Liquidity in the Danish Government Securities Market
8.1 Summary
The government debt policy is based on the assumption that investors are willing to pay a premium for high liquidity. By building up large liquid series in certain maturity segments in the market for government bonds the central government can thus achieve a liquidity premium and cover the borrowing requirement at a lower interest rate.
There is no clear, objective measure of liquidity and the price significance of liquidity. To obtain an accurate picture of liquidity in a market, various dimensions of liquidity, and indicators thereof, must be considered.
This chapter reviews dimensions of liquidity, indicators of liquidity and various alternatives concerning the price effect of liquidity. The focus is on measurable liquidity indicators (bid‑ask spread, turnover, rate of turnover, issue size, etc.) in the Danish government securities market, rather than more theoretical explanations of factors affecting liquidity. There are various proposed alternatives concerning the liquidity premium for Danish government securities compared to government-guaranteed issues and to less liquid government securities. Finally, techniques and results from the most recent studies of liquidity in other bond markets are discussed.
The review indicates that Danish government securities are highly liquid, and confirms the existence of a liquidity premium in the Danish government securities market. Securities with a large outstanding volume and benchmark status are typically the most liquid. Moreover, on-the-run securities quickly become more liquid in the expectation of benchmark status. The relative liquidity premium for benchmarks and on-the-run securities in relation to less liquid off-the-run government securities is assessed to be positive, but relatively modest.
8.2 Liquidity in brief
It is characteristic of a liquid market that market participants can quickly and at any time undertake substantial transactions that have no significant impact on prices. A liquid bond is furthermore characterised by a high turnover rate at low transaction costs.
The concept of liquidity is often divided into the dimensions of tightness, depth, immediacy and resiliency.
- Tightness indicates the costs of selling a bond immediately after purchase
- Depth is the size of the amounts that can be traded without affecting the price
- Immediacy shows how quickly an agent can effect a trading decision (find a counterparty)
- Resiliency indicates how quickly the market is "restored" after a trade.
A liquid market is thus characterised by great depth, a lack of tightness, high immediacy and high resiliency. However, it is difficult to compile the four dimensions individually, since they may be mutually dependent.
The aforementioned characteristics will typically have an impact on price, since high liquidity is attractive to investors who will therefore be willing to pay a liquidity premium for holding more liquid rather than less liquid bonds.
8.3 Indicators of liquidity in the danish market for government securities
The liquidity indicators applied most frequently are issue size, bid‑ask spread and turnover. These are reviewed below with regard to the Danish market for government securities. The relationship between indicators and dimensions of liquidity, and the various factors affecting liquidity, are also briefly discussed.
Box 8.1 describes the data applied to the different liquidity indicators.
|
Liquidity indicators – data |
Box 8.1
|
|
In this chapter day‑to‑day data for prices, outstanding volumes and turnover in the Danish government securities market in the period 1996-2002 is used. Furthermore, measures of non‑residents' ownership share and non‑residents' purchases of government bonds in 2001/2002 are used:
|
|
| Sources: 1 Danmarks Nationalbank and the Copenhagen Stock Exchange. 2 Bloomberg. 3 The Copenhagen Stock Exchange. 4 The figures are based on Statistics Denmark's sector distribution of the circulating securities based on data from VP Securities Services. Danmarks Nationalbank has adjusted for repurchase transactions between Danish banks and non‑residents. Furthermore, estimated adjustments are made for residents' holdings in custody accounts abroad. |
|
Issue size
The size of a loan primarily affects the depth of the market. The larger an issue, the less the price is affected by trades of a given size. Issue size is also of great significance to institutional investors and many major foreign investors. They often have an internal investment policy, or are subject to legislative requirements, prescribing a minimum turnover, a minimum outstanding volume, or a maximum ownership interest in the securities in which investment takes place. The objective of this investment policy is for the investor to move out of a given position without affecting the price.
Issuers often focus on issue size to support liquidity, since to some degree the issuer can determine the size of an issue.
