By concentrating government bond issues in a relatively limited number of popular, standard maturities, governments can assist the development of liquidity in those securities and thereby lower their debt-issuance costs. Markets, in turn, can also use such liquid issues as convenient benchmarks for the pricing of a range of other financial instruments. In addition, spreading the relatively few benchmark issues across a fairly wide range of maturities—building a “benchmark yield curve”—can facilitate more accurate market pricing of financial instruments across a similar maturity range and more generally facilitate better risk management in financial markets. Most industrial countries and some emerging economies have succeeded or made significant progress in developing a benchmark government bond yield curve that spans short-term bills to long-term bonds. Many other countries are in the early stages of developing a yield curve, and often do not have much freedom to issue securities in the full range of maturities needed for a complete yield curve. These countries may seek to develop a more limited number of benchmark securities in those maturities for which there is a market.
4.1 Introduction
Governments can often reduce their debt-service costs through measures that promote the development of deeper, more liquid government securities markets by concentrating their government bond issues in a relatively limited number of maturities. Market participants are typically prepared to pay a premium for a security that can be subsequently traded in a more liquid market—i.e., where sellers or buyers can more readily and cheaply conduct transactions without moving the price of the security against themselves—compared to an otherwise similar security for which the market is less liquid. Debt managers can capture this liquidity premium by standardizing debt instruments in as simple a form as possible and by issuing relatively more of this standardized debt at key maturities. Goldstein and Folkerts-Landau (1994) cite the views of several developed-country debt managers to the effect that such savings for their countries (the difference in rates that had to be paid on benchmark bonds, as opposed to less liquid, non-benchmark issues) were on the order of 5 to 15 basis points.
As long as the maturity and other underlying features of the issues are in reasonable accord with market preferences, larger issues tend to be more liquid because market participants’ typical transactions (including those of the government) are smaller relative to the total supply. Therefore, they can buy or sell desired amounts with less chance of moving the market against them. The market recognition of this liquidity tends to be reinforcing: once some part of the market recognizes an issue as being very liquid, other participants also tend to concentrate their activities in these issues, which makes the security even more liquid. Indeed, since liquid benchmark securities can be readily bought or sold as a vehicle for hedging other financial transactions, many market participants use them as a hedging vehicle in corporate asset and liability management.
Beyond the cost savings to governments, there are also beneficial side effects of building benchmark government bonds in terms of developing a broader financial market. In particular, markets tend to use yields on benchmark government securities to price a range of other financial instruments.59 In pricing a corporate bond, for instance, a spread reflecting default risk, liquidity risk, and other specific risks can be added to the yield curve of the benchmark security with the same maturity as the corporate bond.60 Likewise, by spreading its benchmark issues across a relatively wide maturity spectrum to construct a benchmark yield curve, governments can indirectly facilitate more accurate market pricing across a similarly wide range of maturities for other instruments. Indeed, in many countries the benchmark yield curve underpins the pricing of repo, interest rate futures, and other derivatives markets, such as the interest rate swap market. The existence of these derivatives markets provides additional interest rate hedging options for domestic securities market participants, thereby further enhancing the demand for government securities. In short, building benchmark issues and a benchmark yield curve can contribute importantly to broader securities market development.
These broader benefits may be particularly relevant in emerging and less-developed markets where the development of institutional investors and improved market transparency has further to go. Many transparency problems in the market and difficulties in implementing market-oriented regulation are a direct result of securities-pricing inefficiencies. Accurate pricing of less-liquid debt instruments, such as corporate bonds, is needed for fairness among different groups and generations of investors, and this would be difficult without the reference points provided by benchmarks. Benchmarks can be helpful pricing guides in the case of private pension funds of the unitized type, which are popular in many Latin American countries and increasingly elsewhere. In unitized pension funds, monthly contributions are invested at a net asset value—based price, and accumulated contributions are transferred at the same price. Benchmarks offer similar price guidance for mutual funds, shares of which can be bought and redeemed at frequent intervals. A number of countries are developing matrices for the valuation of less-liquid debt instruments based on net present value, where the rate of discount is calculated by reference to the relationship between the rating of a particular bond and the rating of a risk-free benchmark government bond.
Governments, however, need to be conscious of the trade-offs that, at least in principle, underlie any chosen debt management strategy. In smaller or less-advanced markets such trade-offs may often be more marked. In such markets, even though some of the benefits of the strategy might actually be higher than in other countries, there may nevertheless be important constraints on the extent to which, or the speed with which, a strategy that seeks to develop benchmark issues and a benchmark yield curve can be implemented.
In particular, if the total government debt stock that is tradable is small relative to the size of issue that the market would need before it considers that issue to be liquid,61 the number of benchmark maturities that the debt manager can potentially build up is obviously more limited than otherwise. The fewer the benchmarks that can be built up—implying the debt stock is concentrated in a limited number of maturities—the more likely that rollover risk becomes a concern for the debt manager. Concerns about rollover risk may cause the debt manager to question the desirability of striving to build benchmark government securities issues.
In normal conditions, the trade-off between building liquidity and minimizing rollover risk can be minimized, to some extent at least, by reducing the number of large issues, concentrating issues on maturities that the market considers key, and, if possible, spreading them across a relatively wide range of maturities. However, the ability to issue longer-term securities can often also be constrained where uncertainty about future macroeconomic stability is substantial, since this may significantly raise the term premium that the government will be required to pay when selling longer maturity securities. In less extreme cases, the government will need to weigh how much more it is willing to pay (as a form of investment) to begin developing a longer-term instrument. In a highly unstable macroeconomic environment, there may be no demand for long-term instruments, even at very high rates of interest.
Authorities in many industrial countries and some emerging economies have worked to create a government securities benchmark yield curve that spans short-term bills to long-term bonds. Authorities in less-developed countries, where there may be an unstable macroeconomic environment or a limited domestic government securities market, have been unable to do so. In countries facing such constraints, the current focus is to develop a small number of benchmark securities rather than trying to issue securities across the yield curve.
4.2 Country Experiences with Benchmark Issues
Some insights into how to weigh and reconcile, in practice, the benefits and trade-offs of developing benchmark government securities issues can be gained by examining country experiences in this area. Box 4.1 illustrates some of the essential differences between more liquid and less liquid issues, even in the deepest government securities market in the world, and explains some of the common terminology associated with benchmark issues.
Most industrial countries have successfully developed a benchmark yield curve in their domestic government securities markets.62 (See Figure 4.1 for the benchmark yield curve for Germany and the United States.) Nevertheless, within the overall benchmark yield curve in these countries, the number of benchmark issues varies quite widely, as shown in Table 4.1, while the relative size of benchmark issues compared to nonbenchmark issues—i.e., the extent to which the debt structure is concentrated in benchmark issues—also varies considerably. (see Table 4.3.)
