Abstract
This paper describes the Danish mortgage system by comparing it with the UK mortgage system. The Danish mortgage system has attracted attention in the literature on the design of mortgage systems. It is a type of narrow banking model where mortgage loans are financed by specialised institutions that issue bonds with cash flows that match that of the mortgage loans. Thereby, the Danish mortgage system outsources many of the risks that are usually kept on the balance sheet of banks to bond investors. Measured in terms of four criteria, the Danish mortgage system performed better during the financial crisis than the UK mortgage system. However, both in Denmark and the UK, alignment of the mortgage and the pension systems could offer significant advantages.
Introduction
The purpose of this paper is to explain the design of the Danish mortgage system, comparing it with that of the UK and specifically focussing on the allocation of risk among the different stakeholders. The stakeholders are the borrowers, the intermediaries, those who finance the intermediaries and society at large.
Commenting on the US mortgage system in a recent editorial, the Financial Times (2013) argued for winding down the government-sponsored mortgage lenders, among other things noting that “no other countries allow borrowers to lock in fixed-rate 30-year mortgages and then to refinance their loans when rates fall”. Denmark and the UK are only separated by a small ocean and not divided by the use of a common language, but the Danish mortgage system actually allows borrowers to lock in fixed-rate 30-year mortgage loans and then to refinance their loans, when rates fall.
A crucial element in the design of mortgage systems is the allocation of risk among the different stakeholders. A 30-year fixed-rate mortgage loan is a big risk for a depositary institution whose funding is mostly variable-rate and can disappear at short notice. For a depositary institution with such liabilities, the ideal mortgage contract would be variable-rate and allow for redemption at short notice. However, for the borrower this would not be an ideal loan.
The UK was among the countries that were particularly hard hit by the financial crisis. Major high-street banks failed. As a result, extraordinary steps towards reforming financial regulation have been taken. One of the most prominent of these is the Vickers Committee's recommendation of structural separation of financial institutions (ICB, 2011). Very little attention has been paid to the role of mortgage financing in the UK, and the UK Treasury's review of covered bonds (UK Treasury, 2011) essentially suggested continuing as before. 1
The financial crisis also had a significant impact on Denmark, but the mortgage system emerged from the crisis relatively unscathed. Nevertheless the financial crisis has given rise to a lot of soul searching, especially on how to make the mortgage system even safer. Mortgage systems are remarkably idiosyncratic, making it very difficult to understand the choices made with respect to their design. By making comparisons with the UK system, the implicit choices made in relation to both will hopefully be clearer.
Probably the most significant difference between the UK and the Danish mortgage system is that in Denmark, mortgage loans are not granted by deposit-taking banks, but by specialised institutions akin to narrow banks. More generally, there are two sides to a mortgage system: The financing that it provides and how it is funded. In the Danish system these two sides are intimately connected, as it is essentially a pass-through system where the loans provided are matched by the covered bonds that fund them. In the UK, financing is more diversified, and there is no direct link between the assets side and the liabilities side of the intermediary. UK intermediaries also use covered bonds, but they play a lesser role than in Denmark. However, given the importance of covered bonds in the Danish system, particular attention will also be paid to how this instrument is used in the UK.
The paper is structured as follows. The first section will review the related literature. The second will focus on the implications of the choice of deposit-taking banks and specialised non-deposit taking institutions, respectively, as the key institutions granting mortgage loans. The third section will compare the other elements in the design of the two systems. The fourth section will elaborate on the various risks in the two systems and on who bears them. The fifth section will compare the performance of the two systems relative to a set of criteria. The sixth section will put the respective mortgage systems into the context of the overall balances of the financial systems and link the design of mortgage system to that of pension systems. Section 7 contains our conclusions.
The optimal design of mortgage systems depends on the characteristics of a particular country. Nevertheless, we conclude that the Danish model's performance as well as some of its features could be a source of inspiration for the UK. The features include that mortgage loans can be funded by institutions that are structurally separated from deposit-taking banks, the transfer of the more transparent risks to investors and the safety valves in relation to credit risks that are opaque and therefore more difficult to transfer.
1. A brief review of the literature
The design of mortgage systems has attracted increasing attention following the financial crisis, which was rooted in the US subprime market. Campbell (2012) reviews issues in relation to mortgage design. He argues that the US can learn from other countries, especially Denmark. He lists four appealing features of the Danish system. First, the distribution of risks in the Danish system between borrowers, the intermediary and the investors in covered bonds that limits the risks of the intermediary to those that make sense from an incentive angle. Second, the option to prepay or refinance a loan by buying back the bonds in the market, thus eliminating the effects of idiosyncratic moving uncertainty on individual welfare. Third, the large liquid pools of mortgage loans that create liquidity and few incentives to obtain private information. Fourth, the strict underwriting standards and the use of recourse mortgages that carried the Danish system through the financial crisis without many casualties despite a fall in house prices similar to that in the US. We will devote most space to the first point, but we will also touch on the other points.
The choice between fixed and adjustable/variable rates is an important question in the design of mortgage systems. Campbell and Cocco (2003) conclude that adjustable-rate mortgage loans can be attractive compared with fixed-rate mortgage loans. The Miles Review (2004) took almost the opposite view, i.e. that more mortgage loans in the UK should be fixed-rate. We will argue that the choice between fixed and variable rate also plays a role in terms of the possibility of linking the mortgage system and the pension system.