Danish government bonds are typically built up to an issue size of DKK 40‑80 billion, making them the largest issues in the Danish bond market. In October 2002, there were 8 series in the Danish bond market with an outstanding volume exceeding DKK 50 billion, of which the seven largest were government bonds. The 10‑year bonds typically have the largest outstanding volume. Due to their benchmark status in the Danish market, and in view of market‑maker agreements, the strategy is for the 10‑year bonds to be built up to a volume of at least DKK 60 billion. In a situation with a surplus on government finances, the central government has maintained a certain issue size, and thereby liquidity, in the on‑the‑run government securities by conducting an active buy‑back policy, keeping on‑the‑run securities open for a longer period of time, and by issuing in fewer maturity segments.
A simple comparison of issue sizes and turnover shows that the largest issues are typically also the most traded, cf. Chart 8.3.1. However, it is difficult to draw more specific conclusions regarding liquidity in series with an outstanding volume of DKK 80 billion compared to issues with an outstanding volume of e.g. DKK 50 billion.
| Issue size and turnover |
Chart 8.3.1
|
 |
|
| Note: Maximum outstanding volume in the period 1996-2002 indicates the nominal issue size achieved prior to commencement of buy‑backs. Average daily turnover is based on data in the period 1996-2002. Source: Copenhagen Stock Exchange. |
|
Since both size and turnover contribute to market depth, the largest and most traded securities will usually also be the most liquid, although issue size alone does not ensure liquidity. A large outstanding volume can thus be termed a necessary, but not sufficient,condition for liquidity.
Bid-ask spread
Bid‑ask spreads, i.e. the difference between bid and ask prices, are an indicator of market tightness. The narrower the bid‑ask spread, the less expensive it is to quickly leave a position.
In Denmark, a market‑maker agreement in the government bond market exists whereby market‑makers are mutually obliged to quote two‑way prices for a particular amount. An agreement of this type ensures a certain immediacy in the overall market, and limits tightness.
For a given maturity, bid‑ask spreads will in theory be narrower for highly liquid securities than for less liquid securities. This is because a market-maker must hold a stock, or be certain that it can buy the securities quickly in order to honour the obligation to sell a given paper. For more liquid securities, the probability of being able to buy the paper for the purpose of further sale is greater, and the need to have a stock, and thereby stockpiling costs, is lower. The central government's securities lending facility e.g. contributes to reducing the stockpiling requirement and thereby supports liquidity. The spread ventured by a market-maker for a given maturity thus reflects an assessment of liquidity in the market.
| Bid-ask spreads in danish government securities, october 2002 |
Chart 8.3.2
|
 |
|
| Note: Bid-ask spreads are based on average bid and ask quotes at close of business. Average for October 2002. Source: Bloomberg. |
|
Bid‑ask spreads in the Danish market primarily reflect the outcome of market‑maker agreements as well as a tendency for spreads to increase with higher remaining maturity, cf. Chart 8.3.2.
This relation between spread and remaining maturity stems from the risk of price changes assumed by market makers on quoting prices. This risk increases with price volatility in the securities in which prices are quoted. Volatility typically increases with remaining maturity. The greater risk for securities with longer maturities is reflected in wider bid-ask spreads.
The spread in the Danish market is around 10 tics (difference between bid and ask, 1 tic corresponding to 1/100 price unit) in the 10‑year segment, 8 tics in the 5‑year segment, and 6 tics in the 2‑year segment, cf. Chart 8.3.2. These spreads are based on average quoted prices and are thus indicative, cf. Box 8.1. The actual spreads in the professional market are normally somewhat narrower. For comparison, Swedish 10-year government bonds are quoted in a spread of approximately 12 tics, while spreads in the euro area are narrower as a consequence of electronic market making in MTS, where best bid and ask prices are shown.
Turnover
Turnover, i.e. how much a given bond is traded, is an indicator of both depth and immediacy in the market. High turnover shows that a bond is subject to current price quoting and trading, and thereby indicates the liquidity and tradability of a bond.
