- International Monetary Fund
- Published Date:
- August 2003
Sweden has had a separate agency, Riksgäldskontoret (the Swedish National Debt Office [SNDO]), for government debt management since 1789. Inevitably, the principles and practices of debt management have changed repeatedly over the years. A major reform of the governance system was enacted in 1998. As a result, debt management decisions are made within a more clearly structured framework. There is also a more structured approach to evaluate the decisions after the fact. The outline of this system and the experiences so far are presented in the first section, which also discusses the organization of the debt office.
The new governance system has created a framework for more focused analysis of the debt management strategy and the risks involved. The second section discusses the key features of debt management strategy and the analyses undertaken to get a better understanding of the costs of government debt and the associated risks.
The third section discusses measures for developing the government securities markets.
Developing a Sound Governance and Institutional Framework
The core principles and rules for central government debt management in Sweden are given in the Act on State Borrowing and Debt Management.2 The current legislation was enacted by parliament in 1998. The government’s right to borrow is based on an annual authorization from parliament, which is given as part of the decision on the state budget for the subsequent fiscal year. There is no fixed limit on the annual amount of borrowing. Instead, the act specifies the purposes for which the government may borrow, in particular, to finance the budget and refinance maturing debt.3 The government invariably delegates the mandate to borrow to the SNDO.
The objective of debt management is also formulated in the act. It stipulates that the state’s debt shall be managed so that the long-term costs are minimized while taking risks into account. Debt management shall also respect the demands of monetary policy. This is basically the same rule that governed debt management before the reform in 1998. The difference is that the objective now is stated in a law enacted by parliament, whereas it previously was set out in documents issued by the government.
Finally, the act contains procedural rules. First, it stipulates that the government each year shall decide guidelines for debt management. This decision shall be based on a proposal submitted by the SNDO. The proposal shall be sent to the Riksbank, the central bank for comments, to ensure that the demands of monetary policy are taken into account. Second, the act instructs the government to submit an annual report to parliament in which it makes an evaluation of the management of the debt. The SNDO’s proposal, the Riksbank’s comments, and the government’s guidelines, as well as the evaluation report, are all public documents.
The statutory rules create a framework for delegation, reporting, and evaluation. Parliament—at the top of the system—has established the objective. On the basis of this objective, the government is mandated to set guidelines. In preparing the guidelines, the SNDO assists the government. This reflects the fact that the SNDO staff work full-time with debt policy, whereas the government (the ministry of finance) has many other obligations and is confronted with debt policy issues infrequently.4
The implementation of debt management on the basis of the guidelines is then delegated to the SNDO. The guidelines define in broad terms how the debt should be structured. They typically include ranges around target values, leaving scope for the SNDO to make more detailed decisions on the management of the debt.5 There are two decision levels within the SNDO. The first level is the board, which is made up of external members (with the exception of the director general).6 For example, it makes strategic decisions on how to use the ranges given in the government guidelines and benchmark portfolios. The second level is the operative management of the debt within the frame set by the board, which is in the hands of the SNDO’s staff, led by the director general.
The governance system also puts emphasis on evaluation. Each decision level is evaluated by its immediate superior body. This means that the board monitors and evaluates operative debt management and reports to the government. At the next level, the government evaluates the overall result of the SNDO’s decisions. Finally, parliament evaluates debt management as a whole, including the government’s guidelines. This evaluation is in the form of an annual written statement, adopted after a debate and vote in parliament. This statement is published in time for the SNDO to consider comments and recommendations from parliament when preparing the next guideline proposal, closing the loop of delegation and monitoring.
This governance framework applies to central government gross debt, or the debt portfolio managed by the SNDO. The Swedish government sector also has financial assets, for example, in funds in the public pension system and in equity holdings. Indeed, at end-2001, the general government net financial debt was negative, despite a central government debt-to-GDP ratio of about 54 percent. There is no debt management strategy at the level of general government, partly because the public pension funds are managed separately from the state budget, based on objectives derived from their role in securing future pensions, and because local authorities have a significant degree of independence. However, attempts are made to broaden the perspective to include the central government’s balance sheet in the analysis of risks relevant to central government debt management.
The governance framework in practice
The first guidelines constructed within the new framework were adopted in 1998, covering debt management in 1999. When the guidelines for 2002 were adopted in November 2001, it was thus the fourth time the exercise was repeated. It was recognized when the new system was put in place that the objective was vague and that further analysis was needed to translate it into a practical framework for debt management. In the preparation of the guideline proposals, emphasis has been put on complementing the statutory target with appropriate definitions of costs and risks. In addition, the SNDO has presented analyses of how the composition of the debt portfolio can be expected to affect costs and risks. The main conclusions and the resulting debt management strategy are summarized in the second section.
The changes in the way debt policy is governed have not, at least so far, altered the practices of debt management. The most significant effect is perhaps that the time perspective in the guidelines has been extended from one year to include tentative plans for rolling three-year periods, consistent with the time frame used in the budget process. However, the importance of the form and structure of governance should not be underestimated.
First, the new system has increased the attention given to debt policy in both the government and parliament. Considering the size of the annual debt costs, this attention is justified.
Second, the procedure surrounding the annual guideline decision has affected the perception of debt policy. All strategic proposals and decisions must be explained to the public in terms of their impact on costs and risks. The decisions are then evaluated in terms of their impact on costs, and the results are made available to the public. The transparency of the system helps to commit the policymakers to the stated objective of long-term cost minimization with due regard to risks. In particular, borrowing strategies that reduce short-term costs, but also significantly raise medium- or long-term costs or the risk of the debt are difficult to implement.
Third, the system provides a clearer distribution of responsibilities between the parties concerned. The government (the ministry of finance) delegates debt management to the SNDO for one year at a time. The government can change the guidelines during the year if the circumstances underlying the decision have changed materially, but this has to be done in the form of an amended guideline, that is, in a public document.
Similarly, the central bank’s views are brought into the process in a formalized way when the Riksbank is invited to comment on the annual proposal. This is in contrast to the situation 10 or 15 years ago, when debt management often was used as an instrument of monetary and exchange rate policy. Now it is clear that monetary policy, at most, acts as a constraint on the cost minimization problem, and it is not part of the objective. An objection from the Riksbank to a proposal from the SNDO must be linked explicitly to “the demands of monetary policy” (i.e., the suggested guidelines would interfere with the Riksbank’s ability to reach its statutory objective or price stability). In an economy with well-developed financial markets, this will only rarely be the case. A gradual separation between monetary policy and debt policy had begun before the governance system was reformed, but the new process has contributed to an even clearer decoupling of monetary policy and debt policy.
