Sharon Donnery: Communication, calibration and coordination - challenges implementing macroprudential policy in the euro area
Speech by Ms Sharon Donnery, Deputy Governor (Central Banking) of the Central Bank of Ireland, at the second ECB Macroprudential Policy and Research Conference, Frankfurt am Main, 12 May 2017.
I would like to thank Mark Cassidy, Martin O'Brien, Eoin O'Brien, Mícheál O'Keeffe and Jean Quin for their contribution to my remarks.
Vice-President, distinguished guests, it is a pleasure to speak on this policy panel today.
Significant progress has been made in recent years in developing the EU's macroprudential framework and much credit for this must go to institutions like the ECB, ESRB and European Commission, as well as national macroprudential authorities.
The manner in which we now conduct macroprudential policy emerged in the context of the global financial crisis and recession. The relative impact of these and the interaction with diverse domestic factors has led to a policy framework which, in broad terms, shares common themes and objectives, while also reflecting important country and region-specific characteristics.
It is timely that we now consider our broader framework in the EU as the economic recovery has become more firmly established, and we see a normalisation of the financial cycle.
At the same time, we are potentially presented with broader and more diverse sources of risk than when the parameters of the current framework were being designed. These include the increasing role of non-banks in financial intermediation and the challenges arising from the decision of the United Kingdom to leave the EU.
To date, we have learned 'a lot by doing', and indeed a lot has been done in the establishment of the framework and the activation of macroprudential policy instruments across the EU.1
Today, I would like to focus my remarks on some of the challenges in implementing macroprudential policy through borrower-based measures for the mortgage market.
The Central Bank of Ireland introduced caps on loan-to-value (LTV) and loan-to-income (LTI) ratios in early 2015. Both the introduction of the measures and the first review in 2016, threw up some interesting challenges, some of which I would like to share with you today. These include the need to frequently communicate on the objectives of the measures and the need to regularly review their impact and effectiveness. More generally, some other challenging issues with respect to these instruments include whether there should be a more coordinated European framework for their introduction.
Borrower-based measures such as loan-to-value and debt-service-to-income, debt-to-income or loan-to-income ratios have been introduced in a number of European Member States over recent years.
Prior to the financial crisis, these type of measures were deployed in mainly non-European countries (including South Korea, Hong Kong, Singapore and others) and there is empirical evidence regarding their potential effectiveness in reducing risks to financial stability.2 The evidence however is rarely clear cut and the impact of such measures can vary depending on, among other things, characteristics of national housing markets and the position in the financial cycle.
Whilst subject to intense debate, when we introduced the mortgage measures in 2015, the legacy of the crisis was still at the forefront of many people's minds. This, to some degree, facilitated the subsequent broad societal understanding and acceptance of the measures.
Our two primary objectives when introducing these measures were, first, enhancing the resilience of households and banks to economic and financial shocks and, second, reducing in a structural way the pro-cyclicality that can be inherent in housing and credit markets by capping the amount of high LTV and LTI mortgages allowed at any time.
However, while we have emphasised in all our public communications that it is not our objective to target house prices, this has proven in fact to be a very difficult message to get across to the public. Underlying much of the public discourse on the mortgage rules is the central issue of affordability. In this context, it has been challenging to explain (i) that house prices are determined by a complex interaction of supply and demand side factors, (ii) that although our measures may impact on prices - that is not their primary goal, and (iii) that housing market policy issues like taxes, building measures and the shortage of supply of housing for buyers and renters are outside of our remit.
Our view is that not only would it be extremely difficult to choose an appropriate target for house prices but also extremely difficult to hit this target. Financial sector regulations cannot address these issues, which must instead be addressed by other targeted policies.3 To banks, we have also had to carefully explain that these are limits, not targets. We continue to monitor their risk appetites in this regard.
Figure 1 shows the recent history of property prices in Ireland. It also shows some uptick in mortgage lending across some segments of the market. Although the overall numbers remain modest because of the scale of deleveraging following the crisis, the volume of new lending is going up. In terms of transactions, cash buyers continued to account for a large relative proportion of activity, at 42% in 2016.4
Given these market dynamics, existing supply constraints and strong demand factors are unlikely to moderate significantly in the short term. This requires us to reinforce our communication of the objectives of these measures.
In terms of their design, the measures create a framework which allows us to better take into account the balance of risks to the economy, by adjusting the parameters according to current market conditions.5 Having this flexibility is particularly important for a small open economy like Ireland.
