver, would be flooded with cheap silver for minting. As a consequence of Germany's planned transition to the gold standard its neighbouring countries, including the Scandinavian countries, came under pressure to make a quick transition to the gold standard, cf. p. 22. However, for various reasons the gold standard in Germany was not applied until 876. The USA had been on a bimetallic standard from 792 to 86 and introduced convertibility into gold in 879. The gold standard was not formally introduced in the USA until 900. The period from the beginning of the 870s to the mid-890s was characterised by prolonged deflation. One explanation is based on the limitation which the volume of gold reserves imposes on the growth in the money supply. This limitation prevented growth in the money supply from keeping up with the growing need for money for transactions generated by world production. This led to deflationary pressure. As from the mid-890s new gold strikes meant that the deflationary pressure was replaced by an inflationary tendency. Gold as a form of commitment By the end of the 9th century most leading countries defined their currencies in relation to gold, and their central banks had committed themselves to unlimited purchase and sale of gold at the fixed exchange rate. At the same time, rules were fixed for the proportion of banknotes in circulation that were to be covered by the gold reserves held by the cen- Among the other explanations for the falling prices are lower production and transport costs, see Hansen (976) for further details.

27 22 tral banks. This created a worldwide system whereby the individual currencies, via their gold parity, were implicitly defined in relation to each other. Gold thus constituted the nominal anchor of the international monetary system. For example, a Danish krone contained grammes of gold, while the pound sterling contained 7.32 grammes of gold, so the parity of the pound sterling was 7.32/0.403=kr. 8.6 per pound sterling. However, in practice there were minor deviations in the actual exchange rate due to such factors as the costs of transporting the gold between countries and external imbalances. Besides the convertibility into gold the system comprised a set of "rules of the game" which central banks were expected to adhere to. In the event of an outflow of gold as a result of an external imbalance the central bank had to raise its interest rate in order to reduce the domestic money supply and demand, while at the same time the higher interest rate would attract foreign capital. Similarly, in the event of an inflow of gold, the central bank had to lower the interest rate. This passive interest-rate adjustment restored the external balance and brought the exchange rate back to parity. Together with free trade in gold this interest-rate adjustment had a stabilising effect on the system. The clearly defined rules pertaining to the classical gold standard made it a binding mechanism which contributed a high degree of credibility to the system. The classical gold standard coincided with the period in which Great Britain was at the zenith of its power, and London was the gold capital of the world. Great Britain's long adherence to the gold standard was the cornerstone of the system and many countries on a gold standard held sterling as a foreign-exchange reserve. Holding sterling was perceived as equivalent to holding gold, and sterling was a more appropriate means of transaction. The classical gold standard lasted until 94. DENMARK MOVES FROM SILVER TO GOLD The establishment of the Latin Currency Union in 865 stimulated Denmark's interest in moving from a silver to a gold standard. Prior to this, Denmark, Sweden and Norway had discussed the possibility of forming a currency union. The currencies of all three countries were based on a silver standard, and their coins had an almost identical silver content and were in circulation in all three countries. In the deliberations on the introduction of the gold standard it was first considered to introduce the gold franc as the currency unit in Scan- See Bordo and MacDonald (997) for further details.

28 23 dinavia, but after Germany defeated France in the Franco-Prussian war in 87, and in view of Germany's planned transition to the gold standard, the idea of the gold franc was abandoned. Interest in the gold standard for Denmark grew when Germany decided to introduce the gold standard, since this would mean that the two most important trading partners, Britain and Germany, would now be on the gold standard. In view of the risk of a strong drop in the price of silver the Scandinavian countries, as already stated, came under pressure to adopt the gold standard without delay. Already in 872 agreement was reached on a Scandinavian currency system based on gold. On 23 May 873 King Christian IX signed the new Currency Act according to which gold would be introduced as legal tender in Denmark by January 875 at the latest. A transition period of maximum 4 years was allowed, after which the rigsdaler would no longer be legal tender. The Currency Act set the value of kg of fine gold at kr. 2,480, i.e. kr. was equivalent to grammes of fine gold. The new krone coin was exchanged at a rate of kr. 2 to rigsdaler. The Currency Act also adopted the Convention on the Scandinavian Currency Union, which entered into force in 875 for Denmark and Sweden, while Norway joined the Currency Union in 877. The Scandinavian Currency Union introduced the gold standard and the same unit of account (the krone) in all three countries. In the early days of the Currency Union only coins were accepted at par by the central banks. In 894 Sweden and Norway agreed to accept each other's banknotes at par without restrictions. Denmark's acceptance of the joint circulation of banknotes in 90 marked the full establishment of the Scandinavian Currency Union. However, full union lasted only until 905 when Norway gained independence and restrictions on the joint circulation of banknotes were imposed. Although Denmark was not involved in World War I in 94 many people wished to convert their banknotes into gold on the outbreak of the war. This was perceived as a serious threat to the Nationalbank's gold reserve. The convertibility into gold was therefore suspended in August 94, followed by a ban on gold exports. THE INTER-WAR YEARS The gold standard system was reintroduced in the inter-war years. In this period the international monetary system differed from the classical gold standard in a number of respects. One significant difference was that gold was withdrawn from circulation and placed as reserves with the central banks. A case in point was Great Britain which in 925 rein-

29 24 troduced the convertibility into gold at the gold parity from before World War I. Sterling was overvalued, which led to pressure on the gold reserves, and Britain therefore introduced a gold bullion standard. Under the classical gold standard the central bank was obliged to convert every single banknote on demand. Under the gold bullion standard the central bank was only obliged to issue whole bars of gold, i.e. to convert amounts corresponding to the value of one whole bar of gold. This made it possible to economise on gold holdings and less gold cover was required. This also reduced the role of gold as a means of transaction and took gold out of the hands of citizens. Another important difference from the earlier period was that Britain had lost its role as an economic and political leader. As a result, sterling's importance in the international monetary system diminished, while the dollar was increasingly used as the reserve currency. Finally, prices and wages were less flexible than before and new considerations governed the management of currency flows. This was e.g. due to increased pressure from trade unions and political parties, which now played a greater role than was the case before World War I. A failure of cooperation, a protectionist policy and the absence of a de facto leader undermined the system. The absence of mechanisms to ensure equilibrium meant that the system was never as stable as the classical gold standard. With the depression of the 930s, the gold standard system of the inter-war years collapsed. Britain abandoned gold in 93, and the USA in 933. However, already in 934 the USA reintroduced the gold standard at a devalued exchange rate of 35 dollars per ounce of gold. Apart from a few months during World War II the USA maintained the pegging of the dollar to gold. The gold peg and the USA's economic influence provided a sound foundation for the dollar's future role as an international reserve currency. Reintroduction of the gold standard in Denmark After World War I a declared objective of Denmark's economic policy was to reintroduce the gold standard and to restore the convertibility into gold. However, it was considered more important to ensure an acceptable level of economic activity, so the decision to return gradually to the gold parity was not adopted until 924. On January 927 the convertibility into gold was reintroduced at the previous gold parity, i.e. kr. 2,480 per kilo of fine gold. The "honest krone" had thus been reestablished. As in Britain, a gold bullion standard was introduced, initially for a period of three years. One gold bar was worth approximately kr. 28,000, today troy ounce = grammes.

30 25 equivalent to approximately kr. 700,000. At the end of both 929 and 930 the gold bullion standard was extended for one year. The depression put pressure on Denmark's foreign-exchange reserve due to such factors as the repatriation of assets from abroad. On 22 September, the day after Britain abandoned the gold standard, Denmark imposed a ban on gold exports. During the period 2-29 September an amount of kr. 7.5 million was converted into gold at a deliberately slow pace. On 29 September 93 the Rigsdag (Parliament) decided to abandon the gold bullion standard. THE BRETTON WOODS SYSTEM In 944, in Bretton Woods, USA, the Western powers established a new international monetary system. The two principal countries the USA and Britain generally disagreed on the design of the system and on the future role of gold. Britain had exhausted its gold and currency reserves during the war and therefore called for a system which gave access to international liquidity in which gold thus played only a minor role. A major contributing factor was that Britain was severely affected by recession when it returned to the gold standard during the inter-war years. The USA, on the other hand, which had maintained its gold peg since 934, had substantial gold reserves and called for a fixed attachment to gold. At the negotiations the British delegation was headed by the economist J.M. Keynes, and its proposal was based on a principle of clearing between the countries. A key element of the plan was to create a new reserve currency, the bancor, which would be freely available to countries with a capital-account deficit for payments to countries with a surplus. This proposal would have removed gold completely from the international monetary system. However, the American proposal, based on a close pegging to gold, was adopted. The countries in the system could choose to peg their currencies either directly to gold, or indirectly by pegging to the dollar, which had been defined in relation to gold (35 dollars per ounce of gold). It was thus possible to hold foreign-exchange reserves in both dollars and gold. Most countries chose to peg their currencies to the dollar. The USA was obliged to buy and sell gold in order to maintain the gold price, while the other countries in the system were obliged to keep their exchange rates within a margin of +/- per cent of the agreed parities by buying and selling dollars. Officially, sterling was also an intervention currency, but in reality it played a very limited role.

31 26 Under the Bretton Woods system the convertibility into gold, which had been a fundamental element of the classical gold standard and was still an important element of the inter-war system, was limited to apply to central banks only. Ordinary citizens thus no longer had general access to the gold reserves. Most central banks chose to hold the reserves in dollars since dollar-denominated assets could be placed in interestbearing US government bonds. In contrast to the previous gold standard systems, realignment of exchange rates was possible under the Bretton Woods system, although the major participant countries were reluctant to devalue, even though economic policy did not show the degree of convergence necessary to maintain a fixed-exchange-rate system. Speculative buy-up of gold at the beginning of the 960s caused a number of countries to establish a gold pool for intervention in the gold market. The intervention of the gold pool kept the market price close to the official price of 35 dollars per ounce, making the gold market more a place of distribution than a price setter. In November 967 the devaluation of sterling against the dollar undermined confidence in the system's stability and in the dollar. This was followed by speculative demand for gold in the market. During the first months of 968 the central banks, especially the Federal Reserve, were forced to sell gold for 3 billion dollars. In March 968 it was decided to dissolve the gold pool and disconnect the central banks' gold stock from the world market. This was achieved by the central banks committing themselves not to supply gold to or receive gold from the world market. Those central banks that did so would be excluded from buying gold from other central banks. This gave rise to two gold prices: the official price of 35 dollars per ounce for trading among central banks and a market price, which quickly rose. The Bretton Woods system became more and more unstable due to the ever-growing volume of short-term dollar assets outside the US relative to the USA's gold reserves. France wished to solve the problem by returning to a true gold standard. This wish was in part also motivated by a desire to reduce the economic power of the United States. As from 965 France consistently exchanged its dollar assets for gold. France's preference was not supported by other countries and in 968 an international reserve currency, Special Drawing Rights, SDR, was created. SDR were to ensure sufficient international liquidity, while at the same time reducing holdings of dollar assets outside the USA. SDR were reminiscent of the proposed bancor and were defined as 35 SDR = 35 dollars = ounce of gold. Calculated at 35 dollars per ounce.

