Response to submissions on adjustments to restrictions on high-LVR Residential Mortgage Lending

Ref #6220011
Response to submissions on adjustments to restrictions on high-LVR Residential Mortgage Lending. Aug 2015

2
Ref #6220011
Introduction

1. On June 3, the Reserve Bank published a consultation paper with proposals to tighten LVR
   restrictions on Auckland residential property investors while easing restrictions outside Auckland. Submissions on the initial package of proposals closed on 13 July 2015. 13 submissions were received, principally from registered banks that offer residential mortgage lending. After discussions with the banking industry, an alternative calculation method was also outlined to registered banks for feedback in a letter on July 13.
2. This document reports on the submissions received and explains the anticipated changes to
   BS19. In Annex One, we provide a regulatory impact assessment for the proposed policy.
   Annex Two describes the rationale behind some changes we have made to our proposals for
   the investor asset class in BS2A/B which relate to customers that have a mixture of owner￾occupied and investment collateral. Annex Three lists the names of respondents to the 3 June consultation paper. Rationale for restrictions on investor lending
3. The consultation paper presented evidence on the following points:
   a. House prices in Auckland have become very elevated relative to income, unlike in the rest of New Zealand. While this is true of other cities such as Sydney, Hong Kong and London, they are much larger and tend to have much lower interest rates. House prices to rents are also more elevated in Auckland than elsewhere in New Zealand. In the Reserve Bank’s view, these developments increase the risk that Auckland house prices could fall relatively sharply in the future. b. Analysis of the GFC experience in the US and Ireland suggested that loss rates in those countries tended to be higher for investors at any given LVR during severe downturns. We also noted that rating agencies also have some evidence of this for Australia, and tend to assume in their rating of mortgage-backed securities that investor loans are riskier at any given LVR.
4. Respondents provided few comments on our discussion of the Auckland market. A couple of
   respondents suggested that a number of mitigants limit the risk of a large price correction in Auckland, including sustained housing supply shortfalls and strong/regionally-focussed immigration.
5. With regard to investors being riskier at any given LVR, there was a variety of views. Many
   respondents referred back to their submissions on this matter in response to the consultation paper on the establishment of a residential investor capital class. Several reiterated the view that the Irish and US experiences differed from New Zealand origination practices in important ways.
6. Respondents argued that tighter New Zealand credit/origination policies, together with
   capital buffers (including higher capital that will be required following the establishment of a new asset class for residential investment properties), make banks better equipped to deal with a housing downturn than is the case in other jurisdictions/episodes.
7. Some respondents argued that historical New Zealand housing downturns have tended to be
   shallow and short-lived. Others noted that residential investment loan portfolios have

3
Ref #6220011 tended to have a similar or better performance than that of owner occupiers (particularly if the investor’s own house was part of the collateral package).
8. The resilience of the major banks to the 2014 banking industry stress tests was also
mentioned with some submitters noting the tests had revealed that banks could withstand a very large fall in house prices together with a substantial deterioration in other economic variables such as unemployment. Respondents also noted that sustained house price inflation since the test has tended to create larger equity buffers for the stock of housing loans.
9. Some commentators suggested other tools (relating to debt to income limits, or limiting
loans to a maximum multiple of rents, or requiring banks to hold additional capital) would be a more appropriate policy response. Response
10. The Reserve Bank agrees that migration flows and the apparent shortage of housing supply
differentiate the Auckland housing market from many other areas of New Zealand and have been important factors underpinning a sustained increase in house prices in the region. However, the Bank remains of the view that these factors could change in the future and make high LVR lending at the current level of Auckland house prices relatively risky.
11. The Bank acknowledges that there is little New Zealand evidence that investors are riskier at
any given LVR, but consider this to be a natural consequence of the housing market having had only mild downturns since the early 1990s. The absence of a severe housing market downturn in the last 20 years is not evidence that one could not occur. House prices have reached unprecedented levels relative to income in Auckland, so historical New Zealand downturns may not be a good guide to the consequences of a future severe downturn. This leads us to look at other countries that have had severe downturns. While origination practices in the US and Ireland differ from New Zealand in important ways, we still consider that the empirical evidence from those downturns is relevant to considering what would happen in a severe New Zealand downturn.
12. Some commentators on our arguments in this area have noted that investors do not appear
to be materially riskier than owner occupiers in some Irish studies after controlling for other factors. For example, if investors were more active in the last years of the boom and in areas where house prices were most overvalued, these effects may help to explain their high default rates. It is plausible that these factors help explain relatively high investor default rates, but this is not inconsistent with our view that investor lending is riskier on average at any given LVR, it simply provides some more evidence about possible causal factors. The Kelly (2012) results show that investors are more likely to default at any given LVR, particularly when they have reached a position of negative equity (controlling only for loan vintage and unemployment rate).
13. One respondent provided additional rating agency data, showing that investors had higher
default rates than owner occupiers in the UK downturn (controlling for LVR) but that this effect only really started at LVRs above 80%. This evidence is useful, but given that the UK housing downturn was not as severe as in the US and Ireland, does not necessarily imply that 70-80% investor LVRs pose little risk in a severe downturn.

