Summary of submissions and final policy decisions on the Consultation Paper: Review of bank capital adequacy requirements for housing loans

Summary of submissions and final policy decisions on the Consultation Paper: Review of bank capital adequacy requirements for housing loans (stage two) Please note that this is not a consultation document. December 2013

Background

1. On 26 March 2013, the Reserve Bank announced a staged review of bank capital adequacy
   requirements for residential mortgage loans. Stage one of the review addressed the calibration of the correlation coefficient for ‘internal models’ banks.1 As of 1 September 2013, internal models banks have to apply a correlation factor of 20 percent for high loan-to-value ratio (LVR) loans where the LVR is greater than 80 percent but less than 90 percent, and 21 percent for LVR loans greater than 90 percent.
2. Between 20 September and 25 October 2013, the Reserve Bank consulted on stage two of the
   review. The consultation proposed refinements to definitions in the banking capital requirements, as well as the formalisation of the internal models approval process. The Reserve Bank received eleven submissions – seven from banks and four from other stakeholders.
3. This paper summarises the feedback received on the consultation, as well as the Reserve Bank’s
   policy decisions. In finalising our policy, the Reserve Bank has considered the points raised by submitters and amended its proposed policy in a number of areas, in part reflecting a number of concerns raised by industry.
4. The policy changes to existing capital requirements are expected to take effect from the middle of
   2014, following consultation on the draft changes to BS2A and BS2B early next year. We are releasing a summary of policy decisions at this point, as there are potential system changes for registered banks to consider, and as such we understand that early feedback on policy decisions would be welcome. Moreover, as we are changing some of the requirements we originally consulted on we think it is important to consult on the new handbook wording.
   Part I Definition issues
5. The consultation paper noted that ‘standardised’ and ‘internal models’ banks currently do not
   apply the same definition of the loan amount when calculating the loan to value ratio of residential mortgages. The Reserve Bank’s goal was to ensure that both groups of banks use the same definition as far as possible and to ensure that the capital treatment reflects economic substance.
6. The consultation paper proposed to extend the current BS2A definition, which is the one that
   applies to standardised banks, to internal models banks. The main reason for this proposal was to support the macroprudential LVR restriction policy. A further consideration was that under the common “all obligations” mortgage banks take security over the residential property, which they can then enforce for any claim they have against the borrower. For internal models banks, that would have meant including unsecured lending such as, for example, credit card limits or personal loans in the definition of the loan amount. Standardised banks already have to include those unsecured loans in the LVR calculation.
7. The proposal would have met the Reserve Bank’s objective of having a consistent definition
   across the two sets of requirements, and reduced the scope for avoiding the LVR restriction by taking out personal loans or borrowing on one’s credit card as a substitute for mortgage borrowing. Theoretically it would have also reflected the economic substance of an all obligations mortgage, assuming that banks could enforce the all obligations mortgage for the purpose of recovering other loans. However, as noted above, reducing the scope for avoidance of the 1 ‘Internal models’ banks calculate their capital requirements according to the Banking Supervision handbook BS2B, whereas ‘Standardised’ banks use BS2A:
   http://www.rbnz.govt.nz/regulation_and_supervision/banks/banking_supervision_handbook/