Yield-to-Maturity Curves in German and U.S. Government Bond Markets
(March 29, 2001)
Yield-to-Maturity Curves in German and U.S. Government Bond Markets
(March 29, 2001)
Yield-to-Maturity Curves in German and U.S. Government Bond Markets
(March 29, 2001)
Benchmarks and Their Selections in G–10 Government Bond Markets
Benchmarks and Their Selections in G–10 Government Bond Markets
| Canada | Italy | Japan | United Kingdom | United States | |
|---|---|---|---|---|---|
| Number of benchmarks | Seven (3- and 6-month, 1-, 2-, 5-, 10- and 30-year) | Five (2-, 3-, 5-, 7-, and 10-year) | One (10-year) | Four (5-, 10-, 20-, and 30-year) | Seven (3-, 6-month, 1-, 2-, 5-, 10-, and 30-year) for fixed-coupon securities Two (10-and 30- year) for index-linked bonds |
| How is benchmark selected? | No official rules On-the-run issues for each maturity are usually regarded as benchmarks | No official rules On-the-run issues (with the highest trading volume) for each maturity are usually regarded as benchmarks | No official rules The 10-year bond issue, with large issue amount, long remaining maturity, and being traded near par value, is regarded as the benchmark | No official rules On-the-run issues for each maturity are usually regarded as benchmarks The government targets to create benchmarks for 5-, 10-, 20-, and 30-years | No official rules On-the-run issues for each maturity are usually regarded as benchmarks |
Benchmarks and Their Selections in G–10 Government Bond Markets
| Canada | Italy | Japan | United Kingdom | United States | |
|---|---|---|---|---|---|
| Number of benchmarks | Seven (3- and 6-month, 1-, 2-, 5-, 10- and 30-year) | Five (2-, 3-, 5-, 7-, and 10-year) | One (10-year) | Four (5-, 10-, 20-, and 30-year) | Seven (3-, 6-month, 1-, 2-, 5-, 10-, and 30-year) for fixed-coupon securities Two (10-and 30- year) for index-linked bonds |
| How is benchmark selected? | No official rules On-the-run issues for each maturity are usually regarded as benchmarks | No official rules On-the-run issues (with the highest trading volume) for each maturity are usually regarded as benchmarks | No official rules The 10-year bond issue, with large issue amount, long remaining maturity, and being traded near par value, is regarded as the benchmark | No official rules On-the-run issues for each maturity are usually regarded as benchmarks The government targets to create benchmarks for 5-, 10-, 20-, and 30-years | No official rules On-the-run issues for each maturity are usually regarded as benchmarks |
| Belgium | France | Germany | Netherlands | Sweden | Switzerland | |
|---|---|---|---|---|---|---|
| Number of benchmarks | Two (5- and 10-year) | Seven (3- and 6-month, 1-, 2-, 5-, 10-, and 30-year) Two (10- and 30-year) for index-linked bonds | Four (2-, 5-, 10- and 30-year) | Two (10- and 30-year) | Twelve (2-, 3-, 4-, 5-, 6-, 7-, 8-, 9-, 10-, 11-, 12- and 16-year) | Seven (7-, 9-, 10, 11, 12, and 13-year) |
| How is benchmark selected? | No official rules On-the-run issues for each maturity are usually regarded as benchmarks | No official rules On-the-run issues for each maturity are usually regarded as benchmarks | No official rules On-the-run issues for each maturity are usually regarded as benchmarks | No official rules On-the-run issues with a size of at least Dfl 10 billion for each maturity are usually regarded as benchmarks | All bonds that can be reopened are called benchmarks A new benchmark is created from the first auction of about SKr 4 billion There is also a repo facility of SKr 11 billion so that total available volume outstanding is SKr 15 billion | No official rules Government favors building up liquid benchmarks with maturities of around 10 years |
| Belgium | France | Germany | Netherlands | Sweden | Switzerland | |
|---|---|---|---|---|---|---|
| Number of benchmarks | Two (5- and 10-year) | Seven (3- and 6-month, 1-, 2-, 5-, 10-, and 30-year) Two (10- and 30-year) for index-linked bonds | Four (2-, 5-, 10- and 30-year) | Two (10- and 30-year) | Twelve (2-, 3-, 4-, 5-, 6-, 7-, 8-, 9-, 10-, 11-, 12- and 16-year) | Seven (7-, 9-, 10, 11, 12, and 13-year) |
| How is benchmark selected? | No official rules On-the-run issues for each maturity are usually regarded as benchmarks | No official rules On-the-run issues for each maturity are usually regarded as benchmarks | No official rules On-the-run issues for each maturity are usually regarded as benchmarks | No official rules On-the-run issues with a size of at least Dfl 10 billion for each maturity are usually regarded as benchmarks | All bonds that can be reopened are called benchmarks A new benchmark is created from the first auction of about SKr 4 billion There is also a repo facility of SKr 11 billion so that total available volume outstanding is SKr 15 billion | No official rules Government favors building up liquid benchmarks with maturities of around 10 years |
Maturity Distribution of Debt Instruments and Benchmarks in G–10 Countries
The number of original maturities in the United States dropped to 7, after it stopped issuing 3-year securities in 1995.
Distribution is based on the volume outstanding.
Figures are for the remaining maturity, not for the original maturity.
Maturity Distribution of Debt Instruments and Benchmarks in G–10 Countries
| Canada | Italy | Japan | United Kingdom | United States | ||
| Number of original maturities | 7 | 10 | 8 | 5 | 71 | |
| Original maturities (M – month, Y – year) | 3, 6M, 1, 2, 5, 10, 30Y | 3, 6M, 1, 1.5, 2, 3, 5, 7, 10, 30Y | 3, 6M, 2, 4, 5, 6, 10, 20Y | 3M, 5, 10, 20, 30Y | 3, 6M, 1, 2, 5, 10, 30Y | |
| Distribution by original maturity2 | ||||||
| 1 yr and under | 32% | 17% | 5% | 7%3 | 21% | |
| 1–5 years | 29% | 32% | 8% | 29%3 | 62% | |
| 5–10 years | 27% | 48% | 78% | 34%3 | ||
| Over 10 years | 12% | 3% | 9% | 30%3 | 17% | |
| Number of benchmarks | 7 | 5 | 1 | 4 | 7 | |
| Belgium | France | Germany | Netherlands | Switzerland | ||
| Number of original maturities | 7 | 8 | 6 | 6 | 12 | |
| Original maturities (M – month, Y – year) | 3, 6M, 1, 5, 10, 15, 30Y | 3, 6M, 1, 2, 5, 10, 15, 30Y | 6M, 2, 4, 5, 10, 30Y | 3, 6M, 1, 5, 10, 30Y | 3, 6M, 5, 7, 9, 10, 11, 12, 13, 14, 15, 20Y | |
| Distribution by original maturity2 | ||||||
| 1 yr and under | 19% | 10% | 2% | 4% | 27% | |
| 1–5 years | 6% | 27% | 32% | 10% | 23% | |
| 5–10 years | 43% | 53% | 61% | 74% | 13% | |
| Over 10 years | 32% | 10% | 5% | 12% | 37% | |
| Number of benchmarks | 2 | 7 | 4 | 2 | 7 | |
The number of original maturities in the United States dropped to 7, after it stopped issuing 3-year securities in 1995.
Distribution is based on the volume outstanding.
Figures are for the remaining maturity, not for the original maturity.
Maturity Distribution of Debt Instruments and Benchmarks in G–10 Countries
| Canada | Italy | Japan | United Kingdom | United States | ||
| Number of original maturities | 7 | 10 | 8 | 5 | 71 | |
| Original maturities (M – month, Y – year) | 3, 6M, 1, 2, 5, 10, 30Y | 3, 6M, 1, 1.5, 2, 3, 5, 7, 10, 30Y | 3, 6M, 2, 4, 5, 6, 10, 20Y | 3M, 5, 10, 20, 30Y | 3, 6M, 1, 2, 5, 10, 30Y | |
| Distribution by original maturity2 | ||||||
| 1 yr and under | 32% | 17% | 5% | 7%3 | 21% | |
| 1–5 years | 29% | 32% | 8% | 29%3 | 62% | |
| 5–10 years | 27% | 48% | 78% | 34%3 | ||
| Over 10 years | 12% | 3% | 9% | 30%3 | 17% | |
| Number of benchmarks | 7 | 5 | 1 | 4 | 7 | |
| Belgium | France | Germany | Netherlands | Switzerland | ||
| Number of original maturities | 7 | 8 | 6 | 6 | 12 | |
| Original maturities (M – month, Y – year) | 3, 6M, 1, 5, 10, 15, 30Y | 3, 6M, 1, 2, 5, 10, 15, 30Y | 6M, 2, 4, 5, 10, 30Y | 3, 6M, 1, 5, 10, 30Y | 3, 6M, 5, 7, 9, 10, 11, 12, 13, 14, 15, 20Y | |
| Distribution by original maturity2 | ||||||
| 1 yr and under | 19% | 10% | 2% | 4% | 27% | |
| 1–5 years | 6% | 27% | 32% | 10% | 23% | |
| 5–10 years | 43% | 53% | 61% | 74% | 13% | |
| Over 10 years | 32% | 10% | 5% | 12% | 37% | |
| Number of benchmarks | 2 | 7 | 4 | 2 | 7 | |
The number of original maturities in the United States dropped to 7, after it stopped issuing 3-year securities in 1995.
Distribution is based on the volume outstanding.
Figures are for the remaining maturity, not for the original maturity.
