Limits on debt-to-income as a macro-prudential tool

Transcription

1 Date: 19 August 2016 To: Minister of Finance Limits on debt-to-income as a macro-prudential tool 1. The purpose of this memorandum is to seek your agreement to add an additional class of policy tool to the Memorandum of Understanding on Macro-prudential Policy (MoU). The new class of macro-prudential instrument would constrain the availability of bank mortgages based on the customer s ability to service their mortgage debt. The use of the tool would be subject to the existing governance and consultation requirements set out in the MoU. Possible amendments to the MoU are shown in the Appendix. 2. We believe that a debt serviceability limit could play an important role in promoting financial stability during periods of increasing housing market risks (section I). By reducing the proportion of new mortgage borrowers that are vulnerable to a rise in interest rates or fall in income, a serviceability limit would be expected to: o o o Reduce the risk of a significant rise in mortgage defaults during a severe economic downturn. Mitigate the potential amplification of the downturn due to falling consumption or increased house listings. Temporarily lean against periods of rapid house price and credit growth, thereby reducing the probability of a sharp correction in house prices. A debt serviceability limit has some similarities to the LVR restrictions currently in place, in that it would directly constrain the availability of credit to higher risk borrowers. However, the new class of policy instrument is likely to reinforce, rather than substitute for, the financial stability impact of an LVR policy. For example, some current borrowers have LVRs below 80 percent, but would struggle to service their loans if interest rates rose sharply.

2 3. Although the Bank is not proposing the introduction of a debt serviceability tool at this time, preliminary evidence suggests that many borrowers have debt burdens that leave them vulnerable to a rise in interest rates or fall in income (section II). As a result, we will be investigating the role that a debt serviceability tool could play in maintaining financial stability, particularly if imbalances in the housing market continue to grow. There are a range of calibration and design issues that would need to be considered before implementing such a tool (section III). 4. We would welcome the opportunity to discuss this memo with you and the Treasury. I. An overview of debt-to-income instruments 5. This memo discusses a class of macro-prudential instrument that would constrain the availability of bank mortgages based on the borrowers ability to service the loan. Such a tool could take two broad forms: o o A limit based on the indebtedness of the borrower. A specific example would be a limit on new bank mortgage lending above a certain debt to income (DTI) ratio, defined as the ratio of the debt obligations of the borrower to the income used to service this debt. A limit based on the debt repayment burden of the borrower. A specific example would be a limit on new bank mortgage lending above a certain debt service ratio (DSR), defined as the ratio of required mortgage payments to income. These instruments are closely related, as a DTI effectively caps the debt repayment burden at a given interest rate and loan maturity. We use the term debt serviceability limit when referring to both forms of the tool. 6. As with the macro-prudential tools currently in the MoU, the underlying rationale for a debt serviceability limit is to reduce the pro-cyclicality of the financial system. Individual bank lending decisions fail to take account of the additional credit risk arising from the macroeconomic effects of those decisions. For example, a coordination failure can emerge during periods of heightened competition for mortgage loans, where banks are pressured to ease serviceability standards to maintain market share. As discussed below, easy lending criteria can in turn amplify the potentially damaging effects of boom and bust cycles in house prices and mortgage lending. 7. A key objective of a debt serviceability limit is to mitigate the rise in mortgage defaults during a severe economic downturn. All else equal, higher DTIs reduce the uncommitted income a borrower has available as a buffer against a period of lost income or increase in interest rates. This increases the likelihood that a household becomes unable to service its loan if economic and financial conditions worsen, and consequently needs to sell its property. If these sales occur in an environment of falling house prices, there would be a risk of a sharp rise in bank losses on loans to highly indebted households. Distressed sales and falling consumption by highly indebted households are also likely to reinforce any downturn in the housing market and wider economy. 2