32 27 The use of SDR as from 969 could not prevent the collapse of the system since the economic policies of the leading countries in the system continued to diverge. Despite several attempts to contain the crisis the Bretton Woods system collapsed in August 97 on the announcement by President Nixon that the dollar could no longer be exchanged for gold. SDR were still formerly defined in relation to gold until 976 when gold was replaced by a currency basket. Gold had thus outplayed its central role in the international monetary system. There are numerous different explanations for what happened in the Bretton Woods system. One explanation was offered by Robert Triffin in 960. According to Triffin, demand for international reserves would grow faster than the supply of gold, creating a global shortage of liquidity, which would lead to monetary contraction. This could only be avoided by a sustained current-account deficit in the USA. The problem with a permanent deficit was that this would undermine the dollar's credibility as a reserve currency since the ratio between the oustanding volume of dollars and the USA's gold reserve would rise. The system would collapse if the rest of the countries in the Bretton Woods system demanded the exchange of their dollar holdings for gold. A possible solution to the Triffin dilemma was the establishment of a new reserve currency such as SDR. Another possible explanation applies to currency systems comprising two currencies with a fixed mutual exchange rate, where the central banks are willing to exchange currency at the fixed exchange rate. Should the circulating volume of one of the two currencies, e.g. the dollar, be excessive in relation to the other currency, e.g. gold, the market value of gold (measured in dollars) will rise, as will private demand for gold. Since the authorities are willing to convert at the fixed exchange rate, gold, which has a higher value in the market than in the monetary system, will disappear from the monetary system, to be used for private purposes instead (e.g. industrial use and jewellery). The dollar will thus replace gold. This is a rough outline of what happened in the 960s. The volume of dollars rose far more than the supply of gold. The gold pool established by the central banks maintained a constant gold price, but had to supply the required amount of gold to the market. Consequently, the central banks' total gold stocks fell by 8 per cent from 960 to 970. In the same period the total reserves increased by 5 per cent. Gold had thus begun to disappear from the monetary system before the latter's collapse in 97. Triffin, (960).

33 28 THE NATIONALBANK'S GOLD STOCK Chart 2 Tonnes Source: Danmarks Nationalbank. THE NATIONALBANK'S GOLD STOCK In 936, when the Danmarks Nationalbank Act was prepared, a return to the gold standard at a later stage was anticipated. The Act stipulates that the Nationalbank shall own a gold fund which shall cover at least 25 per cent of notes in circulation. The Danmarks Nationalbank Act provides for a dispensation from the gold coverage requirement. Since 939 the gold coverage requirement has been suspended. Hereafter gold has only played a role as part of the foreign-exchange reserve. Until 969 the Nationalbank's gold stock was entered to the balance sheet at the krone's gold value according to the Coin Act of 873. From 969 to 978 the gold stock was booked at the parity notified to the IMF, and from 979 to 987 at an exchange rate calculated on the basis of the market price. As from 988 the market price has been used. The Nationalbank's gold stock fluctuated considerably under the Bretton Woods system in step with the need to buy or sell gold in order to adjust the amount of liquid reserves 2. The gold stock has been almost constant since 970, cf. Chart 2. The fluctuations in the gold stock during the 970s primarily reflect transactions with the IMF, while the small 2 The parity as notified to the International Monetary Fund was used in the compilation of Denmark's international liquidity during the same period. The most recent considerable fluctuations in the gold stock took place in when gold was sold in order to increase the amount of liquid reserves after the foreign-exchange crisis up to May 969.

34 29 GOLD'S SHARE OF THE TOTAL RESERVES, APRIL 999 Chart 3 USA Italy Netherlands France Portugal Switzerland Germany Austria UK Belgium Sweden Spain Greece Finland Japan Denmark Ireland Per cent Source: International Financial Statistics, July 999, IMF. Data for the UK is from March 999. Luxembourg is excluded. increase since 987 can be attributed to revenue from gold lending by the Nationalbank, cf. Box, p. 30. By tradition gold is an asset of the central bank, and is used for certain transactions among central banks. In the event of Danish membership of the Economic and Monetary Union, EMU, the Nationalbank would for example have to transfer foreign-exchange reserves to the European Central Bank, ECB, of which 5 per cent must be in gold, equivalent to almost 24 per cent of Denmark's gold stock. The Nationalbank's gold stock is relatively small. As Chart 3 shows, gold constitutes a considerably smaller proportion of the total reserves in Denmark than is the case for the EU other member states, Switzerland, the USA and Japan. Only Ireland's gold stock makes up a smaller proportion of the total reserves. The location of the Nationalbank's gold stock For many years the Nationalbank's gold stock was held for safekeeping at various locations all over the world, e.g. Switzerland, the UK, Canada and the USA. The gold was placed in a number of countries for security reasons and to ensure that the risk was spread. At the same time the gold was held at locations where it was easy to divest. Just before World War II, from November 939 to February 940, Denmark transported a large share of its gold stock to the Federal Reserve Bank in New York. The gold was

35 30 GOLD LENDING Box Since the mid-980s a market for gold lending has developed. This market is used primarily by central banks which lend gold in order to achieve a return on their gold stock. The gold is lent to bullion banks which specialise in trading in precious metals. Gold lending by the central banks has increased considerably in recent years. At the end of per cent of the total official gold reserves had been lent. The end-borrowers of gold are typically gold companies wishing to hedge against future price drops. They can do this by e.g. selling their future gold production to the bullion banks (forward sale). The bullion banks hedge the forward purchase by borrowing gold from central banks, selling it in the spot market and investing the proceeds from the sale at a market interest rate which is typically higher than the rate of interest for gold. When the forward contract with the gold producer expires, the bullion bank will divest its investment. The bank receives the purchased gold and pays the agreed forward price to the gold producer. The borrowed gold is then returned to the central bank and the gold interest rate is paid. If the spot and forward prices are far apart, considerable gains or losses can be realised from gold hedging. This risk is assumed exclusively by the bullion banks. In 987 the Nationalbank began to lend gold to bullion banks from its gold stock deposited with the Bank of England. Lending to the bullion banks is subject to the same credit assessment as the Nationalbank's other placements. Gold deposited with the Bank of England is not necessarily physically moved when it is lent. For a number of years interest on lending by the Nationalbank has been settled in gold and added to the reserve. shipped via Bergen in Norway. This was gold held for safekeeping in Denmark, Sweden and Norway. In total 9.6 tonnes of gold were shipped in this way. In view of the political and military situation in Europe at that time the USA was considered a safe place to store the Nationalbank's gold. Moreover, this would make it possible to continue to purchase goods in the USA, provided that imports from the USA were still feasible. During the occupation years the gold held for safekeeping in the USA was administered by the Danish legation in Washington which sold some of the gold in order to make interest payments on the national debt. Since the Nationalbank had moved almost its entire gold stock to the USA there was no gold left for the occupying forces to seize from the Nationalbank during the occupation, as they did in several other countries, e.g. Belgium and the Netherlands. In 987 the Nationalbank began to lend gold to foreign banks against interest paid in gold, cf. Box. This made it necessary to move the gold to London. In August tonnes of the Nationalbank's 66.6 tonnes of gold was in London, corresponding to a share of 94 per cent. However, the primary source of financing of interest payments was the sale of cryolite from Greenland.

36 3 Almost 99 per cent, or 93 per cent of the Nationalbank's total gold stock, had been lent. SALE OF GOLD BY CENTRAL BANKS In 998 global gold production amounted to more than 2,500 tonnes and the world's total gold stock was estimated at 37,400 tonnes at the end of 998, of which the central banks held 26.3 per cent. By tradition central banks hold considerable gold reserves, but the proportion held by central banks has diminished gradually due to the reduced role of gold in the monetary system. In March 999 gold's share of the total official foreign-exchange reserves was 5.2 per cent, which is considerably less than in earlier years, cf. Chart 4 2. Several central banks have sold gold reserves in recent years. Since 988 Banque Nationale de Belgique has reduced its gold stock by 74 per cent, while De Nederlandsche Bank has sold 26 per cent of its gold stock since 992. The underlying reasons included the wish to bring gold's share of the total reserves closer to levels in other EU member states. Other central banks have also sold gold in recent years, including the central banks of Argentina, Australia and Canada. At 2,590 tonnes Schweizerische Nationalbank holds one of the world's largest gold reserves. The large reserve is a consequence of the continued close relationship between the Swiss franc and gold. The value of the Swiss franc is thus defined in relation to gold, and the gold coverage of banknotes in circulation is laid down in the Swiss Constitution. In 997 a group of experts was established to prepare a new basis for the Swiss central bank, including the pegging to gold. On the basis of the group's conclusions and a referendum in April 999 the gold coverage requirement will be removed from the Constitution as from January The abolition of the gold parity awaits approval by parliament. As a result of the abolition of the gold parity Switzerland is considering selling,300 tonnes of gold over a period of several years. In May 999 the UK Treasury and the Bank of England announced their intention over a period of several years to sell almost 60 per cent of the gold stock, corresponding to 45 tonnes of gold, as an element of the restructuring of the foreign-exchange reserve. The intention is to achieve a better balance between risks and returns on the foreignexchange reserve. In tonnes of gold will be sold. The first auction took place on 6 July Including the International Monetary Fund and the Bank for International Settlements. The value of the gold reserves in 960 and 970 is compiled at 35 dollars per ounce and at the market price as from 980.

37 32 GOLD'S SHARE OF THE WORLD'S OFFICIAL RESERVES Chart 4 Per cent Source: International Financial Statistics, IMF, various issues. At the G7 meeting in Cologne in June 999 it was agreed to sell a proportion of the IMF's gold in order to finance initiatives in support of debt-ridden poorer countries. The proceeds from the sale would be placed in interest-bearing assets and the return used to finance such THE MARKET PRICE FOR GOLD (DOLLARS PER OUNCE) Chart 5 USD per troy ounce Source: Danmark's Nationalbank.

38 33 initiatives. This sale was strongly opposed by a number of gold-producing countries, including South Africa, and by the US Congress where the proposal is awaiting approval. The current plans to sell gold can be viewed as a continuation of the sales of gold by several central banks in recent years. After the UK's announcement of sale of its gold the gold price has dropped by per cent in three months, cf. Chart 5. Since the gold price peaked in 980 it has fallen by 70 per cent to dollars per ounce on 9 August 999. However, it is important to bear in mind that most of the present gold reserves were purchased before 97 at 35 dollars per ounce. CONCLUSION Although gold no longer plays a central role in the monetary system, it has retained its traditional role as a store of value in many countries, especially in Asia. Even in countries where gold plays a lesser role, it will continue to hold a significant symbolic value which distinguishes it from other metals, due to its central role in the history of the monetary system.

39 34 BIBLIOGRAPHY Andersen, Poul Nyboe, From Gold Standard to Clearing (in Danish), Nyt Nordisk Forlag-Arnold Busck, Copenhagen, 942. Andersen, Svend, The International Monetary System in Transition, in Current Economic Problems (in Danish), Memorial volume for Carl Iversen, Nationaløkonomisk Forening, 969. Banque Nationale de Belgique, Report 989, Brussels, 990 Bordo, Michael D. and Eichengreen, Barry, The Rise and Fall of a Barbarous Relic: The Role of Gold in the International Monetary System, Working Paper 6436, National Bureau of Economic Research, Cambridge, USA, 998. Bordo, Michael D. and MacDonald, Ronald, Violations of the "Rule of the Game" and the Credibility of the Classical Gold Standard, , Working Paper 65, National Bureau of Economic Research, Cambridge, USA, 997. Clementi, David, Deputy Governor's Speech, 22nd Annual FT World Gold Conference, London 999. Danmarks Nationalbank, Annual Report and Accounts (various issues), Copenhagen. De Nederlandsche Bank, Annual Report 992, Amsterdam, 993. International Monetary Fund (IMF), International Financial Statistics (various issues), Washington D.C. Foreign & Commonwealth Office, Nazi Gold, The Stationary Office, London, 998. Gold Fields Mineral Services, Ltd., Gold Survey 999, Gold Fields Mineral Services Ltd., London 999. Grauwe, Paul de, International Money, Post-War Trends and Theories, Clarendon Press, Oxford, 989. Hansen, S. Aa. and Svendsen, K.E., Danish Monetary History, Volume, Danmarks Nationalbank, Copenhagen, 968.