In summary, the governance framework introduced in 1998 makes it clear that debt management is a policy area in its own right, with its own objective. The government is answerable to parliament for achieving this target, but major responsibilities are delegated to the SNDO, in both the preparation of guideline proposals and operative debt management. As a result of the guideline procedure, the operative independence of the SNDO has increased. However, through the reporting and evaluation mechanisms, there has been a corresponding increase in the possibilities to hold the SNDO accountable for its decisions. As is appropriate, delegation and accountability go hand in hand.
The arguments for delegating the implementation of monetary policy to an independent central bank are well known, and the practice of doing so is wide spread. The Swedish governance framework for debt management illustrates how debt policy can be delegated in a similar fashion to an independent debt office. However, the degree of delegation is less far reaching than in monetary policy.
First, the debt management objective is two-dimensional, in that there may be trade-offs to make between cost and risk. Second, the attitude to risk is presumably not invariant to the overall fiscal position of the country. Debt must be seen as part of the overall balance sheet of the government. It must be assessed in relation to expenditure commitments, defining another class of liabilities, and future tax revenues, which are the most important assets on the balance sheet. It seems inevitable that an optimal debt policy is state contingent in that the attitude to risk will vary depending on the overall outlook for government finances. Strategic decisions on debt management are thus closely linked to fiscal and budget policies, which fall within the realm of the government and parliament. An independent debt office may not have the necessary information to make such assessments. Moreover, and more important, it is not be possible to make a separate debt office accountable for decisions that are ultimately political in nature. It is therefore difficult to delegate strategic decisions to an administrative agency to the same extent as the implementation of monetary policy can be delegated to an independent central bank.
The organization of the SNDO
For the first 200 years, the SNDO was responsible directly to parliament. This meant, for example, that it had an external board appointed directly by parliament. However, in 1989, it was turned into an agency of the government. This reflected primarily the assessment that parliament’s influence was secured via the budget process and that the implementation of debt policy was handled more efficiently by an agency reporting to the government.
The operative responsibility for the SNDO is in the hands of the director general and the deputy director general, both appointed by the government. The director general is also the chairman of the board. The government also appoints the other board members, none of whom is employed by the SNDO. Reflecting the previous link to parliament, four of the eight board members are also members of parliament. The board decides on strategic issues related to debt management (e.g., guideline proposals and risk control).
Within the SNDO, there is a clear organizational separation between front, middle, and back office responsibilities. Front office activities, such as auctions, debt, and cash management transactions, are executed within the debt management department. The head of the debt management department reports to the director general.
Debt management is monitored by the risk control department, the middle office. Its responsibilities include monitoring positions relative to benchmarks and observance of credit limits on counter-parties. The risk control department is also responsible for monitoring operative risks.
Confirmation and settlement of debt management transactions are handled by the back office department. Reports on the results of debt transactions are handled by the accounting department. The heads of risk control, back office, and accounting report to the deputy director general to achieve a further separation between operative debt management and follow-up activities.
In addition, there is the internal auditing department. It reports directly to the board on results of audits of the activities of the SNDO.7
The SNDO is funded by the state budget. The SNDO submits budget proposals to the government, covering rolling three-year periods, that explain its financing needs. Current expenditures for salaries, rents, and similar expenditures are covered by one budget title, interest payments on the debt by another. Consequently, the SNDO cannot use savings on interest costs, for example, to hire additional staff.
Within the limit set in the budget, the SNDO has considerable flexibility in allocating funds between uses, including hiring decisions. Swedish government agencies, in general, are controlled not by detailed specifications on how they may use their funds, but by whether they manage to achieve the objectives set up for them. The evaluation system for debt management is thus unusual only in that it is more elaborate and formalized than in other areas.
This affects, for example, decisions on salaries. The government appoints the director general and the deputy director general and sets their salaries. The SNDO, ultimately the director general, makes hiring decisions and establishes the salaries of all other employees. There is no fixed salary structure, based on seniority or other set criteria, within the Swedish central government. Instead, merit and degree of competition from the outside for a particular skill are key factors. This means that salaries can be adjusted to the SNDO’s need to recruit or retain a particular individual—that is, of course, within the limits in any organization to maintain a wage structure perceived to be fair and reasonable.
Still, the SNDO cannot directly outbid private financial firms in terms of wages. The key to recruiting and retaining staff with appropriate financial market skills is to offer challenging and interesting tasks to work with. In particular, it is important to involve also relatively junior staff in discussion and decisions on policy issues. Even though debt management involves portfolio decisions analogous to those made in private financial institutions, there are connections to broader economic policy issues not present elsewhere. By building on this aspect, a debt office can create an advantage relative to the private sector, offsetting some of the differences in salary levels. In areas such as the back office and information technology, the differences between working in a policymaking institution and a private firm are less pronounced. The SNDO therefore tends to meet tougher competition when it comes to retaining staff in administrative functions than in the core policy areas.
Debt Management Strategy and Risk Management Framework
The statutory objective—long-term cost minimization with due regard to risk—is obviously sound. However, more precise concepts are needed to translate the objective into a strategy for actual debt management. An important element in the analyses conducted within the new governance framework has been to define what one should mean by costs and risks.
The next question is how the debt portfolio should be structured. The SNDO has concluded that the composition of Sweden’s current debt portfolio, in particular in terms of the share allocated to foreign currency debt, differs from what is desirable. The discrepancy is so significant in relation to the speed with which the debt composition can be modified that it has been deemed unnecessary at this stage to define a target portfolio in terms of percentage shares. The guidelines have instead pointed out the desired directions in which to move the portfolio. The resulting debt portfolio strategy is discussed in this section, as are various aspects of active debt management.
The concepts of costs and risks
It was acknowledged when the new law was enacted that additional analyses were required to give the concepts of costs and risks more concrete interpretations. Although much remains to be done, some tentative conclusions have been drawn by the SNDO and confirmed in the government’s guideline decisions.