3 8. A large body of evidence suggests that both ability to pay and loan-to-value ratio are important determinants of loan default. Factors determining ability to pay such as regional unemployment, loan-to-income ratios and interest rate structure played a significant role in the rise in mortgage defaults after the GFC in Ireland and the US. 1 A debt serviceability policy would therefore tend to complement LVRs in reducing credit risk on mortgage lending, particularly for owner-occupiers (who tend to continue servicing their loans if they can, even when they are in a position of negative equity). 9. The financial system is also indirectly exposed to risks associated with high-dti households. The consumption of indebted households tends to fall sharply in response to shocks to income or wealth, as attempts are made to continue servicing loans and precautionary saving increases. Weak demand can be exacerbated by reduced availability of credit as banks tighten lending standards. A number of country studies have found that sharp falls in consumption by indebted households reinforced the economic impact of the GFC (Thornley (2016)). 2 Increased house listings by indebted households could also amplify a downturn in the housing market, particularly if investors with multiple properties were to come under stress. 10. By constraining credit availability for some borrowers, a debt serviceability limit could also have some impact in lowering credit demand and house price inflation. Although the size of the impact would depend on the specific calibration of any limit, a potential advantage of a debt serviceability policy is that the borrowing capacity of restricted borrowers would be limited by growth in incomes. In contrast, LVR restrictions may be less binding when house prices are increasing rapidly, since rising equity provides existing owners with scope to periodically increase their loan balance. Incomes may thus be a more effective and enduring anchor for a macro-prudential policy during periods where the house price-to-income ratio is increasing. 11. A debt serviceability limit could have unintended consequences, which would need to be monitored carefully. By restricting the availability on bank loans, a serviceability limit could decrease financial system efficiency if institutions that are not subject to the policy are incentivised to increase riskier lending. The experience with the current LVR policy suggests that this risk is currently limited due to the small size of the non-bank sector. The risk could be further mitigated by (i) making any limit temporary and (ii) allowing for a speed limit of loans above the restricted threshold to be allocated by banks. 12. Tighter borrowing constraints under a debt serviceability policy would affect the allocation of mortgage credit, thereby potentially constraining the housing choices available to some households. This could enhance the allocative efficiency of the financial system, to the extent that the policy is helping to account for systemic risks that are not considered by individual banks. Nevertheless, it is unlikely that a 1 For example, see Kelly R and T O'Malley, A Transitions-Based Model of default for Irish Mortgages," Research Technical Paper 17RT14, Central Bank of Ireland 2014; and Gerardi K, K Herkenhoff, L Ohanian & P Willen (2015) "Can't Pay or Won't Pay? Unemployment, Negative Equity, and Strategic Default" NBER Working Papers Thornley (2016) Financial stability risks from housing market cycles, Reserve Bank of New Zealand Bulletin. 3

4 debt serviceability policy could reflect the full range of special borrower circumstances assessed by banks. Unintended consequences on housing purchases could be mitigated by allowing for a relatively large speed limit under a debt serviceability policy, or exempting certain categories of borrower. 13. A debt serviceability limit would be more complex to implement and monitor than the other macro-prudential tools currently in the MoU. For example, it would require reliable data on a borrowers income (a variable which is unlikely to be measured identically by banks) and would ideally be calculated on the basis of the borrowers total debt commitments (thereby requiring banks to measure any debt held at other institutions). The Reserve Bank has confronted some of these issues as part of its experimental DTI data collection, but further work would be required before any debt serviceability limit can be implemented. 14. Limits on DTI or loan serviceability are becoming increasingly common internationally, and have commonly been used in conjunction with LVR restrictions (table 1). Both the UK and Ireland have recently applied a limit on the proportion of loans that can be originated by banks above a specified DTI threshold. Restrictions on debt servicing burdens are also increasingly common. To date, most tools have been applied either to owner-occupiers (O-Os) specifically or to all mortgages (we are not aware of instances where such instruments have been used to specifically target investors). 15. A number of countries have introduced guidance for banks on mortgage origination standards related to serviceability, including Australia. These guidelines tend to influence lending standards at a more granular level than the debt serviceability tool that the Reserve Bank is proposing to add to the MoU. However, a prudential practice guide would be difficult to monitor and enforce in New Zealand, given that the Reserve Bank does not have a history of on-site supervision or differentiating regulatory requirements (such as capital requirements) for individual banks. Table 1: Selected interventions related to mortgage serviceability Country Limit Coverage Debt-toincome Ireland No more than 20% above 3.5 Owner-occupiers only limits United Kingdom No more than 15% of lending above 4.5 Owner-occupiers only Other examples: Singapore. Debt service ratio limits Canada Maximum around 40 to qualify for government insurance. Hong Kong Limit of 50 for owner-occupiers and 40 for investors Other examples: US, Korea, Israel, Lithuania, Estonia. All mortgage lending receiving government insurance Bank mortgage lending Prudential practice guide Australia Minimum standards for origination tests (eg assumed interest rate, living expenses). Bank mortgage lending Other examples: Switzerland, Germany, UK. 4