40 35 Hansen, Svend Aage, Economic Growth in Denmark (Volumes I and II), Akademisk Forlag, Copenhagen, 976. Hoffmeyer, Erik and Olsen, Erling, Danish Monetary History, Volume 2, Danmarks Nationalbank, Copenhagen, 968. Lidegaard, Bo, In the Name of the King (in Danish), Samleren, Copenhagen, 996. Newman, P., Milgate, M. og Eatwell, J. (ed.), The New Palgrave Dictionary of Money and Finance, Macmillan Press Limited, London, 992. Nielsen, Axel, The Scandinavian Currency Union (in Danish), Børsens Forlag, Copenhagen, 97. Roth, Jean-Pierre, A View on Switzerland in the Run up to the Demonetarisation of Gold, 22 nd Annual FT World Gold Conference, London, 999. Rubow, Axel, The History of Danmarks Nationalbank , Gyldendalske Boghandel-Nordisk Forlag, Copenhagen, 920. Schweizerische Nationalbank, 9st Annual Report 998, Bern 999. Triffin, Robert, Gold and the Dollar Crisis, Yale University Press, New Haven, 960. Wilcke, J., The Silver and Gold Standard (in Danish), G. E. C. Gad, Copenhagen, 930. Williamson, John, The Failure of World Monetary Reform, 97-74, Thomas Nelson and Sons Ltd., Southampton, 977.

41

42 37 The Market for Government Bonds in the Euro Area Lars Krogh Jessen and Anders Matzen, Financial Markets Department INTRODUCTION The commencement of the 3rd stage of Economic and Monetary Union (EMU) on January 999 brought major changes in the framework for the bond markets of the euro area. The lapse of exchange-rate uncertainty and the transition to a single monetary policy eliminated some of the most important factors behind government-bond yield differentials among the euro-area member states. The expectations with regard to the development in the European bond markets after January 999 were to a great extent based on experience from other currency unions, e.g. the USA and Canada. Comparisons of the size of the bond markets in respectively the euro area and the USA gave rise to expectations of increased liquidity in the European bond markets taken as one. It was also expected that a certain yield differential among the government bonds of the individual euroarea member states would continue to exist, due to differences in credit risk. The work to prepare for EMU by government debt offices indicated that competition among government issuers would intensify considerably after January 999. A central issue was which government securities would achieve benchmark status for the entire euro area. As a consequence, most government issuers focused on building up large liquid bond series. The purpose of this article is to examine whether the expectations of the government bond market in the euro area have been fulfilled. First a description is given of the market for government securities in the euro area. To a degree, the bond markets are still segmented according to national borders. It is also observed that there are still considerable yield differentials among the countries. Cf. Bertil From, Yield Differentials in the Future EMU, Danmarks Nationalbank, Monetary Review, May 997.

43 38 The yield differentials are explained in the following two sections, with focus on respectively credit risk and liquidity. As expected, credit risk has a certain influence on yield differentials. The liquidity of the individual bond issues is another important element in the explanation of yield differentials. In this connection various aspects of importance to liquidity are outlined, including the work by government debt offices to build up liquidity, and the market conditions. The transition to the euro has also led to considerable changes in the market for non-government bonds. This subject is not considered here, since the article focuses solely on the development in the market for government bonds in the euro area. THE MARKET FOR GOVERNMENT SECURITIES IN THE EURO AREA The redenomination of almost all domestic government debt entails that as from the commencement of EMU the basis was created for the establishment of a large-scale market for government securities for the entire euro area. Furthermore, the market conventions were harmonised in connection with the transition to the third stage of EMU. Finally, the importance of regulatory requirements of placements in national currencies by institutional investors has lapsed. The euro area's bond market is the second-largest in the world, exceeded only by the US bond market, cf. Table. Public issuers are central governments, federal states, etc. Government bonds account for a good 3,000 billion euro of total general-government issues in both the USA and the euro area 2. Although the markets for government securities in the euro area and the USA are of similar magnitude, there are major differences. There is still some segmentation into national markets of the government securities market in the euro area, while in the USA it constitutes one single market. This is related to the fact that the government securities of the euro-area member states are not perfect substitutes. One explanation is variations in the member states' credit standing and a continuing tendency for government securities to be purchased particularly by domestic investors. Despite the overall magnitude of the government securities 2 The consistency provisions of the 3rd EU life assurance directive whereby 80 per cent of the technical reserves must be in the same currency as the future disbursements are no longer limited to investments within the euro area. Outstanding government securities in the euro area are calculated for the following countries: Austria, Belgium, Finland, France, Germany, Italy, the Netherlands and Spain. This compilation is based on all negotiable domestic debt instruments, i.e. both government bonds and Treasury bills. In the USA, apart from government bonds the federal government has debt in non-negotiable debt instruments amounting to 2,63 billion euro. Source: National government debt offices.

44 39 OUTSTANDING BONDS AT END-998 Table Euro billion Government issuers Private issuers EU-... 3,875 3,33 EU ,763 4,38 USA )... 7,632 6,405 Japan )... 3,626,66 Source: Bank for International Settlements; International Banking and Financial Market Developments. Note: The figures comprise both domestic and foreign issues by the stated countries or country groups. ) Converted into euro at the exchange rate on 30 June 999. market in the euro area this segmentation entails less liquidity in the market as a whole than is the case in the USA. Yield differentials between government securities in the individual countries continue to exist, despite the lapse of the exchange-rate risk. This appears from Chart which illustrates yields to maturity for government securities in the euro area. The chart also shows a tendency for yield differentials to widen in step with maturity. Chart 2 presents the development in yield differentials in the 0-year maturity segment. Prior to the commencement of the third stage of EMU the expectation was that yield differentials would continue to exist among the member states. In mid-998 the -year forward yields on 0-year maturities thus indicated yield differentials equivalent to the actual differentials in July YIELDS TO MATURITY AND MATURITIES OF EURO-DENOMINATED GOVERNMENT BONDS Chart Per cent Maturity, years Note: 4 July 999. Source: Reuters.

45 40 0-YIELD DIFFERENTIALS TO GERMANY OF SELECTED EMU COUNTRIES, 999 Chart 2 Per cent January February March April May June July August France Netherlands Austria Spain Belgium Italy Portugal Note: Weekly averages Differentials between bonds with the same maturity can be attributed primarily to variations in credit risk and liquidity. Benchmark bonds for the euro area Benchmark bonds are used as a reference for the pricing of other bonds. They are very liquid, which is usually reflected in a relatively high trading volume and a small spread between bid and offer prices. The focus on liquidity entails that only bonds with relatively substantial outstanding amounts can function as benchmarks for the euro area. Whether a bond achieves benchmark status is determined by the leading bond dealers. Achieving benchmark status is of interest to the issuer since this implies lower borrowing costs. The yields to maturity on the benchmark securities in the individual maturities form a "benchmark curve". This might indicate that all maturity segments are equally important. In practice, however, most of the issues in the euro area are in maturities of less than 0 years. Issuing and trading activity is concentrated particularly in the 2-, 5- and 0-year maturity segments. By tradition, German government bonds have been viewed as the benchmark securities for the European countries. Prior to the transition to the third stage of EMU there was some speculation as to whether other countries' government bonds might achieve the status of benchmark securities in a common yield curve for the euro area. The back-

46 4 SECURITIES IN THE EURO BENCHMARK CURVE Table 2 Maturity Reuters Bloomberg 2-year... Bund Schatz 3 per cent 0 BTAN 3 per cent 0 3-year... BOBL s2 4.5 per cent 02 BTAN 4.5 per cent 02 4-year... BOBL s2 4.5 per cent 03 OAT 6.75 per cent 03 5-year... BOBL s per cent 04 BTAN 3.5 per cent 04 6-year... OAT 7.75 per cent 05 OAT 7.5 per cent 05 7-year... OAT 6.5 per cent 06 OAT 7.25 per cent 06 8-year... OAT 5.5 per cent 07 OAT 5.5 per cent 07 9-year... BUND 5.25 per cent 08 BUND 4.25 per cent 08 0-year... BUND 4 per cent 09 BUND 4 per cent year... BUND 6 per cent 6 OAT 8.5 per cent 9 30-year... OAT 5.5 per cent 29 BUND 4.75 per cent 28 Source: Reuters and Bloomberg on 5 July 999. Note: German securities: Bund Schatz, BOBL, BUND; French securities: BTAN, OAT. ground included that certain countries had larger government bond series than Germany. However, the first half-year has shown that no unequivocal yield curve has arisen for benchmark securities in euro. This is due to such factors as the continued segmentation of government bond markets in the euro area. In the important 0-year maturity segment the relevant German government bond is the undisputed benchmark. With regard to other maturities the government securities of several different countries are named as benchmarks for the euro area. This appears from the euro benchmark curve, cf. Table 2, compiled by the two financial news agencies, Reuters and Bloomberg. Market participants also mention Italian government bonds as benchmark securities for the euro area at the short end of the maturity spectrum. This is related to the large series volumes, as well as to the fact that the credit risk is of less significance for the shorter maturities, cf. the next section. Despite the diverging views on which securities are included in the euro benchmark curve it is noteworthy that the curve includes government bonds from other countries besides Germany. However, only government bonds from the major euro-area member states have achieved benchmark status. Before January 999 the swap yield curve in euro was named by several market participants as the most obvious reference for price-fixing of euro-denominated bonds. The basis for this expectation was that the A swap is an agreement between two parties to exchange payment flows during the maturity of the swap. A simple example of an interest-rate swap is an agreement between two parties, A and B, to exchange payment flows based on respectively fixed and variable interest rates. For example, according to the agreement payments from B to A are subject to a fixed interest rate, while payments in the other direction are subject to a floating interest rate. The swap yield curve is based on the interest rates for the fixed interest payments for the various swap maturities.