The first step in the process was to consider whether costs (and related risks) should be measured on the basis of a complete mark-to-market of the debt or using interest rates set when bonds were issued. The conclusion was that market value changes do matter. However, the bulk of the debt is not—and indeed cannot be—refinanced at short notice. As a first approximation, therefore, it is reasonable to assume that debt instruments are left outstanding until maturity. This means that short-term fluctuations in market values resulting from changes in market interest rates are of little consequence for the realized costs of the debt. This view is also reflected in the accounting practice of not revaluing the debt on the basis of current market interest rates.
It should be noted, however, that the foreign currency debt is consistently valued in terms of current exchange rates. One reason to treat interest rate and exchange rate movements differently is that the latter can be expected to lead to realized losses or gains even if the bond is left outstanding to maturity, because payments are made in foreign currency. Moreover, the current exchange rate is probably the best available indicator of the rate at maturity.
In the bill presenting the new legislation, the government indicated that real (as opposed to nominal) measures of costs seemed appropriate from a general economic perspective. However, the government noted that the understanding of real measures of financial risk was limited and that nominal costs therefore would be used pending further analysis. A second step in the analysis has dealt with the question of how to go beyond nominal measures of costs and risks.
In its simplest form, a real measure could be obtained by deflating nominal costs with a price index, for example, the Consumer Price Index. This would, for example, make the cost of an inflation-linked bond predetermined and, hence, risk free. It is intuitively clear, however, that there is more to “real risk” than inflation adjustment, and, more concretely, that an inflation-linked bond is not in general risk free for the government. This line of thinking led to a broadening of the perspective beyond the government debt as usually defined. Inspiration also came from the practice, in particular in financial firms, of making risk analyses in terms of the entire balance sheet, leading to a perspective akin to asset and liability management (ALM).
The starting point is to note that debt is just one item on the government’s balance sheet, broadly defined. First, there are nondebt obligations of all kinds, including entitlement programs and other future expenditures. Second, the government has assets. The most important of these is the right to charge taxes, which in balance-sheet terms could be measured as the present value of future tax revenues.
Risks arise when assets and liabilities are not perfectly matched. To manage its risks, the government must therefore consider the entire balance sheet and try to limit the mismatch between assets and liabilities.8 A complete balance-sheet analysis of the government is an inordinately complicated undertaking. An ALM-based approach to debt management can also be helpful if one does not have a complete quantitative picture, however. In particular, it becomes clear that the risk of government debt should be assessed on the basis of whether it exacerbates or mitigates strains on the balance sheet.
One simple measure of the (current) strains on the balance sheet is the budget balance. For example, a debt portfolio that typically has high costs in recessions, that is, when public finances are strained for other reasons, must be considered riskier than a portfolio for which the opposite is true. This translates into treating deficit smoothing as an operative objective of debt management.
An ALM perspective also modifies the assessment of inflation-linked bonds. Debt costs linked to inflation mitigate swings in public finances as long as inflation is high when tax bases are high and expenditures low, that is, if inflation is positively correlated with the business cycle. However, if the economy is hit by a supply shock leading to stagflation—high inflation combined with low growth—inflation-linked debt adds to the strains on public finances.
Acknowledging that ALM provides an appropriate frame for thinking about debt management risks is one thing; translating it into a complete debt management strategy is quite another. The most visible effect so far on Swedish debt management is that in qualitative and quantitative analyses, debt costs are set in relation to GDP. GDP is here seen as a measure of other business cycle–related influences on the budget, or a debt portfolio perceived as having a relatively stable cost-to-GDP ratio is regarded as less risky. Using the cost-to-GDP ratio as the criterion for ranking debt portfolios is a step in the direction of ALM.
The SNDO has built a stochastic simulation model, which is used, jointly with qualitative reasoning, in the work on guidelines for debt management. The model generates paths for interest rates, exchange rates, GDP, and the borrowing requirement for up to 30 years. These time series are then used to simulate the costs of a set of debt portfolios with different characteristics, making it possible to rank portfolios on the basis of their expected costs and the variability of costs. The primary metric used is the cost-to-GDP ratio. Because GDP is generated in the model, it is possible to capture correlations between interest rates, exchange rates, and GDP in an internally consistent manner.9
Debt portfolio strategy
For the purpose of discussing debt strategy at the portfolio level, the Swedish central government debt can be broken down into three parts, nominal krona debt, inflation-linked krona debt, and foreign currency debt. Figure II.16.1 shows how the debt and its composition have developed since 1992.10
Figure II.16.1.Central Government Debt, 1990–2001
The current debt portfolio is dominated by nominal loans in domestic currency, made up of bonds and bills, but less so than in most Organization for Economic Cooperation and Development countries. First, Sweden has an unusually large share of inflation-linked loans, although still less than 10 percent of total debt. Second, and more significant, more than a third of the debt is in foreign currencies. The foreign currency debt is largely a legacy from the early 1990s. As can be seen in Figure II.16.1, total debt more than doubled between 1990 and 1995, when Sweden experienced the deepest recession since the 1930s. With an annual net borrowing requirement corresponding to 14 percent of GDP at the peak, it was useful to divert some of the borrowing to foreign capital markets. This reduced the pressure on long-term interest rates in the domestic market and diversified the debt portfolio.
The duration of the combined nominal krona and foreign currency debt is approximately 2.7 years. The inflation-linked debt has a duration (measured in terms of real rates) of close to 10 years.
The key question in the analyses conducted within the new governance framework has been how a portfolio consistent with the objective of long-term cost minimization with due regard to risk should be structured. Special attention has been given to foreign currency debt, reflecting that this is the aspect in which the Swedish debt portfolio stands out, and to inflation-linked debt, as a relatively new instrument.
Based on qualitative and quantitative analyses, the SNDO has concluded that it is desirable to reduce the share of foreign currency debt.11 It adds risk without offering expected long-term cost savings. First, the government has few foreign currency assets (i.e., the foreign exchange exposure is basically unhedged).12 Second, it is not unlikely that the domestic currency weakens in recessionary periods, because the costs of foreign currency debt would tend to add to swings in the deficit-to-GDP ratio. Third, at a somewhat subtler level, the simulations model illustrates that (under flexible exchange rates) domestic short-term interest rates are negatively correlated with the business cycle, because the central bank will vary short rates in a counter-cyclical manner. This tends to stabilize the ratio of debt costs to GDP. For a small country, foreign interest rates will be unaffected by domestic events, making foreign currency debt less attractive than domestic currency debt, other things being equal.
