In this section:

• FSA Library
  • Communication documents
    • Speeches
      • 2009

12 May 2009

Speech by Adair Turner, Chairman, FSA
The FSA Mortgage Conference

Today’s conference brings together a wide range of people – from the industry and from consumer representatives – interested in the mortgage market, and at a time when the problems of the mortgage market and the housing market are very much to the fore.

The global financial system has gone through its worst crisis for at least 70 years and developments in housing and mortgage markets played a significant role in the origins of the crisis. Rapid growth of mortgage credit in a number of countries – the US, the UK, Ireland and Spain – drove a house price boom that has now turned to bust. Securitisation of mortgages grew rapidly in scale and in complexity, creating new sources of funding which rapidly dried up when confidence disappeared. And confidence was undermined by lending to uncreditworthy customers, with a proliferation of risky practices – subprime lending, self-certified mortgages – and new categories of mortgage, such as buy-to-let. Many other factors unrelated to the mortgage market have contributed to the crisis, but it was problems in US subprime lending which first triggered it, and in the UK we have faced particular problems in our  rapid growth mortgage lenders, Northern Rock and Bradford and Bingley.

So it is not surprising that the agenda for regulatory reform includes options related to the residential mortgage market. And in March, the Turner Review raised the issue of whether we should consider, for instance, limits such as several other countries have in place, on maximum loan-to-value (LTV) or loan-to-income (LTI) ratios. But the Review also made clear that the FSA was not at this stage making firm recommendations, but rather posing questions for consideration, which we will address in a Discussion Paper to be published in September, a Discussion Paper which we hope will stimulate a wide-ranging debate.

We need that debate and careful consideration, because the issues involved are complex – spanning macroeconomics, financial stability, housing availability and consumer protection. Today I would like to set out some of that complexity and define some of the issues which we need now to consider.

It is obvious that homeownership and mortgage credit play a vital role in individual lives, in the financial system and in the macro economy. Net housing equity – the value of people’s houses after mortgage debt – even after the 20 to 25% fall in prices we have now seen, amounts to about £ 2.4 trillion – easily the largest segment of household sector net worth.¹ Residential mortgage debt in turn accounts for over half of all credit extended by banks and building societies in the UK, and is something like 50% larger than credit extended to all businesses – large, medium and small.² Residential mortgage bad debts can therefore have an important influence on the health of the banks and building societies. The trend in property prices and thus in household wealth has a vital influence on consumer confidence and spending. And trends in the availability of mortgage credit have a significant impact on house prices. Both over the next few years – as we pull through this recession – and over the long term, the role of the housing market is central.

Back to top

And it’s obvious that in the UK, as in the US, developments in the mortgage and housing markets in the years running up to the financial crisis created significant risks:

• The UK, like the US, saw very rapid growth of household debt as percentage of GDP, with that growth (Slide 1) dominated by mortgage credit, up from about 55% of GDP in 1998 to over 80% by 2007.
• Measures of income leverage increased (Slide 2) and an increasing supply of mortgages was available at very high initial LTV ratios, as some borrowers and lenders assumed that debt burdens were likely to fall with continuous property price appreciation. And mortgage credit was extended (Slide 3) – not to the same extent as in the US, but still significantly – to social categories that had not previously enjoyed access to it and which were exposed to high credit risk. At the peak of the market (Slide 4) higher risk loans had come to account for a significant share of the total market.
• Partly as a result (Slide 5) house prices soared, but have now fallen significantly, as mortgage credit availability has suddenly declined (Slide 6).
• And mortgage arrears and repossessions are now rising (Slide 7) creating fears that we might see losses on the scale of the early 1990s.

A now familiar story – and therefore not surprising that in response to the Turner Review some people have asked why we consider constraints on mortgage lending an open issue – isn’t it obvious that we need to stop this happening again? And not surprising that to some people it’s obvious that the trends in arrears, repossessions and losses will be as bad as or worse than those seen in the early 1990s.