47 42 swap yield curve is independent of a single issuer's credit rating and that price formation is comparatively predictable in view of the ample liquidity in the market for interest-rate swaps in euro. However, the euro swap yield curve has not achieved benchmark curve status. The background is the great volatility of swap interest rates as a result of the unrest on international markets in the wake of Russia's financial crisis in the autumn of 998. A bond which is included in the benchmark curve for the euro area does not necessarily have the lowest yield in the relevant maturity. The yield spreads between government securities are determined by several factors, of which credit risk and liquidity are the most important. The significance of credit risk and liquidity to yield differentials is described in further detail in the following sections. CREDIT RISK The credit risk is the risk of suffering a financial loss in the event of default by a counterparty on payment obligations. The credit risk associated with bonds relates to the issuer and is assumed to be relatively constant from year to year. The credit risk related to domestic government debt is normally considered to be very limited due to the central government's ability to levy taxes. By tradition another factor is the access to monetary financing of the debt. In practice, its significance is limited. EU member states are subject to Article 0 of the EU Treaty which prohibits borrowing from the central bank, and thereby monetary financing of debt. It follows that in principle it is a member state's willingness to pay its debt by levying taxes which determines the credit risk related to the EU member states. The responsibility for fiscal policy continues to rest with the individual euro-area member states. The responsibility for financing government expenditure, including the issue of government bonds, is thus a national concern. Article 03 of the EU Treaty, the "no bail-out clause", stipulates that each member state is liable for its own national debt. As a consequence, the assessment of the credit risk related to the government issuers is of importance to the yield spreads among the individual euro-area member states. The credit risk related to a bond issuer is often measured in terms of a rating, which is an overall assessment of the issuer's ability to meet payment obligations. All euro-area member states have a relatively high rating. Standard & Poor's has given four euro-area member states the top rating of AAA, which means that the credit risk associated with in-

48 43 YIELD DIFFERENTIALS TO GERMANY AND RATING Chart 3 Per cent year yield differentials 0-year yield differentials AA AA+ AAA AA AA+ AAA Portugal Italy Spain Finland Belgium Ireland Austria Netherlands France Source: Bloomberg, Standard & Poor's. Note: Yield differentials in July 999. "Generic" 5- and 0-year securities have been used, i.e. typically the most recent issue in a given maturity segment. As a result, the maturity is not necessarily identical to the maturity stated here. Furthermore, there may be considerable variation in the size of the issues. These two factors may explain e.g. Ireland's relatively wide yield differential in the 5-year maturity segment. It should also be noted that the remaining maturity of the German 5-year bond is 4½ years. The likely implication is that the yield differentials are overvalued. vestment in bonds issued by these member states is almost non-existent. Two member states are rated AA+ and four are rated AA. For comparison, Denmark is rated AA+ for foreign-currency denominated debt and AAA for krone-denominated debt. According to Standard & Poor's principles the former rating is most comparable with the ratings of euro-area member states. Chart 3 shows the relationship between rating and yield differential. The chart shows 5- and 0-year yield differentials to Germany and to a degree confirms the assumption that the yield differentials tend to increase in step with the credit risk. However, this does not apply in all cases. For example, Belgium is rated AA+ but has a higher yield than Spain, rated AA. Moreover, France and the Netherlands, also rated AAA, have yield differentials to Germany of 0-5 basis points. This may be attributed to the market participants' perception that Germany is nonetheless more creditworthy. However, this explanation probably holds true for only a small proportion of the yield differential. These examples show that other factors, primarily liquidity, also play a role. Cf. the description in Kristian Sparre Andersen and Anders Matzen, The Use of Ratings in the European Capital Markets, Danmarks Nationalbank, Monetary Review, 3rd quarter 998.

49 44 0-YEAR YIELD DIFFERENTIALS TO GERMANY Chart 4 Per cent May June July Italy Belgium Note: May-July 999. The chart also shows generally wider yield differentials for the 0-year maturity than for the 5-year maturity. This indicates that the market participants tend to attach greater importance to the credit risk for the longer maturities. Furthermore, market participants' perception of the credit risk associated with a given country may vary over time, even if the rating is unchanged. This is illustrated by an example from May 999. On 25 May the EU ministers of finance approved Italy's upward adjustment of its budget deficit for 999 from 2.0 per cent of GDP to 2.4 per cent of GDP, if the economic conditions in Italy make this appropriate. After this decision the euro weakened and Italy's yield differential to Germany widened. Chart 4 shows the 0-year yield differentials to Germany of Italy and Belgium. It appears that the differential to Germany widened in connection with the agreement on Italy's budget deficit. The differentials of other euro-area member states with a relatively low rating and a relatively high government debt also widened. In the chart Belgium is a representative example. The widening of the yield differentials for these countries can also be taken to indicate that for a time there was greater uncertainty concerning the future of EMU. The yield differentials subsequently narrowed to previous levels. To some extent the above confirms the expectation that credit risk would be of great significance to yield levels for bonds issued by gov-

50 45 ernment borrowers. It is also clear, however, that credit risk is not the only explanation for yield differentials among the individual countries. A high credit rating is thus not enough to warrant low yield differentials. Experience from the st half of 999 shows that liquidity is at least as important to yield levels. LIQUIDITY Liquidity expresses the degree of negotiability of the bonds. In contrast to credit risk, liquidity is to a high degree related to the individual bond series. Moreover, in the short term an issuer may also influence the liquidity of a given paper and a given market. Issuers do not have the same degree of influence on market participants' perception of credit risk. Bonds which are liquid are normally characterised by a large volume in circulation, a high trading volume and a small spread between bid and offer prices. Liquidity also depends on whether the bond market is generally perceived to be well-functioning and efficient. A more transparent issuing policy among government issuers may contribute to this perception. The efficiency of the government bond market also depends on the existence of well-functioning repo markets, and on whether the bonds can be delivered in a futures contract subject to high turnover. In general, investors are willing to pay a premium for a more liquid bond. The influence of government issuers on liquidity Government issuers can influence liquidity via size of series, choice of instrument, method of issuing, etc. The intensified competition among government issuers after the transition to the euro has therefore led to an adjustment of the government-debt strategy of most euro-area member states. In recent years building up large bond series has become an established element of the issuing strategy of government borrowers in Europe. This creates the right preconditions for ample liquidity and thereby lower financing costs. The commencement of the third stage of EMU has put greater focus on large series. In the st half of 999 this was demonstrated by a number of member states' buy-back of outstanding bonds in order to compress issues into fewer series. This particularly applied to smaller member states. A case in point is the Netherlands, which intends to reduce the number of outstanding series from 35 to between 2 and 5 by buying back existing loans.

51 46 OUTSTANDING 0-YEAR GOVERNMENT BONDS ON 5 JULY 999 Table 3 Euro million Portugal... 3,337 Finland... 5,087 Ireland... 5,574 Belgum... 9,22 Netherlands... 9,63 Spain...,26 France... 22,522 Italy 23,028 Germany )... 20,000 Source: Bloomberg, ) New 4.5-per-cent 2009 bond, opened on 6 July 999. Both Bloomberg and Reuters apply the German 4-per-cent 2009 as the 0-year benchmark for the euro area, Table 3 shows that the major euro-area member states have 0-year bonds with an outstanding volume of around 20 billion euro or more, while the small member states' outstanding 0-year bonds amount to less than 0 billion euro. A large outstanding volume does not necessarily imply a narrow yield differential since liquidity depends on other factors besides the size of the series. This appears from Chart 5 which shows the relationship between yield differentials to Germany and the size of the series of current 0-year government securities of the euro-area member states. With regard to choice of instruments, the euro-area member states almost exclusively tend to use fixed-interest bullet issues, primarily in the 0-year maturity segment. As a supplement to the fixed-interest bullet issues, all euro-area member states offer stripping, i.e. splitting up each bond into interest and instalment payments. With the exception of a French issue of a long-term index-linked bond and a few structured Italian loans, there have been no innovations with regard to types of loan. Innovation by government issuers has been dedicated to increased use of buy-backs and interest-rate swaps. The purpose of these measures is to concentrate borrowing on a few large issues in a situation with a small financing requirement and greater focus on risk in relation to government debt. The use of interest-rate swaps in particular offers an opportunity to concentrate issues on a few maturity segments without affecting the desired risk profile for the government debt. Buy-backs also make it possible to convert the debt to securities at coupon rates which conform to market levels. Auctions are the most frequently used method of issuing government securities. This method makes it possible to quickly build up a relatively

52 47 CURRENT 0-YEAR BENCHMARK AND AVERAGE YIELD DIFFERENTIAL Chart 5 Per cent ,000 0,000 5,000 20,000 25,000 AA AA+ AAA Euro million Note: July 999. Source: Bloomberg. large volume of outstanding bonds. Due to their limited financing requirements the small member states cannot compete with the liquidity of issues by the large member states. Several of these countries have resorted to syndicated bond issues rather than auctions, when opening new series. Syndicated issues are targeted more towards the endinvestor and enable the issuing countries to reach a wider group of investors. All euro-area member states apply a more or less formalised "primary-dealer" system on a varying scale. Primary dealers are financial institutions holding the exclusive right to buy bonds at issue. They are normally committed to creating liquidity and trading volume in specific government securities. In the run-up to EMU there was a general tendency to include primary dealers from other countries. Today, most countries thus use a mixture of national and international financial institutions as primary dealers. This reflects the wish to reach a wider group of investors, especially foreign investors. Finally, a number of government issuers offer bonds in open government securities series. The purpose of these schemes is to assure investors that there will always be a "lender of last resort" to prevent locking-in of the market. Such measures are normally assumed to contribute to the liquidity of the securities open for new issues and to enhance the efficiency of trading in government bonds.

53 48 NUMBER OF FUTURES CONTRACTS BASED ON 0-YEAR GOVERNMENT BONDS IN EURO Table 4,000 contracts End-July 998 End-July 999 EUREX MATIF LIFFE ) Source: Bloomberg. Note: The number is compiled as the total number of open positions in a given futures contract. ) Trading in the Bund future on LIFFE was suspended at the beginning of 999. Market-related factors of significance to liquidity Factors such as the distribution of investors, the markets for financial derivatives and the design of trading systems also affect the degree of liquidity. Issuers have limited scope to influence these factors. Special factors concerning the placement of outstanding securities may affect the liquidity of the market. For example, despite large issues, the French market for government securities is regarded as less liquid than the German market. One explanation given by market participants is that French government securities are held primarily by domestic investors. Non-residents thus own less than 20 per cent of the total outstanding volume of French government bonds, while the corresponding figure for Germany is around 50 per cent. Since foreign investors are generally more active traders of bonds the small proportion in foreign ownership indicates a relatively lower trading volume in the French market, and thereby poorer liquidity. The poorer liquidity helps to explain the yield spread between France and Germany of more than 0 basis points for the 0-year maturity. The existence of a highly-traded futures contract makes it easier for investors to hedge their positions and may thus contribute to the liquidity of the underlying government bonds. Prior to the transition to the euro there was intense competition among the derivative exchanges for the dominant position in the important futures contract based on 0-year government bonds. The competition primarily involved EUREX (previously Deutsche Termins Börse) of Germany and Switzerland, LIFFE of the UK and MATIF of France. At present EUREX is clearly the leader in the dominating futures contract, cf. Table 4. While at MATIF both French and German bonds can be delivered in the most important 0-year contract, at EUREX only German government bonds can be delivered in the 0-year contract. The development on the futures markets MATIF has also introduced a multi-issuer futures contract for 0-year maturities which can deliver bonds from all euro-area member states. However, the activity in this contract is very modest.