The structure and performance of the Danish mortgage system have also been the subject of a number of other studies. Shin (2010) describes the Danish system as a resilient institutional framework where the intermediary is akin to narrow banks. Hancock and Passmore (2009) state that there would be many benefits from implementing a Danish-type system in the United States. According to the IMF (2006), through the implementation of a strict balance principle the system has proved very effective in providing borrowers with flexible, transparent and close-to-capital markets funding conditions. Simultaneously, as pass-through securities, mortgage bonds transfer market risk from the issuing mortgage bank to bond investors. Frankel et al. (2004) states that in the Danish case, the institutional structure, the regulatory approach and monetary policy together have resulted in a market which, relative to the US market, has shown little or no stress in periods with significant refinancing. Boyce (2008) describes how the balance of payments in the Danish mortgage system creates transparency and the right incentives, and the associated possibility of redeeming a fixed-rate loan at market value reduces the risk of negative equity for borrowers. Gyntelberg et al. (2011), Berg et al. (2013) and Nielsen and Berg (2013) compare the performance of the Danish mortgage system to mortgage systems in other countries, with a special focus on performance during the recent financial crisis. We build on the approach in the two latter papers.
Mortgage systems come in many different varieties. Lea (2011) and CGFS (2006) compare mortgage systems across a number of developed countries. ECB (2009) is an extensive comparison of housing finance within the Euro Area. There are many different parameters associated with a mortgage loan and they can undertake a wide variety of values and settings. The parameters include interest rate, maturity, loan-to-value restriction and calculation, repayment profile, early repayment option, non-interest cost, loan purpose, taxation, and bankruptcy and foreclosure rules. To illustrate the wide variety of settings, take the interest rate. This could be fixed, variable or a mix and apply to different periods, and if variable linked to different benchmarks and capped in different ways. The funding of a mortgage loan could take the form of deposits, bank senior debt, covered bonds or securitisations. If the funding is with covered bonds, these could be issued by a bank, or a specialised institution; they could be backed by a variety of assets, subject to different valuation criteria, have restrictions on asset-liability management, have varying degrees of cover pool transparency, have different types of cover pool monitoring; segregation and bankruptcy remoteness of assets could be different; and they could comply with different levels of EU or other legislation. The permutations are almost infinite. We will compare the choices made in Denmark and the UK.
Comparisons are often made between the European bank-based model and the US capital markets-based system (Allen and Gale, 2001). ECB (2002) compares the financial systems across the Euro Area countries, drawing heavily on the financial accounts. We will also use financial accounts in this paper to put the role of the mortgage system into a larger context.
In the UK the ICB (2011) analysed the UK financial system and made proposals for structural separation of certain activities. The UK Treasury (2011) reviewed UK covered bond legislation, and made a few very limited proposals for reform. We will show that the Danish mortgage system and its use of covered bonds is an example of structural separation of mortgage lending from deposit taking.
2. Banking instability and structural remedies
The structure of the financial intermediaries that provide mortgage lending has implications for the stability of the financial system. The different choices made in Denmark and in the UK are probably the single most important difference between the two mortgage systems.
Financial intermediation may be illustrated by a balance sheet of a very simple bank that has deposits and capital as liabilities and loans and liquid assets as assets (chart 1). The balance sheet can be used to show the benefits and risks, the externalities associated with bank failure, and a number of different ways to contain these risks.
A simplified balance sheet
Banks can pay interest on liquid deposits and make loans available at comparably low rates, because of the maturity transformation and credit transformation banks perform. A bank relies on withdrawals and deposits normally – more or less – cancelling each other out. The liquidity of deposits can therefore under normal circumstances be maintained at low cost. Furthermore, banks’ skill sets and scales give them a comparative advantage as monitors of credit quality.
As described by Bagehot (1873) and modelled 110 years later by Diamond and Dybvig (1983), there is an inherent instability to banks. If a run starts, there is an incentive to be among the first to get out. Or in a slight rephrasing of Mervin King's statement, while it is not smart to start a bank run, it is definitely smart to be among the first in the pack running. The reason is that a bank is worth much more as a going concern than in a forced liquidation. The first to get out gets paid from the liquid assets. At some stage the less liquid loans have to be sold. In a market under pressure that is likely to happen at prices below par, and the capital of the bank will be reduced. When the capital is gone, there are not enough assets left to pay the remaining depositors. A depositor with perfect foresight therefore has an incentive to be among the first to get out.
The failure of a bank is costly, for others besides those who have contributed capital and deposits. There are negative externalities.
Most countries have deposit insurance that protects ordinary depositors. The deposit insurance entails costs for those that finance the payouts. This creates negative externalities. The case for deposit insurance is twofold. One, banks’ financial statements are opaque and the ordinary depositor has little chance of understanding them. Two, deposit insurance reinserts some stability in the unstable banking model by lessening the incentive to run.
Bank failures also have social costs because credits are cut and projects abandoned. Other banks have difficulties in stepping in as new lenders due to information asymmetries, or in bankers’ language the lack of credit history. In some cases the forced liquidation of assets depresses the valuation of assets owned by other banks, and starts a financial accelerator effect. Bernanke (2004) describes in his work on the Great Depression how the financial accelerator drove falling collateral values and limited banks’ overall capacity to lend. This, much more than losses on counterparty exposures, was a main driver of the recent financial crisis (Hanson et al., 2011; Schleifer and Vishny, 2011).