The mortgage measures are complementary to existing microprudential supervision and to lenders' own risk management practices. They are not intended to capture all aspects of credit risk associated with the borrower, nor to replace or substitute for a bank's existing internal credit assessment policies and procedures. Rather they are designed to reinforce and strengthen the existing suite of credit risk mitigation tools employed by prudent lenders.6
Following the crisis, we introduced comprehensive loan-level reporting requirements on the main Irish banks. This data provides information on a wide range of loan characteristics including outstanding balances, loan terms and loan repayment for the population of mortgage and consumer loans and is critical in the calibration and evaluation of the measures. The activation of our Central Credit Register in June will further enhance our analytical capacity with respect to our evaluation of the measures. It may also facilitate us in moving toward a debt-to-income ratio in the future.
Our measures, as initially calibrated, set out a loan-to-income ratio of 3.5 as the anchor of the framework, with the value of loans exceeding the LTI limit of 3.5 not to be greater than 20% of new lending in a given year. This only applies to primary dwelling homes and there is no LTI limit for investment purchases. A maximum LTV of 80% was set for non-first time buyers. For first time buyers a higher cap of 90% applied for house purchases for the first €220,000 and the 80% LTV then applied to that part of the value of the house above €220,000. Banks were allowed to issue up to 15% of their primary dwelling home (PDH) loans above these LTV limits on an annual basis. We apply a more stringent limit to the purchase of houses for investment purposes and here the limit is 70% LTV, with up to 10% of new lending on this basis allowed above the limit.
At the time of introduction of the measures, we committed to regular reviews of their impact and effectiveness. This we believe represents good practice, and mitigates against any potential inaction bias. It is also consistent with the commitment for quarterly reviews of counter-cyclical capital buffer (CCyB) settings and annual reviews of the identification of, and buffer setting for, other systemically important institutions (O-SIIs) as prescribed in CRR/CRD IV.
The review process was overseen by a new Macroprudential Measures Committee and involved a range of analytical projects looking at issues such as borrower impact, the early performance of the measures against the stated objectives, and any side effects, including in relation to impact on the rental market, or leakages. To enhance transparency and understanding, research outputs on all these issues and records of the meetings are published on the Central Bank's website.7
In addition, feedback from external stakeholders was gathered through a call for evidence on the impact of the measures. Submissions were received from a variety of property and construction industry stakeholders, financial services firms and groups, political parties, government departments, academics and individual members of the public. The submissions were a very useful input for our review, and were also published on the Central Bank of Ireland's website.
It is obviously the case that it was very early after their initial introduction to be able to find much causal impact from the introduction of the measures, but still our review threw up some interesting findings and resulted in some changes to the design of our policies.
We saw a reduction in high LTV mortgages and simulations from loan-loss forecasting models indicate that resilience of banks and households increased. We found that the introduction of the measures had an immediate and material impact in moderating price expectations. Actual price increases also moderated although it is more difficult to ascribe any causality to our measures.
In general, we found that the overall framework of the measures is appropriate and effective in meeting the objectives. Nonetheless, following our 2016 review some changes to these parameters were applied to improve the sustainability and effectiveness of the overall framework. Most notably the higher 90 per cent LTV available to first time buyers is now available at all house prices. The change was in part to avoid a situation whereby the €220,000 cut-off point would have to be recalibrated at regular periods as house prices increased. This change was also supported by new research findings of lower credit risk for first time buyers at all house price levels.
The other main change arising from the review was to the structure of proportionate caps such that instead of a LTV allowance of 15 per cent for all owner-occupier mortgages, separate allowances for first time buyers (5 per cent) and second and subsequent buyers (20 per cent) are now allowed.
High levels of household indebtedness and a large number of households in mortgage arrears and/or with negative or low positive equity in their houses, are prominent features of the Irish residential real estate market. Whilst not materially affecting the amount of allowances available, differentiating the allowances of first time buyers and second and subsequent buyers allows for a more precise calibration of the rules by borrower type if in future they need to be tightened or loosened in response to emerging risks and developments in the property market.
Negative equity borrowers are exempt from the measures. As more second and subsequent buyers move from negative equity into small levels of positive equity, the higher level of exemptions will ensure the measures would not act as an excessively binding constraint.