54 49 has thus supported the dominating role of the German 0-year government bond. The introduction of the euro has brought considerable changes with regard to trading in government bonds. There has been a tendency for the size of individual transactions to increase. According to market participants German government bonds are now traded in the same nominal denominations in euro as was previously the case for D-mark bonds, i.e. transaction sizes have doubled. However, there are no indications of an increase in total trading volume. Furthermore, an increasing degree of integration can be observed for the marketplaces for bond trading. This is for example shown by the establishment of a number of alliances between stock exchanges in EU member states, and by the establishment of joint trading systems for the largest bond series. Previously, government bonds were traded mainly in national trading systems. After the introduction in April 999 of "EuroMTS", an electronic trading system for the most liquid European government bonds with an outstanding volume exceeding 5 billion euro, the foundation has been established for concentrating trading of the large benchmark bonds in one system. EuroMTS is owned by a group of private banks which decides which bonds may be traded in the system. So far these are government bonds issued by Germany, France and Italy. The establishment of EuroMTS will probably support the grouping into large and small countries which has been observed since the introduction of the euro. SUMMARY The transition to the third stage of EMU reinforced the tendency for increased integration of the European bond markets. Experience from the st half-year after the transition can be summarised as follows: Despite the creation of preconditions for the establishment of a single bond market for the euro area, the market is still to some extent segmented according to national borders. No euro-area wide benchmark curve has been established. The issue of government bonds continues to be concentrated on the 2-, 5- and 0-year maturity segments. Yield differentials between government bonds in the individual countries continue to exist, but for some countries have narrowed considerably. The factors underlying the current yield differentials are differences in credit risk and liquidity. Competition among government issuers has intensified. The most important parameter of competition is a good level of liquidity in the

55 50 bonds issued. This is because even in the short term the liquidity can be affected by the government issuers. In contrast, the market's perception of credit risk is more difficult for issuers to influence. It is sought to achieve a good level of liquidity primarily by building up large bond series. This has led to a grouping of the euro-area member states into small and large countries. The trend towards greater integration of the bond market of the euro area is expected to continue in the coming years. Yield differentials among government bonds in the euro area will continue to exist in the future due to variations in credit risk and liquidity. Liquidity will be the most important parameter of competition. Together with the continuing trend towards greater integration of marketplaces and trading systems, this reinforces the grouping into large and small issuers. The small countries must thus be expected to resort to innovative use of instruments and issuing strategies.

56 5 Is Last Autumn's Financial Crisis Over? Leif Lybecker Eskesen, Financial Markets Department INTRODUCTION In the autumn of 998 the global financial markets were subject to strong tension. Yield differentials between government bonds and bonds with a higher credit and liquidity risk widened considerably, while stock markets fell strongly. The unrest also affected the Danish markets, thereby among other things causing a substantial widening of the yield differential between mortgage-credit and government bonds. This article takes a closer look at the financial markets during and after this turbulent period. On the basis of the development in market indicators of credit, liquidity and market risk it is evaluated whether the crisis on the financial markets is now over. All the indicators show that the strong tension in the market has dispersed and the actual market crisis therefore seems to be over. However, last autumn's unrest on the financial markets has not disappeared without trace. The measures of market risk and especially of credit and liquidity risk still stand higher than before the crisis in Russia. The background may be investors' greater understanding of the risks they are facing, a higher risk aversion and the still prevailing risks of negative shocks to the financial markets. THE UNREST ON THE GLOBAL FINANCIAL MARKETS In August 998 Russia suspended payments on its foreign debt and abandoned stabilisation of the rouble against the dollar. In the wake of the crisis in Russia came a turbulent period on the global financial markets. The crisis in several Asian countries and growing concern at the development in Russia during the summer of 998 caused many investors to resort to safer securities. This supported particularly the US and European government-bond markets. After the Russian crisis broke out in August demand for safe government bonds issued by the major industrialised countries soared, which led to a strong drop in international interest rates. During the 3rd quarter, 0-year yields in the USA, Japan and Germany thus fell by between 0.75 per cent and.0 per cent, cf. Chart.

57 52 THE CRISIS SPREADS ACROSS MARKETS Chart Per cent YEAR GOVERNMENT BOND YIELDS YIELD DIFFERENTIAL VIS-Á-VIS THE USA Percentage points Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 USA Germany Japan 0 Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 Emerging markets June '98 = SHARE PRICES Yen per dollar 50 FOREIGN EXCHANGE F Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 USA Germany Emerging markets 00 Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 USD/JPY Note: Daily observations. Yield differential: yield differential between the J.P. Morgan interest-rate index (EMBI+) for emerging markets and the US zero-coupon yield curve. Share prices: index for emerging markets based on IFC's capital-weighted index denominated in dollars. This drop in interest rates was accompanied by a significant widening of the yield differentials between major industrialised countries' government bonds and bonds with higher credit or liquidity risk. The widening of the yield differentials was especially pronounced for bond issues from the emerging markets (e.g. Brazil, Argentina, Thailand, Indonesia, China, Russia), cf. Chart, but was also significant for issues by industrialised countries. The yield differentials of EU member states vis-à-vis Germany thus widened. The widening was most pronounced for the member states which were not to participate in EMU, with the exception of the UK, however. The yield differential between mortgage-credit/corporate bonds and government bonds widened, as did the differential between e.g. corporate bond issues with varying credit quality. The stock markets were also affected, which offset the share-price increases earlier in 998. September and the beginning of October saw a further increase in the credit and liquidity risk. A significant underlying factor was the news of problems faced by the major US hedge fund, Long Term Capital Management (LTCM). LTCM had built up very substantial geared positions across several markets, financed by borrowing from a large number of US and European financial institutions. LTCM had speculated in a gen- Gearing is the result of (partly) financing purchases of e.g. securities by direct or indirect leveraging by means of financial derivatives (futures, options, etc.). This increases the potential return.

58 53 eral narrowing of yield differentials and was therefore severely affected by the significant widenings. Even though at the end of September the largest creditors announced a rescue of LTCM the market unrest increased, and yield differentials widened further. The greater unrest could be attributed partly to the prospect of unwinding of several of LTCM's spread trades, and partly to concern that other hedge funds/financial institutions with similar investment strategies might be in difficulties and would therefore be forced to close out their positions. The unrest in September and October moreover reflected investors' decision generally to reduce their exposure, in reaction to the growing uncertainty. This was e.g. illustrated by extensive unwinding of yen-financed positions ("yen-carry-trades"), which contributed to a significant strengthening of the yen at the beginning of October, cf. Chart. Another factor contributing to this general closing-out was the requirement for provision of collateral and margin payments 2, which likewise made it necessary for positions to be liquidated. As from October the lowering of interest rates by several central banks and important political measures in Japan vis-à-vis the banking sector and the economy contributed to gradual stabilisation of the financial markets. As a consequence yield differentials narrowed considerably, but are still higher than immediately prior to the crisis in Russia. On the stock markets too this stabilisation signified a reversal, and on several markets share prices have subsequently risen to record levels. Consideration of various market indicators can give an impression of the extent to which the market unrest has subsided and the crisis ended. IS THE CRISIS OVER? WHAT DO THE MARKET INDICATORS SAY? The crisis on the global financial markets in the autumn of 998 had four main characteristics: Investors generally re-assessed the credit risk on various assets. Liquidity was very low for certain periods, even on the largest financial markets. A strong rise in variations in market prices and in uncertainty of future trends. Willingness to assume risks was reduced. 2 In the "yen-carry-trades" financial institutions, hedge funds, and even certain business enterprises, borrowed in yen at a low rate of interest and then made placements in US, European and emerging-market assets. The low interest rate for yen and the weakening of the yen had made these positions profitable in preceding years. In general, there are two types of margin payment: initial margin and variation margin. The former is a deposit paid from the buyer to the seller of e.g. a futures contract, to hedge against losses. A variation margin is ongoing payments between buyer and seller on gains and losses on e.g. a futures contract.

59 54 Thus, as tension on the financial markets intensified the credit, liquidity and market risk increased in step with greater risk aversion. The development also showed that the three types of risk were interdependent. This exacerbated the market turbulence. In the following the development in the three types of risk during and after the period of turbulence is described in more detail. For that purpose various market indicators, which reflect whether market participants perceive the situation to have stabilised, are used. Credit-risk indicators The credit risk is the risk of loss on an investment due to the default of a counterparty or deterioration in that counterparty's ability to pay. The crisis in Russia and especially the country's suspension of debt payments questioned fundamental credit-quality assumptions. This led to general reconsideration of credit risk across the global financial markets. In conjunction with increased risk aversion this led to more stringent credit terms for all but the most creditworthy issuers. A credit spread is an indicator of credit risk and credit terms. The credit spread is the yield differential between issues with varying credit quality, but typically the same maturity. The wider this credit spread, the greater the market's perception of the difference in credit risk between two issues, all other things being equal. Chart 2 shows the development in various credit spreads. The more stringent credit terms affected both emerging markets and western financial markets. In the western markets the concern focused to a great extent on the financial sector. This is shown by the relatively more pronounced widening of the credit spreads for this sector. The concern regarding the financial sector was also reflected in widening spreads for interest-rate swaps 2, which are often concluded by banks and thereby indicate the market's evaluation of the credit risk on financial enterprises. 3 The reduced confidence in the financial sector was furthermore indicated by a wider yield differential between the 3-month eurodollar 4 and 3-month T-bill rates (the Ted spread). The Danish markets were also However, part of the widening of the spread can also be attributed to a preference for the more liquid benchmark government bonds, cf. below. A swap is an agreement between two parties to exchange future payment flows. A simple example of an interest-rate swap is an agreement between two parties, A and B, to exchange payment flows based on respectively fixed and variable interest rates. For example, according to the agreement payments from B to A are subject to a fixed interest rate, while payments from A to B are subject to a floating interest rate. This enables e.g. a bank to convert a floating-interest loan to a fixed-interest loan. The swap spread is the yield spread between the resulting fixed-interest loan and a fixed-interest bond with the same maturity. The widening of the German swap spread in August and September not only reflected a growing credit risk, but also a period of greater preference for swaps rather than futures to hedge long positions. A eurodollar is a US dollar deposited with a bank outside the USA, typically in Europe.

60 55 CREDIT-RISK INDICATORS Chart 2 YIELD DIFFERENTIAL VIS-Á-VIS THE USA Percentage points SWAP SPREAD AND "TED SPREAD" Percentage points Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 Russia Brazil South Korea USA Germany Japan Ted spread YIELD DIFFERENTIAL VIS-Á-VIS Percentage points GOVERMENT BONDS 2,80 2,40 2,00,60,20 0,80 0,40 0,00 Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 Financial enterprise rated A2 Financial enterprise rated BBB2 Industrial enterprise rated A2 Industrial enterprise rated BBB2 DANISH GOV-MC BOND SPREAD Percentage points Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 Option-adjusted spread for Danish mortgage-credit bond Note: Daily observations for swap spreads, "Ted spread" and Danish Gov-MC bond spread, weekly observations for the other series. Yield differential to the USA: the yield differential between the Russian/Brazilian/Korean 0-year government bonds denominated in dollars and the US 0-year government bond. Swap spreads: the yield differential between 0-year interest-rate swaps and 0-year government bonds in the respective countries, cf. the footnote. Ted spread: the yield differential between the 3-month eurodollar and the 3-month US government paper (T bills). Yield differential to government bond: the yield differential between 0-year issues with different ratings from US financial and industrial enterprises and 0-year US government bond. Danish Gov-MC bond spread: yield differential between a Danish 30-year mortgage-credit bond and the Danish 0-year government bond. affected by the unrest and the yield differential between e.g. mortgage-credit and government bonds widened. This widening was also related to the LTCM crisis, since LTCM held positions in Danish mortgage-credit bonds. Credit spreads have generally narrowed since October. The narrowing has been most pronounced for issues with a higher credit quality. However, most spreads are still above the level before the outbreak of the Russian crisis. It is still too early to say whether the credit spreads will return to the pre-crisis levels, or whether the higher spreads reflect a new level. Nevertheless, the credit spreads traded up to the outbreak of the crises in both Southeast Asia and Russia were very narrow in historical terms. The current higher levels may therefore reflect a certain normalisation of credit spreads. Liquidity-risk indicators All other things being equal a high degree of liquidity is an advantage to an investor since this makes it easier to build up or close out a position without excessive costs. High liquidity and thereby a low liquidity