Turnover is a frequently used indicator of liquidity for which there is well-documented data in Denmark, since all trades are reported to the Copenhagen Stock Exchange. In liquidity terms, however, turnover figures must be interpreted with some caution, since they express only the scale of executed trades, but not potential trades. A bond can e.g. be liquid and tradable without necessarily being traded substantially, as long as the market is willing to execute a substantial number of trades.
The central government is a leading player in the Danish bond market, and the large government bond issues are the most traded in the Danish market. This in itself bears witness to high liquidity in the market for government bonds.
During the past 5 years, the absolute bond turnover in the Danish market has been declining, cf. Chart 8.3.3. The number of trades in the overall Danish bond market has also fallen from almost 1.6 million in 1997 to about 1 million in 2001, while the circulating volume of bonds and the number of securities codes traded on the stock exchange is by and large unchanged.
| Monthly turnover in the danish bond market, 1998-2002 |
Chart 8.3.3
|
 |
|
| Source: Copenhagen Stock Exchange. | |
One explanation for the decline in turnover is consolidation in the financial sector, while another is market developments and stability vis-à-vis the euro area. The consolidation in the financial sector has e.g. meant that today there are fewer, but larger, active banks and securities dealers in the bond market. The individual transactions have become larger, and to an increasing degree the banks take bonds onto their own books and hedge risks via other instruments. The stability vis‑à‑vis the euro area has resulted in fewer trades into and out of krone-denominated bonds, conditioned by market expectations of the krone rate and interest-rate differential, and thereby lower turnover.
The distribution of turnover among the various government bonds reflects a tendency for concentration in the benchmark securities. The concentration is most significant in the 10‑year segment, where the benchmark bond in each period is the most traded bond, cf. Chart 8.3.4. Moreover, on‑the‑run issues also quickly achieve good liquidity, in expectation of achieving benchmark status.
| Average daily turnover in the 10-year segment, 1996-2002 |
Chart 8.3.4
|
 |
|
| Note: The dashed lines indicate change of benchmark. The current benchmark in each period is noted at the top of the Chart. Source: Copenhagen Stock Exchange. |
|
A factor that contributes to generating turnover and liquidity is non-residents' interest in and trading of government securities. Non-resident players are typically willing to trade more frequently than resident players, and for larger amounts.
Non-residents are key investors in the Danish government securities market. Non-residents own approximately one third of all Danish government securities, cf. Table 8.3.1. The Table also shows that during the past year (from the 3rd quarter of 2001 to the 3rd quarter of 2002) non-residents primarily sold old off‑the‑run securities and bought on‑the‑run securities. Non-residents' activity in the market thus supports the build-up of and liquidity in the government's on-the-run issues.
|
Liquidity indicators, october 2002 |
Table 8.3.1 |
|||||
| Unit |
Benchmark/
on‑the‑ run status in 2002 |
Nominal
issue size, DKK billion |
Turnover
ratio, per cent |
Non‑
resident ownership share, per cent |
Non‑
residents' purchases in the past year, DKK billion |
|
| 4 per cent Treasury notes 2004 |
on/benchmark
|
39.5
|
83.7
|
39
|
15.1
|
|
| 4 per cent bullet loans 2008 |
on/benchmark
|
20.0
|
48.4
|
23
|
4.1
|
|
| 5 pct. per cent bullet loans 2013 |
on/benchmark
|
40.6
|
115.5
|
22
|
9.0
|
|
| 5 per cent Treasury notes 2003 |
off/benchmark
|
36.7
|
41.0
|
43
|
-3.8
|
|
| 8 per cent bullet loans 2006 |
off/benchmark
|
57.5
|
15.0
|
24
|
-3.0
|
|
| 6 per cent bullet loans 2011 |
off/benchmark
|
60.5
|
32.7
|
27
|
13.4
|
|
| 6 per cent bullet loans 2002 |
off‑the‑run
|
27.5
|
12.6
|
28
|
-7.3
|
|
| 8 per cent bullet loans 2003 |
off‑the‑run
|
56.7
|
7.22
|
36
|
-10.2
|
|
| 7 per cent bullet loans 2004 |
off‑the‑run
|
67.1
|
21.4
|
35
|
-3.1
|
|
|
5 per cent bullet loans 2005 |
off‑the‑run
|
57.5
|
32.0
|
55
|
0.6
|
|
|
7 per cent bullet loans 2007 |
off‑the‑run
|
52.1
|
23.2
|
19
|
-2.1
|
|
|
6 per cent bullet loans 2009 |
off‑the‑run
|
66.6
|
22.9
|
25
|
-0.7
|
|
|
7 per cent bullet loans 2024 |