5 II. Evidence on debt-to-income ratios in New Zealand 16. The deployment of some form of DTI constraint would be based on financial stability grounds i.e. the case that DTI trends were posing risks to future financial system soundness. Accordingly, there is a need for comprehensive data enabling DTI trends to be reliably monitored and analysed. Comparability of data collected from banks on borrower DTIs is particularly important in drawing conclusions about system-wide trends. 17. The Reserve Bank has been collecting experimental data on debt-to-income ratios from major banks since Our focus below is on the total debt-to-income (TDTI) ratio. The numerator of TDTI captures all of the borrowers contractual debt obligations, including any amount disclosed by the borrower that is owed to other banks. The denominator measures the borrowers before-tax income, largely measured according to the approach used by each bank when applying mortgage origination tests The data suggests that around 30 percent of residential mortgages in New Zealand have a TDTI exceeding six times the borrowers income. Anecdotal evidence and household survey data suggests that the availability of high-dti loans increased after around 2010, after falling sharply in the wake of the GFC. 4 Sustained declines in interest rates and renewed competition among banks for profitable mortgage loans have been significant factors enabling borrowers to obtain larger loans relative to income. 19. Assuming that a borrower is not constrained by the Reserve Bank s LVR policy, the maximum debt-to-income available to a prospective borrower is determined by the bank s assessment of their loan servicing ability. These assessments are designed to ensure customers have sufficient residual income, after mortgage and other commitments are made, to meet an estimate of essential living costs. Based on a survey of major banks during 2014, banks tend to cap maximum DTIs at less than 5 for O-Os with moderate to low incomes. By contrast, the maximum DTI tends to be much higher for high-income O-Os or investors (figure 1). Figure 1 Range for maximum debt-to-income ratio by total gross income 3 The one exception is rental income, where all banks have supplied data on the basis of a 25 percent haircut to rental income in order to account for rental expenses and the possibility of a period without a tenant. 4 See Dunstan A and H Skilling (2015) Vulnerability of new mortgage borrowers prior to the introduction of the LVR speed limit: Insights from the Household Economic Survey, AN2015/02. 5

6 Ratio $75,000 $120,000 $250,000 $135,000 Owner-occupier Investor Ratio Source: RBNZ Hypothetical Borrower Exercise, completed by major banks during The implied LVR for the loans is around for owner-occupiers, and around 60 for investors. 20. Consistent with this finding, most high-tdti lending is undertaken by borrowers with a relatively high income. Around 70 percent of all high-tdti lending is to borrowers with income above the New Zealand household average of about $90,000, and around 40 percent of lending is to borrowers with incomes of above $140,000. While high incomes help to mitigate the risks associated with elevated TDTI ratios, the underlying assumption that high income households could reduce expenditure to essential levels typically seen in low income households appears questionable. 21. Based on data on the stated purpose for taking out the loan, high-tdti lending is also concentrated among investors (figure 2). Investors tend to have more free cash flow at a given DTI than owner-occupiers, which makes banks more comfortable providing high-dti investor loans. This is largely due to investors having significantly lower tax bills relative to a homeowner with a similar gross income, and because most investors do not need to meet living expenses out of the rental income they earn. Nevertheless, the elevated proportion of investor lending at very high-tdtis suggests that many investors could be vulnerable to a rise in interest rates or a period without rental income. Figure 2 Summary of DTI distribution by purpose of lending (January to June 2016) % 35 Investor First home buyer Other owner occupier < >