Central banks, acting as lenders of last resort and standing ready to provide liquidity against illiquid assets, are another way of safeguarding the system. Bagehot as well as Diamond and Dybvig described this avenue.
Supervisors have traditionally focused on capital buffers that could cover losses on the loans (chart 2). In the most recent proposals for a new regulatory framework, international liquidity standards are introduced for the first time.
Banking regulation 101
There are many other ways to address the inherent risks of banks. Narrow banking is one possibility (chart 3). In narrow banking the assets of banks are restricted – in the most restrictive version to short-term government bonds (Cochrane, 2014). The liquidity of the assets eliminates the problems associated with a run and thereby the incentives that can create a run. Loans will have to be granted by other institutions that are not deposit-funded.
Narrow banking
In less restrictive versions only certain assets or activities are ‘no go’. These are the assets or activities that are either considered very risky or less worthy of potential government support. The Volcker rule and the recommendations of the Vickers Commission are examples of such less restrictive versions. Both the Volcker rule and the recommendations of the Vickers Commission face difficult delineation issues. In both cases mortgage loans are allowed to be funded by deposits despite the very significant maturity transformation.
The risks can also be reduced by changing the deposit contract (chart 4). The deposit contract can be made similar to mutual funds, where there is a right to a share of the assets rather than a fixed amount. The practical difficulty is to value the assets. The run on mutual funds during the financial crisis and the stigma associated with ‘breaking the buck’ (redeeming investment certificates at less than par) also suggest that the model was not waterproof.
The mutual fund model
The riskiness of the bank construction also depends on the more general features of the surrounding economy. For instance, if creditor protection is limited, defaults on loans are more likely and the costs in case of default will be higher.
The Danish mortgage system is a special version of the loan part of a narrow-banking model. The mortgage loans are funded by covered bonds rather than liquid deposits (chart 5). Whenever loans are granted, covered bonds are issued to fund the disbursement of the loans. The Danish mortgage system is a pass-through system, where payments on loans pass through to bondholders. The bondholders cannot withdraw their funds as depositors can. There is no maturity transformation and the intermediary is not exposed to interest rate risk. Credit risk is contained through personal liability as opposed to the no-recourse loans that caused so many problems in the US. The legal system ensures that foreclosure is unusually quick, and the social safety net ensures that in most parts of the country, families can service their debt even if one family member is unemployed. ‘Originate to hold’, i.e. the loans stay on the balance sheet of the institution that grants the loan, as opposed to ‘originate to distribute’, where the credit risk is transferred to the investors, results in sharp credit assessments.
The specialised mortgage bank
In the UK, mortgage loans are funded by a mix of deposits, securitisations and covered bonds (chart 6).
The typical structure of a UK regulated covered bond
The owner of the mortgages is an SPV 2 set up by the bank issuing the covered bond. The bank receives the proceeds from the sale of covered bonds and funds the owner's purchase with a loan. The owner guarantees the payments on the covered bonds – typically via a trustee – in case the issuer cannot pay. The issuer may also provide overcollateralisation (OC), i.e. additional assets that protect the covered bond investors. In the UK minimum OC requirements are set by the supervisor based on stress tests. Only deposit-taking credit institutions set up in the UK may issue covered bonds under the UK framework.
The UK framework involves subordination of depositors to covered bond holders (chart 7). The covered bonds have a priority claim not only on the mortgages but also on the OC. In addition they have a similar claim to that of senior creditors on the remaining assets in the bank.
The structure of a UK covered bond at default of issuer
Covered bonds issued out of a bank make guaranteed deposits subordinated. Guaranteed deposits do not need to worry about their subordination, as deposit guarantee funds bail them out. Thus, the issuance of covered bonds out of deposit-taking institutions is subject to moral hazard. The potential problems are amplified by two factors. One, non-performing loans in the SPV are typically replaced by performing loans from the bank. Two, the required OC moves assets available to cover creditors in the bank to be first and foremost available for the covered bonds.
The requirement for OC is justified by the objective of making the covered bonds safer. However, the possibility of enhancing the quality of a covered bond could also create incentives for more risky lending and asset liability mismatches in the cover bond structure. 3
3. Other aspects of mortgage loans in the UK and Denmark
There are other significant differences between mortgage loans in the UK and in Denmark (table 1).
Mortgage characteristics in the UK and Denmark
Source: EMF and national mortgage organisations.
The UK is predominantly a variable-rate market. This was very much the subject of the Miles Review (2004). The background for the review was the risk that variable rates would entail in case the UK joined the monetary union and monetary policy rates were no longer set according to UK economic conditions. Historically, Denmark was a market where interest rates were fixed for the full duration of the loan, i.e. 30 years, but with an option to convert loans by buying back the underlying bonds at the lower of par and market price. The borrowers pay the equivalent of an insurance premium for the option to convert, as the price of covered bonds reflects the conversion option. 4
The interest setting in Denmark makes for a transparent price structure. The base rate of Danish mortgage loans reflects 1:1 the rate on the underlying bonds. On top of the base rate borrowers pay a margin to cover expenses of the mortgage institution, including loan losses. At present, the average margin is approximately 70 basis points (bp). It varies with the loan-to-value ratio (LTV) and mortgage form (fixed or floating, interest-only (10) 5 or amortisation payments), but until now not with borrower characteristics.