But actually, these conclusions are not obvious, because a closer look at the data suggests a more complex reality which we need to take into account, both in assessing likely short-term trends and in designing appropriate long-term regulatory policy.

A key fact which should give us pause for thought (Slide 8) is shown on this slide drawn from the CML’s database and the FSA’s Product Sales Database. It shows that the percentage of new mortgages extended with LTV ratios of over 95% was in the mid teens in the 1970s, soared to over 40% in the 1980s and early 1990s, and then fell significantly in the fifteen years running up to the crisis – which seems to completely contradict the story of irresponsible lending expansion. Indeed if these were the only figures we had available and we knew nothing else about what happened, we would be telling ourselves a story of how borrowers and lenders, burnt by the 1990s recession, had learned their lessons and become more prudent.

So what’s been going on?  Well it’s not entirely clear, there may be oddities of data coverage and meaning still to be understood³, and that’s why we need a period of careful analysis before leaping to any conclusion. But part of the story seems to be a major change in the balance between different types of mortgage borrower, and a major divergence between what’s been happening to initial LTV ratios and initial LTIs.

• One apparent trend, which has been commented on by, for instance, the National Housing and Planning Advisory Unit, which is charged with looking at the issue of housing affordability for first time buyers⁴, (Slide 9) is that initial LTVs for first time buyers, while still relatively high, have come down since the 1990s and had done so even before the significant falls of the last year. First time buyers now make somewhat larger deposits as a percentage of house purchase price than twenty years ago. This may be partly because they buy slightly later in life, and therefore have accumulated more savings, but may also reflect increased reliance on parental support or inheritance from grandparents in providing the initial deposit. Faced with higher house prices, first time buyers have either needed to delay purchase or have needed more help to buy, but as a result have needed lower initial LTVs (though higher LTIs) than twenty years ago.
• An equally, if not more important feature reducing average LTVs on all mortgages, however, is that first time buyers have made up a declining percentage of all mortgage borrowers, and the other two categories of owner-occupier mortgagees typically take out lower LTV loans (Slide 10). Home movers – people selling one home and buying another sometimes larger one – have a distribution of LTV’s significantly below that of first time buyers. And remortgagers – which covers a complex mix of some people simply seeking a better mortgage deal and some people extracting equity from their house by increasing the size of their mortgage – have a lower LTV distribution still. Equity withdrawal has accounted for an increasing percentage of the total owner-occupied mortgage market (Slide 11) and house purchase a declining share, and first time buyers a declining percentage of the house purchase element. As a result average LTVs have actually fallen.
• But average LTIs (Slide 12) have been increasing. First time buyers have made slightly larger deposits, but the impact of higher property prices relative to average incomes has been a more significant effect, requiring higher LTI multiples. And while home movers have more equity than first time buyers, often as a result of price appreciation on their existing house, the stretch to the larger house has become larger relative to income, driving up required LTI. Though note that at the macro level, it is possible that the causation may be the other way round, and that one driver of the house price boom of the 1990s to 2007 may have been the transition to a low inflation and therefore low nominal interest rate environment. In the 1970s and 1980s, high inflation and high nominal interest rates meant that an apparently long-term mortgage was in real repayment terms much shorter in duration than first appeared – the real repayments were frontloaded. As nominal interest rates fell, that front-loading effect disappeared. People may therefore have felt able, in cash flow terms, to afford larger mortgages and that in turn may have driven price rises. If that is the explanation of what occurred, the long-term equilibrium house price/earnings ratio may have increased, for reasons which are, at the individual homeowner level, perfectly rational.

But that development, whether rational or not, still creates significant risk, since the flip side of low interest rates, low consumer price inflation and thus lower nominal earnings growth, is that the burden of mortgage debts is less rapidly eroded by earnings growth over time.  And with higher LTIs today than in the 1990s, we might conclude that mortgage debt affordability is more vulnerable to today’s economic downturn than it was in the early 1990s, despite lower average LTVs.