macroprudential LVR policy was the main factor in proposing the use of the current BS2A definition for internal models banks.
8. Stakeholders generally agreed that the calculation of the LVR for residential mortgage lending
should be consistent across the banking sector as far as possible. Some submitters, most notably the current internal models banks, raised a number of concerns about applying the BS2A definition to the whole sector and argued for the harmonisation to be done on the basis of BS2B, i.e., excluding unsecured lending from the LVR calculation.
9. The main objection to the Reserve Bank’s proposal was that credit card and personal loans were
already treated as unsecured lending and as such given a risk weight accordingly. Their inclusion in the calculation of the LVR would lead to a form of “double counting” and not reflect the way these loans are treated by banks. Internal models banks estimated that the impact on risk weights for housing loans would be in the region of 0.5 to 0.6 percent.
10. Some banks argued that the legal link under all obligations mortgages would be difficult to
enforce in practice, particularly if it is not explicitly disclosed when the unsecured loan is taken out. Others made the point that they could inform their customers that the all obligations mortgage did not extend to any unsecured loans the customer has with the bank or change mortgage contracts to fixed sum mortgages. It was also argued that such a move would not be desirable from a credit risk management point of view.
11. Several submissions expressed a concern that the inclusion of unsecured lending could lead to
customers splitting their banking services across multiple banks and that banks would lose their single customer view. This, it was argued, would make lending to customers more risky, and actually not achieve the objective of reducing the avoidance potential of the LVR policy.
12. Internal models banks also argued that the inclusion of unsecured lending in the LVR calculation
would necessitate changes to internal models and IT systems and staff retraining.
13. Those banks opposed to the inclusion of unsecured loans in the LVR calculation generally
accepted the Reserve Bank’s objective of minimising the scope for avoidance of an LVR restriction. However, some banks suggested that objective could be achieved by other means. Suggestions included the obligation on banks to abide by the spirit of the LVR policy, and the proposal that the Reserve Bank track the development of other loan commitments by extending the reporting requirements under BS19. Policy decision
14. The Reserve Bank has analysed the points raised by submitters carefully and reconsidered its
policy position.
15. With respect to the objective to establish a common approach for calculating LVRs for both
standardised and internal models banks, we have been persuaded by the arguments of submitters. As such, we will now harmonise industry requirements on BS2B, rather than a version of BS2A.
16. This means that most unsecured loans such as credit card limits or personal loans do not have to
be included in the loan amount when calculating the LVR of a residential mortgage loan.
17. The exception is where the terms of a personal loan or credit card are directly linked to a
mortgage on residential property. An example of this would be a credit card that is contractually tied to a customer’s mortgage loan and is effectively an extension of the mortgage. Any undrawn commitments that are a part of this lending facility must also be treated as secured by way of first ranking mortgage over the residential property. Characteristics expected of the contractual arrangements, would include (but not be limited to) a customer losing the lending facility upon

transferral of the mortgage to another lender, or an interest rate charge similar to that of the mortgage loan and below the standard interest rate for such products.
18. The Reserve Bank will also take into account the suggestions made in the submissions to this
consultation and will deal with potential LVR restriction avoidance issues under BS19. This may include additional reporting of information to the Reserve Bank or depositor verification requirements. It should be stressed that revisiting the proposal to include unsecured lending in the LVR calculation or having a separate definition that includes it in BS19 remain options should the Reserve Bank detect any avoidance activity or non-compliance with the spirit of the LVR policy.
19. The Reserve Bank will discuss transitional arrangements to the new requirements with banks.
Definition of the value amount
20. The consultation paper pointed out that standardised banks currently have a more prescriptive and
clearer requirement as to the definition of how the property value of a residential policy is to be determined, compared to internal models banks. Secondly, the Reserve Bank stated that valuations should not be updated more frequently when prices rise than when they fall. The Reserve Bank’s overriding aim was to ensure that the valuations policy is robust and invariant to the state of the economic cycle.
21. There was agreement amongst respondents that banks should have a valuation policy, although a
small minority of submissions considered that a valuation policy was too detailed to required board approval (as proposed in the consultation paper). There was also support for ensuring that valuations are independent. One submitter emphasised that valuations should be carried out by parties that have no direct interest in the transaction process.
22. Banks provided useful feedback on their use of valuations, with most banks making use of the
purchase price, registered valuers, electronic valuations, desktop valuations by professional valuers and property specific valuations by independent valuers. As a general rule, the higher the risk, the more detailed a valuation would be required. For example, high LVR loans tend to require a full valuation done by an independent valuer.
23. Several submitters expressed concern that the proposed wording on who would be eligible to
carry out a valuation, with the emphasis on independent valuers, would rule out, for example, modelled valuations and so-called desktop valuations. This, it was argued, would increase the costs of obtaining a mortgage for borrowers and, in any case, there were not enough independent valuers to do all property valuations. Moreover, developments in valuation products such as desktop-based valuations and modelled valuations had improved their accuracy to make them viable alternatives to full valuations for less risky loans. Policy decision
24. The overriding objective of this part of the consultation was to ensure that the valuations banks
use for calculating their capital requirements are robust. The Reserve Bank considers that valuations should be done by appropriately qualified professionals who are independent and that the same rules should apply to standardised as well as internal models banks. Moreover, the Reserve Bank wishes to ensure that a bank’s valuation policy specifies the triggers for, and frequency of, valuation, and, importantly, behaves symmetrically through the housing cycle. That is, valuations should not be updated more frequently when prices are increasing than when they are falling.
25. With that in mind, the Reserve Bank accepts that there are different valuation products available
to banks for different loan risk profiles that can achieve these objectives. The wording of the