off‑the‑run
|
25.0
|
19.0
|
12
|
0.6
|
|
| Note: Non‑resident ownership share is compiled as of the end of the 3rd quarter of 2002, while non‑residents' purchases are compiled as the difference between non‑residents' holdings in the 3rd quarter of 2002 and in the 3rd quarter of 2001. The rate of turnover is calculated as turnover at market price/outstanding volume at market price*100. Source: Copenhagen Stock Exchange and Danmarks Nationalbank. |
||||||
The above review indicates that Danish government securities are generally characterised by high liquidity. Securities with a large outstanding volume and benchmark status are typically the most liquid. On‑the‑run securities moreover quickly gain good liquidity, in anticipation of achieving benchmark status.
Liquidity in government securities implies that investors typically will be willing to pay a liquidity premium. An often used measure is the yield spread between less and more liquid issues. This implies a relative measure of the liquidity premium that varies according to how illiquid a bond the comparison is made with. The issues that are compared should have (by and large) the same coupon and remaining maturity. Otherwise, differing payment characteristics can drive the spread. Differing coupons can e.g. mean that the spread is affected by asymmetrical taxation of coupon payments and capital gains. This can disturb the interpretation of the yield spread. Methods to estimate liquidity premiums, as well as results from other countries, are described in Box 8.2.
|
Liquidity premium - methods and results from other countries |
Box 8.2
|
|
There are typically two approaches to measuring the liquidity premium on government securities. The first approach estimates the yield spread between liquid (on-the-run/benchmark) issues and less liquid issues. The second approach applies the yield spread between government securities and government-guaranteed issues. On‑the‑run versus off‑the‑run In a study from the Spanish central bank (F. Alonso et al., 2002, Estimating liquidity premia in the Spanish government securities market, Banco de España working paper no. 0017), the liquidity premium for benchmark bonds in the Spanish government securities market is estimated by including a liquidity parameter in the estimation of the zero‑coupon-yield curve for government securities. The parameter is zero for benchmark bonds in the estimate of the zero‑coupon-yield curve, and 1 for non‑benchmark bonds. This makes the co‑efficient for the liquidity parameter a measure for the displacement of the zero‑coupon yield curve that non‑benchmarks entail, and thereby a measure for the liquidity premium. In Fleming's study of the spread between off‑the‑run and on‑the‑run issues in the American market the liquidity premium is estimated at between 4 and 10 basis points for 2‑, 5‑ and 10‑year issues. In the study for the Spanish government securities market the liquidity premium is determined at approximately 5 basis points. Government securities versus other government-guaranteed issues |
|
A simple measure of the liquidity premium in government securities compared to a relatively illiquid bond is estimated by comparing a government bond with an alternative issue that otherwise has the same characteristics as the government bond. Chart 8.4.1 shows the yield on 4 per cent Treasury notes 2004 compared with the yield on the Ørestaden 4 per cent bullet loan, likewise maturing in 2004. Ørestaden is partly owned by the central government[1]. Therefore, there is no further credit risk related to the Ørestaden bond. The two bonds have the same coupon and around the same duration. The Ørestaden bond must be assumed to be relatively illiquid with a volume of only DKK 461 million and no daily price-quoting[2], while 4 per cent Treasury notes 2004 has an outstanding volume exceeding DKK 40 billion, and is traded on a daily basis.
| Yield and yield spread for Ørestaden's 4 per cent 2004 and the central government's 4 per cent treasury notes 2004 |
Chart 8.4.1
|
 |
|
| Source: Copenhagen Stock Exchange. | |
The difference between the two yields gives an estimate of the liquidity premium that varies between 5 and 50 basis points – approximately 30 basis points on average in 2002. In view of periods with no price quotes for the Ørestaden bond, the spread shows relatively substantial fluctuation. An aggregate consideration for the entire period is therefore a more appropriate indicator.