Back to top

But again that might be leaping too hastily to a conclusion. For in one key respect, this recession is likely to be much more favourable to high-income leverage borrowers than that of the early 1990s. Between 1988 and 1990, mortgage interest rates rose from about 9% to 14% and stayed there until autumn 1992. That rise produced a very large increase in the percentage of income devoted to mortgage interest, well above present levels (Slide 13) and therefore, alongside rising unemployment, played a crucial role in driving the dramatic rise in arrears and repossessions between 1988 and 1992. In this recession, conversely, interest rates have collapsed. The extent to which mortgagees have benefited from this fall has been highly variable. Some – with tracker mortgages – are enjoying spectacular decreases in mortgage interest payments, some much milder benefits, but only a minority of people (for instance, among subprime borrowers coming off very low fixed-rate deals) are likely to face a material rise in mortgage payments. And on average the effect is very significant, with total household disposable income after taxes and mortgage interest payments, actually up 6% over the year to Q4 2008.⁵  Falling house prices may shift many people into negative equity, and a rise in unemployment will produce an increase in arrears and defaults, but compared to the early 1990s, we are less likely to see mortgage repayment problems among the vast majority of people who, even under the most extreme forecasts for unemployment, will still be in a job.

It is therefore at least possible that overall arrears and defaults from owner-occupier mortgages may be no worse or even somewhat less bad in this recession than in the early 1990s, even if the fall in GDP is equally large or indeed larger, and even if the peak to trough house price fall is more severe. There could still be very severe problems, in particular with segments and firms – as we have seen with Northern Rock and Bradford and Bingley – but those severe problems might be combined with an average performance which is less extreme than some fear. It is important to keep that perspective. In the current environment it is possible to harm financial stability and confidence, either by failing to anticipate how large the problems could be or by assuming that that they are more severe than likely, ignoring positive factors.

But let me shift from the short-term outlook to the long-term picture, and to the issue of regulation. What if any changes in our regulator approach should be introduced to address mortgage market problems? Before answering that question, we should note four other significant changes which have occurred in the mortgage market over the last ten years:

• Firstly, the growth of the buy-to-let mortgage market (Slide 14) from a trivial proportion of mortgage lending in the late 1990s to 26% of all new mortgage lending in 2007. It is possible that this segment of the market may produce significantly higher arrears and defaults than the owner-occupied segment and therefore possible that mortgage providers in aggregate will face total losses as bad as in the early 1990s even if experience in owner-occupied mortgages is less bad. But even if that is the case, we do not yet have a clear picture of how much worse. This market is not at present regulated by the FSA; available aggregate data on levels of leverage versus value or rental income is less good than it is for owner occupiers and we don’t have the historic record of a past recession to serve as at least a starting point for thinking about likely trends. We will need to observe evolving experience carefully and learn lessons from it.
• Secondly, the growth and then rapid disappearance of funding through securitisation (Slide 15).  Securitised mortgages hardly existed in the UK market before the mid 1990s; by 2005 they accounted for about 25% of all mortgage credit extension. There is nothing necessarily harmful about a significant element of mortgages being securitised. The US had an effective market in securitised mortgage credit for decades before the problems of the last 5-10 years. Mortgages, even when they do end up on balance sheets, are largely credit assessed on the basis of scoring systems rather than via individual judgement, and therefore are more naturally suited to securitised credit form than some other categories of loan; and many mortgage backed securities of good quality mortgages still look likely to prove sensible investments even in the face of a severe recession. But there are clearly issues about how to ensure that the originate and distribute model does not absolve the originating firm from the responsibility for careful credit analysis. And at the macro level, this rapid expansion of securitised credit helped drive the house price boom, and its disappearance in the face of confidence and liquidity effects is a very big macro shock.
• Thirdly, changes in overall bank funding strategies (Slide 16) with UK banks in aggregate relying to an increasing extent on non-customer deposit sources of funding, including a significant element of interbank borrowing from abroad. That funding of course supported the whole of the asset side of the balance sheet – corporate lending as much as residential mortgages. But residential mortgage growth would not have been as rapid without that increased reliance on wholesale funding.
• And fourthly, a very major change in mortgage-selling patterns, with the emergence of so-called ‘specialist lenders’, which according to Bank of England data now account for 34% of mortgage balances outstanding, up from 4% in 2000 (slide 17). Approximately half of the lending in this category is done by subsidiaries of familiar high street names. [erratum: erroneous text removed] These subsidiaries in many cases bundled their higher-risk lending (such as subprime, self-cert, high LTI/LTV, interest only and buy-to-let) in legal entities that are separate from their respective parent company. The other half of this segment is comprised of so-called ‘non-bank’ lenders, such as GMAC, GE Money and others. These are lenders that do not have any high-street branches, which are not deposit takers, and rely entirely on intermediaries for the distribution of their mortgage products. Their funding model, in turn, is based on their packaging-up mortgages and selling them on, as securities or whole-loan sales, to other investors such as hedge funds, but also to other banks and building societies.⁶