Taking communication and calibration together, people broadly understood the rationale for the limited refinements and minor recalibration of the framework. However, with such levels of house price increases, we have found the need to communicate that policy stability is important for both households and banks. As central banks, we have the responsibility to reduce regulatory uncertainty and therefore annual reviews are most appropriate.8 In the case of the LTV/LTI caps, it is intended that the annual reviews will allow for both evaluation of the structural design of the framework, analysis of any side effects or unintended consequences, and assessment of whether the parameters are calibrated appropriately for current housing and credit conditions.
The experience so far in Ireland corresponds with the broad international experience as to the effectiveness of borrower-based measures for the mortgage market given country-specific features. A framework for transparent and accountable policy making in this space, supported by consistently communicated, well-defined objectives and related evidence base, are necessary conditions for their acceptance and effectiveness.
At present national macroprudential authorities play the central role implementing borrower based measures. The ESRB however has an important coordinating role via the notification process. National authorities also benefit in their deliberations from interactions with the ECB. The ECB also has an important 'top-up' power with regard to capital-based instruments like the CCyB and O-SII buffers.
The impact of borrower based measures when they are binding is much more evident to the public at large than capital and liquidity restrictions imposed directly on banks. Borrower-based measures can have important distributional and welfare effects in society, particularly in relation to access to home ownership. These factors underline the importance of getting the design of the measures right, taking into consideration country-specific factors, and also the need to communicate the overall framework and ensure it becomes generally acceptable in society as a permanent feature of the mortgage market.
As well as the wide range of national-specific institutional features which influence the design and calibration of such instruments, this suggests that responsible authorities must be accountable at national level. Although, we should be humble enough to acknowledge that the interactions between macroprudential policy measures may not be fully or correctly identified ex ante.
Significant progress has been made in the analytical toolkit to support macroprudential policy across the euro area, as was outlined in some of the presentations yesterday. As policy-makers we should continue to draw on this input, encourage its further development, and understand where judgement in evaluating policies impact, effectiveness and interactions is necessary. In this regard, particularly as we are still in the early stages of an active macroprudential policy framework in Europe, we should aim to avoid underlaps to ensure systemic risk is mitigated. As the framework in the EU develops, national designated authorities should continue to have a sufficiently broad set of tools to target identified systemic risks. We should continue to learn by doing. Combined with a commitment to regularly, rigorously and transparently review policy measures, we will ensure the most effective tools are in operation, both in isolation and jointly, to mitigate systemic risk.
Today, I have touched on a number of challenges that we have before us when considering the appropriate macroprudential policy framework for Europe. As we well know, past performance is no guarantee of future returns. However, I believe we can look back at the first generation of the EU macroprudential policy framework and how it has been adopted at both national and at Union level in a broadly positive light.
Important features should remain, such as the prominent role national authorities have in identifying systemic risk and designing policy measures, the effective cooperation across the euro area and ESRB and the appropriate acknowledgement of the implications of our policies in other dimensions and countries. While the challenges we face are ever-changing, the foundations we have laid and the commitment shown over the past few years can give us some confidence going forward.
1 By the end of 2016, legislation formally establishing national macroprudential authorities had been passed in 26 Member States and there has been a very high level of effective operationalisation of the framework. National authorities in the EU and Norway had notified the ESRB of 83 significant measures in the three years up to end-2016. See 'A Review of Macroprudential Policy in the EU in 2016', ESRB, April 2017.
2 Cerutti, Eugenio, Claessens, Stijn and Laeven, Luc, (2015), The use and effectiveness of macroprudential policies: new evidence. Journal of Financial Stability. 2016, November: 11.
4 For a fuller discussion see Coates, Dermot, McNeill, Joe and Brendan Williams, (2016), Estimating Cash Buyers and Transaction Volumes in the Residential Property Sector in Ireland, 2000-2014, Central Bank of Ireland Quarterly Bulletin 03 / July 16. The full year 2016 estimate is derived using data on mortgage drawdowns from the BPFI and the Residential Property Price data from the CSO.
5 See 'Lessons from the past, safeguarding stability for the future' Address by Sharon Donnery, Deputy Governor of the Central Bank of Ireland, at the Centre for Economic Policy Research (CEPR) Economic History Symposium, Dublin, 9 June 2016.
7 Records of the Macroprudential Measures Committee meetings. Accompanying research papers regarding the mortgage measures, submissions and feedback statements are also published.
8 See 'Macroprudential policy: action in the face of uncertainty'. Address by Sharon Donnery, Deputy Governor of the Central Bank of Ireland at the Dublin Economic Workshop Annual Economic Policy Conference. 24 Sep 2016.