The liquidity premium can also be determined as a relative measure between the most liquid and less liquid government securities.The spread between a zero-coupon yield based on the most liquid issues and a zero-coupon yield based on less liquid issues will give this relative measure of the liquidity premium[3]. On-the-run issues and issues with benchmark status are considered to be the most liquid. The zero‑coupon yield curves for the most liquid issues are therefore based on Treasury bills, on‑the‑run securities and benchmark securities in 2002, while the curves for the less liquid issues are based on the other off‑the‑run securities that are bullet loans[4].
The method gives a relative liquidity premium of 0-15 basis points for on-the-run securities/benchmarks in relation to less liquid issues. Chart 8.4.2 shows the difference between the two curves at different maturities.
| Spread between less and more liquid zero-coupon yields at varying maturities, 2002 |
Chart 8.4.2
|
 |
|
| Source: Prices from the Copenhagen Stock Exchange and own calculations. | |
The same method can be applied to a specific bond where the observed yield to maturity in the market is compared with an estimated yield to maturity calculated on the basis of a zero‑coupon yield curve based on less liquid issues.
This calculation
is seen in Chart 8.4.3 where the yield to maturity on 5 per cent bullet loans
2013 is compared with a theoretically calculated interest rate for 5 per cent
2013 based on off‑the‑run securities. The theoretical interest rate
based on off-the-run issues is generally higher than the observed interest
rate, equivalent to a positive liquidity premium of approximately 0-20 basis
points. 5 per cent 2013 achieved benchmark status at the beginning of September
2002, and hereafter the spread tends to be higher than before.
| Spread between theoretical and observed yields for 5 per cent bullet loans 2013, 2002 |
Chart 8.4.3
|
 |
|
| Source: Prices from the Copenhagen Stock Exchange and own calculations. | |
Chart
8.4.3 also shows the spread between the market interest rate and a theoretical
interest rate for 5 per cent bullet loans 2013, where all government securities
are included in the estimate of the zero‑coupon interest rate. This
spread can be used to assess whether a paper is over‑ or underestimated
in relation to the overall government securities market. A positive spread
indicates that the paper is overestimated. This may partly be related to the
fact that the paper entails a liquidity premium[5].
[1] Ørestaden is a partnership owned by the central government and the City of Copenhagen.
[2] Daily prices are not quoted in the loan. In cases where there is no price in the market, the preceding observation is applied.
[3] This method is applied by M.J. Fleming, Federal Reserve Bank of New York in Measuring Treasury Market Liquidity, June 2001, and most recently by the Banco de España Estimating liquidity premia in the Spanish Government Securities Market, 2002.
[4] Zero‑coupon curves are estimated on the basis of an expanded Nelson Siegel's interest‑rate model. The most liquid curve includes: Treasury bills, 5 per cent 2003, 4 per cent 2004, 4 per cent 2008, 6 per cent 2011, 5 per cent 2013 and 7 per cent. 2024. The less liquid curve includes: Treasury bills, 6 per cent 2002, 8 per cent 2003, 7 per cent 2004, 5 per cent 2005, 7 per cent 2007, 6 per cent 2009 and 7 per cent 2024. Treasury bills and 7 per cent bullet loans 2024 are thus part of both curves, in order to fix ends of the curves. It is therefore not relevant to measure the difference between the curves at the ends of the maturity spectrum.
[5] The spread likewise depends on the method applied to calculate the zero‑coupon-yield curve to determine the theoretical yield to maturity. By comparing yields to maturity based on respectively off‑the‑run securities and all government securities a rough indication of the liquidity premium excluding the effect of the estimation method is found.