The issuer has strong incentives to minimise the credit risk on the mortgage loans rather than maximising the value of the bond, as the bonds are sold on to investors and not kept on the books of the issuer. When market rates fall, issuers will help borrowers convert, as this reduces the credit risk that stays on the balance sheet of the issuer. In the US there is an incentive to maximise the value of the bond rather than minimise credit risks, and this has hampered refinancing at lower rates and thereby the soothing effect of lower monetary policy rates (Boyce et al., 2012).
Over the past 20 years adjustable-rate mortgage loans (ARMs), typically with annual adjustments, have come to play a larger role in Denmark. Today the split between fixed, adjustable and variable is 30:50:20 per cent. 6 However, the ratios shift over time and, as elsewhere, are very much driven by the shape of the yield curve.
It is sometimes said that the Danish version of ARMs creates a liquidity risk. In the 1990s, ARMs were introduced for which interest rates were reset anywhere between once a year and once every ten years. The ARMs are funded by auctioning mortgage bonds at the same time as the interest rate is reset. The borrower pays the rate at which the auction settles. The mortgage institutions are exposed to a liquidity risk should they not be able to sell the bonds. However, that risk is much more limited than in systems where the interest rate reset on mortgage loans is not aligned with the refinancing. In the Danish system, mortgage institutions are obliged to pass on the rate set at the auctions to the borrower. Therefore, they are never left with an interest rate risk that can create a liquidity risk, if buyers of the bonds are reluctant to acquire bonds issued by institutions that are potentially at risk in the event of increases in interest rates.
Nevertheless, earlier this year the authorities and the mortgage institutions implemented legislation that introduces an option for the issuer of extending the maturity of the bonds, where the maturity of the bonds is shorter than the underlying loans. In the unexpected situation that a refinancing auction fails, the bonds are automatically prolonged for one year at an interest rate 5 percentage points higher than the rate set at the previous refinancing auction. The higher interest rate as well as the price set by the investors for the insurance is passed on to the borrower. The new legislation follows the general principle that the borrower pays an insurance premium and the intermediary outsources the risk to the investor. Furthermore, new products are being introduced, e.g. adjustable-rate mortgage loans with longer refinancing periods.
Even for variable-rate loans, the spread the borrower pays relative to the reference rate is set in the market for the variable-rate bonds. Thus, the Danish system did not experience the problems associated with tracker rate products, where the spread to the reference rate was locked in for the borrower, whereas the spread on the financing of the lender was not locked in.
The margin to cover expenses of the mortgage institution may be adjusted during the term of the loan. The possibility of adjusting to the margin reflects that mortgage institutions historically were cooperatives and that borrowers were jointly liable. The joint liability meant that de facto margins reflected loan losses. Today the largest mortgage institution, Nykredit, is still governed like a cooperative, but the joint liability has been replaced by adjustable margins. The cooperative structure ensures a governance structure, where raising margins reflects as negatively on management as lowering dividends in a joint stock company. In addition, refinancing with another mortgage institution is fairly inexpensive. Thus, competition as well as the price setting of Nykredit keeps a sufficient pressure on the setting of margins within the industry to allow for adjustable margins.
As opposed to the UK, teaser loans have never been available in Denmark. Furthermore, IO loans were not introduced until 2003. IO loans are only granted with no amortisation payments for an initial period of ten years. If the LTV of a loan after ten years is above 80 per cent, amortisation has to commence.
In Denmark there is an ongoing discussion on the advantages and disadvantages of IO loans. Proponents argue that IO loans allow more intertemporal shifting of consumption, which is one of the major purposes of a financial system. Opponents believe – among other things – that IO loans increase the risks of financial instability as debt levels and LTVs stay high.
Available detailed microdata, including the possibility of linking loan data with income and other statistics at the level of the individual, suggest that there is a good match between loan characteristics and the robustness of the borrower (Andersen et al., 2012). Nevertheless, the industry is generally pushing borrowers towards a more balanced use of adjustable-rate mortgage loans and IO mortgage loans by changing margins so that they favour traditional fixed-rate loans and amortisation payments.
Apart from pricing and refinancing options, the major difference for the borrower compared with the UK is the acceptance of the use of foreclosure. Foreclosure in Denmark takes about six months from start to end. Together with the Netherlands this is the fastest in the EU. In some countries in the EU the process can take five years, and in many countries it is – as in the UK – something which is not generally accepted or for other reasons is very difficult. The quick process in Denmark reflects several factors. Denmark historically has had a legal regime under which creditors were very well protected. The origin of mortgage institutions as cooperatives with joint liability also internalised the costs of defaults among borrowers. Finally, Danish communities have a rehousing obligation, which includes families that have been exposed to foreclosures.
As a final point, mortgage brokers are almost nonexistent in Denmark and there has been no subprime lending. The two things could be related.
4. Risk distribution in the UK and in the Danish mortgage system
In the UK system as well as in the Danish system, credit risk rests primarily with the originator. As illustrated by the financial crisis, this is a sensible design. Credit risk is almost by definition opaque. Selling off the credit risk to investors could result in the ‘originate to distribute’ problem. However, in the Danish system, there is a safety valve that allows passing-on costs related to significant increases in credit risk to borrowers.
