The Danish Mortgage-Credit Market – a Liquidity Dilemma


Claus Johansen, Financial Markets and Elena Kabatchenko Nielsen, Statistics

INTRODUCTION

The Danish market for mortgage-credit bonds1 has undergone con siderable development over the last few decades. The total outstanding volume has risen in step with housing prices, and the issuers' introduction of new mortgage products has increased the number of bond types available.

In addition to having a major impact on the structure of the mortgage-credit market, product development has affected pricing patterns. Liquidity is dispersed over more bond series for each issuer, and the wider range of remortgaging options entails faster shifts in the liquidity of the individual series. Scarcity premiums on illiquid bond series would indicate that the outstanding volume and ownership concentration are significant to pricing in the market.

This article investigates whether borrowers experience so-called "lock-in effects" in relation to mortgage-credit bonds because investors are able to demand a higher price for series concentrated on a few invest ors. The conclusion is that lock-in effects for borrowers do exist in certain bond series. For investors, it is found that the ownership concen tration has a major impact on pricing, as scarcity premiums tend to apply to illiquid bond series. These effects for borrowers and investors are seen in bond series with small outstanding volumes and in series that are targets for buy-back in connection with the borrower's remortgaging at a higher coupon rate.

On the basis of the findings in this article, new statistics are introduced showing the ownership concentration for the individual bond series. In future, these statistics will be published as part of Danmarks Natio nalbank's securities statistics by ISIN codes.

STRUCTURAL TRENDS IN THE MORTGAGE-CREDIT MARKET

Over the years, there has been a high degree of consolidation among issuers in the Danish mortgage-credit market, which has concentrated issues on fewer bond series and enhanced liquidity2. Since 1996, when the first adjustable-rate mortgage-credit loans were introduced, the palette of loan types on offer has, however, been widened considerably. In 2003, deferred amortisation was introduced, followed by capped adjustable-rate loans in 2004, cf. Hansen and Thuesen (2006). Several of these product types had been tried before, but not with the same success. For example, loans with unlimited deferred amortisation, i.e. perpetual loans, were introduced as early as in the 19th century, cf. Ebbesen (2008). Overall, no concentration has been observed at series level, as the range of products offered by the issuers has also increased considerably. The introduction of adjustable-rate loans is the single most important factor influencing the structure of the Danish mortgage-credit market in recent decades, cf. Chart 1.

STRUCTURE OF THE DANISH MORTGAGE-CREDIT MARKET OVER TIME
Chart 1

chart 1

Note: Adjusted for a higher outstanding volume in December of each year.
CONSEQUENCES TO THE MORTGAGE-CREDIT MARKET OF A WIDER RANGE OF MORTGAGE-CREDIT PRODUCTS

Issuers
On the issuer side, product development in the market for financing real property reflects competition to offer borrowers new, more sophis ticated products. Consequently, issuers must develop new bond types.

The incentive to innovate lies in the fact that a credit institution can win market shares from its competitors by introducing a new product that meets the requirements of a large group of borrowers. For this reason, competitors are often quick to replicate successful new products. The product development seen since the mid-1990s, combined with the balance principle, has increased the number of open series offered by each issuer, cf. Chart 2.

NUMBER OF OPEN SERIES IN FOUR MORTGAGE-CREDIT INSTITUTES, 1990-2007
Chart 2

Chart 2

Note: The increases in 2007 are partly attributable to the shift to SDO legislation. At the turn of the year 2007-08, all series under the previous legislation were closed.
Source: RIO

Callable mortgage-credit loans are financed via bonds with charac teristics resembling those of the loans. Practically all recent callable mortgage-credit loans are offered in versions with and without amor tisation. The number of underlying bond series is thus considerable.

In contrast, the bonds used to finance adjustable-rate loans do not have characteristics similar to those of the loans. Instead, financing takes place via issuance in series of fixed-rate bullet loans. In recent years, such issuance has primarily taken place in bonds with a coupon of 4 per cent, whereas previously, when interest rates were lower, the coupon was 2 or 3 per cent. Adjustable-rate loans thereby generate a great many bond series in the Danish mortgage-credit market. Unlike other loan types, deferred-amortisation loans are not based on special bond series; issuance is merely adapted to the borrower's chosen redemption profile. Adjustable-rate loans come in many varieties. For a more detailed re view, see Christensen and Kjeldsen (2002).

Issuance of non-callable fixed-rate bullet bonds for financing adjust able-rate loans ("fixed bullets") is greatest in very short-term series with a remaining term to maturity of up to one year. However, a certain volume is also issued in series with a remaining term to maturity of up to three years, cf. Chart 3.