Size Differences Between Average Issues and Benchmark Issues
(US$ billions)
The following issues were used: Belgium, 6.25% due 3/2007; Netherlands, 5.27% due 2/2007; Sweden, 6.5% due 5/2008; Switzerland, 4.25% due 1/2005; United Kingdom, 5.25% due 12/2007.
Size Differences Between Average Issues and Benchmark Issues
(US$ billions)
| Average Issue Size | Benchmark Size | |
|---|---|---|
| Belgium | 1.7 | 5.9* |
| Canada | 2.3 | 6.7 |
| France | 11.5 | 13.0 |
| Germany | 4.8 | 5.3 |
| Italy | 5.5 | 12.3 |
| Japan | 5.2 | 5.7 |
| Netherlands | 3.5 | 6.2* |
| Sweden | 4.0 | 3.8* |
| Switzerland | 0.8 | 2.5* |
| United Kingdom | 5.6 | 15.2* |
| United States | 13.9 | 15.5 |
The following issues were used: Belgium, 6.25% due 3/2007; Netherlands, 5.27% due 2/2007; Sweden, 6.5% due 5/2008; Switzerland, 4.25% due 1/2005; United Kingdom, 5.25% due 12/2007.
Size Differences Between Average Issues and Benchmark Issues
(US$ billions)
| Average Issue Size | Benchmark Size | |
|---|---|---|
| Belgium | 1.7 | 5.9* |
| Canada | 2.3 | 6.7 |
| France | 11.5 | 13.0 |
| Germany | 4.8 | 5.3 |
| Italy | 5.5 | 12.3 |
| Japan | 5.2 | 5.7 |
| Netherlands | 3.5 | 6.2* |
| Sweden | 4.0 | 3.8* |
| Switzerland | 0.8 | 2.5* |
| United Kingdom | 5.6 | 15.2* |
| United States | 13.9 | 15.5 |
The following issues were used: Belgium, 6.25% due 3/2007; Netherlands, 5.27% due 2/2007; Sweden, 6.5% due 5/2008; Switzerland, 4.25% due 1/2005; United Kingdom, 5.25% due 12/2007.
Benchmark Bonds, Benchmark Issues, On-the-Run and Off-the-Run Issues
An on-the-run issue is the most recently issued bond (for a certain term-to-maturity), and is usually treated by the market as a benchmark bond because it tends to be the most liquid or actively traded issue around that term-to-maturity. The new issue will be considered the on-the-run issue until it is replaced by a new bond of the same original term-to-maturity. Once an issue ceases to be on-the-run, it becomes a seasoned issue or an off-the-run issue. A well-seasoned issue is called an off-off-the-run issue. A when-issued security is an on-the-run security that has yet to be auctioned but trades on a when-issued basis in the brief period between the official announcement of a forthcoming auction and the date at which it is actually delivered to the market.
A good illustration of the difference between the various types of securities described above can be found by perusing the various U.S. Treasury note securities with original terms-to-maturity of approximately 10 years that were trading on October 6, 2000. On that date, the on-the-run security was the 5.75 percent August 2010 security, yielding 5.82 percent because this was the security that had been most recently issued by the U.S. Treasury in that part of the yield curve. Meanwhile, several other Treasury notes with similar terms-to-maturity, which had been on-the-run when they were first issued—such as the 6.5 percent February 2010 and the 6 percent August 2009—are no longer being issued by the U.S. Treasury. These latter two securities are thus considered to be off-the-run; they yielded 5.91 and 5.93 percent, respectively, on that date.
The liquidity of off-the-run issues is normally poor compared with on-the-run issues. For off-the-run issues, spreads may be higher and transaction sizes difficult to predict. The table below summarizes key measures of liquidity in U.S. 10-year Treasury note issues (using data between December 1992 and August 1993). Most striking are liquidity differences across different issues. The on-the-run segment has, on average, a transaction every 6.4 minutes, the average volume of transactions is nearly 1,100 million, and the average number of transactions per day is 215. The most recently auctioned securities (on-the-run) and the to-be-auctioned securities (when-issued) have considerably more depth than seasoned issues (off-the-run and off-off-the run) along all dimensions of liquidity.
Liquidity of U.S. 10-Year Treasury Note
Liquidity of U.S. 10-Year Treasury Note
| Statistics for Daily Transactions | ||||||||
|---|---|---|---|---|---|---|---|---|
| Maturity | Number of Transactions | Number of Days | Number | Volume (in millions) | Time Between Trades (in minutes) | |||
| Mean | SD | Mean | SD | Mean | SD | |||
| When-issued | 8,501 | 42 | 202.4 | 170.5 | 1,091 | 924 | 5.8 | 11.5 |
| On-the-runs | 40,459 | 186 | 215.5 | 96.3 | 1,099 | 536 | 6.4 | 5.6 |
| Off-the-runs | 1,589 | 71 | 22.4 | 12.0 | 98 | 70 | 26.9 | 30.2 |
| Off-off-the-runs | 458 | 59 | 5.8 | 5.9 | 28 | 28 | 49.4 | 62.2 |
Liquidity of U.S. 10-Year Treasury Note
| Statistics for Daily Transactions | ||||||||
|---|---|---|---|---|---|---|---|---|
| Maturity | Number of Transactions | Number of Days | Number | Volume (in millions) | Time Between Trades (in minutes) | |||
| Mean | SD | Mean | SD | Mean | SD | |||
| When-issued | 8,501 | 42 | 202.4 | 170.5 | 1,091 | 924 | 5.8 | 11.5 |
| On-the-runs | 40,459 | 186 | 215.5 | 96.3 | 1,099 | 536 | 6.4 | 5.6 |
| Off-the-runs | 1,589 | 71 | 22.4 | 12.0 | 98 | 70 | 26.9 | 30.2 |
| Off-off-the-runs | 458 | 59 | 5.8 | 5.9 | 28 | 28 | 49.4 | 62.2 |
A number of industrial countries have begun facing a new challenge in maintaining a well-functioning yield curve—a declining debt stock resulting from a number of years of fiscal surpluses.63 To date at least, this has not brought seriously into question the approach of concentrating on benchmark issues, though it could do so in the future.64 At this stage, the situation is being dealt with by gradually reducing the size of benchmark issues, buying back older off-the-run issues to maintain new issuance volumes for benchmark issues, and/or reducing the number of benchmark issues. In the United States and Canada, for example, debt managers have taken steps to reduce the number of benchmark maturities. They have stopped issuing securities at the three-year term, and are considering whether to eliminate or substantially reduce issuance volume for other benchmark maturities. They have also been buying back older off-the-run issues in order to maintain new-issuance volumes in the remaining benchmark terms.
The government of Singapore apparently has considered the benefits of benchmark government securities for broader securities market development to be so important that it has established a government securities issuance program at least in part to meet that need, even when it has no pressing need to borrow for its own account (because of a long history of fiscal prudence). Hong Kong, China, likewise, has not had a large need to finance government deficits, but it has issued a range of new governmental bonds to allow for more effective monetary management operations, while also recognizing the usefulness of such securities as benchmarks for financial markets. (See Box 4.2.)
In many less-advanced countries, interest rate benchmark development has not been a policy priority, either because they faced only limited and sporadic financing needs or because monetary policies are conducted through direct credit control. In China, for instance, until 1998 both deposit rates and loan rates of banks were fixed by the People’s Bank of China (the central bank) and monetary policy was implemented through credit plans. Recently, as China has started to liberalize interest rates, the need to develop benchmark securities for indirect monetary policy has been recognized.
Development of Hong Kong Bond Market
In Hong Kong, 1990 is an important dividing line in local bond market development. Prior to 1990, secondary market liquidity in Hong Kong’s local bond market was very low and market infrastructure was underdeveloped. The need for a bond market was very limited, because Hong Kong had a well-developed banking sector, the stock market provided the corporate/private sector with necessary funding, and the government rarely had any deficit financing needs. The Hong Kong Monetary Authority, gradually realizing the importance of a healthy local bond market, initiated the Exchange Fund Bills and Notes program in 1990. Under this program, bills of 91-, 182-, and 364-day maturity were launched in 1990 and 1991. Two-year and three-year Exchange Fund Notes were introduced in May 1993 and October 1993, respectively. The inaugural issue of 5-year, 7-year, and 10-year Exchange Fund Notes followed in September 1994, November 1995, and October 1996, respectively. Twenty-eight-day bills were introduced in November 1996. While debt instruments with different maturities were being developed, the government also worked hard to improve market infrastructure and enhance market demand for government securities. As a result of these efforts, the government was able to establish a market-based, liquid, and reliable benchmark yield curve for Hong Kong dollar debt. This benchmark curve covers a maturity spanning from 28, 91, 182, and 364 days to 2, 5, 7, and 10 years.