So given the changing shape of the market, what implications are there for appropriate regulatory response?  Well as I stressed earlier, we are not ready yet to make even initial proposals for debate, let alone firm recommendations. But I will make some comments, first about the different objectives we could be trying to achieve (and some possible tradeoffs between them) and on the different regulatory tools we could consider using.

We could have at least three different objectives:

• First, to protect individual customers against the consequences of over-risky borrowing. We know that people are not fully rational or foresighted in their financial decision making. That is why, for instance, it has been accepted that public policy in the pensions saving arena should include at least a nudge, through automatic enrolment, towards higher saving than would otherwise occur, and the same arguments could well apply in respect to borrowing products. And the FSA already regulates the selling process for first mortgages⁷ – we have a principle of suitability. The question is whether we should decide that there are some products (for instance, loans above a certain LTV or LTI) which are so unlikely to be suitable for any customer that they should simply be banned and/or whether we can and should tighten our regulation of selling processes, with more restrictive and aggressively enforced definitions of suitability.
• Second, our aim could be to protect the banking system from losses, given the large economic harm that occurs when the banking system suffers large losses and has an impaired ability to extend credit.  And the FSA has already made it clear that we will be far more willing in future to make judgments about sustainable business models, rather than relying on the wisdom of markets to discipline and control excessive risk taking. But it would only make sense to pursue this objective via the regulation of allowable product characteristics, rather than constraints applied via firm capital and liquidity requirements, if we believe that a particular product category such as residential mortgages is so particularly important as a source of risk that it requires a specific approach. And that case is not clear. Commercial real-estate lending has in past recessions, and may again in this one, prove as troublesome an asset class as residential mortgages, but we do not regulate the product characteristics of commercial real-estate lending. On the other hand, there are several countries in the world which have chosen to regulate residential mortgage leverage, while not regulating product leverage in other lending markets, and some of them believe that this has achieved important financial stability benefits – we need to understand those arguments.
• And third, we could be proposing mortgage regulation as a means to reduce the amplitude of economic cycles, if we believed that high absolute leverage (asset or income-related) tends to produce greater house price volatility and as a result greater volatility of consumer expenditure. But that proposition must be considered within the wider context of the debate about macro-prudential tools (such as countercyclical capital requirements) and the relationship of macro-prudential regulation to monetary policy.
• The relative priority of these different objectives carries implications for the mix of policy measures and for their appropriate tightness, and potential tradeoffs between different objectives need to be recognised.
• It may well be, for instance, that a policy designed around the consumer protection objective would focus on cutting off the extreme tail of very high asset or income leverage, or clearly bad sales practices, but would as a result have only a marginal impact on future trends of total mortgage borrowing, house prices, total housing wealth and thus macroeconomic volatility.
• Conversely, constraints tight enough to make a difference to macro variables, certainly if pursued via an LTV constraint but perhaps also via LTI constraints, would be likely to have a disproportionate impact on first time buyers with much less impact on  the behaviour of home movers and remortgagers. The social consequences of that would have to be carefully considered.