61 56 RISK MANAGEMENT MODELS AND REBALANCING OF PORTFOLIOS Box Most financial enterprises, including hedge funds, apply risk management models to protect against losses. The most well-known are Value-at-Risk (VaR) models. They apply a number of assumptions to estimate the loss to the enterprise, with a given probability. A financial enterprise's VaR may for example indicate that there is maximum 2 per cent probability that the loss on the portfolio will exceed kr. 50 million in the coming week. If the enterprise finds that the potential loss is too high, it will rebalance its portfolio, including the liquidation of positions, to reduce the VaR measure. As a result of last autumn's turbulence a number of financial enterprises (e.g. LTCM) suffered far higher losses than expected, since the applied assumptions, which are typically incorporated in e.g. VaR models, did not hold. Extreme price fluctuations proved to be more probable than assumed, and many markets took a parallel course, thereby reducing the expected diversification gain from holding positions across markets. The generally higher market volatility and e.g. the parallel market movements were also reflected in higher VaR and gave rise to liquidation of positions in order to keep VaR within the desired limits. risk thus increase the value of an asset. Investors will therefore be willing to pay a premium (liquidity premium) for a highly-liquid asset. There is no unequivocal definition of the concept of liquidity since it encompasses several aspects: Depth: the extent to which the price is affected by major purchases and sales. Width: the difference between bid and offer prices and other transaction costs. Immediacy: how quickly a market participant can effect the decision to trade and find a counterparty. Resilience: how quickly the market will restore equilibrium after a temporary shift in supply or demand. Against this background a high degree of liquidity and thus a low liquidity risk can be defined by great depth, small width, a high degree of immediacy and resilience. At the peak of the financial unrest in September/October even the largest markets were characterised by very low liquidity. The liquidity pressure on the markets which arose in autumn 998 was a consequence of many investors' simultaneous unwinding of positions typically with a high gearing. The unwinding should be viewed in the light of ) a general rebalancing of portfolios, cf. Box, including reduc- Cf. Danmarks Nationalbank, Danish Government Borrowing and Debt 997 (in Danish), pp. 43ff.

62 57 LIQUIDITY-RISK INDICATORS Chart 3 YIELD DIFFERENTIAL VIS-Á-VIS BENCHMARK Percentage points YIELD DIFFERENTIAL VIS-À-VIS GERMANY Percentage points Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 Off-the-runs Kreditanstalt für Wiederaufbau France Austria Note: Yield differential vis-à-vis benchmark: ) yield differential between (Off-the-runs) 29- and (On-the-runs) 30-year US government bonds; 2) yield differential between 0-year bond issued by Kreditanstalt für Wiederaufbau denominated in euro and 0-year German government bond. Yield differential vis-à-vis Germany: yield differential between a 0-year French/Austrian government bond and the 0-year German government bond. tion of the gearing of these positions in the light of the increasing risks, 2) rising margin payments and payment of collateral, and 3) reduction of losses on spread trades and placements in emerging markets. The liquidity risk or liquidity pressure can be expressed by the liquidity spread, i.e. the yield differential between liquid and less liquid issues with the same credit quality and maturity. All other things being equal, an increase in the yield differential and thus a higher liquidity spread reflects that the market is willing to pay more (a higher liquidity premium) to own the more liquid of the two issues. Chart 3 shows the development in various liquidity spreads. As the chart shows, the liquidity pressure e.g. led to a widening of the yield spread between "on-the-runs" and "off-the-runs," i.e. series from the same issuer which are respectively open and closed to new issues. Similarly, the higher liquidity premium entailed a widening of the yield differential between e.g. the 0-year German government bond and the 0-year bond issued by Kreditanstalt für Wiederaufbau, which is a state-guaranteed issue in considerably smaller volumes. The larger liquidity premium also resulted in the widening of spreads between the 0-year German government bond and less liquid issues by other EMU member states with equivalent maturity and credit quality. The liquidity premium has decreased since the autumn, but is still higher than in July 998. This might indicate that the market participants focus more on the liquidity risk and that they have realised that even in the largest financial markets liquidity may vary. Market-risk indicators The market risk is the risk of losing money as a consequence of fluctuations in asset prices. The strong market turbulence after Russia's de facto Open series are traded more actively, and are thus more liquid.

63 58 devaluation and suspension of debt payments seriously aggravated this risk. In order to monitor the development in market risk it is relevant to consider the development in the actual and implicit volatility. The actual volatility is a measure of the actual fluctuations in e.g. the price of a given asset. The implicit volatility is a measure of market participants' expectations of future fluctuations and is derived on the basis of option premiums, cf. Box 2. An increase in the actual volatility indicates that the market risk is now higher than before, whereas an increase in the implicit volatility indicates a rise in the expected future market risk. The difference between actual and implicit volatility can thus be regarded as an indicator of the expected development in the market risk from the present levels. An indication of the expected development in the market risk can also be given by an implicit-volatility forward curve. The forward curve is drawn by deriving the implicit volatility from option premiums on a given date for options with varying maturities, cf. Box 2. Implicit volatility which declines with the maturity of the option yields a downward sloping forward curve, reflecting market expectations of a decrease in the market risk. The left-hand column of Chart 4 presents graphs showing the development in implicit volatility on the bond, stock and foreign-exchange markets. The right-hand column presents graphs which compare the historical development in actual and implicit volatility. Forward curves for implicit volatility are also drawn. The forward curves show the market's perception at the beginning of the 3rd, 6th, 9th and 2th month of the development in volatility over the next six months, i.e. the implicit volatility six months ahead. The graphs in both the left-hand and right-hand columns show that both actual and implicit volatility rose strongly from August to mid-october. At the same time, the two volatility measures appeared to rise in tandem. The exception is USD/JPY, for which the increase in implicit volatility occurred later than the increase in actual volatility. Since the implicit volatility is a forward-looking indicator this suggests that the turbulence caught the market by surprise. The forward curves for implicit volatility likewise indicate that the development was unexpected. The flat forward curves starting as from the beginning of June 998 thus reflect a market expectation at that time of an unchanged market risk in the period up to December 998. Since September/October the actual and the implicit volatility have fallen considerably and are now close to the levels before the crisis. The There are several other indicators besides those described here, cf. Bank of England, Quarterly Bulletin, February 999.

64 59 OPTION PREMIUM AND IMPLICIT VOLATILITY Box 2 The implicit volatility can be derived from the option price. Basically, there are two types of options, a call option and a put option. A call option gives the buyer of the option the right, but not the obligation, to purchase the underlying asset at a given price, the strike price, in return for an option premium to the issuer (writer) of the option. The writer is obliged to sell the underlying asset at the agreed price, should the owner wish to exercise the option. A put option gives the buyer of the option the right, but not the obligation, to sell the underlying asset at a given price. Should the price of the underlying asset rise above the strike price, the buyer of a call option may benefit from exercising the option. Similarly, should the price of the underlying asset fall below the strike price, the buyer of the put option may benefit from exercising the option. A wide range of different models are used to determine the option premium. The most popular models are variants of the Black-Scholes model. In a typical model the premium for e.g. a foreign-exchange option is determined by five factors: ) the current exchange rate, 2) the interest rates in the two currencies, 3) the maturity of the option, 4) the strike price and 5) the expected volatility of the underlying exchange rate. Of the five variables only the volatility cannot be observed directly. For a given expected volatility the option premium can thus be derived. Similarly, the implicit volatility can be calculated if the market price of the option is known. Technically, the implicit volatility is the market participants' expectation of the future standard deviation in the percentage changes in the price of the underlying asset. PROBABILITY DISTRIBUTION AND PAYMENT FLOWS ON EXPIRY FOR TWO CALL OPTIONS ON AN UNDERLYING ASSET WITH VARYING PRICE VOLATILITY Payment on expiry 0,89 0,79 0,69 0,59 0,49 0,39 0,29 0,9 0,09-0,0 Strike price Probability 0,90 0,80 0,70 0,60 0,50 0,40 0,30 0,20 0,0 0,00 Price of unerlying asset Pay-off on option a and b Probability distribution for b Probability distribution for a Note: It is assumed that the two options are based on an underlying asset with the same price and the same strike price. In addition, it is assumed that when bought the options have no positive or negative value, i.e. they are at-the-money. All other things being equal a higher expected volatility will increase the option's value and thus the option premium. The reason is that a higher expected volatility increases the probability of a higher gain on the option. This is illustrated in the chart, which shows the expected probability distribution of prices for the two underlying assets, a and b, and the pay-off curve for a call option on the two assets. The probability that the option yields a gain or is worthless is 50 per cent for both assets. However, the expected price volatility is higher for b than for a, which means that there is greater probability that a call option on asset b will give a higher pay-off than a call option on asset a. It follows that the option premium for b will be higher than for a. The high option premium thus implies expectations of high volatility, which also explains why options are often quoted at implicit volatility. ) Black, F. & Scholes, M. (973), The pricing of options and corporate liabilities, Journal of Political Economy, 8, p

65 60 MARKET-RISK INDICATORS Chart 4 Per cent 2 IMPLICIT VOLATILITY IN THE BOND MARKET OPTIONS TRADED ON FUTURES Per cent 2 ACTUAL VERSUS IMPLICIT VOLATILITY: 0-YEAR US INTEREST RATE Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 USA 0-year German 0-year Japanese 0-year 0 Jun. 98 Dec. 98 Jun. 99 Dec. 99 Implicit Actual Forward curves Per cent 60 IMPLICIT VOLATILITY IN THE STOCK MARKET OPTIONS TRADED ON INDICES Per cent 50 ACTUAL VERSUS IMPLICIT VOLATILITY: S&P Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 S&P 500 DAX Nikkei Jun. 98 Dec. 98 Jun. 99 Dec. 99 Implicit Actual Forward curves Per cent 30 IMPLICIT VOLATILITY IN FOREIGN-EXCHANGE OPTIONS TRADED ON FUTURES Per cent 40 ACTUAL VERSUS IMPLICIT VOLATILITY: USD/JPY Jun. 98 Sep. 98 Dec. 98 Mar. 99 Jun. 99 USD/JPY USD/DEM GBP/USD USD/MXP 0 Jun. 98 Dec. 98 Jun. 99 Dec. 99 Implicit Actual Forward curves Note: Implicit volatility: derived from stock-exchange-traded options on futures in 0-year government bonds and currency crosses. For stock markets derived from stock-exchange-traded options on stock indices. Actual volatility: calculated as the -month volatility. It corresponds to the annualised standard deviation for the daily percentage changes during one month. Forward curves: determined by on a given day at the beginning of the 3rd, 6th, 9th and 2th month deriving the implicit volatility from options on futures with a remaining maturity of up to six months. rapid decline in the implicit volatility reflects that despite the unrest the market expected that the situation would soon normalise, which it did. Moreover, to a degree the two volatility measures appear to have stabilised around the lower levels. The stabilisation of the market risk is also confirmed by a general levelling-out of the forward curves, which indicates that the market expects that the lower volatility levels will be maintained in the 2nd half of 999. According to the market participants the market risk on the financial markets is close to the level before the crisis. Last autumn's turbulence therefore does not appear to have caused any significant change in the investors' perception of market risks.

66 6 CONCLUSION The markets have stabilised since the global financial crisis peaked in October 998. Indicators of credit, liquidity and market risk thus all show that the strong tension in the market has dissolved and the actual market crisis is over. Nevertheless, last autumn's unrest on financial markets has left its mark. The volatility measures, and especially the credit and liquidity spreads, are still higher than before the crisis in Russia. The background may be increased understanding among market participants of the risks they are facing, as well as greater risk aversion and the still prevailing risks of negative shocks to the financial markets.