In Asia before the 1997 financial crisis, most East Asian emerging-market country governments, except for China and the Philippines, had not been active issuers of government securities because they were running fiscal surpluses. When government securities were issued, most of the securities were held to maturity by financial institutions and contractual saving sectors, resulting in an illiquid secondary market. The yields of such issues were, therefore, not market based and not reliable enough to serve as benchmarks.
4.3 Building a Benchmark Yield Curve: Strategy and Implementation
Although it is ultimately the market that judges whether a security is to be treated as a benchmark, governments typically attempt to develop certain securities into benchmarks through a variety of measures that take into account the features that markets desire. The market often takes on-the-run issues of certain maturities as benchmarks. However, on-the-run issues are occasionally not large and liquid enough to become benchmarks when they are first issued, so they may need to be reopened several times.
While all governments need to understand what market preferences are likely to be, in countries at the early stage of market development, governments may need to be more proactive in influencing what the underlying market preferences are. Governments need to examine market circumstances carefully and consult with the market closely when choosing to target particular issues such as benchmark bond issues.
4.3.1 Standardizing Debt Instruments
A broad variety of debt instruments may allow the debt manager to address the preferences of diverse investors. Too many products, however, can “fragment” the market, to the extent that for any given level of bond market trading volume, the debt stock would be divided into too many instruments to support active trading of any one instrument. Experiences in more mature markets suggest that the cost of market fragmentation far outweighs the benefit of product diversification. Consequently, governments should generally narrow the variety of debt instruments by consolidating and standardizing government securities issues, with an emphasis on issuing marketable Treasury bills and bonds.
An important basic step in building benchmark issues is standardizing a large part of the government securities issue. There are many forms of fragmentation that can arise from the existence of different types of bonds, coupon rates, maturities, issue sizes and frequencies, and whether an issue is an on-the-run or off-the-run issue. Also contributing to fragmentation are different investor profiles. From the policymaker’s perspective, fragmentation presents a serious impediment for creating liquidity and developing benchmark bonds. A fragmented debt structure hinders substitutability between bonds (fungibility), reduces the size and trading volume of benchmark issues, and disperses market liquidity over many issues rather than over a more limited number of benchmark instruments. Fragmentation also limits dealers’ market-making capacity by forcing them to hold a larger number of securities in their inventories or to engage in extra, and possibly costly, risk management activities. Fragmentation still exists in many developed government securities markets, although it is declining over time as governments consolidate their securities issues in Treasury instruments. Box 4.3 illustrates the decline in fragmentation that has taken place in Canada.
Reducing Fragmentation in the Canadian Government Bond Market
In most developed markets, government debt is now issued on a regular basis, in a limited set of maturities (i.e., benchmark maturities) and in relatively large sizes. Market fragmentation is minimized by these practices. However, this was not always the case. For example, in Canada before 1992, the size, maturity, and frequency of domestic marketable bonds were based on market preferences. Consequently, the stock of domestic bonds had irregular original maturities. As of end 1991, 112 fixed-rate C$-denominated marketable bonds were outstanding for a stock of C$157 billion, suggesting a highly fragmented debt stock, with negative implications for the liquidity of the bonds outstanding. However, in 1992, Canada began regularizing its issuance around benchmark securities, with positive results. As a result, in July 2000 there were 68 issues outstanding for a stock of debt of C$286 billion; the number of issues outstanding will eventually decline to less than 50 as the previously issued irregular maturities are gradually retired.
Many emerging economies face severe market fragmentation. In addition to Treasury securities, these countries have many other types of government securities issues outstanding. In some countries (e.g., China, Israel, and the Republic of Korea), special purpose government bonds have been issued to finance specific projects. These special bonds typically are not traded in the secondary market; or, if traded, their size and liquidity are limited by the scope and refinancing needs of the project. These bonds are not in the Treasury-bond class and reduce the size and liquidity of Treasury bonds. As a result, many of these economies are addressing fragmentation on two fronts, since they need to standardize both the Treasury bond market and the government bond market as a whole.
Box 4.4 summarizes the Republic of Korea’s experience in dealing with these challenges. Recognizing the problem brought on by fragmentation of too many government debt instruments, the Korean government in 2000 decided to merge the Grain Security Bond into Treasury bonds. In Brazil, as another example, more than 250 different types of federal securities are currently issued, of which 30 are reported to be “relatively actively” traded and only 10 are somewhat liquid.
Conventional Treasury bonds should serve as the main government funding instrument, and issues of special purpose bonds should be discouraged. The special character of the latter fragments the market and complicates budget management by earmarking certain receipts for specific expenditures.
In many emerging economies, development of the government securities market is also hampered by the existence of both Treasury and central bank securities. (See Chapter 2, Money Markets and Monetary Policy Operations.) Brazil, among others, has started to merge central bank–issued securities into Treasury issues.
Reliance on nonmarketable government securities should also be discouraged. Since the mid-1980s, Israel has taken steps to streamline its government debt instruments by gradually reducing the issuance of nonmarketable government securities in order to develop liquidity for marketable government bonds.65
4.3.2 Developing Appropriate Maturity Distribution for Benchmark Issues
In most developed domestic government securities markets, benchmark yield curves have been established with a view toward spreading the government’s maturities over time to alleviate rollover risk. Table 4.2 provides information on maturities of benchmark and other issues in the domestic bond markets of G–10 countries. The number of original maturities existing in a market can differ significantly from that of benchmark maturities. Most countries tend to have between 5 (in the United Kingdom) and 12 (in Switzerland) original maturities in their issues. However, the number that is deemed by the market to be benchmarks tends to be much less. In many countries the main benchmark maturities are 3 and 6 months, and 1, 2, 5, 10, and 30 years.
In Belgium, Japan, and the Netherlands, benchmark issues are concentrated at the long end of the yield curve, 5- and 10-year maturities for Belgium, 10-year for Japan, and 10- and 30-year for the Netherlands. In contrast, in Canada, France, the United States, and many other countries, benchmark issues cover the whole yield curve as described above. The reason for the diverse maturity mix of benchmark issues has much to do with market preference. Most governments like to create benchmarks covering the whole yield curve, but whether an issue can become the benchmark for a particular maturity is ultimately determined by market practice.
Fragmentation in the Korean Government Bond Market
In the Republic of Korea, six types of domestic government securities are issued: Treasury Bonds (one, three, and five years), Foreign Exchange Stabilization Fund Bond (FESF Bond), Grain Security Bond, National Housing Bond (I) and (II), and Public Land Compensation Bond. The outstanding debt of each bond is depicted in the figure below. Treasury Bonds are relatively liquid. However, because of the wide fragmentation of these instruments, the size of the government bonds, including Treasury bonds, is still very limited. Within the class of Treasury bonds, those issued on different dates were all treated as separate issues, fragmenting the market. Consequently, the trading volume for each issue of Treasury bonds tended to be too small to support the amount of trading needed to create liquidity.
In order to establish Treasury bonds as benchmark securities, the government now concentrates debt issues on Treasury bonds, which account for about 70 percent of annual government debt issuance, and to gradually phase out other types of government securities through buyback programs. The government launched its first fungible issue on May 16, 2000, for a five-year government bond, and now most issuance is fungible, reducing market fragmentation.
While long-term, fixed-interest debt instruments are essential for developing a benchmark yield curve, most developing countries do not have the leeway to issue such instruments given the limited size of their public debt or a history of macroeconomic and financial instability. As a result, some of these countries have opted to issue floating-rate notes and bonds with variable rates. While these instruments have contributed to market fragmentation, they have played an important role in extending the maturity of government securities.