7 Source: Experimental DTI data from major banks. 22. Figure 2 shows that first home buyers (FHBs) have significantly lower DTIs than investors. (This contrasts with the situation for LVRs, where FHBs tend to have much higher ratios than investors.) A typical FHB will have less free cash flow at a given DTI than a typical investor, but will also typically experience faster than average income growth following loan origination. We believe that other O-O DTIs are biased upwards in this data, due to investors being classified as other O-Os when they purchase a property for the purpose of owner-occupation. 23. We believe that the introduction of the LVR policy in 2013 has resulted in some reduction in DTI ratios relative to the counter-factual. However, the share of high- DTI loans has trended up for all buyer types since The mean DTI for a FHB in the top half of the distribution has increased from 5 to 5.5, while the same figure for investors has increased from 7.5 to 8. The rise in DTIs for FHBs is likely to reflect significant increases in house price-to-income ratios in many regions. Rising house prices are also likely to be a driver of increased investor DTIs, reflecting falling rental yields and increased scope to borrow against rising housing equity. 24. It is very difficult to compare DTI ratios across countries, reflecting differing tax regimes, definitions of debt and income, and data coverage (eg many countries exclude investors). Notwithstanding comparability issues, our assessment is that a large proportion of lending is at DTI ratios that appear high by international standards, even though mortgage rates are also relatively high. It is notable that both the UK and Ireland have applied limits on O-O DTIs at multiples of , and that these limits do not strongly bind on current lending. While we believe the current O-O TDTI data is biased upwards, there is a significant amount of O-O lending well above these ratios in New Zealand. III. Implementation choices for DTI restrictions 25. If the Reserve Bank were to implement a debt serviceability tool in the near future, it would likely be as a complement to LVR restrictions (rather than supplanting them). This reflects the evidence discussed above that both LVR and ability to pay are important determinants of mortgage credit risk. In addition, there is no mechanical link between a borrower s DTI and LVR. A significant proportion of high-tdti lending is less risky on an LVR metric, and vice versa. 26. The addition of another tool would add to the complexity associated with macroprudential policy. In order to reduce compliance costs for banks and uncertainty for the general public, any new limit should be as simple as possible. Similar to the current LVR policy, any debt serviceability limit would likely allow for a speed limit of loans that banks can allocate at above the restricted threshold. This would mitigate the need to create complex exemptions for special borrower circumstances. 27. If a debt serviceability tool is implemented, there are design questions around how debt and income are measured: Debt can be defined narrowly to focus on mortgage debt at the bank approving the loan, or expansively to encompass all contractual debt obligations, including non-mortgage debt and any debt held at other banks. A narrow definition is 7

8 easier to measure and enforce, but would also potentially be easier for borrowers to avoid (eg mortgage debt could be reduced by splitting a property investment portfolio across multiple banks, or maintaining credit cards at multiple banks). There are also a range of possible ways to measure income. Most policies internationally are based on gross income, often allowing banks to use internal policies for deciding what income sources are stable enough to be counted (and what haircuts should be applied to things like overtime income). However, this would require periodic assessment of whether bank data was sufficiently comparable. Finally, some tools could potentially focus on a particular class of income (eg a rental interest coverage ratio, as is currently being developed in the UK). 28. The appropriate coverage of a debt serviceability limit is another area where further work would be required. Revised TDTI data is expected to shed more light on how risks around high-tdti lending vary across owner-occupiers and investors (see below). We will also be monitoring the impact of the recent tightening of the LVR policy, which is expected to bring about some reduction in high-tdti lending for investors. If our monitoring suggests that this tightening has sufficiently mitigated financial stability risks associated with investor lending, it is possible that a debt serviceability policy would be applied only to owner-occupier loans. 29. There are a range of other design questions that would need to be considered if some form of debt serviceability limit were implemented. If the limit included both investors and owner-occupier lending, it would be possible to calibrate the tool to take account of the fact that investors have a greater amount of free cash flow at a given TDTI. It would also be an option to calibrate the policy and exemptions in a way that allowed some first home purchases and/or relatively low value purchases to proceed without policy constraint. Conclusion and next steps 30. Debt serviceability measures would be a useful addition to the Reserve Bank s macro-prudential toolkit, and we are seeking your approval to add this class of instrument to the MoU. Possible revisions to the MoU are shown in the Appendix. 31. A debt serviceability tool may have a role to play in further mitigating financial stability risks associated with the buoyant housing market. In coming months, the Reserve Bank will be investigating whether a debt serviceability limit should be implemented and, if so, how it should be designed. Part of this work involves monitoring the financial stability impact of the recent tightening of LVR restrictions for investor loans. We are also starting to work with banks on a revised TDTI dataset, with the objective of better measuring risks across different buyer types, improving understanding of how different banks measure and verify debt and income, and widening the coverage of the survey to include smaller banks. We will consult you and the Treasury as our thinking develops. 8