In the guidelines for 2001, the government decided that the share of foreign currency debt should be reduced.13 No percentage target was set, partly because the desired share is so far below the current one that a decision on this point was not urgent. Instead, the government established a plan of annual amortizations corresponding to SKr 35 billion. Annual repayments could be varied within an interval of ±SKr 15 billion. The government instructed the SNDO to take account of the value of the krona when deciding the actual rate of repayment. If the krona is seen as significantly undervalued, it is rational—given a cost minimization target—to reduce repayments until the krona exchange rate has returned to more normal levels. In the guidelines for 2002, the government decided to lower the long-term rate of repayment to SKr 25 billion. Moreover, it set the 2002 target rate to SKr 15 billion, citing the weakness of the krona.
Also pointing to the depreciation of the krona, the SNDO used its mandate to hold back amortizations of close to SKr 15 billion in 2001. It has also announced that repayments will be made at a lower pace than the targeted rate from the start of 2002. In combination with a significant reduction in the krona-denominated debt and the depreciation of the krona, this has led to an increase in the foreign currency share during 2001 (Figure II.16.1). The long-term ambition is still to gradually decrease the foreign currency debt to achieve a debt portfolio more in line with the objective of debt management.
In the guideline proposal for 2002, the SNDO focused on the role of inflation-linked debt.14 The quantitative results from the simulation model were less clear cut than in the analysis of foreign currency debt. The model indicates that there is little difference in terms of costs and risks between nominal and inflation-linked domestic currency debt. One potential explanation is that the model assumes the economy is not subjected to severe shocks. For example, budgetary and monetary policy targets are met on average in all simulations. The debt portfolios are thus not subjected to any stress-tests, because these are hard to handle in a long-term simulation model. In such an environment, there is little reason to expect inflation-linked debt to differ markedly from nominal debt.
Using qualitative reasoning, the SNDO points to other possibilities. For example, in an environment with low growth and low inflation, perhaps even deflation, inflation-linked bonds are helpful for deficit smoothing. Conversely, during a period of stagflation, having a large inflation-linked debt is undesirable. The observation that neither deflation nor stagflation can be ruled out before the fact. is then sufficient to indicate that a portfolio made up of several types of debt is preferable from the point of view of reducing risk. As long as inflation-linked bonds are not markedly more costly than nominal bonds, this diversification effect thus argues for including inflation-linked debt in the portfolio. The tentative assessment is that the share should be higher than the current 8 percent for inflation-linked debt to make an appreciable difference in an actual stress test.
In the guidelines for 2002, the government instructed the SNDO to increase the share of inflation-linked debt in the long term.15 However, given the cost minimization objective, the rate of increase must be weighed against the costs of other types of debt. As in other countries, the inflation-linked bond market has periodically been characterized by limited investor demand and high liquidity premiums. An important task for the SNDO is therefore to continue its efforts to improve the functioning of the market so that the benefits in terms of reduced portfolio risks can be achieved at acceptable costs. (See also the third section.)
The choice of duration involves a trade-off between costs and refinancing risks. Experience indicates that nominal short rates, on average, are lower than long rates. Strict cost minimization would thus argue for having a debt with short duration. As noted before, domestic short rates may also be negatively correlated with the business cycle, which contributes to deficit smoothing. However, a short duration would make debt costs more sensitive to current interest rate levels. Moreover, short rates tend to be more volatile than long rates.
The duration of the nominal part of Swedish debt is 2.7 years, which has been concluded to represent a reasonable trade-off between the considerations discussed. The current guidelines thus indicate unchanged duration over the three-year planning horizon.
The inflation-linked debt is significantly longer, as is appropriate for an instrument aimed at protecting investors from inflation uncertainty. The guidelines state that inflation-linked debt should have at least five years’ maturity at the time of issuance and preferably longer. Inflation-linked debt therefore extends average maturity and helps reduce refinancing risks for the debt portfolio as a whole.
A duration target does not limit refinancing risks. In principle, a mixture of just two maturities can achieve any duration target. In practice, the rollover risk is limited by the SNDO’s overall borrowing strategy, based on a set of nominal benchmark bonds extending to at least 10 years and a set of inflation-linked bonds, some of which have an even longer time to maturity. For the purpose of clarity, the guidelines still set a limit on the permissible extent of refinancing over the short term. The stipulation is that the SNDO should plan its borrowing in such a way that no more than 25 percent of the debt matures over the next 12 months.
Sweden has decided that the concept of risk in the statutory objective shall be interpreted in terms of how debt costs affect the overall stability of government finances. In this regard, Swedish authorities have adopted an ALM approach as the starting point for debt portfolio analysis. Given the time perspectives involved in government debt management, potentially spanning generations, genuine uncertainty will always be a key element in debt management. One should therefore not expect to reach robust, once-and-for-all quantitative conclusions about what is an optimal debt portfolio. The main contribution of the ALM approach is probably as much in the questions it raises as in the formal answers. In particular, the debt manager is forced to address issues related to risk taking in a more consistent manner than if debt costs are seen in isolation from the rest of the government balance sheet.
Active debt management
It is useful to distinguish between two types of active debt management. The first type includes actions allowing a separation between funding decisions, on the one hand, and decisions on the characteristics of the debt portfolio, on the other, achieved primarily through derivatives. Such activities are motivated by a desire to use low-cost methods of funding without necessarily accepting the risks attached to those instruments. The resulting debt portfolio should have lower funding costs or risk or both than an identical portfolio created by direct borrowing. This form of active management is driven by strategic considerations.16
The second type of active debt management refers to positions taken on the basis of views on the future paths of interest rates or exchange rates. This requires a defined benchmark. A position is then created by modifying the actual portfolio so that it deviates from the benchmark. The result of the position can be evaluated by comparing the (market) value of the actual portfolio to the value of the benchmark. This form of active management is typically driven by tactical considerations.
The SNDO uses both forms of active debt management. The following explains the motivations and frameworks used in each.
The separation between funding and portfolio decisions originates from how the foreign currency debt has been managed. Traditionally, the SNDO pursued an opportunistic borrowing strategy, seeking out low-cost funding sources without regard to currency or maturity. To achieve the desired composition of the foreign currency debt portfolio, expressed in a benchmark portfolio, it used derivatives to transform the cash flows.