4.3.3 Determining Appropriate Size and Frequency of Benchmark Issues
Once the maturity distribution is decided, the next question facing the debt manager is to set the appropriate size and frequency for benchmark issues. The size of new government securities issues varies across countries, and, as indicated in Figure 4.2, for many countries appears to be closely related to the size of the government bond market.66
Volume of Government Bonds Outstanding Versus Benchmark Issue Size
(US$ billion)
Source: BIS (1999) and French Ministry of Finance staff (2000)Volume of Government Bonds Outstanding Versus Benchmark Issue Size
(US$ billion)
Source: BIS (1999) and French Ministry of Finance staff (2000)Volume of Government Bonds Outstanding Versus Benchmark Issue Size
(US$ billion)
Source: BIS (1999) and French Ministry of Finance staff (2000)In some countries, as shown in Table 4.3, the size of benchmark issues is considerably larger than that of other issues.67 This reflects the government’s efforts to have large benchmark issues to enhance market liquidity. The relative size of these issues also indicates the degree of market fragmentation and the difficulty governments face to consolidate their issues and focus on benchmark issues.
Issuance size also depends on market conditions. With the launch of the Euro in 1999—and hence the elimination of intra-Euro-zone currency speculation as a trading motivation—market participants began to pay more attention to the liquidity of the zone’s various government debt instruments. Since issuance size and liquidity tend to move hand-in-hand, the size of new government securities issues in the Euro-zone countries is converging despite differences in the amount of government debt outstanding. In general, as the global economy becomes increasingly open and integrated, market influence on debt issuance decisions should strengthen.
While governments should not prejudge what size issues can ensure market acceptance of government securities as benchmarks, they do need some basic projections, based in good part on consultation with the market, to determine the appropriate size. The size question can be partly solved by reducing auction frequency and increasing the total amount offered at each auction. The downside to this approach is that a large auction amount per issue may not have enough bidders. Given the total size of a government’s annual borrowing program and the desirability of having auctions on a regular basis over the course of a year, governments may need to explore the possibility of reopening previously issued securities to make new bonds fungible and to achieve the volume needs of the market.
4.4 Reopening and Buyback Operations to Build Benchmark Issues
Reopening and buyback operations (see Box 4.5 for definitions) are used to increase the fungibility of benchmark issues by reducing the number of issues outstanding and concentrating remaining issues on a few benchmark bonds. Such operations help debt managers achieve the desirable size and life cycle for benchmark issues.68 Reopening the same bond with the same maturity and coupon for several consecutive auctions reduces the need to open another issue, keeps the bond in its life cycle, and gradually builds outstanding volume to the desired benchmark level. Buybacks can be used in combination with reopenings to build the size and lengthen the life cycle of those issues targeted to serve as benchmark issues by eliminating non-actively traded bonds, standardizing current outstanding bonds, and maintaining or increasing the volume of bonds targeted for benchmark issues. Details on the use of reopening and buyback operations in various countries are provided in Annex 4.A and Annex 4.B.
Policymakers need to pay attention to the proper timing of reopening and buyback operations. An on-the-run issue may not always be the most liquid of the benchmarks. Even if the life cycle of a government security is long enough to maintain its on-the-run status, it may lose its benchmark status. If the time between issues is too long, the security may reach the point where its maturity no longer corresponds to the needs of the investment community and it may no longer be the most actively traded of the benchmark instruments even though it may still technically be an on-the-run issue.69 Understanding the limits of continuing on-the-run issues helps the government to decide the timing of reopenings and buy-backs. A government should not continue to reopen an issue that has aged and lost its benchmark status—perhaps an old, large, off-the-run benchmark issue that has rolled down the yield curve. In this situation, buying back the instrument may end the circulation of such illiquid securities, thus allowing the government to concentrate on building new benchmark issues.
Reopening and Buyback Operations
Reopenings
A reopening operation helps overcome limited dealer bidding/subscription capacity at a single auction. A reopening operation means that the same bond (with the same maturity and coupon) is issued on more than one occasion to gradually build its outstanding volume to the desired level. Reopening operations improve debt fungibility and provide a tool for the government to concentrate its issuance on a few benchmark issues. Reopenings also have other benefits. In many OECD countries, they have been used to prevent a market squeeze by issuing additional amounts of the same security when it appears that one or more market participants have cornered the existing supply and are able to use their position to command excessive prices from those who would need to acquire the security. In some countries, they are used to create new benchmarks. In the United States, for example, one-fourth of the benchmark six-month Treasury bills are actually reopened one-year T-bills because the T-bills roll down the yield curve, and as off-the-run issues eventually become a benchmark for original maturity six-month bills.
Buybacks
A buyback operation is the repurchase of government bonds prior to maturity. These operations are conducted in the form of reverse auctions, outright purchases in the secondary market, or through bond conversions. Buyback operations have been used to assist the development of a benchmark yield curve by eliminating unwanted bonds, standardizing current outstanding bonds, and maintaining or increasing new issuance volumes. A common problem faced by the government in its efforts to improve the liquidity of targeted benchmark bonds is the limited stock of such bonds. By buying back less liquid and often off-the-run issues ahead of maturity, the government can increase the amount of benchmark securities being issued, in essence using the proceeds from the new debt issues to finance buyback operations.
In addition to assisting benchmark yield curve development, buybacks are widely used for other related purposes. France uses buybacks to smooth debt maturities and manage rollover risk; Austria and Italy use them to reduce government debt; and Belgium and Finland use them to accelerate the redemption schedule of illiquid government bonds.
4.5 Conclusion
Promoting the development of benchmark issues can offer many benefits. It may reduce government borrowing costs for a given maturity, as governments are able to capture the liquidity premium. It reveals price information in the form of a yield curve that facilitates the issuing and refinancing of government securities and the actions of private investors and borrowers, and it helps underpin the development of associated derivatives markets. By fostering more active and informed markets, benchmark issues help to strengthen the transmission mechanism for monetary policy.
Development of benchmark issues, however, can aggravate rollover risk if the government debt outstanding is concentrated in a small number of maturities. In the short run at least, attempts to build benchmark issues along the yield curve may raise debt-service costs, since term premiums tend to be larger for longer-term maturities. The cost of longer-term government securities obviously will be higher in countries with unstable economic conditions. In the long run, however, by providing a better maturity structure to the market, the presence of benchmark securities may help reduce or limit risk premiums on government securities.
Building benchmark issues requires government commitment to a borrowing schedule that may not always coincide with its current funding needs. It will also lead governments that have relied on direct monetary policy control measures to shift to indirect methods that allow markets to function more efficiently.
The balance between the benefits and costs to be derived from promoting benchmark issues will vary from country to country. Authorities thus must determine the appropriate course of development actions. In some countries, political uncertainties and unstable macroeconomic and financial policies will be too pervasive, thereby preventing the emergence of markets for longer-term securities regardless of steps the authorities might take to encourage the development of these markets. In others, the environment may permit development of only a limited number of benchmark issues.
Proper sequencing of measures in developing a benchmark yield curve is important. First, the government needs to assess the benefits and costs of developing such a yield curve and how far out in the future such a curve should extend. Next, the government should take steps to develop the desired yield curve. In addition to achieving macroeconomic stability and improving market regulation and infrastructure, the government, as the most creditworthy domestic issuer, should take the lead in developing benchmark issues through its issuing activities. Ultimately, however, the market will determine which maturities and issues become benchmarks.
The government’s credibility as a government security issuer is established by developing a credible market-oriented funding strategy that emphasizes enhancing market liquidity by issuing mainly tradable instruments, reducing market fragmentation, and improving issue fungibility. Under such a strategy, the government or debt manager can use a variety of issuing activities to develop the benchmark yield curve. The first step is to improve issuing policy choices, such as standardizing government securities instruments, forming a proper maturity mix of benchmark securities, and determining the appropriate size and frequency of the issues. As development progresses, the government may find that operations beyond issuance, including more technical market operations, can be used to help implement policy choices. Such operations can include reopening on-the-run government bonds and/or repurchasing or buying back government bonds from the market.