67

68 63 Can Inflation Expectations Be Deduced From the Development in Danish Consumption? Michael Pedersen, Economics Department INFLATION EXPECTATIONS The real rate of interest indicates borrowing costs (or the return on an investment in financial assets), disregarding price increases. The expected real interest rate is important to investments, consumption and savings, and is therefore significant in the evaluation of economic prospects. The problem is that since the inflation expectations of investors and borrowers are not known, it cannot be observed directly. Inflation expectations can be estimated in several ways. The actual rate of inflation can be used as an approximation of the expected inflation rate, but since a period of high inflation is often followed by a period of low inflation, the actual inflation rate may be a poor measure of expected inflation. In countries whose central bank has an inflation target, the latter may be the basis for inflation expectations provided, however, that the central bank enjoys absolute credibility. Another possibility is to employ various prognoses as the basis for expectations. These can be prognoses by national authorities or international organisations such as the OECD or the IMF. In addition, a large number of private banks prepare prognoses and an average of these can be taken to express expectations. This method assumes that, on average, borrowers and investors concur with the prognoses' future expectations. The yield on index-linked bonds can be used as an approximation of the real interest rate since this yield is adjusted for inflation. A simple expression of inflation expectations can be derived by subtracting the yield on index-linked bonds from a nominal yield. However, this method entails some elements of uncertainty, due to the special characteristics and taxation terms of index-linked bonds. Questionnaire surveys are another method of charting inflation expectations. A representative selection of households or business enter- See Topp (996) concerning the method applied using Danish data. See Andersen & Gyntelberg (999) for a description of the special characteristics of Danish index-linked bonds.

69 64 prises are asked about their expectations of inflation, whereby an overall picture is achieved. Statistics Denmark has carried out monthly questionnaire surveys for several years. An article in a previous Monetary Review presents a method of translating the households' responses into a measure of inflation expectations. This article presents a measure of inflation expectations in Denmark based on a theoretical economic model, the Consumption-based Capital Asset Pricing Model (CCAPM). According to this model, at the equilibrium point the real interest rate is determined on the basis of growth in private consumption. The inflation expectations found deviate considerably from other results and the conclusion is that the underlying assumptions are so restrictive that the method cannot be applied to the Danish economy without difficulty. Similar surveys for the USA, Germany and the UK appear to yield more plausible results. An obvious conclusion is therefore that the model is more suitable for larger, more closed economies in which domestic factors play a greater role in determining interest rates than is the case in Denmark 2. THE CCAPM MODEL CCAPM is a general equilibrium model, i.e. all markets are assumed to be in equilibrium: supply equals demand via price formation and interest rates. The model's starting point is a representative consumer who is also an investor and who is assumed to act rationally. Within that consumer's budget he seeks to achieve the best possible situation for himself, i.e. to maximise utility. Subject to these assumptions, the representative consumer will distribute his consumption over time, so that the growth in consumption corresponds to the real interest rate, taking into account time preferences, risk aversion and other consumer-specific factors, cf. Box. Technically, this model makes it possible to determine a real interest rate as well as inflation expectations on the basis of the development in consumption per capita and the nominal interest rate. The technical details of the model are described in Appendix A 3. The expected real return on a nominal bond depends on the expected inflation during the period of ownership of the bond. The actual inflation rate will be uncertain, so the investor will require a premium on the yield an inflation-risk premium. 2 3 See Christensen (996). See Ireland (996) and an internal working paper of 997 from the European Monetary Institute. In Ayuso & López-Salido (996, 997) the model is e.g. applied to Spanish data. It should be noted that the model considered here does not take taxation aspects into account. Introduction of taxation may therefore alter the results. Ayuso & López-Salido (996) discusses the introduction of a proportional income tax in the model.

70 65 THE RELATION BETWEEN CONSUMPTION AND INTEREST RATE Box CCAPM assumes a very simple world with only one product. The return on an asset accruing real interest will consist of a number of units of this product. So the price level has no effect on the return. In this simple world the real interest rate will depend on how much the consumer demands for postponing his consumption. This can be illustrated by a simple example. A consumer has 00 grains of wheat. He has two options: he can eat the grains (consumption), or lend them to a farmer in return for interest (investment). Should he choose to lend the grains, he will receive 05 grains of wheat a year later. The return of five grains is the consumer's real interest. So the consumer must consider whether he wants to postpone his consumption for one year and then be able to eat five grains more in the following year. This shows a relation between the development in consumption and the real interest rate, which is formalised in CCAPM, while taking consumer-specific factors into account. Our society has passed from a barter economy to the use of money. In the aforementioned situation this implies that the consumer will give the farmer money instead of grain. The yield paid to the consumer is also money. If the return is a fixed amount, e.g. kr. 5, the number of grains the consumer can buy with this return will depend on the development in the grain price during the investment period. If the consumer wants another product, the real return will depend on the price of the product chosen. In principle, a large number of real interest rates are created. The real interest rate is often measured against an average consumer's choice of products, so that the price development is measured by the consumer-price index. The size of the inflation-risk premium depends on e.g. the maturity of the bond. The longer the maturity, the greater the uncertainty concerning inflation. Moreover, the investor's risk aversion, i.e. the importance the investor attaches to avoiding risk, also plays a role. The size of the risk premium also depends on the stability of the inflation rates. The risk premium can be viewed as a premium on the expected real interest rate and inflation rate, and thus increases the nominal interest rate. This gives the following relation: R r R is the nominal interest rate, r is the real interest rate, e is the expected inflation and RP is the risk premium. In CCAPM the real interest rate is determined on the basis of the growth in private consumption: e RP r f ( c) c is the change in consumption. Since the nominal interest rate can be observed, an expression of inflation expectations and risk premium can be calculated.

71 66 EMPIRICAL APPLICATION OF THE THEORETICAL MODEL Application of the model requires consumption data and a nominal interest rate. The nominal interest rate applied is the yield on a 5-year government bond, which means that a 5-year real interest rate and inflation expectations over 5 years are derived. The consumption data is the data for non-durable goods and services from the national accounts. This consumption concept is often used in relevant literature because the important aspect of the model is the underlying consumption growth. Consumption per capita is found by dividing the value by the population aged between 5 and 74. To illustrate how the model functions, a number of parameters to describe e.g. the consumer's risk aversion must be selected. The parameters chosen here give the greatest possible correlation between the generated real interest rate and a 5-year realised real interest rate calculated as the nominal 5-year interest rate less actual inflation for the duration of the bond. It is also taken into account that the averages of the two interest rates are at the same level. The choice of parameters indicates an almost risk-neutral consumer, i.e. the consumer is not particularly hesitant to assume risks. Charts and 2 show the generated inflation expectations and real interest rates. In Chart the actual inflation is also drawn. The trend for the generated inflation expectation generally matches the actual inflation. The averages for the period considered are by and large at the same level, but the expectation's volatility is considerably greater than for the actual inflation. It is also noteworthy that inflation expectations and actual inflation are only at the same level at the beginning of the period, while there are considerable deviations for by far the greatest part of the period. The deviations from actual inflation, as well as from other measures of inflation expectations, appear too large to be plausible. In particular, the low level in recent years is difficult to interpret. Chart 2 shows that the nominal interest rate followed a generally falling trend during the period under consideration. On the other hand, there were no major fluctuations in the generated real interest rate. Overall, the model implies that there is no close relation between the nominal and the real interest rates. Furthermore, the calibration of the model results in a relatively low inflation-risk premium. For the period under consideration it is generated at between -0.2 and 0.4 percentage points. The calibration of the model is described in Appendix B.

72 67 5-YEAR INFLATION EXPECTATIONS AND ACTUAL INFLATION Chart Per cent (year/year) Inflation expectations Actual inflation Note: The inflation expectation stated here is the mean of the generated range. Fluctuations in the calculated real interest rate and risk premium were relatively limited. On the other hand, according to the model changes in the nominal interest rate must be attributed first and foremost to changes in inflation expectations. 5-YEAR INTEREST RATES Chart 2 Per cent p.a Generated real interest rate Nominal interest rate

73 68 COMMENTS ON THE RESULTS The results concerning inflation expectations, etc. derived from the model naturally depend on the parameters and calibration method chosen. The purpose here was not to state estimates for all the parameters in the model, but to illustrate a specific example of the derivation of inflation expectation and real interest rate. Without becoming too absorbed in the academic discussion it can be stated that major fluctuations in inflation expectations are a characteristic feature, irrespective of the choice of utility function, although the level varies considerably. It can also be discussed whether the relevant expression of the nominal interest rate is to be compiled as the amount after tax at the consumer's disposal. In the first instance this will not affect the generated real interest rate, but will solely reduce the generated inflation expectations. It gives no immediate improvement in the applicability of the model to a description of the Danish economy. As stated, CCAPM is a very simple representation of reality and the question must be asked of whether the model provides a satisfactory description of an economy such as Denmark's. At first sight the calculated inflation expectation seems far removed from the inflation estimates of e.g. the OECD, the IMF and other prognosis brokers. Their estimates were relatively stable in the period considered. The same picture is painted by other methods to derive inflation expectations, such as questionnaire surveys. Use of the aforementioned model gives a positive relation between the increase in consumption and the real interest rate, cf. Chart 3. This can be attributed to the type of consumer chosen, but also to the model's correlation between real interest rate and development in consumption. Since strong consumption growth indicates a low savings ratio, a negative correlation is also established between the real interest rate and the propensity to save, cf. Chart 4. This seems to be a contra-intuitive effect. In principle, it could be defended as an adjustment of the interest rate to reduced savings. This effect from consumption and savings on the interest rate is conceivable in a large economy or A sensitivity analysis considers three different types of consumer. One with a standard utility function, i.e. the utility achieved depends only on the consumer's own consumption in the current period. The second is a habitual consumer where there is a connection between yesterday's and today's consumption. The third is a consumer for whom the utility also depends on the general level of consumption in society as a whole. One can be envious or pleased about the success of others. The sensitivity of the model to the choice of utility function and parameters is shown in Tables and 2 of Appendix C. It can be said that on its own merits the model has the expected characteristics. The more impatient and averse to risk the consumer, the higher the real interest rate he will demand. The more stability consumers want in their consumption, the higher the real interest rate, and the more the consumers are pleased about the success of others, the lower the real interest rate they will demand. Reference is made to Ayuso & López-Salido (996, 997) for further discussion of the model's sensitivity to choice of utility function.