4
Ref #6220011
14. It is true that rising house prices do not immediately increase the risks of losses in a stress
test. Indeed any given percentage fall in house prices will leave house price levels higher in absolute terms if house prices have risen further prior to the downturn (so someone borrowing years prior to the downturn may still have substantial equity). The Reserve Bank is mindful, however, that gross housing credit originations are substantial (in the order of 30% of the outstanding stock of housing credit each year). So elevated levels of house prices tend to lead fairly quickly to higher levels of borrowing and debt to income ratios for many borrowers. Additionally, if house prices rise further relative to fundamentals they are likely to fall further in a downturn.
15. The Reserve Bank’s legislative objective is promoting “soundness and efficiency” of the
financial system. In the context of macro prudential policy, the Memorandum of Understanding on Macro-Prudential Policy (MOU)1 agreed between the Minister of Finance and the Governor creates two specific intermediate objectives, namely providing additional resiliency buffers in the financial system, and also (where possible) dampening extremes in credit cycles. LVRs directly affect credit origination practices (which we consider are currently quite permissive). In a severe downturn, even if the banks did not face a solvency threat, they would be likely to sharply tighten home loan origination standards. The impact of this tightening on the housing market and individual borrowers will be greater if standards have been unusually permissive prior to the downturn. Thus even if we could be entirely confident that the financial system would remain solvent following a sharp housing downturn, a role for macroprudential policy remains.
16. The Reserve Bank acknowledges that capital tools could be considered, as could debt to
income restrictions, although the latter is not currently one of the tools permitted by the MOU. The reasoning behind the choice of LVRs was discussed in the consultation paper (and the paragraph above is also relevant). Policy objectives and expected effectiveness
17. Several respondents felt that the proposed LVR policy would have a limited impact on house
prices given that supply shortages (and low interest rates) appear to be a key driver of Auckland house prices. Respondents also suggested that many investors currently borrowing at 70-80% LVR would be able to obtain more equity by restructuring their borrowing (eg putting more borrowing against the family home or other properties).
18. There was general consensus that there could be some redirection of investment outside of
Auckland and potentially upward pressure on regional house prices as a result.
19. Respondents generally agreed with the loosening of LVR restrictions outside Auckland, with
a number encouraging a further loosening, or a formal regular review process.
20. There was general support for maintaining an Auckland owner-occupied high-LVR speed
limit of 10%.
1 http://www.rbnz.govt.nz/financial-stability/macro-prudential-policy/mou-between-minister-of-finance-and￾governor-of-rbnz

5
Ref #6220011
Response
21. As noted above, we agree that this policy is not likely to dramatically change the landscape
for Auckland house prices. However, we consider that the factors that make housing scarce at present may reverse in the future (e.g. supply may loosen, migration flows may reverse and interest rates are likely to increase from current low levels), and that this could lead to a different demand/supply balance and tighter origination practice.
22. Some investors may be able to restructure their funding to reduce their desired LVRs below
the 70 percent speed limit while still buying additional properties. The RBNZ’s estimates of policy effectiveness from the initial consultation paper assumed investors would have more scope to find additional equity than other home buyers. These investors will also tend to be more resilient in a severe downturn than those without additional sources of equity, so we do not see this as greatly compromising the effectiveness of the policy proposed.
23. LVR restrictions will all be reviewed periodically in the future, and discussed in the FSR every
6 months whether or not a policy change has been made, and through other channels (e.g. policy announcements may be made at any point). Mixed loans and investor status
24. One issue that generated a lot of feedback was the treatment of customers who had mixed
collateral (e.g. an Auckland owner-occupied property and a number of Auckland rental properties). Banks typically treat the combined business of customers with multiple mortgage loans as a single asset. In the original consultation paper, the Reserve Bank proposed a method for splitting new lending to customers across the speed limit classes. As noted above, the Reserve Bank then wrote to registered banks on 13 July 2015 after receiving feedback from a number of banks which suggested the proposed method of ‘apportioning’ new commitments across speed limit classes was overly complicated. The alternative would be to require customers to be placed entirely in the most restricted class (for example, if any of the collateral is an Auckland investor property, the entire loan would be treated as an Auckland investor loan).
25. A number of banks provided useful responses, but there were mixed views about whether
the complexity apportioning created was worthwhile. Some banks considered that apportioning should be retained. Some noted that it made intuitive sense, and others noted that they had already begun to work out how to operationalise it.
26. A majority of banks, including some major banks, suggested an approach where a customer
is placed into a single speed limit class would be much simpler to operationalise and would be preferred. Some noted they would be forced to do it this way (which the original rules allowed) and would be at a competitive disadvantage.
27. Other banks that favoured apportioning being allowed still suggested there might actually be
an advantage to not apportioning, and said they might still choose not to apportion. Not splitting customers will tend to increase the denominator for the Auckland investor speed limit. If the Auckland investor speed limit is the most important constraint on a lender, not splitting customers may thus make the overall LVR limits easier to manage. This partly reflects the ‘combined collateral exemption’ which allows banks to lend up to 80% against