Annex 4.A Reopening Operations and Their Role in Developing Benchmark Issues
4.A.1 Reopening Operations
Reopening operations have become a standard practice in many countries. As shown in Table 4.4, only Japan, among industrial countries, does not reopen previously issued securities. In France, there is a basic rule that every bond, from a few months to 30 years’ maturity, can be reopened. The practice is also well accepted in other markets, and strongly supported by local and international dealers and investors. In New Zealand, it is common to offer the same maturity bond for 12 consecutive auctions over a period of up to 12 months.
Bond Reopening Systems in G–10 Countries
Bond Reopening Systems in G–10 Countries
| When and how are securities reopened? | |
| Canada | There are 2 types of reopening: 1) to build up new issues via regular reopenings, and 2) to keep the integrity of the market. |
| |
| |
| The reasons the government does so, instead of holding a large initial auction are: 1) a consensus has been reached between bond market participants and the authorities on the maximum size for an auction that dealers can bear, 2) the financing requirements of the government during any period tend not to exceed the size of the reopening, and 3) one is able to “freshen up” issues by adding some bonds to actively trading market participants, who are contrary to the buy-and-hold market participants who tend to take the bond out of circulation. | |
| Italy | There are 3 types of reopening: 1) to build up new issues via regular reopenings, 2) to provide specialists with an additional supply of bonds in “reserved reopenings,” and 3) to maintain market integrity. |
| |
| |
| |
| United Kingdom | There are 2 types of reopening: 1) to build up new issues at 5, 10, 20, and 30 years, and 2) to maintain market integrity. |
| |
| |
| United States | There are 2 types of reopenings: 1) to create benchmarks in the 3- to 6-month range, and 2) to keep market integrity. |
| |
| |
| Belgium | There are 2 types of reopening: 1) to build up new issues via regular reopenings, and 2) to maintain market integrity. |
| |
| |
| France | There is 1 type of reopening: to build up new issues via regular reopenings. |
| The reopening is almost systematically used as the Treasury favors successive securities sales that are parts of a single issue in order to build up sufficiently large securities reserves. | |
| Rules for reopening are: | |
| |
| |
| |
| Reopenings are not conducted to prevent market manipulation. | |
| Germany | There are 2 types of reopening: 1) to build up new issues via regular reopenings, and 2) to increase the amount set aside for market management operations. |
| |
| |
| Netherlands | There is 1 type of reopening: to build up new issues via regular reopenings. |
| |
| Reopenings are not conducted to prevent market manipulation. | |
| Sweden | There is 1 type of reopening: to build up new issues via regular reopenings. |
| |
| Reopenings are not conducted to prevent market manipulation. | |
| Switzerland | There is 1 type of reopening: to build up bonds of sufficient liquidity, especially benchmarks (Sfr 2–4 billion) via regular reopenings. |
| Reopenings are not conducted to prevent manipulation. | |
Bond Reopening Systems in G–10 Countries
| When and how are securities reopened? | |
| Canada | There are 2 types of reopening: 1) to build up new issues via regular reopenings, and 2) to keep the integrity of the market. |
| |
| |
| The reasons the government does so, instead of holding a large initial auction are: 1) a consensus has been reached between bond market participants and the authorities on the maximum size for an auction that dealers can bear, 2) the financing requirements of the government during any period tend not to exceed the size of the reopening, and 3) one is able to “freshen up” issues by adding some bonds to actively trading market participants, who are contrary to the buy-and-hold market participants who tend to take the bond out of circulation. | |
| Italy | There are 3 types of reopening: 1) to build up new issues via regular reopenings, 2) to provide specialists with an additional supply of bonds in “reserved reopenings,” and 3) to maintain market integrity. |
| |
| |
| |
| United Kingdom | There are 2 types of reopening: 1) to build up new issues at 5, 10, 20, and 30 years, and 2) to maintain market integrity. |
| |
| |
| United States | There are 2 types of reopenings: 1) to create benchmarks in the 3- to 6-month range, and 2) to keep market integrity. |
| |
| |
| Belgium | There are 2 types of reopening: 1) to build up new issues via regular reopenings, and 2) to maintain market integrity. |
| |
| |
| France | There is 1 type of reopening: to build up new issues via regular reopenings. |
| The reopening is almost systematically used as the Treasury favors successive securities sales that are parts of a single issue in order to build up sufficiently large securities reserves. | |
| Rules for reopening are: | |
| |
| |
| |
| Reopenings are not conducted to prevent market manipulation. | |
| Germany | There are 2 types of reopening: 1) to build up new issues via regular reopenings, and 2) to increase the amount set aside for market management operations. |
| |
| |
| Netherlands | There is 1 type of reopening: to build up new issues via regular reopenings. |
| |
| Reopenings are not conducted to prevent market manipulation. | |
| Sweden | There is 1 type of reopening: to build up new issues via regular reopenings. |
| |
| Reopenings are not conducted to prevent market manipulation. | |
| Switzerland | There is 1 type of reopening: to build up bonds of sufficient liquidity, especially benchmarks (Sfr 2–4 billion) via regular reopenings. |
| Reopenings are not conducted to prevent manipulation. | |
Many emerging economies have started using reopenings to reduce the number of government bond series and to enhance the liquidity of bonds targeted to serve as benchmarks. In 1990, Israel had 261 series of marketable government bond series and a new series was issued each month in that year. This resulted in a large number of small thinly traded bonds. Realizing the size and the liquidity problem, the government has been trying to reduce the number of bonds issued by reopening the same series over a period of a year or longer. By December 1999, only 135 series were outstanding. (See Figure 4.3.) Because of the reduction in the number of bond series, the average size rose from $250 million equivalent in 1990 to $1.25 billion equivalent in 1999. While reopenings enhance liquidity of the benchmark and other issues, the tax treatment of the discount or premium on a bond at the time of issuance could be an impediment to employing the reopening practice. Box 4.6 summarizes the tax considerations for the use of reopenings in the case of the Republic of Korea.
Number of Tradable Government Bond Series and Average Size of Series in Israel
Source: Ministry of Finance, State of Israel (2000)Number of Tradable Government Bond Series and Average Size of Series in Israel
Source: Ministry of Finance, State of Israel (2000)Number of Tradable Government Bond Series and Average Size of Series in Israel
Source: Ministry of Finance, State of Israel (2000)Tax Impediments for Reopening Practices—The Case of the Republic of Korea
In the Republic of Korea, while there was consensus among practitioners and policymakers that reopenings should be employed to enhance the liquidity of the benchmark and other issues, tax treatment of the discount or premium on a bond at the time of issuance was an impediment. The fiscal authority taxed the coupon amount and the income, amortizing the amount of any discount at the time of issue over the life of the bond to obtain the total income from it each year. This was a reasonably simple way to determine the tax liability, but it prevented the establishment of a series of fungible bonds with which to achieve the objective of a liquid benchmark issue.
In building a fungible series of bonds, it is necessary to allow multiple auctions of the same bond maturity, so that individual bonds become indistinguishable in the secondary market. In sequential offerings of the same bond maturity, bonds will be sold at different prices and yields in each auction. Because current tax treatment of the amount of discount on a bond when it is sold into the primary market is different for each auction, bonds from sequential auctions cannot be combined in investor and dealer portfolios to make them fungible. Thus, the tax treatment of the discount or premium on a bond needed to be addressed before the government could achieve liquid benchmark bond issues through sequential selling of the same bond.
This problem could be resolved in two ways. The definition of taxable income could be changed to one based on the market valuation of the bonds, so that the amount of discount at the time of issue was no longer an element in the calculation. This method would, however, effectively create a capital gains tax on bonds, as any increase in their market value would become taxable income. Given the complexity of introducing a capital gains tax, and the likely need to extend its application to other types of investment such as equities, this may not have been an attractive solution for the problem for Korea.
An alternative was to tax only the coupon interest for reopened issues of government bonds and ignore the discount or premium at issuance. This would be revenue neutral in three ways.
First, the increase or decrease in taxes caused by this alternative treatment would be reflected in the price of the bonds, and this would offset the change in tax revenues. For example, suppose interest rates rose following the first issuance of a new bond, so that it is reopened at a lower price. Under this alternative, the tax revenue would be lower than under the current regime. However, the favorable tax treatment of a discount government bond would raise the price of the bond, offsetting the decline in tax revenue. So far as the government was concerned, the change in tax treatment could be considered revenue neutral. In fact, considering the potential decrease in interest cost due to an increase in liquidity, the proposed change in tax treatment may even be revenue enhancing.