proposed valuation policy was not intended to categorically rule these other products out. Finally, the valuation requirements will apply to both standardised and internal models banks. Calculation of the LVR when there are multiple securities or multiple borrowers
26. This part of the consultation asked banks to supply information on how five customer lending
scenarios with multiple borrowers and multiple securities are treated for LVR calculation purposes. The consultation paper also asked banks whether there were other possible scenarios that did not fall within the five broad categories presented in the paper and to give an estimate of the proportion of their residential mortgage lending books that involve these more complex lending scenarios.
27. The consultation paper did not propose a prescribed LVR treatment. Rather, the purpose of this
section was to establish how banks currently calculate the LVR for these scenarios with a view to
requiring banks to seek approval from the Reserve Bank for making any changes to their current practice. The underlying reasons were to ensure that banks calculate their LVRs correctly and that they apply collateral spreading strategies appropriately, and that the macroprudential LVR policy does not lead to undue innovative LVR calculations.
28. By and large respondents confirmed the Reserve Bank’s expectations in terms of how the LVR is
calculated. Where banks provided estimates, they also confirmed the Reserve Bank’s understanding that these complex arrangements are a small fraction of banks’ overall mortgage lending books. Other clarifications to the LVR requirements
29. The consultation paper proposed to clarify for standardised banks that mortgages that have an
LVR of more than 100 percent – for example as a result of having gone into negative equity – still fall within the definition of a residential mortgage. BS2A as currently drafted could be interpreted to mean that these mortgages should be treated as unsecured lending and be given a risk weight of 100 percent. As such, we will reclassify residential mortgages with an LVR above 100 percent as residential mortgages by including an additional row in table 4.11 in BS2A that attaches a risk weight of 100 percent to those mortgages.
30. The Reserve Bank also proposed to further clarify that residential mortgage loans partially
secured by a residential property remain residential mortgage loans by including the words “or partially” in the definition of a residential mortgage loan. This has caused some confusion amongst some submitters who have interpreted this as meaning that any loan that includes a residential property as part of the collateral would be a residential mortgage loan, e.g., an SME business loan and a mortgage over a residential property whereby the residential property is used as part collateral. The Reserve Bank wishes to stress that such cross collateralisations do not make the other loan, in this case the SME business loan, a residential mortgage loan. The clarification was merely aimed at ensuring that loans secured over a residential property where the outstanding loan amount is more than the value of the residential property must be considered as residential mortgage loans.

Distinction between a residential mortgage exposure and other types of exposure
31. The consultation paper proposed a new definition to establish under what circumstances a loan on
a potentially dual use property could no longer be classified as a residential mortgage loan. The definition proposed by the Reserve Bank consisted of two criteria: (i) whether a property was in
part or fully used for farming or business purposes and (ii) whether the borrower was in part or
fully reliant on the income from the farming or business activities to repay the loan, then the loan could not be classified as a residential mortgage loan.

32. The paper also proposed that if a bank had recourse to four or more dwellings owned by the
    borrower, then the loan could no longer be classified as a residential mortgage loan.

33. There was general agreement that further clarification on the boundaries between asset classes
    would be useful. Respondents also mostly agreed that the Reserve Bank had identified the correct classification criteria, although several raised marketability as a residential property as a further criterion in addition to intended use or purpose of the property and the source of income from which the loan is repaid.