DEVELOPMENT IN OUTSTANDING VOLUME OF FIXED BULLETS, BY MATURITY
Chart 3

Chart 3

Note: "Up to and including 1 year" comprises remaining term to maturity <= 13 months.
"1.1-3 years" comprises 13 months < remaining term to maturity <= 37 months.
"More than 3 years" comprises remaining term to maturity > 37 months.
Adjusted for a higher outstanding volume in December of each year.

It is sought to build up liquidity in mortgage-credit bonds by issuing in the same series for a prolonged period. Issuance in callable fixed-rate bond series typically takes place over a period of three years. Bullet loans for financing of adjustable-rate loans can be used from the first issues with a remaining term to maturity of up to 11 years until the last issues in the year up to maturity. However, shifts in the level of interest rates may entail that series are replaced by other series with higher or lower coupons. On the other hand, such shifts may also lead to renewed issuance in old series.

Investors
If investors are to be able to buy and sell positions in a bond series without affecting market prices, it is in their interest that liquidity is built up in the individual series. Investors holding positions in illiquid series may be forced to accept a price compromise if they need to sell a pos ition in the market. Consequently, investors may be willing to pay a li quidity premium when buying bonds in liquid series.

The wider range of remortgaging options for borrowers as a result of the issuers' product development increases the liquidity risk incurred by investors. There is now a greater probability that the status of a series will change from liquid to difficult to sell in the market.

Although investors may be prepared to pay a premium for liquid bond series, smaller series often tend to be more highly priced in the market than large series. The reason is that it is easier for investors to demand a premium for selling out in a small bond series. Consequently, a scarcity premium may apply to illiquid mortgage-credit bond series. This situ ation mainly arises if the ownership concentration in a series is high.

Over the years, a number of initiatives have been taken to address the issue of dispersed liquidity and price differences between identical bond series owing to differences in volumes. For example, members of the Danish Securities Dealers Association have established a voluntary market maker arrangement for a number of unit bonds, cf. Hansen and Thuesen (2006). Most recently, in 2008, it has been sought to introduce forwards on callable mortgage-credit bonds. The first contract under the Danish Securities Dealers Association's market maker arrangement is a December contract for delivery of any chosen bond from an underlying basket comprising the most liquid series in the 30-year 5-per-cent segment.3 If liquidity is built up in this forward, this will contribute to investors buying up relatively inexpensive securities from the underlying basket, thereby ironing out any price differences. Turnover has, however, been very low in the first months, possibly on account of the prevailing market turmoil.

Borrowers
Product development by issuers has given borrowers more choice. In terms of interest-rate risk, the introduction of adjustable-rate loans and capped loans has increasingly allowed borrowers to select loan types to match their risk appetite. Likewise, the introduction of deferred amor tisation has enabled borrowers to structure their redemptions in relation to alternative savings options and their own private finances.

The potential illiquidity of bond series impacts on the price of the loan when it is raised and in connection with any buy-back from the market. When a new series is opened, it is often uncertain whether or not it will become liquid. Consequently, a scarcity premium may be applied to a se curity as investors become convinced that it will close with a low out standing volume, or that it will subsequently become illiquid due to prepayments and/or buy-back in the market. Borrowers issuing early in the opening period in what later turns out to be an illiquid series thus risk missing out on a scarcity premium on issue, as well as having to pay more if they subsequently buy back in the market. The scarcity premium is often referred to as a lock-in effect for borrowers.4

LOCK-IN EFFECTS IN MORTGAGE-CREDIT BONDS

Lock-in effects for borrowers generally occur when a bond series is concentrated on a few investors. The following is an analysis of the extent to which lock-in effects are observed in the largest two types of mortgage-credit bonds in the Danish market, i.e. callable fixed-rate bonds and fixed bullets.

In callable fixed-rate bond series, a lock-in effect can occur in a situation where the borrower wishes to buy back bonds in the market, i.e. when redeeming a loan or remortgaging at a higher coupon rate. For borrowers wishing to exercise their right to redeem the bonds at par or another applicable price, lock-in due to illiquidity will not play any role. The built-in prepayment price thus sets an upper limit to the lock-in premium payable by the borrower as the bonds can always be redeemed at the agreed price.

Borrowers with adjustable-rate loans do not have this call option. Consequently, there is no upper limit to the premium that the borrower risks having to pay upon premature redemption if the ownership con centration is high. On the other hand, the lock-in does not apply beyond the date when the interest rate has to be adjusted. As previously stated, the majority by far of adjustable-rate loans are issued in short-term bond series with a remaining term to maturity of up to three years. For fixed bullets, lock-in is therefore most likely to occur in the longer series, i.e. the least liquid ones.