Second, the decline in tax revenues in the case of a discount would also be offset by the increase in revenue in the case of a premium issuance. If an increase in the interest rate was as likely as a decrease, then the change in tax treatments should not change the expected tax revenue.
Finally, the proposed change in tax treatment would not affect the tax revenue, because interest income and capital gains of a financial institution were taxed as a part of corporate income tax. Since the majority of government bonds were held by financial institutions, they were subject to corporate income tax, and thus the proposed tax change would have little overall effect on tax revenue.
Following this line of argument, the government abolished taxation of discount or premium of a bond, and introduced fungible issues on May 16, 2000.
Source: World Bank, Financial and Corporate Restructuring Assistance Project by Capital Market Department 2000
ANNEX 4.B Buyback Operations and Their Role in Developing Benchmark Securities
A buyback is the repurchase of government bonds prior to their maturity. Using the proceeds of new issues to acquire less-liquid instruments typically uses buybacks to pay down debt in an environment of fiscal surpluses, to disperse redemption concentration and manage rollover risk, and to maintain new issuance volume in benchmark issues.
Box 4.7. highlights the basic principles in conducting buyback operations.
4.B.1 Buyback Operations
Three variants of buyback operations are a reverse auction, outright repurchases in the market, and bond conversions. These operations may be conducted by the government debt manager or by the central bank in its open market operations.
4.B.1.1 Reverse Auction
A reverse auction is often used for large-scale repurchase of government bonds. A reverse auction is conducted in a manner similar to a regular auction, except that it is for purchase instead of sale of bonds. Table 4.5 illustrates how a reverse auction works and how it compares with a regular auction, using Canada as an example.
How a Reverse Auction Works: The Case of Canada
How a Reverse Auction Works: The Case of Canada
| Reverse Auction | Regular Auction |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
How a Reverse Auction Works: The Case of Canada
| Reverse Auction | Regular Auction |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
Basic Principles in Selecting Buyback Candidates
The following are basic principles in selecting bonds to be repurchased that are observed by many countries:
Choose only less-liquid and off-the-run issues for buyback programs. Do not buy back current or targeted benchmark issues.
Consider the maturity and coupon structure when selecting bonds to be repurchased. Buybacks should be based on an evaluation of the current debt portfolio, including currency composition, duration, and maturity distribution.
Be aware of other bond market operations. One should not choose a bond to be repurchased if a great proportion of the bond has been stripped.
Consider the impact of buyback programs on the pricing of secondary market issues and weigh the effectiveness and cost of each potential buyback operation.
The most significant difference between a reverse auction and a regular auction is the uncertain size of a reverse auction. In a reverse auction, only the maximum amount the government plans to repurchase is announced. This stands in contrast to a regular auction, where the amount to be issued is known in advance, but the government reserves the right to reject individual bids. This feature allows the government to reject offers that are below the internally set cut-off yield and to reduce the auction amount. It protects the government from locking itself into purchasing bonds at a significantly higher than fair-market price.
4.B.1.2 Outright Purchase
In an outright purchase, the government contacts market makers to solicit directly their offers to sell government bonds. Outright purchases are often used for much smaller-scale repurchase operations than in a reverse auction.
Two types of outright purchases are the “coupon purchase” and the “reverse tap purchase.” A coupon purchase is the purchase of an unspecified amount of bond issues within a specific maturity range. It is frequently used in the Federal Reserve’s open market operations. In a typical coupon purchase, the Fed requests offers to sell government bonds from its primary dealers. The mechanism of the coupon purchase is similar to that of a reverse auction, but there are also some subtle differences. A coupon purchase does not go through a formal auction process. Unlike a reverse auction, the maximum amount to be repurchased in a coupon purchase is not specified before the operation is finished. In this manner, a coupon purchase is more open ended than a reverse auction. A coupon purchase only announces the government’s interest to purchase bonds within a specified maturity range, while a reverse auction tends to focus on a much more restricted range of bond issues, e.g., one or two bond issues. Another important difference is the very short period between the announcement of a coupon purchase and the completion of the operation. The whole process often takes only 10 to 20 minutes.
Box 4.8 describes how the Federal Reserve Bank of New York conducts a coupon purchase repurchase in an open market operation and offers additional operational details on coupon purchases.
In addition to the coupon purchase, the reverse tap purchase is also used to conduct outright (re)purchases. As its name implies, a reverse tap purchase is the inverse of the tap sales of government bonds. In a reverse tap purchase, the government directly contacts primary dealers or brokers in the secondary market and purchases government bonds from them. The reverse tap purchase is good for a relatively small buyback program.
4.B.1.3 Bond Conversion
Another way to conduct a buyback program is a “conversion” or bond exchange. In a conversion program, the government buys back from bondholders some particular series of government bonds, and the payment for the repurchase is made in terms of another series of government bonds, which are often newly issued, more liquid bonds. A conversion can also be used in combination with reopening operations, i.e., the newly issued bonds that are used to pay for the conversion can be a reopened issue. By replacing a less-liquid issue with a more liquid benchmark issue, a bond exchange may be used as an active market management tool to develop benchmark bonds and maintain market liquidity.
Coupon Purchase Operation in the United States
Staff at the Federal Reserve Bank of New York and at the Board of Governors of Federal Reserve System provide the open market operation desk with estimates of the average supply of and demand for reserves to project open market operation needs. If staff projections indicate a large and persistent imbalance between reserve demand and supply, say for a month or more, the desk may conduct an outright purchase or sale of securities. Such transactions permanently increase or decrease the size of the Federal Reserve’s portfolio. The desk conducts far more outright purchases than outright sales, primarily because it must offset the reserve drain resulting from the public’s increasing demand for currency.
Before 1995, the desk entered the market to conduct outright operations only a few times each year. It would wait until reserve needs were large enough to warrant a sizable transaction, on the order of $3–4 billion, in part because such operations, especially coupon purchases, were time consuming. For a coupon purchase, the desk must review numerous offers on about 200 securities. Automation of the bidding process in 1994 decreased the time needed to evaluate offers, but dealers still had to wait a significant amount of time between submitting offers and learning whether their offers had been accepted. For that reason, dealers, in pricing their offers, took into account the risk that market prices might move adversely while they were waiting.
In November 1995, the desk changed its approach to outright coupon purchases. It now divides a purchase of coupon securities, focusing on only a portion of the maturity spectrum rather than on all maturities at once. This approach has further decreased the turnaround time for such operations and has likely resulted in better prices to the desk. The desk still purchases all maturities of Treasury coupon issues, but it generally spreads its purchases over several weeks in keeping with the size of estimated reserve needs. With this new procedure, the desk is better able to tailor its purchases to reserve needs. In addition, the operations, which had been conducted only in the early afternoon, can now be conducted in the morning as well.
When purchasing securities in the secondary market, the desk entertains bids from primary dealers on all securities of a particular type (Treasury bills or Treasury coupon securities) and, for coupon securities, in a particular portion of the maturity spectrum. In determining which bids to accept, the desk considers the bids that represent the highest yield (for purchases) or the lowest yield (for sales) relative to the prevailing yield curve. To avoid holding too large a share of any one issue, the desk also considers the size of its holdings of the particular issue relative to the total amount outstanding.
Source: Edwards 1997
A bond conversion can be carried out in either of two methods, representing two distinct means to determine the price ratio (or conversion ratio) of the converted bonds to new bonds. The first is a competitive-offer format, where the conversion ratio is determined by the bids submitted by market participants. The participants also submit bids on the amount of the conversion. A competitive-offer format can be implemented by either a reverse auction or a coupon purchase operation. They are the same as a conventional auction and a conventional coupon purchase, respectively, except that their offers are in terms of a price ratio and in essence they are paying for the new bonds by another, old bond.
The second manner of conducting a bond conversion is a fixed-rate format, where the conversion ratio is predetermined by the government according to the prices prevailing in the secondary market at the time of announcement. After the government’s announcement of a conversion program, holders of the targeted bonds have a limited amount of time to decide whether to accept the government’s offer before it closes. The actual volume of conversions is determined by the market; the conversion amount will be open ended and there is no guarantee that the government will be able to buy back all the targeted bonds.