6
Ref #6220011 non-Auckland investor collateral even if it is part of a package with Auckland investor collateral. Respondents generally supported this exemption.
28. One bank suggested an alternative to apportioning, where the purpose of the loan rather
than the nature of the collateral should be used to classify loans. For example, a loan to buy a house outside Auckland would be treated as a non-Auckland loan even if the loan was partly backed by Auckland collateral.
29. Banks noted the policy could still create some incentives to split loans between banks. For
example, a customer with an existing investment property (at an 80% LVR) may be able to borrow more (up to 70% on new collateral) at a new bank, whereas at their existing bank they will generally need to stay under 70% against all Auckland investor collateral if borrowing more.
30. Similarly, it was also noted that a customer who wants to borrow more than 80% to buy a
property outside Auckland may find the loan easier to obtain if they take that loan away from a bank where they also have Auckland investment properties. One bank noted that the incentive to split the lending in this particular instance may be greater if apportioning is not allowed.
31. With regard to investor and owner occupancy status, respondents made a number of
comments, including in relation to the definition of related party. Because BS19 takes its definition of investor from BS2, these issues were considered in the final proposed versions of BS2 (see Annex Two). Response
32. The Reserve Bank appreciates the considered feedback provided by the industry on these
points. We are aware there are arguments on both sides, but we consider overall that it is appropriate to harmonise on a system where customers are placed in the most restrictive relevant category.
33. In our judgement this makes the restrictions easier to implement without greatly affecting
the overall extent to which the policies constrain lending. By ensuring that all banks operate in the same way, there is a more level playing field and the data obtained will be easier to interpret.
34. There are a few cases where customers may face additional incentives to split banking
arrangements if apportioning is not being used. However, we consider that the combined collateral exemption means that cases where split banking will allow customers to borrow significantly more will be relatively rare.
35. We consider that the use of loan purpose codes would have created complexities for BS19
implementation even if they were in use at all banks. Furthermore, most respondents did not object to using the nature of the collateral to determine the type of loan. It would not be possible to allow banks to choose between categorising based on loan purpose or the nature of the collateral without creating significant disparities between the treatment of individual customers at different banks. More generally, we consider that purposes are not always easy to define (money is fungible once borrowed) and prefer the LVR restrictions to be based on the nature of collateral.