9 Graeme Wheeler Governor CC: Secretary to the Treasury Appendix: Possible revisions to the Memorandum of Understanding between the Minister of Finance and the Governor of the Reserve Bank of New Zealand Memorandum of Understanding: Macro-prudential policy and operating guidelines This agreement between the Minister of Finance (the Minister) and the Governor of the Reserve Bank of New Zealand (the Bank) defines macro-prudential policy and the operating guidelines that the Bank shall operate under when considering the use of macro-prudential policy. The international practice of macro-prudential policy is a developing area and it is expected that the Bank s macro-prudential policy framework will evolve over time. Accordingly, this agreement may be amended from time to time. The proper purpose for macro-prudential policy that underlies this agreement is provided for in Section 1A (1)(b) of the Reserve Bank of New Zealand Act 1989 (the Act), which requires the Bank to be responsible for promoting the maintenance of a sound and efficient financial system. In conducting macro-prudential policy, the Bank seeks to reduce or manage the risks to the financial system arising from extremes in the credit cycle or developments in liquidity conditions and global debt markets, through the use of the prudential instruments listed below. 9

10 Effective macro-prudential policy depends on the timely use of instruments. This memorandum of understanding (the Memorandum) provides clarity over the purpose and instruments of macro-prudential policy, so that emerging systemic risks are able to be addressed in a timely manner. This agreement covers the application of macro-prudential policy instruments to the registered banks, which account for the major share of domestic lending to households and businesses in New Zealand. However, it is acknowledged that, in some circumstances, it may be desirable to apply macro-prudential instruments more widely. The Bank will advise the Minister of any proposed changes to the macro-prudential framework that would extend the use of macro-prudential instruments to non-banks, including any changes to the Bank s powers or involvement of other agencies that might be required. 10

11 The Minister and the Governor agree as follows: 1. Objective of macro-prudential policy The objective of the Bank s macro-prudential policy is to increase the resilience of the domestic financial system and counter instability in the domestic financial system arising from credit, asset price or liquidity shocks. The instruments of macro-prudential policy are designed to provide additional buffers to the financial system (e.g. through changes in capital, lending and liquidity requirements) that vary with the macro-credit cycle. They may also help dampen extremes in the credit cycle and capital market flows. As such, these instruments can play a useful secondary role in stabilising the macro economy. As a result, the Reserve Bank will consider any interaction with monetary policy settings when implementing macro-prudential policy and will explain the implications, if any, for monetary policy. 2. Operating guidelines This agreement confirms the guidelines the Bank will operate under, in discharging its obligations under the Act. 2.1 List of macro-prudential instruments The following macro-prudential instruments are considered useful in the New Zealand context for addressing the systemic risks of financial instability: Adjustments to the Core Funding Ratio a minimum core funding ratio requirement that could vary the proportion of lending the banks are required to fund out of stable core funding sources over the cycle, and is intended to reduce the vulnerability of the banking sector to disruptions in funding markets Countercyclical Capital Buffer an additional capital requirement that may be applied in times when excess private sector credit growth is judged to be leading to a build-up of system-wide risk. The buffer would be able to be released when the credit cycle turns down, helping to reduce the risk of a sharp contraction in the availability of credit Adjustments to sectoral capital requirements an additional capital requirement that may be applied to a specific sector or segment in which excessive private sector credit growth is judged to be leading to a build-up of system-wide risk Quantitative restrictions on the share of high loan-to-value ratio (LVR) loans to the residential property sector. These could include: Restrictions on the share of new high-lvr lending that banks may undertake; Outright limits on the proportion of the value of the residential property that can be borrowed to create a minimum equity buffer for the lender. 11