In the Danish system, market risk is outsourced to the investor. The covered bonds match the loans in terms of the financial characteristics. For instance, the covered bonds have the same conversion option as the loans. The fact that the conversion option is part of the bonds also takes out the counterparty risk that would have originated if the conversion option were instead handled through a third party. Outsourcing market risk makes sense as market risk is transparent.
Before the introduction of adjustable-rate loans, there was no maturity mismatch. With the introduction of adjustable-rate loans a potential maturity mismatch risk was introduced. However, the liquidity risk was limited by the obligation of the borrower to pay whatever interest rate that was needed to refinance the loan. The major liquidity risks come from regulatory and rating requirements to maintain additional buffers in the cover pools.
In the first Capital Requirements Directive (CRD) definition of covered bonds, it was stated that every loan has to observe LTV restrictions throughout the life of the covered bond. If house prices fall and a loan no longer observes LTV restrictions, supplementary collateral has to be provided. Prior to the CRD definition, loans behind Danish covered bonds only had to observe LTV restrictions at the time the loan was granted. In a system where loans are funded almost 1:1 by covered bonds, this creates a liquidity risk. When house prices fall, mortgage institutions may be forced to issue bonds that are subordinated to the regular covered bonds in order to generate additional collateral. The likelihood that financial markets would be stressed, when house prices fall substantially, only adds to the problem. However, the good historical record of Danish covered bonds made it possible to issue covered bonds that were junior to regular covered bonds during the financial crisis and generate the necessary additional collateral. If this had not been possible, the covered bonds would have had to give up their CRD status and drop to UCITS 7 compatible bonds (the older and lighter EU definition of covered bonds).
The rating agencies’ requirements for OC generate a similar liquidity risk.
In a bank-based mortgage system like the UK system, banks perform the usual maturity transformation with associated risks. Market risks are managed at bank level rather than by matching assets and liabilities. Even in the UK covered-bond system there is nowhere near the same strict balancing of cash flows as in the Danish system. Risks are instead contained through OC. In the Danish system, refinancing of a loan is done by buying back the bond(s) that fund the loan. Thereby the balance between assets and liabilities is maintained. Both the UK and the Danish covered-bond system are based on national legislation rather than only on contracts. This reduces legal risk.
5. The performance of the UK and the Danish systems during the financial crisis
We focus on four objectives for a mortgage system:
Affordability: A mortgage system should make it possible for households to acquire a home when they need it most.
Robustness: A mortgage system should be robust when house prices fall.
Resilience: A mortgage system should be able to continue to fund mortgage lending during and after a financial crisis.
Government Intervention: A mortgage system should not be dependent on government subsidies, e.g. in the form of guarantees.
There is an argument for including a fifth objective: the costs of moving related to the mortgage system. These costs relate to repaying or transferring the mortgage to the new owner. However, neither in the UK, nor in Denmark are these costs large.
Affordability
A mortgage system should make it possible for individuals to acquire a home when they need it most, i.e. early in an individual's productive life, when income and savings are likely to be lowest. Low income tends to result in relatively high debt-to-income ratios for young borrowers, which lenders normally associate with higher credit risk. The quest for affordability does not allow for high owner-financing requirements, which would otherwise have been an obvious way to reduce the risks in mortgage finance.
Affordability of mortgage loans cannot be assessed in isolation. In some countries – like the US, Germany, Denmark and Sweden – there are large specialised mortgage lenders that will sell mortgage loans to consumers as a stand-alone product. In many countries mortgage loans are provided by universal banks as part of a financial package offer. Sometimes a mortgage loan can be conditional on the customer buying a life insurance contract at the same time. Hence we should be cautious when comparing mortgage rates across countries, as the rates may not reflect the true cost of achieving housing finance. Moreover, many countries lighten the burden of mortgage payment by allowing full or partial tax deductibility for interest paid on mortgage loans. The comparison of after-tax interest expenses on mortgage loans between countries is thus complicated.
The average costs of housing loans are also difficult to compare because they are highly dependent on the interest rate definition, in particular whether it is variable or fixed, the LTV, characteristics of the borrower and other factors.
Nevertheless, if we compare the interest rate on fairly similar house loans (Figure 1), and deduct the swap rate, we can see that the calculated spread on new loans increased substantially during the financial crisis in the UK and was higher than in Denmark (Figure 1).
Spreads on mortgage loans in the UK and Denmark
Instead of looking at interest rates, one can also look at owner-occupancy rates and residential mortgage debt to GDP to get a sense of how easy it is to obtain financing for a home (figures 2 and 3).
Owner occupier rates
Residential mortgage debt to GDP
Both Denmark and the UK are characterised by being at the low end in relation to home ownership rates 8 and at the high end when it comes to residential mortgage debt. This points in opposing directions as to the affordability of home financing. However, the supply of rental housing, the availability of land, cultural norms and many other factors influence the rate of home ownership. The tax system may impact both home ownership and mortgage debt.
We can also look at the cost components in the mortgage system. Two important elements are the costs related to foreclosure and the costs related to OC. The average length of a foreclosure process is very low in Denmark, while it is somewhat higher in the UK (Figure 4).