7
Ref #6220011
Calculation of speed limit and timeframe
36. Banks noted that the implementation of these restrictions would be complex and generally
asked for more time and a longer initial compliance period. Banks noted a range of relevant factors, including the system changes required, the short period from final policy position to start of compliance window, the pre-approval pipeline, and the relatively low speed limit being suggested.
37. Smaller banks (who were defined in the existing rules as those who generally lend less than
$100m per month) have a longer (six month) measurement period for LVR restrictions. Most respondents were comfortable with retaining the $100m boundary for the 3 vs. 6 month compliance window after the initial compliance period ends.
38. The 2% speed limit for Auckland residential investor lending >70% LVR was seen as too
restrictive, with at least 5% recommended.
39. There was support for providing further ‘error tolerance’ by, for example, allowing discretion
in determining materiality of non-systemic breaches, or moving to an anti-avoidance provision to allow for inadvertent errors.
40. A number of banks supported expressing speed limits on a common base (total mortgage
lending) to smooth regional variation in the denominator and reduce complexity. Similarly, one small bank supported just two speed limits – one for Auckland and one for all other regions rather than a separate investor and other borrower speed limit for Auckland. The purpose would be to reduce administrative burden. Response
41. The Reserve Bank has considered feedback about the difficulty of managing a very low
speed limit for Auckland investor lending. Given that something like 50% of Auckland investor loans are currently at LVRs above 70%, we agree that the speed limit on Auckland investor lending can be raised above 2% without compromising policy effectiveness. We have decided that the speed limit for Auckland investor lending will initially be set at 5%, instead of the 2% proposed in the consultation paper. This will cover some errors and exceptional cases. Some additional exemptions (see below) will provide some additional flexibility.
42. The Reserve Bank accepts that implementation of front line systems to achieve compliance
with these LVR restrictions will take some time. While we are modifying some technical aspects of the rules, we signalled the essential characteristics of the rules and required data in our June consultation paper. However, to ensure compliance is feasible, we have elected to push implementation out to 1 November and begin with a 6 month first compliance period for all banks (as we did when LVRs were first implemented in 2013).
43. This means that banks will first need to demonstrate compliance at the end of April 2016
(for the 6 month period finishing then) and attest to this in the following disclosure statements. Ideally, Banks will be able to produce new compliance reporting to the Reserve Bank on a monthly basis beginning with a report on November lending during December
2015. However, if necessary Banks can begin doing this slightly later in that 6 month period,
as long as data for each of the 6 months is eventually provided. Thus these changes provide substantial additional system build time for banks that need it.

8
Ref #6220011
44. While a single denominator would be easier to forecast, we have determined that it would
make the incidence of the Auckland investor restrictions overly variable across banks. For example, a bank that did a lot of mortgage lending outside Auckland but very little lending to Auckland investors (at any LVR) would find it relatively easy to meet an Auckland investor speed limit that was measured against all mortgage lending. In contrast a bank that was more focused on Auckland investors could find it much harder.
45. The calibration of the policy is designed to ensure that most investors in the Auckland
market are providing substantial deposits to reflect the elevated risks associated with the current Auckland housing market pressures. Estimates of the effect of the calibration have been produced and were published as part of the consultation paper. The impact of the policy will be reviewed over time. Unintended consequences
46. The construction exemption, which is already in the LVR rules, means that loans for
dwellings yet to be constructed or at an early stage of construction are exempt from the LVR restrictions. Respondents generally thought this exemption was working well and continued to have merit under the new policy proposals.
47. Some banks encouraged extending the construction exemption to include all new buildings
(i.e. already completed new dwellings yet to be occupied) to stimulate supply. One also mentioned the possibility of extending the construction timeframe permitted by the exemption to 2 years from 18 months.
48. Banks requested broadening the construction exemption to include remediation work, such
as rectifying leaky buildings, across all three proposed LVR speed limits. Lending related to remediation work was seen as ‘lumpy’, as many borrowers in a multi-unit block might need to obtain additional finance for remediation in the same month. This could be difficult to manage given the lower denominators associated with each speed limit in the current proposal, especially for Auckland investor lending.
49. Some banks also saw scope for an exemption on financial hardship cases to assist them in
fulfilling their obligations under the Credit Contracts and Consumer Finance Act. They also noted that they sometimes need to make payments on behalf of borrowers (e.g. rates or insurance) when the borrower has not done so, which could effectively be a high-LVR extension of credit.
50. Smaller banks (especially those that do little lending to Auckland investors) noted they could
inadvertently breach their condition of registration by writing a single high-LVR loan to an Auckland investor in error.
51. The costs of data/lending system changes to facilitate the proposed LVR policy changes was
estimated by banks that responded to the question, with answers ranging between $1.2m - $5m for large banks and between $0.25m - $0.5m for small banks. Response
52. The Reserve Bank accepts the points made around the desirability of exempting remediation
work. This appears to have been manageable under the single 10 per cent speed limit across