4.B.2 Choice of Buyback Program(s)
Effectiveness and cost of the operation are the two basic criteria for choosing a buyback program.
Effectiveness is determined by the government’s ability to repurchase government securities in large volume, to consolidate government bonds outstanding, and to establish liquid benchmarks. The reverse auction and bond conversions can handle larger-scale repurchases of government bonds, the coupon purchase is good for medium-scale repurchases, and the reverse tap works best for small repurchases.
The government suffers a cost when repurchasing the bonds at a price above the fair value. This could happen because the government repurchase announcement has an impact on the price of related bonds in the secondary market. The degree of impact has much to do with the mechanism of the operation. It is negatively correlated to the time between the announcement and the closing of the transaction. Gravelle (1998) examined the secondary market pricing implication of buyback operations and concluded that the reverse auction and bond conversions are most likely to put the government in an unfavorable position when pricing buybacks. The reverse tap tends to give market participants the least opportunity to charge the government a price above the fair price. Table 4.6 summarizes the effectiveness and cost of different types of buyback programs. Reverse auction and bond conversion allow the government to conduct a large amount of buybacks, but also most likely subject the government to unfair prices; the opposite is true for the reverse tap. The coupon purchase results in average effectiveness and cost.
Effectiveness and Cost of Buyback Programs
Effectiveness and Cost of Buyback Programs
| Buyback Program | Effectiveness | Cost | |
|---|---|---|---|
| Reverse Auction | Best | Highest | |
| Outright Purchase | |||
| Coupon Purchase | Average | Average | |
| Reverse Tap Purchase | Worst | Lowest | |
| Conversion | Best | Highest | |
Effectiveness and Cost of Buyback Programs
| Buyback Program | Effectiveness | Cost | |
|---|---|---|---|
| Reverse Auction | Best | Highest | |
| Outright Purchase | |||
| Coupon Purchase | Average | Average | |
| Reverse Tap Purchase | Worst | Lowest | |
| Conversion | Best | Highest | |
Bibliography
APEC (Asia-Pacific Economic Cooperation). 1999. Proceedings of APEC Workshop on Development of Domestic Bond Markets. Hong Kong.
BIS (Bank for International Settlements). 1999a. How Should We Design Deep and Liquid Markets? The Case of Government Securities. Basel, Switzerland.
BIS (Bank for International Settlements). 1999b. Market Liquidity: Research Findings and Selected Policy Implications. CGFS Publications No. 11. May. Basel, Switzerland.
Bennett, Paul, Kenneth Garbade, and John Kambhu. 2000. “Enhancing the Liquidity of U.S. Treasury Securities in an Era of Surpluses.” Federal Reserve Bank of New York, Economic Policy Review 6 (1): 89–119. Available at www.ny.frb.org/.
Campbell, John Y., and Robert Shiller. 1996. A Scorecard for Indexed Government Debt. Yale University Discussion Paper No. 1125, Yale University, New Haven, Conn.
Edwards, Cheryl. 1997. “Open Market Operations in the 1990s.” Federal Reserve Bulletin. 83 (11): 859–74.
Fabozzi, Frank. 1996. Bond Markets, Analysis and Strategies. Englewood Cliffs, N.J.: Prentice Hall.
Fleming, Michael J. 2000. “The Benchmark U.S. Treasury Market: Recent Performance and Possible Alternatives.” Federal Reserve Bank of New York, Economic Policy Review 6 (1): 129–45. Available at www.ny.frb.org/.
Goldstein, Morris et al. 1994. International Capital Markets: Developments, Prospects, and Policy Issues. International Monetary Fund, Washington, D.C.
Gravelle, Toni. 1998. “Buying Back Government Bonds: Mechanics and Other Considerations.” Bank of Canada Working Paper. Ottawa, Canada.
Gravelle, Toni. 1999. “Liquidity of the Government of Canada Securities Market: Stylized Facts and Some Market Microstructure Comparisons to the United States Treasury Market.” Bank of Canada Working Paper 99–11. Ottawa, Canada.
Hong Kong and Shanghai Banking Corporation (HSBC). 1999. A Guide to the Philippine Domestic Debt Capital Market.
Hui, Liu. 2000. Development of Government Bond Markets in DMCs: People’s Republic of China. Asian Development Bank Conference on Government Bond Markets and Financial Sector Development in Developing Asian Economies.
Inoue, Hirotaka. 1999. The Structure of Government Securities Market in G–10 Countries. Bank for International Settlements, Basel, Switzerland.
IMF (International Monetary Fund) and World Bank. 2000. Draft Guidelines for Public Debt Management. Washington, D.C. Available at www.imf.org/external/np/mae/pdebt/2000/eng/index.htm.
Kahn, Michael. 2000. “Promoting Liquid Government Securities Benchmark: The Case of Israel.” Second OECD/World Bank Workshop on Development of Fixed-Income Securities Markets in Emerging Market Economies. Washington, D.C.
Price, Robert. 1997. “The Rationale and Design of Inflation-Indexed Bonds.” Working Paper 97/12, International Monetary Fund, Washington, D.C. Available at www.imf.org.
Sundaresan, Suresh. 1997. Fixed Income Markets and Their Derivatives. Cincinnati, Ohio: International Thomson Publishing.
World Bank. 1999. Financial and Corporate Restructuring Assistance Project (FCRA). World Bank, Washington, D.C.
Benchmark bonds are typically securities of the highest credit quality—i.e., in many cases they are government securities (although this need not always be the case), owing to their sovereign backing, assuming the government in question, is a credible issuer (see Chapter 3, A Government Debt Issuance Strategy and Debt Management Framework). Benchmark securities are also typically the largest and most liquid issues. This is because the prices of liquid and actively traded securities are generally more stable than those of other securities, other things being equal, so that such securities are more reliable as benchmarks for the pricing of other securities.
A yield curve is the relationship between the yield on securities and their maturities at a particular point of time. In graphical terms, it can be represented as the plot of yield to maturity on the y-axis against the time to maturity, shown as the maturity date, along the x-axis. It can serve as an important analytic device for markets and policymakers, as well as a guide for more efficient pricing and risk management. The most useful form of yield curve would reflect the yields and maturities of market-based securities of similar nature, credit quality, and liquidity. A “benchmark yield curve” refers to a yield curve constructed from the yields on benchmark securities.
The debt stock available for market trading may be relatively small because the volume of domestic debt issuance is itself small (e.g., the government may have had a long history of strict fiscal discipline or may have funded itself through foreign borrowings), or because an important proportion of the total amount of the issue is held by institutions that are either required holders of government securities or buy-and-hold investors.
See Fleming (2000) and commentary thereon for recent U.S. experience with benchmark issues.
See Bennett, Garbade, and Kambhu (2000) and commentary thereon.
If the stock of government securities were to become too small in some countries to provide a benchmark yield curve, there may be some other instruments in those countries that can, at least to some degree, approximate the benchmark function of government securities. These might include government agency debt, mortgage-backed securities, good-quality corporate debt, and the swap market. Similarly, foreign investors in some emerging-market countries without well-developed domestic markets sometimes use interest rate and currency swap transactions to construct an approximate local currency yield curve. See Fleming (2000) and commentary thereon.
Before 1990, government-financing instruments in Israel were dominated by nonmarketable government bonds, most of which paid a fixed, higher than market, yield. The process of developing a tradable bond market began in 1986, and the nonmarketable component has steadily declined, thereby significantly improving the tradability/liquidity of the bond market. As a result, the proportion of tradable government bonds has increased from 16 percent of the domestic debt in 1986 to 52 percent by December 1999. The market value of tradable government bonds stood at US$36 billion equivalent in December 1999.
If all issues are benchmarks, the average issue size and benchmark size are identical.
The “life cycle” of a benchmark issue refers to the period between the issue date of a bond and the date it becomes on-the-run.
For example, a large seven-year original maturity off-the-run old benchmark issue rolling down the yield curve becomes a bond with five years of remaining life. Its five-year benchmark status might compete with a five-year original maturity on-the-run issue that now has only a four-year maturity, but still maintains its on-the-run status because of reopenings that lengthen its life cycle.