Typical duration of a foreclosure procedure
Furthermore, foreclosure is not widely used in the UK. Both aspects suggest higher loss-given default rates in the UK.
OC is another factor that adds to the cost of covered bond issuance. As mentioned above, OC is primarily driven by the perceived credit risk in the mortgage loans and the mismatch between cash flows coming from loans and going to bond holders. The Danish covered bond system is characterised by low levels of OC, whereas the UK system is characterised by a high level of OC (Figure 5). The high level of OC in the UK system is partly driven by cash flow mismatches (Figure 6).
Target and actual OC
Relationship between asset-liability mismatch and required OC
Robustness towards falling property prices
A mortgage system should be robust in case of falls in house prices, e.g. a fall in house prices should not put the financial system at risk. This suggests that the risks associated with housing finance should be distributed to those agents who are best suited to handle them.
Falling house prices erode the value of the collateral behind the mortgage loans and hence reduce the credit quality of the mortgage loan portfolios. In combination with increasing unemployment or other factors reducing the borrowers’ ability to meet their mortgage obligations, falling house prices is a key driver of delinquencies 9 and foreclosures. The widespread use of non-recourse or limited recourse mortgage loans in some countries created an extra risk in connection with deterioration in property values, namely an incentive for homeowners with negative equity in their homes to walk away from their mortgage loans. The most extreme examples of this have been in some states in the US during the recent financial crisis, but the phenomenon was also seen in the UK in the 1990s.
Credit risk should at least for some meaningful part be shouldered by lenders, and house price risk should be mitigated by prudent LTV thresholds. There should also be some sort of recourse to the homeowner in order to retain an incentive to make mortgage payments, even after home equity has been depleted by falling property prices.
The patterns of house price movements in Denmark and the UK have some similarities (Figure 7). Prices were exuberant prior to the financial crisis and fell substantially after the financial crisis. However, the price fall in Denmark was more persistent, in aggregate larger, and the turnaround in house price inflation from immediately prior to the crisis was more dramatic.
House price and delinquencies movements UK vs DK
Historically, delinquencies have been at a higher level in the UK than in Denmark. This was also the case during the financial crisis, where delinquencies increased in both countries. However, the level in Denmark in the worst years stayed below the level in the UK during the best years.
In Denmark there has also been a discussion of the pros and cons of variable vs fixed-rate mortgage loans. Interest rate risk arises in a situation where interest rates rise substantially and nominal GDP/the borrower's disposable income does not follow suit. As Denmark conducts a fixed exchange rate policy towards the euro, this could either happen because the Danish economy is out of sync with the Euro Area, or because the Danish currency is under pressure and the Danish central bank has to raise monetary policy rates to defend the peg.
At the beginning of the financial crisis, Denmark had to raise monetary policy rates unilaterally. Delinquencies – also driven by general economic developments – increased in 2009 but, as shown earlier, were small in absolute terms. Furthermore, later in the crisis Danish covered bonds became safe-haven assets, and rates dropped to very low levels, benefitting the Danish economy.
While both Denmark and the UK saw bank failures during the financial crisis, the failures in the UK were of a different dimension. In the UK two of the largest banks failed, while failures in Denmark were limited to a few regional and local banks. None of the mortgage banks in Denmark failed.
Both in the UK and in Denmark the government offered public guarantees and capital to the banking sector. The Danish government ended up making money on the guarantees and capital contributions, while the cost to the UK government is likely to be substantial.
Many other factors than the design of the mortgage system determined the outcome of the financial crisis. However, in this context it is worth noting that the UK had the advantage, compared with Denmark, of a floating exchange rate. Thus, Denmark did not benefit from the depreciation of its exchange rate early in the crisis as the UK did.
Resilience during and after periods of financial stress
The crisis illustrated the detrimental consequences of financial systems that were constrained and could not provide sufficient credit to the economy. Danish covered bond issuers continued to issue covered bonds in the months following Lehman's crash, whereas the European covered bond market stood still, and government-guaranteed bonds became the instrument of choice or necessity (Figure 8). The market for senior unsecured debt for banks also virtually shut down, putting pressure on banks’ balance sheets and their ability to extend credit to households and businesses. In the UK, issuance of covered bonds increased substantially in 2008, but then fell back (Figure 9). According to anecdotal evidence, the increase in issuance reflected changed collateral rules at the Bank of England that allowed for greater use of covered bonds as collateral, whereas there was only very limited market distribution. Substantial funding in the UK also came in through government-guaranteed issuance.
Issuance of covered bonds
Total issuance of UK and DK covered bonds
The Danish experience showed that the banks stopped lending because of lack of funding, while the mortgage banks kept extending mortgage-secured credit to both retail and corporate clients, because the mortgage banks had ample access to funding via the Danish covered-bond market (figures 10 and 11). In the UK, lending to households and non-financial corporations declined following the onset of the financial crisis (Figure 12).
Outstanding loans from DK mortgage institutions to households and non-financial corporations
Outstanding loans from DK banks to households and non-financial corporations
Outstanding loans from UK MFIs to households and non-financial corporations
Government intervention
Both the UK and Denmark are characterised by little government intervention in mortgage markets as opposed to the US housing finance system, which for many years has been characterised by a high degree of government intervention (table 2). The government had to step in and nationalise the government-sponsored entities that are the key providers of mortgage finance in the US, and thereby de facto placing most of the housing finance system under government control. However, even before the nationalisation a number of government schemes were in place to support the housing finance system.