9
Ref #6220011 the whole country (adopted in October 2013), but would be harder to manage with a very low speed limit set on a relatively small portion of lending.
53. In order to operationalise the exemption, the Reserve Bank has drafted a separate “property
remediation” exemption that can be used for non-routine maintenance of a residential property. Some obvious cases the exemption is designed to include are weathertightness problems (not arising due to normal wear and tear or lack of maintenance), seismic or structural strengthening, or reconstruction after a fire or natural disaster.
54. The Reserve Bank also notes the concerns raised by smaller banks about the potential for a
single loan written in error to cause them to breach their LVR limits. In the spirit of the recent regulatory stocktake consultation document, which suggests conditions should be written to reduce the risk of relatively minor errors leading to breaches, the Reserve Bank considers it is appropriate to write an exemption that will effectively allow banks to exempt a single loan each month on the grounds that it was written in error. The Reserve Bank will expect to discuss the circumstances that caused the error with the bank, but the bank will not be in breach of its Conditions of Registration.
55. There are likely to be some other categories of lending that banks wish to undertake which
are not covered by exemptions, but should be manageable within the speed limits. In the Reserve Bank’s view, restructuring in the case of hardship, and compulsory payments of rates/insurance and the like, will typically involve relatively small amounts of additional lending and should be manageable within the speed limit. Hardship restructurings might become much more common in a severe housing downturn, but the Reserve Bank would expect to remove LVR restrictions if a housing downturn arises.
56. The construction exemption will continue to be available, including to Auckland investors.
We will adjust the maximum expected completion length for the construction project to 2 years (from 18 months) as requested. The requirement that the borrower commit to construction at an early stage will be retained. The alternative, where the exemption would apply to any new dwelling, could encourage flipping of partly or newly completed properties to new buyers at high LVRs. Ultimately the intention of the exemption is to fund construction rather than provide a pool of high-LVR buyers for newly built buildings. Our understanding is that most New Zealand builders tend to build houses for a specific buyer, so that this exemption can be used as drafted in most cases.
57. We note there will be significant system costs for banks in complying with the restrictions.
To some degree these should be mitigated by the delayed start and raised speed limit signalled above, which will allow banks a bit more time to work out how to comply and allow for more high-LVR lending. We discuss overall costs and benefits of the policy in the Regulatory Impact Assessment (see Annex One).

10
Ref #6220011
Anti-avoidance
58. The Reserve Bank requested information on lending secured by residential mortgage that
was not currently in the retail residential mortgage asset class. Some banks were able to provide estimates showing that this was fairly small relative to the residential mortgage asset class. Some of this lending (generally significantly less than half) was estimated to have LVR>70%, although LVRs are difficult to compute in some non-retail cases due to complex cross-collateralisation across a variety of security types.
59. Banks indicated a strong preference that this type of lending be classified as an avoidance
issue through anti-avoidance provisions of BS19 (rather than being dealt with more explicitly via formal LVR restrictions).
60. The proposed anti-avoidance clause described ‘lending primarily secured on residential
property that is treated as outside the residential mortgage asset class’ as a potential avoidance risk. Some banks queried the definition of ‘primarily secured’ and also asked if this clause was only intended to apply to Auckland investor loans. Banks also asked about the intended boundary between the residential mortgage asset class and more ‘corporate’ ownership of residential property.
61. Banks suggest monitoring of customer occupancy status outside of credit events would be
difficult. Customers would not necessarily notify banks when owner occupancy status changed, even if required to do so under loan contracts (which in general they are not).
62. Banks prefer to rely on customer attestations at the time of origination. Re-checking owner
occupancy status would occur at the time of credit events such as borrowing additional funds.
63. Banks signalled a strong preference that loans changing categories should not be counted as
new commitments, as they do not represent new lending.
64. There was widespread respondent support for replacing the BS19 conditions of registration
relating to second mortgage requirements with an anti-avoidance clause. Banks urged that this change be made immediately rather than as part of the final changes relating to the regulatory stocktake.
65. One bank sought clarification around the treatment of certain sources of deposits, such as
interest free family loans and social housing shared ownership structures. Response
66. The Reserve Bank accepts that there would be significant practical difficulties in imposing
LVR restrictions on relevant lending within the corporate asset class, and does not plan to do so.
67. It is important that this decision does not lead to bank loan policies changing so that
customers are placed outside the residential mortgage asset class in order to obtain high LVR loans. Indeed, the anti-avoidance clause proposed states that customers should not normally receive high LVR loans primarily secured on residential property if the loan will be placed in a different asset class.