Given this experience of working with derivatives, it was not a major step when the SNDO in 1996 began borrowing in Swedish krona and converting the debt into synthetic foreign currency obligations by use of foreign currency swaps.17 This transformation involves several steps. First, the SNDO issues a long-term krona bond. Second, it does an interest-rate swap (IRS) in kronor in which it receives payments based on a fixed interest rate and pays based on a floating interest rate. Third, the floating krona cash flow is converted to euros, say, via a foreign currency swap. Finally, there is an IRS in foreign currency to achieve the desired duration of the foreign currency exposure. At the end of 2001, about 45 percent of the foreign currency exposure, equivalent to SKr 180 billion, was in the form of krona/currency swaps. As can be seen in Figure II.16.2, annual swap volumes have varied between SKr 20 billion and SKr 40 billion, depending partly on the perceived depth of the market. In recent years, at least half of the krona bonds issued have been swapped.
Figure II.16.2.Bond Issuance and Swap Volumes, 1996–2001
Krona/currency swaps have for several years been the cheapest method for creating foreign currency exposure. The reason is that the SNDO—as a representative of the state—has a comparative advantage in long-term borrowing in kronor. In other words, the long-term swap rate is higher than the SNDO’s long-term funding cost. This swap spread is partly offset by the fact that the floating swap rate is higher than the treasury bill rate, which determines the cost for direct short-term funding. As long as the average short-term spread over the term of the swap is lower than the initial long-term swap spread, the SNDO obtains cheaper funding by using swaps than by issuing bills. Moreover, it avoids the refinancing risk.18
The SNDO has announced that IRSs can be used as an alternative to short-term funding, that is, as a complement to bills. This enables the SNDO to borrow in the long end of the yield curve without increasing the duration of the debt. Swaps thus also result in larger issue volumes in the Swedish bond market, which should support liquidity. This aspect has been important in recent years, when gross borrowing needs, following an improved budget situation, have decreased.
It is important that large-scale derivates transactions are handled in a transparent manner to avoid any confusion in the marketplace about the motives for using them. As in the bond market, the SNDO therefore announces its yearly planned swap activities. The volumes are decided based on previous experiences and following comments from market players. The actual volumes can change depending on market conditions. Moreover, a swaps book of the size built up in Sweden presupposes that swaps markets have sufficient depth, so that shrinking swap spreads do not erode the benefits. A gradual expansion of the program, with due regard to the costs of swaps relative to other funding techniques, is therefore advisable.
Swaps and other derivatives give rise to counter-party risks, which must be carefully managed. The SNDO uses credit support annex agreements as a method for reducing credit risks. This is a system of bilateral exchange of cash based on the current market value of the net position. Such a system limits the credit exposure to the daily changes in market value, allowing the SNDO to transact with the counter-parties that offer the best prices at each point basically without regard to the previous contracts written with those intermediaries.19
The SNDO takes tactical positions to benefit from movements in exchange rates and interest rates only in the management of the foreign currency debt. The framework for this activity is a benchmark portfolio determined by the SNDO’s board. The board defines a neutral portfolio in terms of currency and maturity composition and the maximum permissible deviations from this portfolio. Within these boundaries, the SNDO management has the mandate to take positions. The observance of these limits is monitored by the risk control department, which is separate from the debt management department.
The SNDO also engages external portfolio managers (currently five) working with the same mandate, scaled down to a fraction of the total foreign currency debt. This practice gives an additional measuring rod for evaluation of the SNDO’s debt allocation decisions.
The SNDO makes no corresponding debt allocation decisions based on views on interest rates in the management of the krona debt. The main reason is that the SNDO is so dominant a player in the krona fixed-income market that its reallocations could move interest rates. Opportunistic behavior by such a borrower will raise the overall level of interest rates as investors demand compensation for the added risks they face. Given its typical dominance of the domestic currency bond market, a predictable and transparent borrowing strategy is a better means to lower debt costs for a sovereign issuer. (See also the third section.)
As noted, the SNDO bases decisions on the rate of repayment of the foreign currency debt partly on exchange rate assessments. This is also a form of active debt management, affecting the composition of the debt, although there is no defined benchmark for the overall debt portfolio.
It should be emphasized that from the point of view of securing low costs and acceptable risks, active debt management relative to a benchmark portfolio is of secondary importance compared with the decision on the benchmark portfolio itself. For example, it is obvious that choosing a duration target of 2 years instead of 5 years affects costs and risks far more than variations within an interval of ±0.3 years around either central value. Still, the savings from successful active debt management can be significant in absolute numbers. Moreover, instruments used for position taking can be applicable also in conventional debt management, for example, on the use of derivatives instruments, thus giving positive side effects.
Cash management and the links to the central bank
The state payment system in Sweden is based on the single-account principle, that is, all payments are channeled through a system with a single top account, managed by the SNDO. Up until 1994, the balance on this account was held with the Riksbank. This meant that the SNDO was not engaged in cash management. Instead, the Riksbank had to sterilize the changes in bank reserves resulting from swings in the balance on the government’s account.
In connection with Sweden’s entry into the European Union (EU), it was decided that the SNDO should manage the top account outside the Riksbank. One reason was that EU rules prohibit the central bank from lending directly to the government, that is, the SNDO could no longer have a negative balance on its account overnight. Because the government’s cash position fluctuates strongly over the month, it would have had to deposit a sizable sum with the Riksbank to create a buffer that would prevent the balance from ever turning negative. This was deemed to be an inefficient form of cash management.
In the current framework, the SNDO uses its Riksbank account for participation in the payment clearing system, but the balance on the account is set to zero at the end of each day. This is achieved via transactions in the short-term interbank market with such instruments as overnight loans and deposits, as well as repos. Typically, the balance is brought to zero by such interbank market transactions. On occasion, there is a remaining balance (positive or negative), rarely exceeding a few million kronor, at the end of the day. This is then transferred automatically to an account held by the SNDO with a commercial bank. In this way, the SNDO has responsibility for all aspects of government debt management, from overnight loans to 30-year inflation-linked bonds.