To describe the extent of lock-in effects, the correlation between ownership concentration and price is investigated for closed mortgage-credit bond series. The ownership concentration is measured as the maximum ownership share5, calculated as the share of a series owned by the largest investor6 in the series. If one investor owns 60 per cent of a bond series, the maximum ownership share in this series is thus 60 per cent. All bonds are grouped according to ownership concentration:

  1. Low: maximum ownership share < 20 per cent.
  2. Medium: 20 per cent ≤ maximum ownership share < 50 per cent:
  3. High: maximum ownership share ≥ 50 per cent.

To analyse the significance of ownership concentration to pricing of the bond series, an option-adjusted spread (OAS) is calculated between mortgage-credit bonds and the government yield curve. OAS indicates the spread between a bond's yield to maturity and a risk-free interest rate, taking into account the value of the option element in the bond. The higher the OAS, the higher the excess interest rate and the lower the price of the bond.

Besides describing the impact of ownership concentration on pricing, the following investigates the correlation between ownership concen tration and liquidity. In this context, liquidity is measured as the average nominal outstanding volume, and as the average number of days in the period of observation on which securities within each group were traded.

For callable fixed-rate bonds, the analysis of lock-in effects is based on the period end-May to end-October 2008, during which, inter alia, 30-year bond series with a coupon of 7 per cent were opened. For fixed bullets, the analysis is purely based on data for June 2008, as the number of trade observations is higher for this group of bonds.

Lock-in effects in callable bonds7
In step with the assumption that it is easier for an investor to obtain a high share in a small series, all coupon groups for callable fixed-rate bonds tend to have a higher ownership concentration in series with low outstanding volumes. This is also reflected in the lower number of trade days for bond series with the highest ownership concentration, cf. Table 1.

OWNERSHIP CONCENTRATION AND LOCK-IN EFFECTS FOR CALLABLE FIXED-RATE BONDS
Table 1
  Outstanding volume kr. billion1 Number of trade days2 OAS (bp) 3
Coupon: 4 per cent
Low concentration
4.02
20.23
71.18
Medium concentration
0.74
4.80
67.41
High concentration
0.16
1.00
63.00
Coupon: 5 per cent
Low concentration
5.12
23.76
97.54
Medium concentration
0.47
3.25
89.60
High concentration
0.23
1.83
68.18
Coupon: 6 per cent
Low concentration
1.20
11.35
140.59
Medium concentration
0.12
2.19
141.19
High concentration
0.06
1.21
157.87
Coupon: 7 per cent
Low concentration
0.09
2.25
160.99
Medium concentration
0.04
1.06
183.04
High concentration
0.02
0.67
186.44
1 Average outstanding volume as at end-June 2008.
2 Average over 104 banking days (30 May - 23 October 2008).
3 Average June-October 2008. The maximum ownership share shows a high degree of persistence over time. It is therefore reasonable to use the calculated maximum ownership share as at end-June in the comparisons over the full period of observation.

The excess interest rate is highest for the high coupon groups. For the lowest two coupon groups, the correlation between ownership concen tration and average OAS is negative, which means that bond series with high ownership concentrations tend to be traded at higher prices than those with low ownership concentrations. For the 5-per-cent coupon group, this negative correlation is significant at a significance level of 5 per cent8, thereby verifying the existence of lock-in effects in this coupon group. The correlation is insignificant for the 4-per-cent coupon group. No similar tendency is observed for the 6- and 7-per-cent coupon groups, presumably because these bond series have not seen large volumes of buy-backs as the price has been relatively close to par.

Lock-in effects in fixed bullets
In recent years, adjustable-rate loans have primarily been financed via issues in bond series with a coupon of 4 per cent. Consequently, only this segment is considered in the following. Moreover, bonds with a remaining term to maturity of less than one year are disregarded as calculations of yield spreads9 to the government yield curve for securities in this maturity segment are highly sensitive to the choice of underlying model for the government yield curve.

As was the case for callable bonds, it is seen that fixed bullets in series with medium or high concentrations are characterised by lower out standing volumes and a lower average number of trade days, cf. Table 2.

OWNERSHIP CONCENTRATION AND LOCK-IN EFFECTS FOR FIXED BULLETS
Table 2
  Outstanding volume kr. billion1
Number of trade days2
OAS (bp)3
Remaining term to maturity: 1-3 years
Low concentration 9.40
16.29
48.73
Medium concentration 1.29
4.25
52.01
High concentration 0.26
1.31
40.75
Remaining term to maturity: more than 3 years
Low concentration 5.25
10.22
54.82
Medium concentration 0.92
4.93
47.97
High concentration 0.18
1.81
47.12
1 As at end-June 2008.
2 Average over 20 banking days in June 2008.
3 Average, June 2008.