11
Ref #6220011
68. The Reserve Bank is aware that loans to small businesses may be made primarily based on
an analysis of expected cashflows. Security will be taken where feasible, but is a secondary consideration. Even if the main collateral is residential property, the Reserve Bank does not consider a loan of this nature to be a residential mortgage loan and has clarified in BS19 that these loans should not be regarded as ‘primarily secured’ on residential property for the purposes of 7)2)f.
69. The Reserve Bank appreciates the information provided by the industry about how owner
occupancy status is identified. The general approach appears to be that it is not regularly checked, but is re-determined during credit events. Our key concern is that banks take reasonable steps to check buyer’s attestations about which properties they intend to occupy if they appear unusual or implausible. For example, if someone says they plan to owner￾occupy two homes in the same suburb, while there could be valid reasons for this, it would be suitable for the bank to discuss this further with the prospective borrower.
70. While the Reserve Bank will not alter BS19 to make a change in owner occupancy status a
‘new commitment’ we will be interested in discussing how common changes in occupancy status are with banks and would be concerned if it appeared to be being used as an avoidance channel.
71. The Reserve Bank is aware that banks regard breaches of conditions of registration as
serious. At the time the LVR restrictions were put in place, we considered a variety of ways the restrictions could have been avoided that would have been strictly permitted if not proscribed. Many were put into the anti-avoidance clauses, while some were seen as easier to define in formal conditions of registration, such as the second mortgage and broking rules. We have observed careful consideration by banks of the anti-avoidance clauses in BS19 since it was put in place in 2013. As the regulatory stocktake consultation document notes, it is appropriate for conditions to be written to avoid relatively trivial errors leading to breaches if that is practicable. Accordingly, we intend to remove several of these conditions of registration and replace them with similar anti-avoidance clauses in section 7 of BS19.
72. BS19 does not attempt to control the sources of the borrower’s deposit, and the Reserve
Bank recognises that borrowers will sometimes be assisted by parents or other parties. This is not ruled out by BS19. However, the anti-avoidance clauses note that the Reserve Bank would have concerns if a registered bank was directing customers to third parties so they could obtain deposits or knowingly providing unsecured funding that a customer was then using to obtain a deposit at that bank.
73. Various other minor amendments to BS19 are contained in the final draft text. For example,
the material on property valuation practices has also been removed because that is now covered in more detail in BS2A/B.

12
Ref #6220011
Annex One: Regulatory Impact Assessment
The Reserve Bank published a substantial Regulatory Impact Assessment for the initial LVR restrictions during 2013. The case for retaining LVR restrictions has been discussed since then in documents such as the Financial Stability Report. The proposed rule changes described above are incremental adjustments to the existing LVR restrictions, which focus the LVR restrictions more onto Auckland and in particular onto investors in the Auckland region. Because of the more incremental nature of these changes, this regulatory impact assessment is considerably shorter than the original one. The original one remains a useful for reference for understanding the original decision to implement LVRs. This regulatory impact assessment contains the following sections: a problem definition (section 1), an assessment of potential policy responses and, a summary of the costs and benefits of the preferred option (section 2), and a discussion of the consultation undertaken and the expected future monitoring of the policy (section 3).
1. Problem definition
The initial sections of our June 2015 consultation paper described the issues that led to our proposed changes, describing the experience to date with LVR restrictions, and the Reserve Bank’s perception of increasing risks associated with current high LVR lending to investors and to the Auckland region. Some key material from that paper is summarised in this section. New Zealand experienced one of the strongest increases in house prices in the OECD in the period prior to the global financial crisis (GFC), with house prices more than doubling between 2002 and
2007. Over the same period, household debt increased from 114 percent to 160 percent of
household disposable incomes (including debt held against rental properties). In contrast to many other countries, New Zealand did not experience a significant housing market downturn following the GFC, leaving house prices at relatively stretched levels relative to both incomes and rents. While the current LVR speed limits appeared to have a significant impact on housing market activity and growth in house prices, surprising strength in net migration and falling interest rates have boosted housing demand more recently. Since September 2014, there has been a significant increase in both housing market activity and house price inflation in the Auckland region. Auckland house prices have increased by around 24 percent over the year to July. At the same time that pressures in the Auckland housing market have increased, the Canterbury housing market has eased, and price pressures in the rest of the country have remained more subdued (figure 1).

13
Ref #6220011
Figure 1: House price-to-income ratios
Source: REINZ, QV, Statistics New Zealand.
As well as rising quickly, Auckland house prices are also very elevated relative to incomes and rents. The Auckland housing market is now one of the least affordable housing markets in the world, with a house price to household income ratio of around 9. By some estimates, this surpasses ratios seen in London and Melbourne, and approaches Sydney multiples. The consultation paper discussed why the Reserve Bank considers that these valuations create downside risks to borrowers and lenders. Rental yields in Auckland have been falling rapidly as house prices rise (figure 2), and data suggests investors have been particularly active in the Auckland market recently (figure 3). 0 2 4 6 8 10 0 2 4 6 8 10 1998 2001 2004 2007 2010 2013 Ratio Ratio Auckland Rest of New Zealand