74 69 CONSUMPTION GROWTH AND GENERATED REAL INTEREST RATE (5 YEARS) Chart 3 Per cent p.a. 6.0 Per cent (year/year) Real interest rate (left-hand axis) Consumption growth (right-hand axis) currency area. However, this interpretation does not match Danish data, where the interest rate to a high degree is determined by external factors. It is conceivable that lower savings increase inflation and reduce the real interest rate at an externally-determined nominal interest rate. However, this is not incorporated in the model, which considers equilibria for consumption and capital market at an exogenous inflation rate. SAVINGS RATIO AND GENERATED REAL INTEREST RATE Chart 4 Per cent p.a. 6.0 Per cent of disposable income Real interest rate (left-hand axis) Private savings ratio (right-hand axis)

75 70 As stated, in an open economy such as Denmark's with a fixedexchange-rate policy the nominal interest rate is to a great degree determined by external factors. In this situation the propensity to save can barely affect the nominal interest rate. This means that there is no adjustment of the interest rate to consumption behaviour, but solely an adjustment of consumption behaviour to the interest rate. In the short term, this adjustment to a higher interest rate reflects the consumers' desire to increase their savings ratio. Once the savings ratio has been established at a higher level an equilibrium has been reached with less consumption today and more consumption in the future. This corresponds to an increase in the planned consumption growth. A high real interest rate and strong consumption growth will thus be related in the long term, even with an exogenous interest rate. Against this background the CCAPM relation between interest rate and consumption growth can probably survive since CCAPM is an equilibrium model. Unfortunately, the necessary adjustment takes up both time and significance. The economy is characterised by cyclical fluctuations which often seem to reflect an adjustment to changes in interest rates, cf. the drop in interest rates prior to the upswing in the 980s and 990s. It follows that the dominating pattern is the negative relation between consumption growth and the real interest rate as perceived by consumers. This impedes or perhaps even rules out, an empirical application since account must be taken of persistent, dominant cyclical fluctuations in consumption growth and savings. Besides the influence on interest-rate formation, another effect of the open economies in reality is that consumption can be financed abroad. This leads to a current-account deficit, which the model does not take into account. This problem is probably also relevant on calibration of the model for other countries such as the USA, whose current-account deficit is currently approximately 3 per cent of GDP.

76 7 APPENDIX A We consider a CCAPM which describes an exclusive barter economy with no explicit formulation of the production technology. It is assumed that the output cannot be stored certain authors call it a fruit-tree model. In addition, the model is based on the following assumptions: All agents are identical with an infinite lifetime. This means that only one representative agent (consumer or household) is considered. The agents have rational expectations and wish to maximise the expected utility within their budget constraints. Business enterprises in the model are identical, operating in full competition. This means that they can be represented by one single enterprise. It manufactures one product with a very short lifetime, so it cannot be stored. The entire output is consumed within the same period. Output is given exogenously. There are two types of bond in this economy: a nominal bond and a real bond. The nominal bond has a price of kr. at the time t and yields (+R t ) kroner in the following period. The real bond costs consumption unit at the time t and yields (+r t ) units at the time t+. R t and r t are the nominal and real interest rates respectively. The bonds are perfectly divisible. The markets are perfect and thus without friction no transaction costs. The model is formulated in discreet time t=0,,...,. The timing is stated in subscript. The remaining notation in the model is as follows: y t is the consumer's income, P t is the price of the product, B t is the consumer's holdings of nominal bonds, b t is the consumer's holdings of real bonds, and U(c t ) is the consumer's utility function, which is dependent on consumption c t. is the consumer's subjective discounting factor, which is assumed to be constant over time. Since the output cannot be stored, it will all go to consumption: in other words: consumption is determined by output. In each period the consumer must decide how many (nominal and real) bonds he will buy. At the time t the consumer therefore faces the following problem: u. b. y t ( r t max B, b t t ) b t j Et U ( c j0 ( Rt ) B P t t j t ) c t b t Bt P t (A)

77 72 E t [] is the mathematical expectation. (A) is a dynamic programming problem which can be solved by means of Bellmann's principle of optimality. The first-order conditions give the following relations: E t t MRS ( r ) t t t t t MRS ( R ) (A2) E (A3) MRS t+ is the marginal rate of substitution between consumption in the period t and t+, while t+ is the relationship between the price level in the period t and t+: MRS U '( c t t, U '( ct ) ) t P t P t (A4) (A2) is the real interest rate generated by the model. At the equilibrium point it equals the reciprocal of the marginal rate of substitution. The following relation is achieved by using (A2) and (A3) : Cov MRS Et t t t, t Rt rt (A5) Cov t [,] is the covariance. By using the statistical definition of the correlation coefficient, which is between and, and using the approximation /E t [/ t+ ]E t [ t+ ], it is now possible to delimit inflation expectations: ( R ) E t ( MRSt tmrst Et[ R ) E MRS MRS t t t t t t t ] (A6) t () is the standard deviation. The conversion to (A6) includes the assumption that the volatility of the (reciprocal) inflation rate is limited: E t t t t t E t t t (A7) In empirical terms this assumption is rather conservative. The fraction is far below for the entire period under consideration. Knowing the marginal rate of substitution for consumption the model can now be used to generate a real interest rate, cf. (A2), and to estab- The conversion takes into account that E[ab]=cov[a,b]+E[a]E[b], where a and b are stochastic variables.

78 73 lish limits for inflation expectations, cf. (A6). The variables to be used in the model are consumption per capita and a nominal interest rate. Data is available for both of these variables. Consumer preferences Empirical tests of CCAPM yield mixed results, depending on the specification of preferences and on the country on which the model is tested. Relevant literature often uses the concept of an isoelastic utility function with a constant relative risk aversion. Besides this type of preference other types are considered here, i.e. a consumer forming habits and a consumer whose individual consumption is determined in consideration of total consumption in the economy. Constant relative risk aversion The most frequently used utility function in the literature is an isoelastic function with a constant relative risk aversion: ct U ( ct ), (A8) For = the utility is logarithmic in consumption. is the coefficient of the relative risk aversion, which is assumed to be constant over time. One of the properties of this utility function is that the coefficient of the relative risk aversion equals the reciprocal inter-temporal elasticity of substitution for consumption. With this type of utility function the marginal rate of substitution is as follows: MRS t t g t, g t (A9) ct Habit formation In the isoelastic utility function consumption is independent of time. There is often a certain relation between today's and yesterday's consumption, i.e. a certain inertia in the development of consumption. This can be captured in a utility function in various ways. A simple method, inspired by e.g. Heaton (995), is to subtract a weighted measure of yesterday's consumption from today's consumption: c U ( c c t t c, t ) (A0) is the degree of habit formation where the requirement is <min{g t } for the function to be well-defined. Furthermore, it is assumed that >0,

79 74 which introduces habit formation into the utility function. It is noteworthy that the relative risk aversion is not constant in (A0), as it is a function of the absolute risk aversion, the degree of habit formation and the consumption growth rate g t. Given this utility function the marginal substitution rate can be calculated as follows: ( gt gt gt ) ( g gt ) ( g t ) MRS t (A) ( g ) g g ) t t t ( Relative consumption The last type of preference to be considered here is the assumption that the consumer considers not only his own consumption, but also that of his neighbour. This can be captured in a utility function in a simple way by following Galí (994), where the aggregate consumption is part of the individual utility function: t ct U ( ct, Ct ) Ct, (A2) C t- is the total consumption in the economy in the preceding period, while c t is the individual consumption. is the extent to which the utility to the consumer depends on total consumption in the most recent period. If is positive, the consumer is pleased at the success of others (altruism). If is negative, on the other hand, the consumer is envious of the consumption of others. In this model using one single representative consumer, at equilibrium individual consumption and consumption in society will coincide. The marginal rate of substitution can thus be derived as follows: MRS g gt (A3) t t

80 75 APPENDIX B Expected values and standard deviation for the marginal rate of substitution must be used in the calibration of the model. Ireland (996) is adhered to here, and an auto-regressive process is assumed for the rate of substitution: MRSi MRS i ( L) i (B) is a constant, (L)= 0 + L+ 2 L k L k is a polynomial in the lag operator L, and i+ is a noise element which meets the conditions E[ i+ ]=0, [ i+ ]=, E[ i+ i-j ]=0 and E[ i+ MRS i-j ]=0 for all i,j=0,,2,. By estimating (B) it is possible to establish expected values and standard deviation for MRS. On the basis of the relatively limited amount of data a lag length of one is applied to the estimation of (B). Quarterly data is used for the period st quarter 977 to 4th quarter 998. Since a 5-year nominal interest rate is used, each period in the model is 5 years. A 5-year real interest rate and inflation expectations over 5 years are thus generated. Initial choice of parameters The first parameter chosen in the model is the subjective discounting factor. Here two values of are considered: 0.9 and 0.7. The first value represents a patient consumer with an annual discounting rate of.9 per cent, while the other represents an impatient consumer with an annual discounting rate of 5.4 per cent. Four values of the relative risk aversion coefficient are considered: : 0,; 0.5; 2 and 5. They represent consumers ranging from almost risk-neutral to a relatively high degree of aversion. Furthermore, two types of habit formation are considered: weak (=0.2) and strong (=0.8), and four values for the degree of altruism (envy) shown by the consumer ={0.75, 0.25, -0.25, -0.75}. Choice of utility function and parameters The highest correlation between the real interest rate generated by the model and a realised real interest rate appears where the consumer is very altruistic and almost risk-neutral. The discounting factor is determined on the basis of simple linear interpolation between the average generated real interest rates and the realised real interest rate. This gives a factor equivalent to an annual discounting rate of 4.09 per cent. The model is calibrated using the parameters (,, )=(0.778; 0.; 0.75).

81 76 APPENDIX C GENERATED REAL INTEREST RATES ( = 0.7), AVERAGES Table Risk aversion = 0. = 0.5 = 2 = 5 Isoelastic utility function Habit formation = = * * Relative consumption = = = = Note: * means that the selected parameter gives unrealistic real-interest-rate values. GENERATED REAL INTEREST RATES ( = 0.9), AVERAGES Table 2 Risk aversion = 0. = 0.5 = 2 = 5 Isoelastic utility function Habit formation = = * * Relative consumption = = = = Note: See Table.

82 77 REFERENCES Andersen, J.V. & Gyntelberg, J., Index-Linked Mortgage Bonds, Danmarks Nationalbank, Monetary Review, st quarter 999. Ayuso, J. & López-Salido, J.D., What Does Consumption Tell Us About Inflation Expectations and Real Interest Rates?, Banco de España, Servicio de Estudios Documento de Trabajo n o 9633, 996. Ayuso, J. & López-Salido, J.D., Are Ex-post Real Interest Rates A Good Proxy for Ex-ante Real Rates?, Banco de España, Servicio de Estudios Documento de Trabajo n o 970, 997. Christensen, A.M., Households' Inflation Expectations, Danmarks Nationalbank, Monetary Review, November 996. Galí, J., Keeping Up with the Joneses: Consumption, Externalities, Portfolio Choice, and Asset Prices, Journal of Money, Credit and Banking 26(), pp. -8, 994. Heaton, J., An Empirical Investigation of Asset Pricing with Temporally Dependent Preference Specifications, Econometrica 63(3), pp , 995. Ireland, P.N., Long-Term Interest Rates and Inflation: A Fisherian Approach, Federal Reserve Bank of Richmond, Economic Quarterly 82(), pp. 2-35, 996. Topp, J., Indicators of the Market's Interest-Rate and Inflation Expectations in Denmark, Danmarks Nationalbank, Monetary Review, May 996.

83

84 79 More Information Available on the Nationalbank's Web Site Liselotte Perch Lind, the Secretariat INTRODUCTION Danmarks Nationalbank s Web site at became available on the Internet in May 998. The Web site is kept up-to-date with new publications and information, and provides an overview of the Nationalbank s areas of activity. The Web site also gives access to press releases, publications, statistics, etc. The material is presented so as to meet the need for general information of ordinary users and the requirements of professional users for more detailed information. The Nationalbank s Web site has an ever-increasing number of users, cf. Chart. Statistics show that the Web site is visited primarily by professional users in the financial sector. However, the information is also accessed by other users. The number of searches under Notes and coins has increased considerably in connection with the issue of new banknotes.