Government mortgage market support
Source: Lea (2011) and own research.
Table 2 does not take into consideration the implicit government support for mortgage systems from deposit-insured funding, which recent experience shows has extended well beyond the insured deposits. In the EU, most unsubordinated bank creditors have been bailed out, with the prominent exception of two Danish cases and the debacle in Cyprus. 10 If we take the explicit government support into consideration, there is a strong case for specialised mortgage banks that do not use deposits as funding. Furthermore, given the preferred creditor status of covered bonds, covered bond systems with extensive OC pose a potentially greater risk to other creditors with implications for the costs of a bail-out. As demonstrated above, OC is partly driven by market risk in covered-bond systems. Covered-bond systems that limit market risk therefore deserve promotion.
Both in the UK and Denmark, the government provides support for social housing. However, this support falls outside the scope of this paper.
It is something of a paradox that the Danish covered-bond system, despite its performance, has been under pressure from the preferential status given to government bonds in new international regulatory standards. Under the Liquidity Coverage Ratio (LCR), covered bonds were initially only given status as Tier 2 liquidity. As such, covered bonds could only make up 40 per cent of the liquidity buffer used for calculating the LCR. The remainder had to be made up by Tier 1 liquidity, primarily government bonds. The first draft of the LCR came out in December 2009, i.e. prior to the government debt crisis and the discovery of the potential doom loop between banks and governments. The government debt crisis has not favoured giving government debt any preferential treatment.
Studies using actual transaction data have also shown that Danish covered bonds are at least as liquid as Danish government bonds. Dick-Nielsen et al. (2012) and the European Banking Authority's own study on covered bonds across the EU (EBA, 2013) showed that covered bonds are as liquid as government bonds in the EU as a whole. As government debt is limited in Denmark after many years of sound fiscal policies, the option of using covered bonds in the liquidity buffer is very important for the banking system. Fiscal repression would have reached a new peak if Danish banks had been forced to include in their liquidity buffers less liquid government bonds from other countries with added foreign exchange risk and credit risks. Fortunately, the EU Commission – at the time of writing – seems to have concluded that covered bonds deserve better treatment.
6. Mortgage systems in the context of the overall financial system
The financial accounts allow us to see mortgage financing in the overall perspective of the financial system. The financial balances of both Denmark and the UK are high in an international perspective, with aggregate domestic financial assets totalling roughly fourteen times GDP in both countries (table 3).
Financial balances
Source: ECB.
Note: Data as of end-2013.
The Euro Area average is ten times GDP. 11 Looking at the sub-balances for sectors of the economy, the monetary and financial institutions (MFI) sector has financial assets of more than 430 per cent of GDP, which is high compared with the Euro Area average of 332 per cent. In addition, Denmark and the UK share the rather unusual feature, in an international context, that their insurance corporations and pension sectors are very large, totalling more than 150 per cent of GDP (the eurozone average is 81 per cent).
Though Denmark and the UK are similar in these respects, there are also differences between the two countries. For instance, Danish households are relatively more indebted and have smaller net financial savings (144 per cent of GDP against 207 per cent for the UK).
The high gross indebtedness of Danish households has attracted some attention. It primarily reflects the Danish tax structure, where interest payments are partly tax deductible and pension savings also receive favourable tax treatment (Andersen et al., 2012). This is very similar to the situation in the Netherlands. At a net level, Danish households look similar to the EU average, while at the national level Denmark is a substantial net saver, running current account surpluses of around 7 per cent of GDP.
The MFI sectors in the UK and Denmark differ – for the purposes of this paper – in two important respects. One, deposit funding is much more important in the UK than in Denmark (table 4). Two, bond funding is much more important in Denmark than in the UK. Bond funding in Denmark is primarily covered bonds.
Financial balance sheets for the MFI sector
Source: ECB.
Note: Data as of end-2013.
In both the UK and Denmark, bonds are an important asset for insurance corporations and pension funds as well as other financial intermediaries (tables 5 and 6).
Financial balance sheets for insurance corporations and pension funds
Source: ECB.
Note: Data as of end-2013.
Financial balance sheets for other financial intermediaries
Source: ECB.
Note: Data as of end-2013.
Looking at the aggregate bond market (table 7), the total amount of bonds issued is roughly the same relative to GDP in Denmark and the UK (246 per cent and 220 per cenet, respectively). The bonds primarily serve to fund MFIs in Denmark, with covered-bond issuance as the largest instrument. The Danish covered-bond market is more than four times larger than the government-bond market. In the UK, the government-bond market exceeds the bonds issued by MFIs, reflecting both that deposits are the main funding source for MFIs and greater non-tax public financing needs in the UK than in Denmark. It is also worth noting that the corporate-bond market is almost three times larger in relative terms in the UK. Finally, the main bond investors in both countries are insurance corporations and pension funds, MFIs and other financial intermediaries.
Investors and issuers of bonds
Source: ECB.
Note: Data as of end-2013.