14
Ref #6220011
Figure 2: Rental yields
Source: REINZ, DBH, RBNZ Calculations
Figure 3: Net investor purchases by region
Source: Corelogic NZ
Note: Percentage of sales to investors less percentage of sales by investors. As also explained in the consultation paper (points 24-31), the Reserve Bank considers that there is adequate evidence that loans to investors carry greater risk in severe downturns than loans to owner occupiers at any given LVR. This is discussed further in the response to submissions above. -9 -6 -3 0 3 6 Jun-13 Dec-13 Jun-14 Dec-14 Jun-15 -9 -6 -3 0 3 6 % % Auckland Rest of New Zealand

15
Ref #6220011
2. Possible Policy Responses
The consultation paper also discussed potential alternatives to LVR restrictions, including restrictions related to borrower income multiples, or capital based tools. Restrictions on debt to income would be administratively complex, because the definition of borrower income is not well standardised (unlike loan to value ratio). The Reserve Bank has been working with banks to obtain fairly well standardised statistical data, but it is not clear the data is sufficiently harmonised to allow for prudential limits at this point. Capital based tools such as sectoral overlays on housing risk weights would help to make banks more resilient in a downturn, but in our view would not materially affect lending practices (unlike LVRs). Thus borrowers would continue to be exposed to the risk of a sharp change in lending policies in a downturn. With regard to LVR restrictions, the conditions that motivated the imposition of LVR restrictions have intensified within the Auckland region. We also consider that investors are riskier at any given LVR and that the current cluster of Auckland investor lending around 70-80% LVRs would create financial stability risks in a downturn. Our proposed policy intensifies LVR restrictions on Auckland, while easing them in the rest of the country. Alternatives could have included status quo LVR restrictions, or tighter LVR restrictions on the entire country. The Reserve Bank is aware that LVR restrictions can make it difficult for people to undertake some transactions that they would otherwise be able to undertake (such as purchase of a first home without boosting their available deposit). We consider that the case for these restrictions outside of Auckland has significantly diminished, so the Reserve Bank would like to be able to ease these restrictions and potentially remove them completely in the future. In order to be able to achieve this, it is necessary to ask the banks to develop data systems so that they can rigorously identify the location of lending (as well as the investor/owner-occupier status, which is also required for capital purposes). Banks have provided estimates of the costs of obtaining this data and designing processes to comply with the LVR rules. These estimates varied fairly widely but were significant. Banks commented that the speed and complexity of the policy implementation exacerbated costs. The delayed start data and simplified calculation methods should thus mitigate the costs involved. However, they could still be in the region of $10 million dollars for the system as a whole. The system work will allow mortgage credit to the Auckland market to be monitored much more precisely than in the past. Given the divergence between the Auckland market and the rest of New Zealand, this has the potential to be useful in other contexts (capital modelling, stress testing, internal bank decision making). The Auckland housing market has remained heated for longer than the Reserve Bank and most commentators expected
when LVRs were introduced, so LVR restrictions are likely to remain in place for longer than expected in 2013. We think the costs this imposes on prospective home purchasers are larger for owner occupiers than for investors, as investors have the option of buying outside of Auckland or buying a different sort of asset (while prospective owner occupiers have to buy a home in the region they live). Investors may also use the construction exemption, ensuring that the policy does not reduce the supply of new dwellings (relative to the status quo). The key costs and benefits of tightening aggregate LVR restrictions, or the proposed tightening on Auckland investors are evaluated relative to the status quo in the table below.

16
Ref #6220011
Table One: Key costs and benefits of nationwide and targeted LVR restriction tightening
Possible Policies
Tighter nationwide LVR restrictions
Proposed three speed limit restrictions.
Key benefits relative to status quo
Policy provides further restraint on Auckland market Policy provides restraint on Auckland market, and more substantial restraint on investor lending. Systems lead to better data on lending types. Key costs relative to status quo Harder to buy houses (for all buyers, nationwide) Not well targeted: Policy imposed nationally despite regional nature of issue Required system development by banks (perhaps $10m) Harder to invest in Auckland property with low deposits (unless using construction exemption) The risks of the housing market becoming dysfunctional if lending policy became much less permissive in a severe downturn appear substantial. The costs that a tighter aggregate LVR policy would impose on owner occupiers and outside of Auckland also appear substantial. These factors led us to conclude that it is worthwhile to adjust the LVR restrictions to focus on Auckland investor lending specifically.
3. Consultation and monitoring
The Reserve Bank has consulted the Minister of Finance on our proposed approach, including discussions with the Treasury. We have also consulted the public on our proposed approach, and had a series of exchanges with affected banks (including a policy workshop, a number of meetings and conference calls, and an additional exchange of letters described above). This paper summarises that feedback and the changes we propose to make to our initial proposal based on that feedback. The adjusted LVR restrictions will continue to be monitored through the Reserve Bank’s prudential data collections, and other macroeconomic statistics relating to the housing market and credit. The sort of analysis shown in this regulatory impact assessment and the initial consultation paper on investor LVRs will be updated alongside other analytical work. This work will be discussed with other agencies with an interest in the housing market (including the Treasury). The Reserve Bank will publically describe its view of the case for maintaining or adjusting LVR restrictions in the Financial Stability Report and other periodic communications. Relevant data will also be released in the Macroprudential Indicator Report (which is currently being refined, partly to include more housing market indicators).