What this range of possible objectives and possible tradeoffs in turn implies, is that we need to think carefully about the relative merits of alternative policy instruments, and in particular the choice between product specific regulation, sales regulation and firm level regulation.

• A key choice in any product regulation would be between LTV and LTI limits. There is a prima facie case that LTI limits are more likely to be appropriate, given that it is LTIs that have risen over the last fifteen years, while LTVs have actually fallen, and given that it is the relationship between income and loan repayments rather than the existence of negative equity, which ultimately determines whether a household can service it’s debt.  But LTVs do have a major role in determining the lender’s loss given default. Detailed empirical understanding of the relationship between loan losses and either high LTVs or high LTIs should inform our debate, and that understanding will grow as the actual experience of the next few years emerges.
• But we also need explicitly to compare the merits of product level regulation, with firm regulation.  We need to look in detail at the appropriate capital treatment of mortgages of different LTIs and LTVs ratios.  But it is also possible that our most important policy tool will turn out to be tighter liquidity policies. We certainly do not know this, but I have suggested earlier that it is not impossible that this recession could produce arrears and losses less severe on average than the 1990s recession, but more concentrated in specific firms and markets segments. And those firms may well turn out to be those which took the greatest liquidity risks, because taking liquidity risks enables rapid growth, and rapid growth of market share or of total assets is in general a very good warning sign of likely credit losses. The FSA has already published specific proposals on liquidity⁸ which will significantly constrain the ability of banks to grow rapidly on the basis of wholesale market funding. And we have questioned whether a still more radical change in policy, involving the introduction of a Core Funding Ratio, may make sense.⁹ It is possible that these measures will prove to be the really powerful tools by which to constrain rapid growth in aggregate mortgage credit (with possibly beneficial macroeconomic effects) and will more effectively address the problem of specifically risky institutions than would product regulation.
• Finally, it may be that policy intervention should focus on specific selling approaches and specific institutional risk, rather than on product regulation or in addition to it. If the bad debt experience which actually emerges in this recession is concentrated in mortgages where the income was self-certified, or in mortgages originated by the specialist non-deposit taking lenders, rather than by banks or building societies, it may make sense to focus policy interventions on those specific problems.

So should the FSA end up recommending limits to LTV or LTI – the headline issue on which the debate about the future of the mortgage market sometimes focuses?  I do not at present know and I make no apology for that lack of certainty. Because what I have tried to do today is to indicate that the issue is a complex one, which requires careful consideration and further empirical analysis, running up to the FSA September Discussion Paper, and indeed subsequently, in a wide-ranging debate with many of you in this room and many not here today. We do not need to rush to decision. We do not face today, nor are we likely to face anytime soon, the danger of irrationally exuberant behaviour by either borrowers or lenders.  We have time to get it right. And getting it right is very important given the huge importance of the mortgage and housing markets, to individual households, to banks, building societies and other credits intermediaries, and to the macro economy.

Back to top

^1.ONS, ‘Capital stocks tables for publication’, Table 5.10, Bank of England and FSA calculations

^2.See ONS Financial Statistics, Table 3.1.G, bank and building society sterling lending to non financial corporates, unincorporated businesses and households.  Note that this table understates the scale of total mortgage debt since it excludes “Other specialist mortgage lenders” – see Table 3.2C.

^3.Lending associated with but separate from a mortgage (e.g. the Northern Rock “together” product) is not included in these figures.

^4.See “A rapid assessment of the economic and social consequences of worsening housing affordability”, Final Report for the NPHAU , February 2009 , for a discussion of possible drivers of this declining LTV.

^5.ONS UK National Accounts, Table A37-A41

^6.In October 2007 the FSA requested that building societies cease purchasing non prime assets

^7.The FSA does not at present, however, have any role in the regulation of second mortgages, which can be used to achieve high levels of total leverage.

^8.FSA Consultation Paper 08/22: Strengthening liquidity standards

^9.See the Turner Review Section 2.2.vii