This arrangement also means that the Riksbank does not have to offset the swings in the government cash position via market operations. Because the SNDO is part of the interbank market, the reserves available to the banking system are not affected by the government’s cash position. Separation of government cash management from the central bank thus also simplifies the Riksbank’s task of managing its balance sheet to set the overnight interest rate at the target level.
Although the SNDO has been responsible for domestic currency cash management since 1994, it has continued to make all exchanges between kronor and foreign currencies with the Riksbank. The Riksbank makes the exchanges needed to cover the SNDO’s purchases of foreign currencies in a predetermined pattern to avoid confusion with interventions for exchange rate policy purposes. Specifically, the bank buys a preannounced sum of foreign currency during a certain period on each trading day.
As of July 1, 2002, the SNDO will also have the right to make such exchanges with other counter-parties. The motivation for the government’s decision is that this will allow greater flexibility in the handling of the transactions. Although the SNDO is also instructed to act predictably and transparently in the foreign exchange market—that is, short-term speculative transactions are ruled out—it need not adhere to such a strict calendar as the Riksbank, thus making it possible to also use the timing of purchases of foreign currency as an instrument for reducing the costs of the debt.
Management of contingent liabilities
Contingent liabilities in the form of guarantees can be issued only on the basis of authorization from parliament. Four government agencies are in charge of special guarantee programs related to export credit, housing, international aid, and deposit insurance. The SNDO issues other guarantees based on specific authorizations by parliament in each case. The budget law stipulates that a risk-related fee should be charged for guarantees. If parliament decides that the recipient of the guarantee does not have to pay, budget means must be reserved to cover the fee. The SNDO sets the fee, that is, there is a clear separation between the decision to issue the guarantee and the pricing. The accuracy of the SNDO’s pricing decisions can be evaluated after the fact by checking whether the fees accumulated over long periods match the payments made to cover guarantee claims.
With an ALM approach to debt management, it is clear that guarantees (and other contingent liabilities) must be considered in analyses of the risks in public finances. In a consistent risk management framework, it should be possible, for example, to consider whether a risk reduction should be achieved via a change in the government debt portfolio or by transferring a guarantee to a guarantor in the private sector.
Again, this principle is easier to state than to implement. A first step would be to present consistent aggregate information on government guarantees. This should include expected losses on the total guarantee portfolio, but also capture the magnitude of unexpected losses. Reports on the guarantee portfolio should be presented to parliament in connection with the budget proposal. If the expected or unexpected guarantee losses increase, this should be considered before decisions on expenditures and taxes are taken, because this would be equivalent to a weakening of the underlying budget position. Such a framework is not in place in Sweden, but steps in this direction are being taken to improve the analysis and management of contingent liabilities.
Developing the Markets for Government Securities
In addition to a well-balanced allocation among types of debt, a smoothly functioning market for government securities is important to achieve the objective of minimizing costs. As a dominant borrower and participant in the Swedish fixed-income market, the SNDO has a responsibility for ensuring the efficiency and development of the market. The SNDO has therefore taken an active role in the discussions and development of the market place. This includes measures to improve the secondary market to enhance liquidity and transparency. Some examples of measures taken in the different market segments are discussed in this section.
The overall strategy is to concentrate borrowing in a few fairly large issues. Currently, there are around 10 such benchmark bonds. Large issues make trading easier, and the risk of short squeezes in specific issues is lowered. The SNDO previously aimed at ensuring that there were bonds maturing every year. This strategy was slightly changed during 2000, when a new 10-year issue was introduced, maturing in 2011, leaving 2010 without any maturity. Considering the large budget surpluses, it was more important to concentrate borrowing to support liquidity than to have yearly maturities of benchmark bonds. The maturity profile at the end of 2001 is illustrated in Figure II.16.3.
Figure II.16.3.Maturity Profile for Nominal Bonds, December 2001
When a new issue is introduced, normally with a 10-year maturity, switches from old issues are carried out to quickly build up the new issue. In addition, a special repo facility of some SKr 20 billion is in place immediately after the first auction of the new bond. The repo rate is typically 15 basis points below the Riksbank’s target for the overnight interest rate. This repo facility ensures that no single investors can squeeze the market for a new issue. As a result, the SNDO has not set any limitation on how large a share a single investor may hold of a single bond or on the share allocated to a single dealer in an auction.
Bond auctions are held biweekly. The issue for auction and the volume are announced one week in advance. The authorized dealers are committed to enter bids on behalf of investors and for their own account. The result is presented 15 minutes after the auction is closed.
One important way to enhance liquidity in the secondary market is to offer repo facilities to the authorized dealers. This lowers the risk of shortages and, hence, supports liquidity. To ensure that the SNDO does not assume too dominant a position in the repo market, these activities are limited. The SNDO has set a maximum repo volume of SKr 500 million, which the authorized dealers can use at a penalty rate of 60 basis points below the Riksbank’s target for the overnight rate. This spread ensures that the facility is regarded as a “repo of last resort.” Although these repo facilities are rarely used, they are important because they make it impossible to create a squeeze. With a penalty rate as high as 60 basis points, supply and demand in the market still govern pricing in normal circumstances, without interference from the SNDO.
In addition, repos are a natural instrument in the SNDO’s liquidity management. When the SNDO needs to borrow in the short term, it might, for instance, lend a security to an investor in a repo to get short-term funds. In this case, the repo rate is 15–25 basis points below the targeted overnight interest rate. Other alternatives are to borrow directly in the deposit market or issue short-term treasury bills on tap.
The Swedish treasury bill market is relatively large in an international perspective. About 20 percent of the total debt is in treasury bills. This market is built around the same principles as the bond market. Borrowing is made through biweekly auctions and is concentrated in eight bills, normally of up to 12 months’ maturity (see Figure II.16.4). The repo market is less liquid and deep than the bond market, but the SNDO has similar repo facilities as those in the bond market, with slightly more generous conditions.
Figure II.16.4.Maturity Profile for Treasury Bills, December 2001
When a bond has less than one year to maturity, investors are offered opportunities to switch into a package of three or four bills. The package is constructed so that the exchange is duration neutral. The operation leaves the investor with more liquid securities and lowers the refinancing risk of the SNDO, because the redemption is spread over several dates.
Inflation-linked bonds were introduced in 1994. The share of the total debt is some 8 percent. This makes the Swedish inflation-linked market one of the largest in the world in relative terms. There are currently seven inflation-linked bonds ranging from 2 to 26 years in maturity.