34. Several respondents, however, raised issues with the proposed wording and pointed out possible
    unintended consequences. One respondent argued that the wording would turn all self-employed people’s residential mortgages into SME lending. This was a reference to the words “in part” as this could be interpreted as rendering that are partly used for other purposes but primarily as a residential property ineligible for residential mortgage treatment.

35. Others pointed out that the definition of four dwellings was insufficiently precise and that one
    could have more than one dwelling on a single title, e.g. units or flats. A few argued that drawing the boundary at a specific number would encourage customers to split their borrowing across more than one lender, potentially limiting a bank’s ability to assess the customer’s credit risk and increasing overall risk, and disadvantage borrowers with considerable other income which makes repayment of the mortgage not dependent on the rental income those properties generate. Some proposed an income related definition, so that if the borrower uses the income from the properties to pay off the loan, then the loan should not be considered as a residential mortgage. Others agreed with the Reserve Bank’s proposed wording or made no specific comments. Policy decision

36. As a result of the feedback received on this part of the consultation, the Reserve Bank has decided
    to amend the wording of the definition to address many of the potential issues raised by respondents. Marketability will be added to the boundary definition of whether a loan is a residential mortgage loan or a farm or business lending, there is greater emphasis on the income source used to service the loan, and the purpose component of the definition will be amended.

37. However, we are retaining a count based element in the determination of the boundary between
    residential mortgage lending and commercial property investor lending, although this will also include an income test. The count threshold, however, has been increased to five properties and greater clarity will be provided as regards the treatment of multiple dwellings within a property. The Reserve Bank is of the view that anyone with more than five properties, regardless of whatever other income sources or revenue streams may exist, should be treated as running a small business. To avoid further confusion, this would mean treating those loans as corporate property loans.
    Part II Internal models requirements

38. The Reserve Bank proposed to formalise existing internal model change requirements and to
    agree a “Compendium of Models” with each of the banks that have been accredited to use internal models.

39. Internal model change requirements are currently specified in a letter all accredited internal
    models banks have received from the Reserve Bank. The proposed wording mirrored the latest update of the letter from July this year. All respondents welcomed the formalisation of the internal model change requirements and the establishment of the model compendium.

40. A minority argued against the requirement to supply four data points of the proposed model
    change impact as part of the model change submission documentation. One submission in particular contended that wholesale model changes should not require four data points as the customer information was only updated annually and therefore not frequently enough.

41. Most respondents also were against the requirement to calculate the capital outcome under the
    standardised approach when making a model change submission and to record that in the compendium. Banks argued that IT systems were not currently configured to do standardised calculations and manual calculations were costly. An in-system calculation component was costly to establish, according to some internal models banks. Some submissions wondered why the information was needed given that they had been accredited as internal models banks. Policy decision

42. The Reserve Bank does not consider it onerous for internal models banks to supply four data
    points as a general rule. As already pointed out in the consultation paper, the requirement clearly allows for alternatives to be agreed to by the Reserve Bank on a case by case basis, such as when four data points do not provide meaningful information, as long as the Reserve Bank can obtain a similar level of comfort as regards the model’s long term impact.

43. The calculation of the capital outcome under the standardised approach is mainly for monitoring
    purposes and as a reference point. While the Reserve Bank accepts that it might take some time for banks to change their IT systems to do the calculation within their systems, it is not convinced that it would be too onerous or costly to implement the relatively simple standardised calculation. Therefore the Reserve Bank intends inserting a new section into BS2B along the lines of that consulted on. However, the precise wording of BS2B will also be consulted on in early 2014.
    Part III Next steps

44. The Reserve Bank intends to consult on the updates to BS2A and BS2B to reflect the policy
    decisions laid out in this summary of submissions in early 2014. This will be accompanied by a Regulatory Impact Assessment. Following that, changes to the capital adequacy requirements in BS2A and BS2B are expected to come into effect on 30 June 2014.