With the increased popularity of adjustable-rate mortgage loans in Denmark, the banking system has increased its role as investor in covered bonds relative to longer-term investors. Insurance and Pension funds were volatile investors during the financial crisis. The volatility was driven by the combination of their nominally guaranteed pension commitments and the associated sensitivity to the choice of discount rate in their asset liability regulation. Fortunately, the impact of the regulation has seen a shifting of pension savings away from guaranteed products. This should allow for insurance and pension funds to become more significant holders of adjustable-rate and variable-rate mortgage loans, see below.
From the perspective of this paper, the salient similarities are that both the UK and Denmark have a very large insurance and pension system, where long-term savings accumulate, and a large market for bonds issued by non-public entities. The insurance and pension funds, as well as MFIs, other financial intermediaries and foreigners are large investors in bonds. The difference is the type of bonds they invest in. In Denmark, covered bonds issued by specialised mortgage institutions dominate the bond market. In the UK, there is a larger variety of bonds, and MFIs are to a much larger extent funded by deposits.
Financial intermediaries serve to bind together savers and those in need of finance. If they have different preferences, the financial intermediary may end up with assets whose characteristics do not match their liabilities. This creates risks for the financial intermediary with possible repercussions for the public at large. A benign social planner would therefore try to match savers and those in need of finance so that their preferences match each other. Those who prefer long-term finance should be matched with those who prefer long-term investments. Those who prefer fixed-rate assets should be matched with those who prefer fixed-rate liabilities.
Mortgage loans are long-term obligations, and it seems sensible to match them with long-term savers. The most obvious intermediary is insurance and pension funds. These have in many instances committed themselves to defined benefit obligations or other forms of nominally guaranteed returns. They fit best with fixed rate assets. If borrowers prefer variable-rate mortgage loans, there is a mismatch.
However, if you were the social planner in charge of redesigning the financial system, you might ask if guaranteed nominal returns make sense for pensioners. The acceleration in inflation in the 1970s and the subsequent deceleration caused large fluctuations in the real value of pension savings among those who had nominally guaranteed returns. If your desire as a pensioner is to keep up with those who have a job, you want a pension that is closely correlated with nominal GDP. If central banks follow a Taylor rule, short-term rates will be closely correlated to growth in nominal GDP. Investments in adjustable-rate products may therefore make a lot of sense. Thus, there might be a possibility of linking the mortgage system and the insurance and pension system in a variable-rate environment in a manner that would reduce risks in the financial intermediaries and provide better products for the customers.
7. Tentative conclusions
There are five tentative conclusions.
First, a remarkable feature of the Danish mortgage system is that it is an implementation of narrow banking. It is somewhat surprising that the Vickers Committee did not address the issue of mortgages in relation to structural separation, and that the UK covered bond review allows for structural subordination of guaranteed deposits, when the Danish model has shown that structural separation in this area is actually possible.
Second, in the Danish model fewer risks remain in the intermediary than in the UK model. The transparent risks are outsourced to investors, and the more opaque risks, where there could be informational asymmetries, are kept on the balance sheet. Furthermore, the Danish model contains safety valves in relation to the credit risks kept on the balance sheet of the intermediary.
Third, the Danish model proved less fragile than the UK model during the financial crisis. While the Danish system has performed well during the financial crisis, it is challenged on many fronts, not least by international regulatory and rating standards that are designed for a universal banking model and occasionally are counterproductive when it comes to the stability of the Danish model.
Fourth, the Danish and the UK financial structure are both examples of a hybrid structure that falls between the pure-banking model of Southern and Central Europe and the US Capital Markets-based model.
Fifth, the optimal design of mortgage systems on the one hand depends on the characteristics of the particular country, and on the other hand should also raise questions as to whether there is scope for changes to other elements of the financial system that could allow for a better mortgage system. The most obvious synergies seem to be between the design of the mortgage system and the insurance and pension system. Both are there for the long run. As opposed to the Miles Review, the trick may not be to turn more mortgages into fixed-rate mortgages, but to turn more savings toward variable-rate investments. How this is done is a question we leave for future research.
Footnotes
1
Covered bonds are debt instruments secured by a cover pool of mortgage loans (property as collateral) or public-sector debt to which investors have a preferential claim in the event of default. Different from asset backed securities, covered bond assets remain on the issuer's consolidated balance sheet. The investor in covered bonds has recourse to both the pool and the issuer.
2
Special Purpose Vehicle, i.e. a legal vehicle set up for the purpose of the particular transaction.
3
According to the Basel Committee's recent paper on large exposures (2014), covered bonds with a minimum of 10 per cent OC qualify for lower weights. This reflects the misunderstanding that OC is a sign of quality, rather than the reality that OC is there to make up for the poorer quality of the underlying credits and the underlying structure. As in many other cases, everything is not equal.
4
The pricing of the insurance premium in the market for bonds is the subject of extensive literature that falls outside the scope of this paper.
5
Interest-only refers to loans, where there are no amortisation payments.
6
Adjustable includes loans for which the rates are adjusted up to every ten years. Many of these loans would be considered fixed in other countries. Fixed means fixed for the entire duration of the loan.
7
Undertaken for the collective investment in transferable securities.
8
If you include cooperative housing, home ownership rates in Denmark are about average.
9
The UK definition is payments past due for more than 3months, whereas the payments must be past due for more than 3.5 months in Denmark.
10
Here we exclude a small UK savings bank.
11
The ECB is the source for statistics for the Euro Area.