Two main arguments have been put forward to support issuance of inflation-linked bonds. First, the long-term funding cost should be lower, because the state assumes the inflation risk. If the inflation risk premium is higher than other premiums that might work in the other direction, such as liquidity premiums, the funding cost of inflation-linked bonds should be lower than for nominal bonds. The second argument—discussed in the second section—is that inflation-linked bonds contribute to diversification of the debt portfolio.
In the first years of the program, the cost argument seems to have been valid because investors bought inflation-linked bonds at breakeven levels higher than the official inflation target set by the Riksbank. Probably the inflation target at that time did not have full credibility, which made it rational for investors to pay to avoid the inflation risk. Inflation fell below the inflation target, and, as a consequence, the SNDO has calculated that so far the inflation-linked borrowing has saved some SKr 8 billion in accumulated funding cost since 1994. However, in recent years, the credibility for the 2 percent inflation target has been established, and breakeven inflation priced by the market has been below 2 percent. This has raised the question whether issuance of inflation-linked bonds is cost efficient.
The government and the SNDO have concluded that the cost saving might vary substantially over time. When inflation risks are regarded as small, it might be less favorable to issue inflation-linked bonds and vice versa when inflation risks rise. Also, it is important for the SNDO to support an efficient market for inflation-linked bonds to bring down the liquidity premium.
Developing a market for a new type of instrument is a challenge. Theoretical arguments support inclusion of inflation-linked bonds in long-term asset portfolios. However, most large asset portfolios in Sweden still have small shares of inflation-linked assets. There are several reasons behind this. One is probably that many investors use nominal accounting and benchmarking, making inflation-linked instruments appear more, rather than less, risky. Fund managers also have a tendency not to deviate very much from their competitors, implying that there may be thresholds that need to be passed to increase aggregate holdings of inflation-linked bonds.
One lesson from Sweden’s experience is that a pragmatic approach is warranted. For example, the SNDO has issued inflation-linked bonds through both auctions and on tap. On-tap issuance—issuance of securities at the request of authorized dealers—worked well when the large domestic investors were building up strategic holdings in inflation-linked bonds. The on-tap method made it possible to meet this demand in a flexible way. As the market grew, auctions were introduced, because this method increased transparency and predictability.
Although inflation-linked bonds are sold through auctions, the SNDO offers authorized dealers an on-tap switching facility in the secondary market, making it possible for authorized dealers to switch between two bonds on a duration-neutral basis. The SNDO sets the price in a way that makes switches expensive for the dealers. Like repo facilities for nominal bonds, the switching facility should be regarded as a “last resort” offer. In addition, the SNDO has repo facilities for inflation-linked bonds. Each authorized dealer can repo SKr 200 million 25 basis points below the Riksbank overnight rate.
Ahead of auctions, the SNDO takes advice from authorized dealers. The SNDO is more inclined to listen to advice in the inflation-linked market than in the nominal bond market; dialogue with investors and dealers is more important in a new market.
Naturally, when deciding what and how much to issue, the SNDO also takes prices into account. In particular, the it tries to avoid funding at breakeven inflation rates, which are too low. However, the SNDO finds it important to support the market by also issuing at least small volumes when funding costs seem less favorable. In the longer run, this should help bring down liquidity premiums and, hence, make inflation-linked funding less expensive.
The SNDO has three separate dealer agreements: one for nominal bonds, one for treasury bills, and one for inflation-linked bonds. The reason for having formal agreements is that it enables the SNDO to form dealer groups committed to take part in both the primary and secondary markets on an ongoing basis. A commitment from dealers is of added importance in Sweden, given its fairly small market.
The agreement for nominal bonds was changed in 2001, when an electronic trading platform was started. The SNDO used the agreements as an instrument to make a uniform change of the market structure possible. According to the new agreement, the authorized dealers will, apart from taking part in the primary market, quote binding two-way prices in 2-, 5-, and 10-year bonds in the electronic system. The agreement specifies minimum volumes and maximum bid-offer spreads. As part of the new agreement, the SNDO began to pay commissions to dealers to encourage participation in the electronic trading system. One part is fixed, and one is related to how active the dealer has been in the primary and secondary markets. In total, commissions amount to between SKr 15 million and SKr 20 million per year (equivalent to less than US$2 million).
An advisory board governs the new electronic trading platform. The authorized dealers, the SNDO, and the exchange are represented on that board.
The SNDO does not pay commissions in the treasury bill market, because the commitment needed—especially in the secondary market—from the dealers is not regarded as important as in the bond market. However, the authorized dealers have the advantage of being the only ones allowed to bid in the auctions and have access to repo facilities.
The agreement for inflation-linked bonds is more extensive in the sense that it requires the candidates to apply once a year and present a business plan for their activities. By requiring business plans, the SNDO wants to stress that it is important that dealers are active in promoting inflation-linked bonds to help broaden the investor base. Commissions are also paid to authorized dealers in the inflation-linked market—in total, about SKr 12 million per year.
As the fixed-income markets become more global and integrated, the importance of investor relations increases. Previously, the debt policy of a country was of interest for only a limited number of mostly domestic investors. Now, investors can and will choose from a number of fixed-income markets. Consequently, sovereign borrowers are in competition with each other and with other large issuers.
The main element in the SNDO’s investor relations strategy is to have a transparent and predictable borrowing strategy. When investors understand the framework for debt management and know what variables are important, there will be less uncertainty and risk premiums will be lower. However, transparency and predictability do not exclude changes in borrowing plans or the set of instruments used. The issuer must have the opportunity to adjust its plans, for example, to unforeseen changes in the borrowing requirement. Therefore, the objective should be to communicate strategic principles—for example, duration of the debt and the desired debt composition—and explain what factors are important when formulating policies.
In this respect, the guideline system serves as a good basis. In the guidelines, investors and others can find the motives behind a certain strategy. The fact that all documents in the guideline process are public also allows dealers, investors, and other concerned parties to offer their comments.
The SNDO publishes a report three times a year in which it presents the latest forecast for the borrowing requirement as well as plans for future issuance. In addition, the report discusses different topics related to debt management. This may include articles about swap strategies, proposals for changes of the market structure, and similar topics. The purpose is to provide information and stimulate the debate on debt management. The Internet (www.rgk.se) is used extensively to make information available.