For bonds with a remaining term to maturity of more than three years, there is a significant negative correlation between OAS and ownership concentration at a significance level of 5 per cent. The price for the high-concentration series is thus significantly higher than for the low-concen tration series. This correlation indicates lock-in effects in the high-con centration series for long-term fixed bullets. For bonds with a remaining term to maturity of 1-3 years, the tendency is not significant.

The tendency for illiquid securities – callable fixed-rate bonds with a low coupon, as well as fixed bullets with a remaining term to maturity of more than three years – to be traded at higher prices than liquid se curities indicates that investors are able to retain positions until another investor or a borrower is willing to pay a scarcity premium to buy the bonds.

OWNERSHIP CONCENTRATION AS A NEW VARIABLE IN DANMARKS NATIONALBANK'S SECURITIES STATISTICS

The above analysis shows a significant correlation between ownership concentration and bond price for several segments of the mortgage-credit market.

For issuers, this correlation between ownership concentration and pre mium for a bond indicates that the product development incentive should be balanced against the value to borrowers of liquidity in the bond series.

For investors, the ownership concentration has a significant impact on pricing of bonds, as a scarcity premium tends to apply to illiquid series. This premium relates to the higher concentration on a few investors generally seen in these series. On the other hand, the analyses show that borrowers can benefit from, whenever possible, issuing in series that are already liquid or very likely to become liquid. Publication of the max imum ownership share variable is therefore assessed to be relevant to issuers, investors and borrowers alike.

On the basis of the findings in this article, Danmarks Nationalbank's securities statistics by ISIN codes are expanded to include information about ownership concentration, expressed by the maximum ownership share variable, cf. Box 1. This variable contributes to greater transpar ency in the market for mortgage-credit bonds.

OWNERSHIP CONCENTRATION INCLUDED IN SECURITIES STATISTICS
Box 1

As from January 2009, the securities statistics by ISIN codes will be expanded to include an "ownership concentration" variable. To each bond series in the statistics, an indicator will be applied, reflecting its ownership concentration:
Group 1: Low, maximum ownership share < 20 per cent.
Group 2: Medium, 20 ≤ maximum ownership share < 50 per cent
Group 3: High, maximum ownership share ≥ 50 per cent.

The new securities statistics by ISIN codes are in accordance with the applicable non-disclosure principles and will only be published for bonds.

LITERATURE

Christensen, Anders Møller and Kristian Kjeldsen (2002), Adjustable-Rate Mortgages, Danmarks Nationalbank, Monetary Review, 2nd Quarter.

Danmarks Nationalbank (1998), Danish Government Borrowing and Debt 1997.

Ebbesen, Martin (2008), Everything has been Tried Before – Almost (in Danish only), The Danish Association of Chartered Estate Agents, Ejen domsmægleren, February.

Hansen, Lars Jul and Jesper Ulriksen Thuesen (2006), Trends in Mortgage-Credit Financing: the Market and its Players, Danmarks Nationalbank, Monetary Review, 1st Quarter.

Jacobsen, Svend (2007), Lock-In Effects on Housing Loans (in Danish only), Finans/Invest, No. 8.


[1] In this article, the term "mortgage-credit bonds" is used as a collective term for bonds issued to finance loans against real property as collateral. It therefore comprises traditional mortgage-credit bonds, as well as covered bonds and covered mortgage-credit bonds issued by banks and mortgage-credit institutes.

[2] The liquidity of a security is used as a generic term for negotiability in the market. The term is often applied as an expression of the total outstanding volume in a bond series, as a large outstanding volume is typically linked to high negotiability, cf. Danmarks Nationalbank (1998), Chapter 8.
[3]For further information about the forward, see the press release of 28 August 2008 (in Danish only) published at the website of the Danish Securities Dealers Association, www.dbmf.dk

[4] See e.g. Jacobsen (2007).

[5] The use of this concentration measure is verified by comparison with other concentration measures, including the Herfindahl index. Unlike the maximum ownership share, the Herfindahl index does not express a single investor's share of a series, but is a weighted average concentration.

[6] In the calculations, private individuals and foreign investors are disregarded as only aggregated data is available for these groups.

[7] The analysis is limited to callable bonds with amortisation.

[8] In these statistics, the significance level indicates the probability of finding an incorrect statistical correlation between maximum ownership share and OAS.

[9] For non-callable bonds, OAS simply corresponds to a yield curve spread as there is no option elemets.
Go to buttom
Publication in PDF-format.
 
PC: Press the right mouse-button, choose "Save Link As", then choose where to save the file.
 
MAC: Hold down the mouse-button, choose "Save Link", then choose where to save the file.
 
Download
Acrobat Reader here:

 
 
 
Go to previous chapter               Go to top              Go to next chapter