12 .... Information withheld under.. section 9(2)(f)(iv) Development of any additional macro-prudential instruments will be undertaken in consultation with the Treasury, given the Treasury s role in advising the Government on risks to the Crown s balance sheet. 2.2 Operation of macro-prudential instruments The Bank will assess financial system developments, and monitor risks to the system. The Bank will publish information on its risk assessment framework, including the macro-prudential indicators that are used to guide its macro-prudential policy settings. Where significant risks are judged to be emerging, a case for macro-prudential intervention in the form of deployment of a macro-prudential policy instrument or instruments will be considered by the Bank. Macro-prudential instruments do not replace conventional prudential regulation but may be used from time to time to help manage the risks associated with the credit cycle. In most instances macro-prudential instruments will reinforce the stance of monetary policy. The selection of macro-prudential instrument(s) will depend on the type of risk being addressed. The decision on macro-prudential intervention will be taken by the Governor. 2.3 Relevant legislation This section sets out the Bank s prudential powers over the registered banks. Under section 67 of part 5 of the Act, the Bank is charged with undertaking prudential supervision of registered banks. Under section 68 of part 5 of the Act, the Bank is conferred with powers for the purpose of promoting the maintenance of a sound and efficient financial system. Under section 74 of part 5 of the Act, the Bank may impose conditions of registration relating to a range of specified matters, including carrying on business in prudent manner. Section 78 of the Act Carrying on business in prudent manner. The Bank is confined to considering, inter alia, the following matters: 12

13 (1)(c) capital in relation to the size and nature of the business or proposed business allows the imposition of a counter-cyclical capital buffer and/or sectoral risk weights in the conditions of registration; (1)(fa) risk management systems and policies or proposed risk management systems and policies allows the imposition of the Core Funding Ratio in the conditions of registration. Section 78(1)(fa) of the Act provides the basis for the implementation of quantitative restrictions on housing loan-to-value ratio limits. Under section 68B of the Act, the Minister may direct the Bank to have regard to a government policy that relates to the Bank s functions under Part Consultation The Bank will keep the Minister and the Treasury regularly informed on its thinking on significant policy developments relating to macro-prudential policy, and of emerging risks to the financial system. The Bank will consult with the Minister and the Treasury from the point where macroprudential intervention is under active consideration, and will inform the Minister and the Treasury prior to making any decision on deployment of a macro-prudential policy instrument. The Bank will consult with the registered banks prior to deployment of a macroprudential policy instrument in the manner required under Section 74(3) of the Act. The Bank will advise the Minister if it considers further legislative change is required to give full effect to any of the instruments outlined in Section Reporting and accountability The Bank s Financial Stability Report will report on matters relating to the soundness and efficiency of the financial system including any build-up of systemic risk, and the reasons for, and impact of, any use by the Bank of macro-prudential policy instruments. The Bank shall be fully accountable to the Board, Minister and Parliament for its advice and actions in implementing macro-prudential policy, under the normal conventions outlined by the Reserve Bank Act. The appropriateness and effectiveness of macro-prudential policy decisions will be reviewed on a regular basis. This will include an assessment of the key judgements that led to decisions on whether or not to adjust macro-prudential policy. The Bank will report the results of its assessment in its Financial Stability Report. 13

14 The Minister and the Bank agree that a review of the macro-prudential framework shall be conducted after five years. Hon Mr Bill English, Minister of Finance Mr Graeme Wheeler, Governor, Reserve Bank of New Zealand 14