Apart from written material, the SNDO finds it important to be available to investors who want to discuss debt policy. The SNDO also takes initiatives to meet investors both in Sweden and abroad.
Clearing and settlement
All tradable debt instruments used by the SNDO are registered electronically with the Swedish Central Securities Depository and Clearing Organization (VPC), which is a corporation controlled by the major domestic banks. VPC handles the clearing and settlement of all transactions in government securities, as well as interest payments and repayment on maturity. The normal settlement date for treasury bills is T+2, and for bonds, it is T+3.
Clearing in the Swedish system is done on a net basis, making it sensitive to unwinding problems in the event of a major player failing to pay or deliver securities. To cope with this problem, all market players, including the SNDO and the Riksbank, have agreed to support the market through repo arrangements. However, this is not binding, that is, there is still a risk that unwinding problems would occur, which implies that the system does not fully comply with international standards. There is a systemic risk that is implicitly covered by the state. The Riksbank, as responsible for the payment system in Sweden, has demanded that the system be changed. At present, the introduction of a central counterpart seems the most likely solution, but other options, such as gross settlement, have been discussed.
Trends shaping the future market for government securities
Looking ahead, debt management and funding strategy in Sweden will continue to focus on broadening the investor base. This includes attracting new international institutional investors to offset the international diversification by domestic portfolio managers. However, it also involves making government debt instruments available to smaller investors, including retail. In recent years, such a development has to some extent been hindered by the extreme stock market performance; for private individuals in Sweden, fixed-income instruments have not seemed attractive. Following the correction in the stock market, the SNDO has noticed an increased appetite for fixed-income savings.
This raises issues related to how to reach retail investors. More efficient distribution channels might change the role of the authorized dealers and other intermediaries. One scenario could be that investors, retail as well as institutions, enter bids in auctions directly through the SNDO’s web site. The advantage from the investors’ point of view would be that they could enter bids without intermediaries and without revealing information to other market participants. Moreover, intermediaries might not find it worthwhile to invite retail investors to the primary market for government securities, because this may be less profitable than selling other products.
Therefore, the SNDO believes that to reach smaller investors, it is important to develop direct distribution channels. The drawback with such a strategy is that it might be costly to handle small lots. The trend toward straight-through processing should bring down this cost in the coming years, however. Also, the costs involved need to be valued against costs of using intermediaries, such as commissions and underwriting fees. In the more competitive environment faced by sovereign issuers, the fees required for such services may tend to rise.
It is far from certain what rapid technological change will mean for the distribution of debt instruments. However, the SNDO considers it important to be prepared in the event that new distribution channels, for example, based on Internet solutions, turn out to be attractive to investors. Changes are hard to anticipate, making it all the more important not to rely on only one strategy.
If the primary market changes with more direct selling to investors, what will then happen to the secondary market? Today, the secondary market is built around a market-maker system, with banks quoting two-way prices. With a less central role in the primary market, some intermediaries might not find it profitable to commit resources to secondary market activity. Either such a development will go hand in hand with a trend where debt markets, similar to the equity markets, are primarily order driven and market makers are less important; or the issuers will have to find new ways to get support from intermediaries in the secondary market. In either case, it is rational to have alternative distribution channels. However, these alternatives should be developed taking into account the effects on the secondary market. Also, for retail investors, it is important that their bond holdings can be converted to cash at reasonable costs.
Continued technical change is bound to change fixed-income markets as it is changing other financial markets. The separation of the primary and secondary markets, and the strong role for intermediaries in both segments, are two areas where new solutions might come up. Therefore, it is advisable for debt managers to make possible changes in the traditional market structure into account when forming strategies for how to improve the functioning of government securities markets.
The case study was prepared by Lars Hörngren and Erik Thedéen from the Swedish National Debt Office.
This report deals with central government debt management. For brevity, the term “central” will be left out, unless it is needed to avoid confusion with other aspects of public debt.
Expenditures are controlled via the budget, not by ceilings on government borrowing or the size of the debt.
This also reflects a long-standing tradition in Sweden of working with small ministries, which are responsible for policy decisions, and delegating operative functions to agencies that have separate management and are at arm’s length from the ministries.
Concrete illustrations of the contents of the guidelines are presented in the second section.
For more on the organization of the SNDO, see the second section.
The national audit office, an independent agency for central government auditing, also audits the activities and accounts of the SNDO.
The ALM perspective was introduced in the guideline proposal presented in October 2000, available at the SNDO’s web site (http://www.rgk.se/files/upl497-Guidelines_2001.pdf).
The model and the simulation results are described in papers available at the SNDO’s web site (http://www.rgk.se/files/upl553-Teknisk_Rapport.pdf).
The SNDO has used derivatives and other debt management instruments actively for a number of years. The numbers in Figure II.16.1 refer to the exposures when account has been taken of derivatives.
The analysis is presented in the guideline proposal for 2001, available at the SNDO’s web site (http://www.rgk.se/files/upl497-Guidelines_2001.pdf).
The foreign exchange reserves are owned and managed by the Riksbank, reflecting its responsibility for implementing foreign exchange policy. Because the foreign exchange reserves are set aside for a special purpose, they give no hedging effect from the point of view of the central government. The foreign currency debt and the foreign exchange reserves are therefore managed separately.
The government’s decision for 2001 can be found at the SNDO’s web site (http://www.rgk.se/files/upl546-statsskuld_eng.pdf).
The guideline proposal for 2002 is available at the SNDO’s web site (http://www.rgk.se/files/upl1037-Riktlinjer_Eng.pdf).
The government’s decision for 2002 can be found at the SNDO’s web site (http://www.rgk.se/files/upl1115-riktlinjebeslut_2002_eng.pdf).
One could include exchanges and buybacks of outstanding debt in active debt management, because they imply deviations from a plain “issue-and-leave-outstanding” strategy. These instruments are discussed in the third section, because they are also meant to enhance market liquidity.
For an in-depth review of the use of swaps by debt managers (including the SNDO), see Gustavo Piga, Derivatives and Public Debt Management (Zurich: International Securities Market Association), 2001.
A long-term bond combined with an interest-rate swap is equivalent to a (synthetic) floating-rate note.
The SNDO works with symmetrical credit support annex agreements, that is, the it also transfers cash if a counter-party has a net claim, in line with market practice.