17
Ref #6220011
Annex Two: Comment on proposed changes to BS2A and BS2B announced on 29 May
2015

1. The Reserve Bank published a Summary of Submissions and Final Policy Position, and a
   Regulatory Impact Statement on the asset class treatment of residential property investment loans on 29 May 2015. In considering feedback from the earlier consultation, one issue that arose was the capital treatment of those loans that are secured over both investment and owner occupied properties.
2. The policy document and draft BS2A/B contained an option to pro rate a loan for capital
   purposes across both the property investment and non-property investment asset classes, according to the relative values of the types of collateral securing that loan. The proposal also allowed banks to take a customer-level view and allocate that customer to a single asset class if, for example, that bank’s systems did not allow a pro rata approach.
3. A number of banks provided useful feedback on the draft BS2A/B published on 29 May, both
   in formal submissions and in bilateral meetings. Similarly to the responses received on the BS19 consultation, views were mixed on the merit of introducing a pro rata approach. In theory, if the capital requirements for property investment loans were applied in proportion to the share of loans that are secured by investment properties, capital and the risks in banks’ portfolios would be more closely aligned. The splitting of loans for capital purposes would also limit the incentives of customers to split their affairs across banks. A small number of banks supported the option to apply a pro rata approach on these grounds.
4. However, it was also argued that the pro rata approach would introduce an ‘unwarranted’
   level of complexity for banks. Several banks indicated that to enable this functionality in their systems would require significant investment for what they perceived as little benefit other than maintaining competitive neutrality. These banks anticipated that they would be placed at a somewhat arbitrary competitive disadvantage due to the historic approach taken to the design of their systems. It was argued that using a ‘single customer in single asset class’ approach would maintain current banking practice.
5. Additionally, this treatment would greatly simplify the building of internal models for the
   two residential mortgage loan asset classes. For model construction, IRB banks would not need to be able to dynamically portion their historic loan data according to the collateral types and values. A greater proportion of a bank’s total residential mortgage exposures may be allocated to the property investment asset class than would be the case under the pro rata approach, resulting in a small increase in the capital requirement outcome. However, IRB banks will be able to take the nature of the collateral into account in the models they submit for Reserve Bank approval, so that having some owner occupied properties in the investment asset class could ultimately reduce average risk weights for investor mortgages.
6. Based on the industry feedback, the Reserve Bank believes that, on balance, the costs of the
   complexity of a pro rata approach to the capital treatment of residential mortgage loans would not be justified, and given the decision not to allow the apportioning of loans under BS19, it would be more appropriate to maintain a ‘single customer in a single asset class’ approach in BS2A and BS2B as well.
7. The Reserve Bank understands that the system changes required for compliance with the
   changes to BS19 are substantially linked to those required for BS2A or BS2B, and therefore believes it would be sensible for the change in the asset class treatment of residential

18
Ref #6220011 property investment loans come into effect from 1 November 2015 for new lending. The 12 month transition period for the reclassification of existing lending will remain unchanged, therefore all residential mortgage loans will need to be reclassified from 1 November 2016.
8. Several banks commented on the proposed definition of an owner-occupied residential
property, in particular on the definition of a related party. It was argued that extended family members such as parents or children should be included within the definition, as in many cases there would be no stream of rental income. However, the Reserve Bank considers that to do so would unjustifiably expand the scope of the definition. The Reserve Bank does not intend to alter the definition of related party from the draft wording communicated in the Final Policy Position.

19
Ref #6220011
Annex Three: Submission respondents
Below is a list of respondents to the Reserve Bank of New Zealand consultation paper “Adjustments to restrictions on high-LVR residential mortgage lending” dated 3 June 2015. ANZ Bank New Zealand Limited ASB Bank Limited Bank of New Zealand Chris Moore Hongkong and Shanghai Banking Corporation Limited, New Zealand Branch Jennifer Goldsack Mark Strong Michael Reddell New Zealand Bankers Association Real Estate Institute of New Zealand SBS Bank The Co·operative Bank Ltd Westpac New Zealand Limited